Redescription and reassessment of the phylogenetic ...

Journal of Systematic Palaeontology 7 (2): 199–239 doi:10.1017/S1477201908002691 Printed in the United Kingdom

Issued 26 May 2009 C The Natural History Museum

Redescription and reassessment of the phylogenetic affinities of Euhelopus zdanskyi (Dinosauria: Sauropoda) from the Early Cretaceous of China Jeffrey A. Wilson Museum of Paleontology & Department of Geological Sciences, University of Michigan, 1109 Geddes Avenue, Ann Arbor, Michigan 48109–1079, USA

Paul Upchurch Department of Earth Sciences, University College London, Gower Street, London WC1E 6BT, UK

SYNOPSIS Euhelopus zdanskyi was the first dinosaur described from China. Both traditional and modern cladistic assessments have found support for an endemic clade of Chinese sauropods (Euhelopodidae) that originated during an interval of geographic isolation, but the monophyly of this clade has remained controversial. The phylogenetic affinity of the eponymous genus Euhelopus is central to this controversy, yet its anatomy has not been completely restudied since the original German-language monograph in 1929. We jointly re-examined the cranial and postcranial anatomy of the holotypic and referred materials of Euhelopus to provide a new diagnosis for the genus and to explore its phylogenetic affinities. Diagnostic features of Euhelopus include: postaxial cervical vertebrae that have variably developed epipophyses and more subtle “pre-epipopophyses” below the prezygapophyses; cervical neural arches with an epipophyseal–prezygapophyseal lamina separating two pneumatocoels; anterior cervical vertebrae with three costal spurs on the tuberculum and capitulum; divided middle presacral neural spines, which in anterior dorsal vertebrae bear a median tubercle that is as large or larger than the metapophyses; middle and posterior dorsal parapophyseal and diapophyseal laminae arranged in a “K” configuration; and presacral pneumaticity that extends into the ilium. Following this morphological study, we rescored Euhelopus for the two most comprehensive sauropod data matrices (Wilson 2002; Upchurch et al. 2004a), which previously yielded vastly different hypotheses for its relationships. Both matrices decisively demonstrate that Euhelopus is closely related to Titanosauria; traditional and cladistic claims that Euhelopus, Omeisaurus, Mamenchisaurus and Shunosaurus formed a monophyletic “Euhelopodidae” endemic to East Asia are not supported. These results suggest that there were at least two clades of very long-necked sauropods in East Asia, occurring in the Middle Jurassic (i.e. Omeisaurus + Mamenchisaurus) and Early Cretaceous (e.g. Euhelopus, Erketu), with the latter group perhaps also occurring in Europe (Canudo et al. 2002). It is probable that the Euhelopus + Erketu lineage invaded East Asia from another part of Pangaea when isolation ended in the Early Cretaceous. The large number of basal titanosauriforms from East Asia has been interpreted to mean that this area may represent their centre of origin (You et al. 2003), but the titanosaur fossil record and phylogenetic studies indicate that the group probably originated prior to the Middle Jurassic and acquired a virtually global distribution before Pangaean fragmentation. KEY WORDS palaeontology, Asia, phylogeny, sauropod, palaeobiogeography

Contents

Email: [email protected] Email: [email protected]

Introduction

200

Institutional abbreviations

201

Collection history Exemplar a Exemplar b Exemplar c

201 201 201 201

200

J. A. Wilson and P. Upchurch

Systematic History

202

Age of the Mengyin Formation

203

Systematic palaeontology Dinosauria Owen, 1841 Saurischia Seeley, 1887 Sauropoda Marsh, 1878 Neosauropoda Bonaparte, 1986 Titanosauriformes Salgado et al., 1997 Euhelopus Romer, 1956 Euhelopus zdanskyi (Wiman 1929)

205 205 205 205 205 205 205 205

Description Skull Vertebral column Cervical vertebrae Cervical ribs Dorsal Vertebrae Dorsal ribs Sacral vertebrae Scapula, coracoid and humerus Pelvis Hindlimb

206 207 210 211 217 218 221 221 222 223 223

Discussion Phylogenetic affinities of Euhelopus Euhelopus and the biogeographical history of East Asian sauropods

227 227 229

Conclusion

233

Acknowledgements

235

References

235

Introduction In contrast to the much longer history of dinosaur studies in Europe (Plot 1677; Mantell 1825), North America (Leidy 1858), India (Falconer 1868), Madagascar (Depéret 1896) and South America (Lydekker 1893), Asian dinosaurs first appeared in the scientific literature in the 1920s with the Central Asiatic Expedition’s discoveries in the Gobi Desert of Mongolia (Andrews 1932). A flurry of descriptions of nowfamous dinosaurs followed, including the ceratopsians Protoceratops and Psittacosaurus, the ankylosaur Pinacosaurus and the theropods Oviraptor and Velociraptor (Granger & Gregory 1923; Osborn 1923, 1924a). In addition to this excellent material, Osborn (1924b) also described much more fragmentary remains of the first sauropod from Asia, Asiatosaurus mongoliensis, now considered a nomen dubium (McIntosh 1990; Barrett et al. 2002; Upchurch et al. 2004a). At the time of these descriptions, the Sino-Swedish Palaeontological expedition was concluding its long tenure in Asia with the excavation of the sauropod Euhelopus zdanskyi, the first of many excellent sauropod skeletons to emerge from China (Wiman 1929; Mateer & Lucas 1985). Despite this somewhat “late start”, Asian dinosaurs now represent 21.3% of all dinosaur discoveries (data downloaded from the Paleobiology Database on 21 August 2006, using the group name “vertebrate” and the following

parameters: time interval = Mesozoic, taxon = Dinosauria, region = Asia). Phylogenetic affinities of the Asian sauropod fauna suggest an interesting temporal pattern: whereas all Jurassic Asian sauropods fall outside the neosauropod radiation, nearly all Cretaceous Asian species belong to the neosauropod subgroup Titanosauriformes (Wilson 2005a). This pattern has been interpreted as the consequence of an interval of geographical isolation lasting from Middle Jurassic until Cretaceous times (Russell 1993; Upchurch 1995; Buffetaut & Suteethorn 1999; Luo 1999; Barrett et al. 2002; Upchurch et al. 2002; Zhou et al. 2003). If a physical barrier prevented neosauropods from dispersing into Asia, then this same barrier should have prevented basal sauropods from leaving Asia, which would imply the emergence of endemic clades. Euhelopodidae was forwarded as an exemplary endemic Asian sauropod clade (Upchurch 1995, 1998; Upchurch et al. 2002), a grouping that remains a central controversy in sauropod systematics (Wilson & Sereno 1998; Wilson 2002; Upchurch et al. 2004a). In this contribution, we test the monophyly of Euhelopodidae following revision of the anatomy and diagnosis of its namesake species, Euhelopus zdanskyi. We begin by providing an introduction to the history of discovery and the taxonomy of Euhelopus, as well as the age of the Mengyin Formation. Next, we redescribe the cranial and postcranial anatomy of Euhelopus, based on our joint examination of

Redescription and reassessment of Euhelopus zdanskyi

201

materials housed in the Palaeontological Museum of Uppsala, Sweden. This redescription forms the basis for the third portion of our contribution, in which we rescore and reanalyze the data matrices of Wilson (2002) and Upchurch et al. (2004a), the two most recent global analyses of sauropod genera. Finally, we explore the implications of the phylogenetic affinities of Euhelopus for Cretaceous Asian palaeobiogeography.

Institutional abbreviations BMNH = British Museum (Natural History), London, UK IGM = Geological Institute of the Mongolian Academy of Sciences, Ulaan Baatar, Mongolia MB = Museum für Naturkunde der HumboltUniversität, Berlin, Germany PMU = Palaeontological Museum, Uppsala, Sweden.

Collection history Euhelopus zdanskyi is known from two specimens that were collected from grey sandstone deposits from the Mengyin Valley of central Shandong Province, northeastern China (Fig. 1) on three different occasions by three different parties. Wiman (1929:5) reported that the initial discovery of a dinosaur in Mengyin was made by a Father R. Mertens in 1913. Around 1916, some bones from Mertens’ specimen were given to Dr V. K. Ting, Director of the Geological Survey of China, by German mining engineer W. Behegel (Young 1935). These were probably the first dinosaur remains accessioned to a modern repository in China, although fossils in general, and dinosaurs in particular, were known to ancient Chinese scholars (Needham 1956; Mayor 2000). The provenance of Mertens’ skeleton was unknown for some time, until Chinese geologist C. H. Tan and Sino-Swedish Expedition leader J. G. Andersson relocated the site in November 1922 (Wiman 1929:5; Mateer & Lucas 1985:15). The majority of Euhelopus material was discovered and excavated later from this locality. In March 1923, Otto Zdansky collected two partial sauropod skeletons, referred to as “exemplar a” and “exemplar b”. These specimens are housed in the Palaeontological Museum of Uppsala under accession numbers PMU 233 and PMU 234, respectively. Zdansky (pers. comm. in Mateer & McIntosh 1985:125) was forced to forgo complete excavation of exemplar a because he was obliged to return to Beijing, and Wiman (1929:7) remarked: “There are probably parts of this skeleton [exemplar a] somewhere else, but since they have apparently been ruined during excavation, I did not intend to fit them in” [translated from the German by J.A.W.]. A decade later, Chinese palaeontologist C. C. Young and geologist N. Bien returned to Zdansky’s quarry in Autumn 1934 and collected additional material that probably pertains to exemplar a. We provisionally refer to this material as “exemplar c”. Its whereabouts are currently unknown.

Figure 1 Map of China showing the position of the Euhelopus locality (bulls-eye) and map showing the position of Asia during the Early Cretaceous (120 Ma; modified from Blakey 2006). The map is a Mollweide projection with latitude and longitude lines spaced at 30◦ intervals.

“Ning Chia Kou” (now Ningchiakou) in Yantai County of Shandong Province, which juts into the Bohai and Yellow Seas (Fig. 1). A “li” is a traditional Chinese unit of measurement that at the time of Wiman’s writing was equivalent to 608.7 yards or 556.6 m (Alexander 1857). Based on this conversion, exemplar a was collected approximately 22 km northeast of Mengyin and approximately 1 km west of Ningchiakou. Exemplar a consists of a partial skull and lower jaws, an articulated vertebral series from the axis to the 25th presacral, a dorsal rib and a left femur (Fig. 2). The vertebral series appears to be continued by a series of articulated dorsal vertebrae described by Young (1935) from the same quarry (exemplar c; see below).

Exemplar b The partial skeleton of exemplar b (PMU 234) was collected approximately 2–3 km from exemplar a (Wiman 1929), but the relative positions of the sites are not known. Exemplar b consists of a series of articulated dorsal and sacral vertebrae, two dorsal ribs, pelvis and left hindlimb lacking some phalanges (Fig. 2). Only the femur and posterior dorsal vertebrae overlap with exemplar a. The strong resemblance in overall morphology and the presence of a crisscross pattern of “K” laminae (see Description, below) on the lateral aspect of the posterior dorsal neural arches clearly indicate that exemplar b is referable to E. zdanskyi.

Exemplar a

Exemplar c

Wiman (1929:7) reported that Zdansky collected exemplar a (PMU 233) 40 li northwest of Mengyin and 2 li west of

As noted above, Wiman (1929) left open the possibility that additional bones remained at the exemplar a locality. Young

202


Figure 2 Schematic representation of preserved elements of the holotype (top) and referred specimen (bottom) of Euhelopus zdanskyi. Skeletal outlines are shown in right lateral view, and light grey tone indicates limb elements from the left side. Exemplar c is considered to be part of the holotype (see the text and Fig. 3). Holotypic and referred skeletons of Euhelopus overlap in the middle dorsal region and in the femur, and autapomorphies in the dorsal vertebrae (see the text) support referral of exemplar b to Euhelopus zdanskyi.

and Bien returned to the area 11 years after Zdansky and collected a left scapula, coracoid and humerus. In the paper describing these forelimb remains, Young (1935:523) also described the partial series of vertebrae “sent to Dr. V. K. Ting by Behagel”, which “probably are also a part of . . . the skeleton a of Uppsala”. The adjoining portions of the dorsal series collected by Behagel and exemplar a of Wiman appear to fit – the first-preserved of Young’s series is a posterior portion of a vertebra that is broken obliquely from the postzygapophyses to the anterior neural arch pedicle, and the last-preserved vertebra of Wiman’s exemplar a is broken obliquely from the postzygapophyses forward to the anterior centrum (Fig. 3). Although we were not able to examine these vertebrae because their location is not known, we believe that they belong to exemplar a. Young’s (1935:523) contention that the limb material that he and Bien collected also pertains to exemplar a is less well supported but certainly plausible. The remains were collected at or nearly at the same locality as exemplar a, there is no duplication of elements between the two specimens, and the size of the elements is consistent. Exemplar c includes a series of four dorsal vertebrae, as well as a left scapula, left coracoid, and a left humerus

(Fig. 2). We consider it highly probable that exemplars a and c pertain to the same individual, but we retain separate names for them in the discussion that follows. Young and Bien also purchased a right coracoid collected by a villager somewhere near Ningchiakou, but the exact locality is not known (Young 1935:528). They ascribed this element to a sauropod, but it is quite narrow transversely and probably pertains to a large theropod dinosaur. We also note that the supposed theropod ulna collected from the nearby locality of “Hsichüfu” (Young 1935:fig. 8) is actually a proximal fragment of a pterosaur wing phalanx 1 (ph IV.1).

Systematic history The original generic name Helopus means “marsh-foot” and refers to the sauropod pes, which Wiman (1929: fig. 2) likened to trugors, a “kind of snowshoe used in the North of Sweden both by horses and by men for walking on marshes or loose snow” (Säve-Söderbergh 1946:401). At the time of Wiman’s writing, no hierarchical structure existed in sauropod taxonomy, which consisted of 5–6 families erected by


Figure 3 Comparison of posteriormost preserved presacral vertebrae of exemplar a (Wiman 1929: pl. 3) and the anteriormost preserved presacral vertebrae of exemplar c, which was described by Young (1935: fig. 1).

Marsh (1882, 1895). Although Janensch’s (1929) dichotomous scheme for sauropod taxonomy, which divided sauropods into the broad-crowned group Bothrosauropodidae and the narrow-crowned group Homalosauropodidae, appeared that same year, it was not popularised until Huene (1956) and Romer (1956) adopted it (Wilson & Sereno 1998; Wilson 2005b). Consequently, Wiman (1929:28–29) presented a relatively limited discussion of the taxonomy of “Helopus”, allocating it to its own Subfamily Helopodinae within “Cardiodontidae” (= Cetiosauridae; Table 1). The second sauropod described from China, Tienshanosaurus, was assigned to Helopodinae by Young (1937:21), who placed the subfamily within Morosauridae (= Camarasauridae). Shortly thereafter, Young (1939) named Omeisaurus and also included it in Helopodinae, which he then regarded as a subfamily of Brachiosauridae – presumably because of the long neck shared by members of the group. Romer (1956:621) followed Young’s taxonomic arrangement but changed the name of the genus and subfamily to Euhelopus and Euhelopodinae, respectively, because “Helopus” had been coined well over a century earlier by Wagler (1832) for the Caspian Tern, which itself was later synonymised with Hydroprogne caspia. The fourth Chinese sauropod, Mamenchisaurus, was described by Young (1954) with little taxonomic discussion, but the description of a more complete skeleton was accompanied by a classification of Chinese sauropods (Young 1958). By this time, Janensch’s dichotomy had been popularised by Huene (1956) and Romer (1956). Young (1958:25) considered Mamenchisaurus distinct at the suprafamilial level from earlier-named genera and placed it in the narrowcrowned group Homalosauropodidae (Table 1). Young & Zhao (1972) maintained the split of Asian sauropods into

203

broad-crowned and narrow-crowned groups, but separated Euhelopus – rather than Mamenchisaurus – from the other three genera, which were now regarded as members of the narrow-crowned subgroup Homalosauropodidae. At this point, dental material was available only for Euhelopus, which was clearly broad-crowned. Bonaparte (1986) and McIntosh (1990) provided the last “traditional” sauropod classifications. Bonaparte ignored Chinese sauropods in his 1986 discussion of the phylogenetic relationships of Jurassic sauropods and in his 1999 treatise on the vertebral anatomy of sauropod dinosaurs, but McIntosh (1990) classified all major Chinese sauropods, favouring their separation amongst the Families Cetiosauridae (Shunosaurus, Omeisaurus), Camarasauridae (Euhelopus) and Diplodocidae (Mamenchisaurus). Cladistic assessments of the phylogenetic affinities of the Chinese sauropods are split between the view that they form a monophyletic group of euhelopodid sauropods and the view that they are a paraphyletic assemblage of basal and derived forms (Fig. 4). Analyses by Upchurch (1995, 1998) gave cladistic valence to the view of Young and others that the Chinese sauropods Shunosaurus, Omeisaurus, Mamenchisaurus and Euhelopus form a natural group called Euhelopodidae (Table 1). In these two analyses, Upchurch presented character evidence that (1) supported euhelopodid monophyly and (2) excluded euhelopodids from membership in Neosauropoda. Wilson & Sereno (1998) and Wilson (2002) presented an alternative view that these four Chinese genera form a paraphyletic series including non-neosauropods (Shunosaurus, Omeisaurus + Mamenchisaurus) and one neosauropod closely related to titanosaurs (Euhelopus). Finally, Upchurch et al. (2004a) presented an analysis of genus-level sauropod relationships that did not recover a monophyletic Euhelopodidae. Although still positioned outside Neosauropoda, the four Chinese genera were resolved as a paraphyletic series.

Age of the mengyin formation The Mengyin Formation was originally considered to be Early Cretaceous (?Neocomian) in age by Wiman (1929). Subsequently, a Late Jurassic age (early Tithonian) was suggested on the basis of the dinosaurian fauna (Young 1958; Mateer & McIntosh 1985; Dong 1992) and conchostrachans (Chen et al. 1982), which has been accepted by most authors (e.g. Weishampel 1990; Barrett et al. 2002; Weishampel et al. 2004). However, X.-C. Wu et al. (1994:227) considered the co-occurrence of the crocodyliform Shantungosaurus and the turtle Sinemys as evidence that the Mengyin Formation was correlated with Early Cretaceous deposits in the Luohandong Formation of Inner Mongolia, which is considered Barremian in age (ca. 130–125 Ma; Averianov & Skutschas 2000). Dong (1995:94) likewise referred the Mengyin Group to the Early Cretaceous Psittacosaurus Complex (ca. 120 Ma; H.-Y. He et al. 2004). More recently, Barrett & Wang (2007) described Euhelopus-like teeth from the Yixian Formation that led them to infer a possible Aptian age for the Mengyin Formation. There is a growing consensus that the Mengyin Formation is Early Cretaceous, rather than Late Jurassic in age, although more specific determination is not yet possible. Accordingly, we ascribe an age range of Barremian–Aptian (ca. 130–112 Ma) to Euhelopus zdanskyi.

204

Table 1


Classification of the Chinese sauropods Euhelopus (= Helopus), Tienshanosaurus, Omeisaurus, Mamenchisaurus and Shunosaurus.

Reference

Higher taxon

Content

Wiman 1929

Helopus

Young 1937

Helopodinae: Cardiodontidae (= Cetiosauridae) Morosauridae: Helopodinae

Young 1939

Brachiosauridae: Helopodinae

Lapparent & Lavocat 1955 Romer 1956

Titanosauridae Brachiosauridae: Euhelopodinae

Young 1958

Bothrosauropodidae: Astrodontidae

Steel 1970

Homalosauropodidae: Titanosaurinae Camarasauridae: Euhelopodinae

Young & Zhao 1972

Homalosauropodidae: Mamenchisauridae Homalosauropodidae

Dong et al. 1983

Bothrosauropodidae: Euhelopodidae Camarasauridae: Cetiosaurinae Camarasauridae: Euhelopodinae

X. He et al. 1988

Mamenchisauridae

Zhang 1988 McIntosh 1990

Cetiosauridae Camarasauridae: Camarasaurinae Diplodocidae: Mamenchisaurinae Cetiosauridae: Shunosaurinae

Upchurch 1995, 1998

Eusauropoda: Euhelopodidae

Dong 1998

Camarasauroidea: ‘Euhelopidae’ Camarasauroidea: Mamenchisauridae

Martin-Rolland 1999

Camarasauroidea: Barapasauridae Somphospondyli Eusauropoda (paraphyletic series) Euhelopodidae: Euhelopodinae

Tang et al. 2001a

Euhelopodidae: Shunosaurinae Mamenchisauridae

Ouyang & Ye 2002

Mamenchisauridae

Wilson & Sereno 1998

Helopus Tienshanosaurus Helopus Omeisaurus Tienshanosaurus Helopus Euhelopus Omeisaurus Tienshanosaurus Helopus Omeisaurus Tienshanosaurus Mamenchisaurus Euhelopus Mamenchisaurus Omeisaurus Tienshanosaurus Mamenchisaurus Omeisaurus Tienshanosaurus Euhelopus Shunosaurus Mamenchisaurus Omeisaurus ‘Helopus’ Mamenchisaurus Omeisaurus Shunosaurus Euhelopus Tienshanosaurus Mamenchisaurus Omeisaurus Shunosaurus Euhelopus Mamenchisaurus Omeisaurus Shunosaurus Euhelopus Mamenchisaurus Omeisaurus Shunosaurus Euhelopus Omeisaurus Shunosaurus Euhelopus (= Tienshanosaurus) Mamenchisaurus Omeisaurus Shunosaurus Mamenchisaurus Omeisaurus Mamenchisaurus Omeisaurus


Table 1

205

Continued.

Reference

Higher taxon

Content

Wilson 2002

Somphospondyli Eusauropoda: Omeisauridae

Euhelopus Mamenchisaurus Omeisaurus Shunosaurus Euhelopus Mamenchisaurus Omeisaurus Shunosaurus Euhelopus Mamenchisaurus Omeisaurus Shunosaurus

Upchurch et al. 2004a

This analysis

Eusauropoda Eusauropoda (paraphyletic series)

Somphospondyli Eusauropoda (paraphyletic series)

Systematic palaeontology DINOSAURIA Owen, 1841 SAURISCHIA Seeley, 1887 SAUROPODA Marsh, 1878 NEOSAUROPODA Bonaparte, 1986 TITANOSAURIFORMES Salgado et al., 1997 Euhelopus Romer, 1956 Euhelopus zdanskyi (Wiman 1929) Figs 5–25; Supplementary Data Figs 1–5 “Supplementary data” available on Cambridge Journals Online: http://www. journals.cup.org/abstract S1477201908002691 HOLOTYPE. PMU 233 (exemplar a) and exemplar c (the latter’s accession number and whereabouts are unknown; X. Xing, pers. comm., 2007). Exemplar a comprises a partial skull (right and left premaxillae, maxillae, lacrimals, quadratojugals and palatines, left nasal, left postorbital, left squamosal, right quadrate, right pterygoid) and lower jaws (right and left dentaries, surangulars, angulars, left prearticular), 28 articulated presacral vertebrae, a left scapula, left coracoid, left humerus and left femur (Fig. 2; Wiman 1929; Mateer & McIntosh 1985; Young 1935). LOCALITY AND HORIZON. Mengyin Formation, central Shandong Province, China (Fig. 1). The age of the Mengyin Formation remains controversial, but correlation with other units in Asia suggests an Early Cretaceous age. REFERRED SPECIMENS. PMU 234 (exemplar b), which includes an articulated series of nine dorsal vertebrae and a sacrum, two dorsal ribs, a nearly complete pelvis and right hindlimb lacking metatarsal V and several pedal phalanges (Fig. 2). Britt (1993:125–128) tentatively referred to Euhelopus an isolated, nearly complete, posterior cervical vertebra (IVPP 10601) from the Shishougou Formation (Junggar Basin) of China. He identified small, thin-walled chambers (camellae) extending throughout the interior of the centrum and neural arch (Britt 1993: fig. 14) and on that basis questioned McIntosh’s (1990) allocation of Euhelopus to Camarasauridae. IVPP 10601 differs from Euhelopus cervical vertebrae in its nearly circular centrum cross-section, large pleurocoels, relatively short centrum and tall neural spine. However, we agree with Britt (1993) that camellate pneu-

maticity is evidence against phylogenetic affinities with Camarasaurus. Ruiz-Omeñaca et al. (1997) and Canudo et al. (2002) described isolated Euhelopus-like teeth from the lower Barremian (Lower Cretaceous) of La Cantalera (Teruel), Spain and proposed an Early Cretaceous geographical connection between Europe and Asia. Canudo et al. (2002: figs 2–3) defended this claim on the basis of prominent “cingular cusps”, which is their term for the lingual crown buttresses that are, thus far, only known in Euhelopus zdanskyi (Wilson 2002: appendix C). Buffetaut et al. (2002) described several isolated euhelopodid teeth from the Phu Kradong Formation of Dan Luang in northeastern Thailand. They drew attention to their close resemblance to Omeisaurus and Mamenchisaurus, genera that are no longer considered to be closely related to Euhelopus (and therefore not euhelopodids; see below). Nevertheless, nearly all of the Dan Luang teeth possess the autapomorphic lingual crown buttress that characterises Euhelopus. Likewise, Barrett & Wang (2007) referred isolated teeth from the Lower Cretaceous Yixian Formation of China that also bear these distinct lingual crown buttresses. This shared unique feature suggests that the La Cantalera, Dan Luang and Yixian specimens are closely allied to Euhelopus, but more material is needed before referral to the genus and species can be justified.

REVISED DIAGNOSIS. Procumbent teeth with asymmetrical crown-root margin (i.e. the mesial margin is closer to the apex of the crown) and well developed crown buttresses on mesiolingual crown surface, axis with postspinous fossa containing three coels, cervical 3 neural spine with laterally compressed, anteriorly projecting triangular process, postaxial cervical vertebrae with variably developed epipophyses and more subtle ‘pre-epipopophyses’ below the prezygapophyses, cervical neural arches with epipophyseal– prezygapophyseal lamina separating two pneumatocoels, cervical pleurocentral openings reduced to foramina, cervical neural spines reduced anteroposteriorly and dorsoventrally, anterior cervical vertebrae with three costal spurs on tuberculum and capitulum, middle cervical ribs hang well below centrum margin due to elongate parapophyses and capitula, presacral neural spines 11–30 divided, presacral neural spines 16–21 “trifid” with median tubercle as large or larger than metapophyses, middle and posterior dorsal parapophyseal and diapophyseal laminae cross to form “K” configuration, presacral pneumaticity extending into the ilium.

206


Figure 4 Cladistic hypotheses of the relationships of Chinese sauropods (in bold-face type) to other sauropod genera. For simplicity, some terminal taxa have been combined into suprageneric taxa.

Description The anatomy of Euhelopus zdanskyi exemplars a and b was carefully described by Wiman (1929), and Young (1935) described and illustrated key limb elements of exemplar c that were not preserved in exemplars a and b. The skull of Eu-

helopus exemplar a was redescribed by Mateer & McIntosh (1985), who reidentified several elements, provided more anatomical detail and reconstructed the skull in lateral view. Although we comment on specific aspects of the skull, our redescription focuses on the vertebral, pelvic and hindlimb elements, which were not discussed by Mateer & McIntosh


207

quadratojugals and palatines, a left nasal, left postorbital, left squamosal, right quadrate, right pterygoid, paired dentaries, surangulars, angulars and a left prearticular. The premaxillae, maxillae and dentaries contain several teeth in situ, but the collection also includes at least 9 isolated teeth that presumably belonged to exemplar a. Wiman (1929:7) provided some information on the state of preservation of the skull when first discovered: “the skull was disarticulated but the appropriate elements were lying on top of each other and side by side within a small limited area in front of the axis. In several cases the bones lay so close to each other that it was difficult to separate them especially because some were very thin. Perhaps it is because the skull disarticulated before being buried and crushed by the overlying sediment that the bones are so undeformed that the skull when rebuilt is only slightly more asymmetrical than it would have been in life.”

Figure 5 Photograph of Otto Zdansky examining Euhelopus zdanskyi exemplar b within its glass-enclosed mount in the 1950s. Photograph kindly provided by Museum of Evolution, Uppsala University.

(1985). In the description that follows, we draw attention to morphological features of the vertebral and appendicular skeleton that will help elucidate its phylogenetic relationships, many of which have not been discussed previously. Where appropriate, we note where our interpretation differs from Wiman (1929), which we have translated from the German. All quoted passages are from the translation by N. Insel, which is available at the Polyglot Paleontologist website (http://www.paleoglot.org/index.cfm). Exemplars a and b have been mounted within a display enclosure at the Palaeontological Museum in Uppsala since the 1930s (Fig. 5). Only the skull, axis, cervical 3 and the pes can be removed; all other elements are fixed in their original positions. We have reproduced the excellent plates illustrating Wiman’s monograph (Supplementary Data Figs 1–4) because they furnish the only views of certain elements that can no longer be accessed due to the constraints of the display enclosure (e.g. top of sacrum) or the way the specimen was mounted (e.g. articular surfaces of vertebrae). We supplement these figures with photographs and diagrams of specific areas of interest (Figs 6–26). Our description uses traditional orientational descriptors and anatomical terminology (i.e. Romerian terms), rather than standardised terms from the Nomina Anatomica Avium or Nomina Anatomica Veterinaria, which apply to birds and domesticated mammals, respectively (see discussion in Wilson 2006).

Skull (Fig. 6; Supplementary data Figs 1, 2) The skull of Euhelopus was described and figured in the original monograph by Wiman (1929: pls 1–2) and then redescribed by Mateer & McIntosh (1985: figs 1–5). Here, therefore, we do not attempt a comprehensive redescription, but focus instead on new observations and amendments.

PRESERVATION OF CRANIAL ELEMENTS. The preserved cranial elements include paired premaxillae, maxillae, lacrimals,

PREMAXILLA. Both premaxillae are preserved virtually intact except for the loss of the middle and upper portions of the ascending process, which would normally form most of the internarial bar. The premaxilla resembles that of other non-diplodocoid sauropods, with a robust tooth-bearing main body, ascending process and posterolateral process. The region where the ascending process meets the main body is slightly damaged on both sides, but it is clear that this process was offset a little posteriorly relative to the anterior margin of the main body. This step-like offset is seen in most non-diplodocoid sauropods (Wilson 2002; Upchurch et al. 2004a), although it is not developed as strongly in Euhelopus as it is in Camarasaurus (Madsen et al. 1995) or Brachiosaurus (Janensch 1935–36). The premaxillary posterolateral process is subtriangular in outline and forms a thin sheet of bone that has been incorporated into the floor of the external narial fossa, which cannot be seen clearly in lateral view. The ventrolateral margin of the posterolateral process contacts the dorsal edge of the anterior ramus of the maxilla and extends posteriorly onto the base of the maxillary ascending process. The dorsomedial margin of the posterolateral process merges into the posterior part of the base of the internarial bar. As a result, the left and right posterolateral processes create a deep, narrow groove extending vertically up the posterior midline of the base of the internarial bar. Thus, although the premaxillae and maxillae form an external narial fossa, the medial portions of these bones do not contact each other on the midline. The bases of the premaxillary teeth are supported labially by a ventral extension of the lateral surface of the premaxillary main body (i.e. the “lateral plate”). This structure is also seen in the maxillae and dentaries. The subnarial foramen is a small, elliptical opening that lies on the premaxilla–maxilla suture. It is visible in lateral view, but lies within the external narial fossa. The foramen faces mainly laterally and a little dorsally. MAXILLA. The maxillary ascending process is directed posterodorsally in lateral view. There is a small preantorbital fenestra that pierces the body of the maxilla below the posteroventral corner of the antorbital fenestra (Fig. 6). The fenestra is matrix-filled and was not noticed previously (Mateer & McIntosh 1985; Upchurch 1995, 1998; Wilson & Sereno 1998; Wilson 2002; Upchurch et al. 2004a). The antorbital fenestra lies flush with the lateral surface of the snout; i.e. there is no antorbital fossa on the maxillary ascending process

208


Figure 6 Euhelopus zdanskyi exemplar a (PMU 233). Stereophotographs of right premaxilla and maxilla in medial view, showing preantorbital fenestra (paof) and lingual crown buttresses (lcb). Scale bar = 5 cm.

or lacrimal. The presence of an articular area for the lacrimal on the posterior part of the dorsal margin of the maxilla suggests that the jugal made very little, if any, contribution to the margin of the antorbital fenestra.

NASAL. We interpret the bone identified as the right frontal by Mateer & McIntosh (1985: fig. 1c–d) to be the left nasal. This element was not described or figured by Wiman (1929). It is very thin dorsoventrally (maximum thickness equals approximately 3 mm) and bowed slightly upwards so that it has a mildly concave ventral surface and correspondingly convex dorsal surface. Sauropod nasals are generally thinner than the frontals, and Mateer & McIntosh (1985:125) stated that: “It is a thinner bone than [the frontal] in Camarasaurus.” There are two projections that we interpret as anteromedial and anterolateral processes. The anteromedial process would have met its partner on the midline to form the posterior part of the internarial bar, and the anterolateral process would have contacted the maxillary ascending process, prefrontal and lacrimal at the corner of the antorbital fenestra. The anteromedial process is relatively long but broken at its anterior end. The anterolateral process is shorter, more slender and subtriangular in dorsal outline. The area between these processes is broadly concave and in our interpretation represents the posterior margin of the external naris. On the ventral surface, a low, rounded ridge extends along the anterolateral process and then divides into two branches at its base. One branch extends along the lateral margin of the bone to the posterior edge, whereas the other follows the underside of the possible narial margin and fades out midway between the bases of the two processes. Mateer & McIntosh’s (1985) identification of this element as a frontal was supported by the presence of a subdued ridge on the ventral surface for articulation with the laterosphenoid and orbitosphenoid. However, the orientations of the ridges on the ventral surface do not conform to those expected in a frontal, and they also lack the irregular transverse ridges and grooves characteristic of the suture between frontal and braincase elements, as well as the orbital ornamentation present on most saurischian frontals. POSTORBITAL. The postorbital is triradiate with a long, anteroventrally-directed jugal process. The orientation of the anteromedial and posterior processes suggest that the upper temporal bar was displaced ventrally in Euhelopus, so that the supratemporal fenestra would have been visible in lateral view, as in most eusauropods (Wilson & Sereno 1998). The posterior process articulates with a corresponding

triangular notch in the lateral surface of the squamosal, and it is clear that the latter element formed the posterior margin of the supratemporal fenestra. Consequently, Euhelopus lacked the derived exclusion of the squamosal from the margin of the supratemporal fenestra by a postorbital–parietal contact, which is a synapomorphy of Nemegtosaurus and Quaesitosaurus (Upchurch 1995, 1998, 1999; Wilson 2002, 2005a). The morphologies of the postorbital and squamosal together suggest that the supratemporal fenestra opened dorsolaterally, was wider transversely than anteroposteriorly, and was relatively large compared to the width across the skull roof. The jugal process of the postorbital has a subtriangular transverse cross-section that is wider transversely than dorsoventrally. This derived state is characteristic of Eusauropoda (Wilson & Sereno 1998). The absence of the jugal and the incomplete preservation of the jugal process of the postorbital make it difficult to reconstruct the shape of the lateral temporal fenestra. However, the postorbital indicates that this fenestra extended forwards below the orbit and Euhelopus was probably as derived in this respect as other neosauropods.

SQUAMOSAL. The left squamosal is mounted upside-down in the skull in the position of the right squamosal (Mateer & McIntosh 1985; see Supplementary Data Fig. 1). The preserved portion includes the ventral process, the articulation for the postorbital and the lateral part of the main body (i.e. a substantial part of the main body is missing medially, contra Mateer & McIntosh 1985). The ventral process is formed from very thin bone that is directed anteroventrally in its proximal part, but distally it extends ventrally. This process tapers to a sharp point in lateral view, although this may have been exaggerated by breakage. The thin sheet of bone is curved in horizontal cross-section, with a mildly convex anterolateral face and corresponding concave posteromedial face. The latter represents the area that covered the anterolateral part of the quadrate shaft in life. The anterolateral surface of the ventral process forms part of a fossa surrounding the lateral temporal fenestra. As the ventral process joins the main body, its anterior part is embayed medially with respect to the posterior part. The latter region is the lateral surface of the main body and this forms a ridge that curves upwards and forwards to define the dorsal margin of the lateral temporal fenestra and the lower boundary of the triangular slot for the posterior process of the postorbital. The region for reception of the postorbital is particularly large and deep and


subtriangular but is thin-walled medially. The medial surface of the postorbital articulation is mildly concave and is continuous with the anterior face of the main body of the squamosal. Ventrally, this medial surface is delimited by a ridge, which projects medially as it extends anteroventrally; this forms the roof of the fossa for the quadrate.

QUADRATOJUGAL. Both the left and right quadratojugals are almost completely preserved, lacking only the end of their dorsal processes. The anterior process of the quadratojugal is very long and slender at its base, where it has a subtriangular cross-section. Its dorsolateral surface is slightly excavated, marking the ventral margin of the lateral temporal opening. Towards its anterior end, this process widens vertically to form a laterally-compressed, rounded plate. The dorsal process projects perpendicular to the anterior one, and the two merge smoothly into each other at the posteroventral corner of the bone. The dorsal process expands both anteroposteriorly and mediolaterally above its junction with the anterior process and then tapers to a point at its anterodorsal tip. Its distal extreme is broken away, and its length relative to the length of the anterior process cannot be stated with certainty. The anterior part of the lateral surface of the dorsal process is slightly excavated. This excavation narrows towards its ventral end and may represent evidence for a contact with the ventral process of the squamosal. Such a quadratojugal– squamosal contact is the plesiomorphic state found in most sauropods except diplodocoids (Upchurch 1998; Upchurch et al. 2004a). QUADRATE. The right pterygoid and quadrate were found in articulation (Wiman 1929:8), although a break just posterior to the ectopterygoid process may mean that the two elements are no longer in their correct relative orientation (Mateer & McIntosh 1985). The shape of the quadrate, its articulation with the pterygoid and the 90◦ angle between the anterior and dorsal rami of the quadratojugal, all suggest that the quadrate in Euhelopus was orientated nearly vertically, as in most sauropods, and therefore did not display the derived anteroventrally slanting orientation found in diplodocoids. Wiman (1929:8) stated: “from the back a big foramen quadrati is visible between the quadrate and the quadratojugal.” As Mateer & McIntosh (1985) noted, this implies that the quadratojugal formed the lateral wall of the fossa, whereas in fact this is formed by the quadrate. Although the true depth of the fossa is obscured by matrix, Euhelopus clearly possesses the derived “deep” fossa also found in macronarians, some diplodocoids such as Limaysaurus (= “Rebbachisaurus” and “Rayososaurus”) tessonei (Calvo & Salgado 1995; Upchurch 1998) and some non-neosauropod eusauropods (e.g. Mamenchisaurus sinocanadorum, Russell & Zheng 1993). Towards the dorsal end of this fossa, the lateral margin curves medially to partially enclose this region. However, the extent of this closure cannot be determined because of breakage. The distal (articular) end of the quadrate is damaged, but its surface slopes ventromedially to an anteroposteriorlyexpanded region, as in other sauropods (Upchurch & Barrett 2000). PTERYGOID. The pterygoid of Euhelopus was described as having a unique shape among sauropods (Mateer & McIntosh 1985). In particular, the anterior process is a flat plate of bone, with a rounded spatulate lateral profile, that projects upwards into the anterior part of the orbit. The anterior and

209

ectopterygoid processes are also deflected laterally at an unnatural angle that would make articulation with the palatine and vomer very difficult if these latter bones had been preserved in situ. This deflection also makes the ectopterygoid process appear much smaller in lateral view than it would have been in life. Anterior to the fractured area, the pterygoid is heavily reconstructed and it is, therefore, difficult to determine its original morphology. However, there is a faint ridge on the lateral surface of the anterior process, close to its ventral margin, that extends forwards from the base of the ectopterygoid process to approximately half way along the anterior process. This ridge may have articulated with the palatine. There is little information on the position and morphology of the fossa for articulation with the basipterygoid process. The region on the medial surface, just above the base of the ectopterygoid process, is damaged and the approximate position of the articular fossa is now occupied by a small broken mass of bone. The finger-like process that curves around the tip of the basipterygoid process in the pterygoids of Brachiosaurus and Camarasaurus (Upchurch 1998; Wilson & Sereno 1998) appears to be absent in Euhelopus, but this may have been caused by poor preservation. The quadrate articulation resembles those found in other sauropods.

PALATINE. The elements that Wiman (1929; Supplementary Data Fig. 2) identified as vomers were reidentified by Mateer & McIntosh (1985: fig. 2c–d) as palatines. Although the palatine of Euhelopus differs in some respects from those of Camarasaurus (Madsen et al. 1995: figs 5, 38) and particularly Brachiosaurus (Janensch 1935–36: figs 33–35), as noted by Mateer & McIntosh (1985), we nevertheless agree with their identification. The palatine is roughly triangular in medial and lateral views, with a rounded ventral portion and a platelike dorsal portion. The right and left palatines are nearly completely preserved, and lack only a portion of their dorsal blade. The left palatine appears to be slightly more distorted than the right. Anteriorly, the rounded ventral portion of the palatine extends forward as a rodlike process that terminates in a flattened, subcircular articulation for the maxilla, as in all eusauropods (Wilson 2002). In medial view, the ventral portion of the palatine is rounded and contacted the pterygoid. In lateral view, the ventral portion of the palatine forms a sharpened edge that contacted the ectopterygoid. The ectopterygoid articulation probably continued dorsally onto the blade of the palatine, but its exact extent is difficult to determine in the absence of an ectopterygoid. The dorsal blade of the palatine is thin and reaches its peak anteriorly, near the neck of the maxillary process. The blade of the palatine is straight in Euhelopus and Camarasaurus (Madsen et al. 1995), unlike the curved blade present in Brachiosaurus (Janensch 1935–36). In lateral view, a thick vertical ridge occupies the anterior portion of the blade; behind it is a subtriangular fossa some portion of which contacted the ectopterygoid. The inverse of this topography can be seen on the medial side of the dorsal blade, which bears a vertical fossa anteriorly and a more prominent posterior portion. Based on comparisons with Camarasaurus (Madsen et al. 1995), the vomer probably articulated in the fossa on the anteromedial edge of the dorsal blade of the palatine.

210


DENTARY. Complete right and left dentaries are preserved. The dentary increases slightly in depth towards the symphysis, as in most sauropods (Wilson & Sereno 1998; Upchurch & Barrett 2000). The long axis of the symphyseal articular surface is orientated at approximately 110◦ relative to the long axis of the jaw. The ventral margin of the mandible, at its anterior end, is gently rounded and lacks the sharp, triangular, chin-like projection found in some diplodocoids (Upchurch 1998; Wilson 2002). The parasagittallyorientated posterior part and transverse anterior part of the dentary merge smoothly into each other to form a mandible that is U-shaped in dorsal view. Thus, Euhelopus resembles Brachiosaurus and Camarasaurus in this regard and does not display the more rectangular dorsal mandibular profile seen in diplodocoids. SURANGULAR. The dorsal margin of the right surangular has been damaged and it is therefore not possible to assess the height of this bone relative to that of the angular. ANGULAR. Most of the left angular is preserved, although the anterior and posterior ends appear slightly damaged and the exact length and outline of the bone cannot be determined precisely. The angular is a long, plate-like bone that is orientated vertically and bears a transversely thickened ventral rim. The ventral margin is slightly concave in lateral view and the dorsal margin is more strongly convex. The ventral thickening is created by a medial expansion that underlies an excavated area on the anterior part of the medial surface. This ventral medial shelf is most prominent anteriorly. There is no indication that the angular, surangular, or dentary formed any part of the margin of an external mandibular fenestra and it seems very likely that the latter was closed in Euhelopus as in most eusauropods (Upchurch 1998; Wilson & Sereno 1998). There are no foramina penetrating from one side of the angular to the other. In dorsal view the angular is slightly bowed laterally although this is exaggerated by the presence of the medially-directed ventral shelf. DENTITION. Wiman’s (1929) description of the teeth of Euhelopus is brief and focuses on macrowear and the relative positioning of the crowns. Mateer & McIntosh (1985) did not describe the teeth at all. A detailed description of the teeth is, therefore, provided below. There are four teeth in each premaxilla, 10 in each maxilla and 13 in each dentary. Previously published figures and photographs of the tooth-bearing elements show the teeth projecting somewhat anteriorly, roughly parallel to the symphysis. The enamel margin at the crown–root junction is asymmetrical – slanting apically towards the mesial side – indicating that the procumbent orientation of the teeth is a genuine feature rather than the result of distortion. The largest teeth are situated at the anterior ends of the upper and lower jaws. As in most sauropods, apart from narrowcrowned forms such as Diplodocus and titanosaurs, adjacent teeth in Euhelopus contact each other and are arranged in a slightly overlapping “imbricate” pattern (Wilson & Sereno 1998; Wilson 2002). The tooth crowns expand very slightly mesiodistally immediately adjacent to the root, but not prominently to form the broad spatulate crowns found in Camarasaurus and several basal eusauropods such as Omeisaurus. The Euhelopus teeth then taper towards relatively narrow apices. As a result, the Euhelopus crowns are more parallelsided in labial view, like those of Brachiosaurus and several

other basal titanosauriforms. The slenderness indices (SI), maximum crown length divided by maximum mesiodistal width (Upchurch 1998), for in situ teeth from the left premaxilla, maxilla and dentary, are close to 2.0 or less. The labial surface of each crown is convex both mesiodistally and towards the apex, with a narrow groove (or change of angle) extending from root to apex near both the mesial and distal margins of the crown. The lingual surface is mildly concave. This concavity is created by the slight lingual deflection of the crown apex and the lingual curvature of the mesial and distal margins. However, within the lingual concavity is a prominent ridge that extends from the root to the apex, virtually filling the lingual concavity: consequently, the crowns are D-shaped in horizontal cross-section (Wilson & Sereno 1998). One of the most distinctive features of these teeth is a rounded boss-like structure on the lingual part of each mesial and distal margin, close to the base of the crown (Wilson 2002; Fig. 6). These lingual crown buttresses have also been observed in unnamed Early Cretaceous sauropod teeth from Spain (Canudo et al. 2002), Thailand (Buffetaut et al. 2002) and China (Barrett & Wang 2007), as mentioned above. As in virtually all other sauropod teeth and several basal sauropodomorphs (Upchurch et al. 2007), the tooth enamel in Euhelopus has a characteristic wrinkled or reticulate texture (Wilson & Sereno 1998; Wilson 2002). The macrowear tends to be in the form of concave facets on the mesial and distal margins close to the apex, creating “shoulders” on worn teeth, as also occurs in Camarasaurus and many basal eusauropods (Upchurch & Barrett 2000). The flat, high-angled apical wear facets observed in Brachiosaurus and many titanosaurs are absent in Euhelopus (Upchurch & Barrett 2000). Apart from changes in size, there are no observable differences in crown morphology along the jaws from mesial to distal, or between the upper and lower jaws. No denticles have been found on the teeth of Euhelopus. Recent examination of microwear on isolated Euhelopus crowns collected with exemplar a shows scratches that are roughly parallel to the tooth axis (P. Barrett, pers. comm., 2007).

Vertebral column (Figs 7–23; Supplementary Data Figs 3, 4) Euhelopus exemplars a–c preserve overlapping sections of vertebrae that together form a series from the axis to the last sacral vertebra. Wiman (1929) did not estimate the position of the cervicodorsal or dorsosacral boundaries, but he did estimate that the first vertebra of exemplar b corresponds to the 22nd vertebra from the skull in exemplar a, which implies 36 precaudal vertebrae. We agree with this estimate and below discuss our justifications for positioning the cervicodorsal and dorsosacral boundaries. Although we were able to establish that there are six sacral vertebrae and 30 presacral vertebrae, the position of the cervicodorsal boundary is ambiguous, because the 18th vertebra from the skull bears features of both the cervical and dorsal region of the axial column. We provisionally suggest that Euhelopus had 17 cervical, 13 dorsal and 6 sacral vertebrae. Although exemplar a and exemplar b are nearly the same size, they exhibit slightly different states of fusion of vertebral sutures. Exemplar a bears no trace of neurocentral sutures nor sutures between the postaxial cervical vertebrae and their associated ribs. Exemplar b, in contrast, bears


211

Figure 7 Euhelopus zdanskyi exemplar a (PMU 233). Axis vertebra in left lateral (A) and right posterolateral (B) views. Abbreviations used for vertebrae figs: acdl, anterior centrodiapophyseal lamina; acpl, anterior centroparapophyseal lamina; ca, capitulum; cpol, centropostzygapophyseal lamina; co, coel; cprl, centroprezygapophyseal lamina; csp, costal spurs; di, diapophysis; epi, epipophysis; eprl, epipophyseal–prezygapophyseal lamina; fl, flange; fo, fossa; nsp, neural spine; p, parapophysis; pc, pleurocoel; pcdl, posterior centrodiapophyseal lamina; pcpl, posterior centroparapophyseal lamina; ppdl; parapodiapophyseal lamina; podl, postzygodiapophyseal lamina; poz, postzygapophysis; prdl, prezygodiapophyseal lamina; prpl, prezygoparapophyseal lamina; prepi, pre-epipophysis; prz, prezygapophysis; spol, spinopostzygapophyseal lamina; sprl, spinoprezygapophyseal lamina; tlp, triangular lateral process; tu, tuberculum; Arabic numbers refer to pneumatic coels described in text. Scale bar = 5 cm.

partially fused neurocentral sutures that can be identified on the lateral aspect of all dorsal vertebrae. However, exemplars a and b overlap only across six vertebrae, and the difference in maturity between the two specimens may be attributed to the fact that exemplar b consists of the posterior portion of the presacral column, and exemplar a consists of the anterior portion. The neurocentral sutures of posterior dorsal vertebrae typically fuse later than those of cervical and anterior dorsal vertebrae in Camarasaurus (Ikejiri et al. 2005). The cervical, dorsal and sacral vertebrae all show signs of pneumaticity. As will be discussed below, slight regional differences in the extent and nature of pneumaticity are present. We employ the nomenclature for vertebral laminae and associated abbreviations proposed by Wilson (1999), with appropriate additions to this system to accommodate the unusual structures found in Euhelopus. The abbreviations for laminae are used in the text after the first usage of each term; abbreviations appear in uppercase, and their plurals are followed by a lowercase “s” (e.g. spinopostzygapophyseal laminae = “SPOLs”). We note here that the Roman numerals written directly on the vertebrae (and visible in photographs) differ by one from those numbers discussed and labelled in Wiman (1929), which are their inferred position in the series. The inked numbers on the vertebrae correspond to their position in the preserved series, which does not include the atlas. The axis vertebra is numbered “I” because it was the first preserved cervical (Fig. 7), but is labelled with a Roman numeral “II” in the figures of Wiman (1929). In the interest of brevity, the description below refers to vertebrae by their region and

number (e.g. “cervical 1”, “dorsal 4”) rather than by more complete descriptors (i.e. “first cervical vertebra”, “fourth dorsal vertebra”).

Cervical vertebrae (Figs 7–13; Supplementary Data Fig. 3) Exemplar a includes a complete, articulated cervical series extending from the axis to cervical 17. All cervical vertebrae show the camellate pneumatic structure that is characteristic of titanosauriform sauropods (Upchurch 1998; Wilson & Sereno 1998; Wedel et al. 2000). Pneumaticity extends throughout the entirety of the centrum and neural arch in all cervical vertebrae. The extent of pneumatisation can be confirmed in fortuitous breaks in the external bone surface where pneumatic chambers are exposed, but it can also be inferred without these breaks, where the external bony surface conforms to the internal pneumatic chambers (e.g. see Figs 9–12, 14, below).

AXIS (Fig. 7; Supplementary Data Fig. 3). The axis is well preserved, but has been damaged posterolaterally on its right side. A portion is missing from the centrum, the right postzygapophysis is absent and the neural spine is distorted and the base of its anterior margin has been reconstructed. The centrum is strongly pinched at midlength and bears expanded anterior and posterior ends. Ventrally, the centrum is gently concave anteroposteriorly and slightly convex transversely. The anterior articular surface is roughened and divided into low, rounded middle and upper portions that are

212


separated from the ventral and ventrolateral region by a deep groove. This upper portion corresponds to the odontoid process (atlantal pleurocentrum). The parapophysis is marked by a low, roughened area at the extreme anteroventral corner of the lateral surface of the centrum. The long, shallow fossa on the lateral surface of the centrum is a rudimentary pleurocoel that is divided by a small, oblique lamina. The diapophysis is best preserved on the left side and forms a process positioned just dorsal to the neurocentral junction. Its base lies near the anterior end of the neural arch, very close to the prezygapophysis, but the process itself projects laterally, posteriorly and slightly ventrally. Low, rounded ridges extend from the base of the diapophysis and correspond to the four main diapophyseal laminae – anterior and posterior centrodiapophyseal laminae (ACDL, PCDL), which extend along the top of the centrum and form the dorsal margin of the pleurocoel, a prezygodiapophyseal lamina (PRDL) and a prominent, curving postzygodiapophyseal lamina (PODL). The prezygapophysis is a small, triangular platform that projects laterally and slopes a little ventrally from the base of the neural spine. It appears that the prezygapophyses were separated from each other on the midline by the tall, thin and anteriorly convex neural spine. The postzygapophysis is very large and positioned high above the centrum. It projects a short distance beyond the posterior margin of the centrum. This is the plesiomorphic state for sauropodomorphs and differs from the derived condition found in certain prosauropods, in which these processes terminate at the posterior margin of the centrum (Sereno 1999; Yates & Kitching 2003; Upchurch et al. 2007). The postzygapophyseal articular surface is large, flat and elliptical. It faces downwards and curls slightly ventrally towards its medial margin. Plate-like SPOLs meet at the summit of the neural spine to form a deep postspinal fossa that is floored by thin intrapostzygoposphyseal laminae (TPOLs) that extend medially from the medial edge of each postzygapophysis. The medial surface of the SPOL has three excavated areas that are separated from each other by low ridges (Supplementary Data Fig. 4B). These excavations become larger towards the anterior region of this surface and appear to be an autapomorphy of Euhelopus (Wilson 2002). The summit of the neural spine is robust, thickened and positioned near mid-centrum. Anteriorly, it slopes ventrally and forms a transversely compressed plate.

POSTAXIAL CERVICAL VERTEBRAE (Figs 8–13; Supplementary Data Fig. 3). The 15 postaxial cervical vertebrae (see discussion of the cervicodorsal junction below) form a natural series that was found in articulation (Wiman 1929). The cervical vertebrae are in excellent condition, but there is minor damage to several parts of the series. The left postzygapophysis of cervical 3 is missing, cervical 4 and the spines of cervicals 5 and 6 are poorly preserved, and the epipophyses of cervical 11 are broken. All postaxial cervical centra are strongly opisthocoelous, with a well-developed sub-hemispherical anterior articulation and corresponding concave posterior articulation. As noted by Upchurch (1998), these articular surfaces are taller than wide, an unusual condition that also occurs in Omeisaurus tianfuensis (X. He et al. 1988), Mamenchisaurus hochuanensis (Young & Zhao 1972), Erketu ellisoni (Ksepka & Norell 2006) and Shunosaurus lii (Zhang

Figure 8 Euhelopus zdanskyi exemplar a (PMU 233). Cervical vertebra 3 in right lateral view. See Fig. 7 for abbreviations. Scale bar = 5 cm.

1988). The anterior articular ball is asymmetrical in lateral view, with its apex positioned dorsally. This condition is most noticeable in the anterior and middle cervical vertebrae (Figs 8–11). The ventral edge of the posterior articular cup projects more posteriorly than the dorsal edge, as in other sauropods. The cervical centra are relatively long and slender throughout most of the series, but they become shorter anteroposteriorly and increase in diameter towards the cervicodorsal junction (Table 2). In cervicals 3 and 4, the ventral surface of the centrum is shallowly concave between the parapophyses but becomes flat posteriorly. The ventral surfaces of cervicals 5–17 are concave both longitudinally, because of the expanded articular ends, and transversely, due to sharp ridges that extend along the ventrolateral edges of the centrum. Posteriorly, these ridges become more flangelike and hang below the level of the centrum. There is a faint midline ridge within the anteriorly placed ventral depression in cervical 3, but this is absent in all other cervical centra until it reappears in cervical 17. The postaxial cervical parapophyses lie at the anteroventral margin of the centrum and are directed ventrolaterally. The posterior margin of each parapophysis merges into the ventrolateral ridge described above. The lateral surface of the centrum of cervical 3 has a small but sharp excavation that is divided by a ridge. The size and depth of this pleurocoel, and the prominence of the anterodorsally sloping oblique lamina that subdivides it, increase in more posterior cervical vertebrae. By cervical 17 the pleurocoel is a deep pit, but the oblique lamina is absent. The postaxial cervical neural arches are relatively low and occupy the length of the centrum apart from the section immediately below the postzygapophyses. As in other sauropods, the height of the neural arch increases towards the cervicodorsal junction (Table 2). The prezygapophyses are large processes that project anteriorly and slightly laterally beyond the margin of the articular ball in dorsal view. They have flat articular surfaces that face dorsomedially at an angle of about 30◦ above the horizontal. They slope so that they face a little forwards. The size of the articular facets and the distance separating them increase in the posterior part of the series, especially


213

Figure 9

Euhelopus zdanskyi exemplar a (PMU 233). Cervical vertebra 8 in right lateral view. See Fig. 7 for abbreviations. Scale bar = 10 cm.

Figure 10

Euhelopus zdanskyi exemplar a (PMU 233). Cervical vertebra 10 in right lateral view. See Fig. 7 for abbreviations. Scale bar = 10 cm.

214


Figure 11 Euhelopus zdanskyi exemplar a (PMU 233). Cervical vertebra 14 and 15 in right lateral view. See Fig. 7 for abbreviations. Scale bar = 10 cm.

cervicals 16 and 17, as in other sauropods. Prezygapophyses are supported from below by a single, stout centroprezygapophyseal lamina (CPRL). Unlike diplodocids (Upchurch 1998), the CPRLs of Euhelopus do not bifurcate dorsally to form lateral and medial branches. These laminae are unusual in Euhelopus because each gives rise to a short, blunt process just below the prezygapophysis. These structures resemble the prominent epipophyses located above the postzygapophyses (see below) and are referred to here as “preepipopophyses”. These structures are also present in other sauropods, such as Erketu (Ksepka & Norell 2006). In the anterior cervical vertebrae, the prezygapophyses extend posteromedially and meet each other on the midline at the top of the neural canal opening via intraprezygapophyseal laminae (TPRLs). Despite the increasing height of the middle and posterior cervical neural arches, the TPRLs maintain their close association with the top of the neural canal, and there is no “anterior midline lamina” (with coels on either side). Euhelopus differs in this respect from other sauropods such as Apatosaurus (Gilmore 1936; Upchurch et al. 2004b) and Cetiosaurus (Upchurch & Martin 2002). In cervical 3, the diapophysis is situated at the neurocentral junction and projects ventrolaterally. It is supported posteriorly by the PCDL, which extends as a low ridge along the neurocentral junction above the pleurocoel, fading out well before the posterior margin of the arch. In subsequent cervical vertebrae, the diapophyses are positioned above the neurocentral junction but below the level of the zygapophyses. They are supported by well-developed diapophyseal laminae (PCDL, PRDL, PODL), as in other nontitanosaurian sauropods. The ACDL is generally poorly developed or absent in the cervical vertebrae, but it can be seen as a short, anteroventrally-directed lamina in cervical 17 and the dorsal vertebrae (see below). The PCDL is prominent and is directed posteriorly and a little ventrally in postaxial cervical vertebrae until cervical 17, where it becomes noticeably steeper. There is a small pit in the infrapostzygapophyseal

fossa, just below the PODL and immediately above and behind the base of the diapophysis in cervicals 3 and 4, but this coel is absent from cervical 5 onwards. The diapophyses themselves remain relatively short processes that are directed laterally and curve ventrally (i.e. they are “pendant”) as far posteriorly as cervical 17. The postzygapophyses are well-developed processes with gently concave articular surfaces that overhang the posterior end of the centrum. Each postzygapophysis is supported from below by a stout CPOL. Despite the increasing height of the neural arch in middle and posterior cervical vertebrae, there is no evidence that Euhelopus possessed the “posterior midline lamina” or associated coels present in other sauropods (e.g. Cetiosaurus; Upchurch & Martin 2002). All postaxial cervical vertebrae bear a prominent epipophysis on the dorsal surface of each postzygapophysis. These structures continue into the dorsal series (see below). In the anterior and posterior cervical vertebrae (cervicals 2–5, 11–17), the epipophyses are short and rounded, but in the middle cervical vertebrae (cervicals 6–10), they extend posteriorly beyond the edge of each postzygapophysis (Figs 10–12). Each epipophysis is separated ventrally from its respective postzygapophysis by a shallow groove. The lateral margin of each epipophysis merges with the free lateral edge of each postzygapophysis, which together extend as a combined ridge across the neural spine to the base of the prezygapophysis. We term this ridge the epipophyseal– prezygapophyseal lamina (EPRL) and consider its presence throughout the cervical series an autapomorphy of Euhelopus that is also present in Nigersaurus (Sereno et al. 2007), Zapalasaurus (Salgado et al. 2006), Camarasaurus (Ostrom & McIntosh 1966: pls 10, 11) and in some theropods (e.g. Rajasaurus, Wilson et al. 2003; Stokesosaurus, Benson 2008). The neural arch laminae on the lateral aspect of the neural spine (i.e. SPRL, SPOL) define a hollow that is divided into an upper and lower coel by the EPRL (Figs 8–13). We term

215


Table 2

Measurements (in cm) of precaudal vertebrae of Euhelopus zdanskyi, taken from Wiman (1929:21). Vertebral centrum

Item no.

Length without anterior convexity a b

Posterior width a b

cv2 cv3 cv4 cv5 cv6 cv7 cv8 cv9 cv10 cv11 cv12 cv13 cv14 cv15 cv16 cv17 d1 d2 d3 d4 d5 d6 d7 d8 d9 d10 d11 d12 d13 s1 s2 s3 s4 s5 s6

9.4 13.0 22.2 23.4 23.8 26.0 26.2 27.4 28.2 28.3 27.6 26.8 26.3 26.3 20.3 18.0 14.2 12.8 10.1 11.6 12.2 12.8 12.7 — — — — — — — — — — — —

3.3 3.6 4.0 4.6 5.5 6.6 7.2 7.4 8.9 9.3 10.0 11.3 11.3 12.3 12.7 14.8 13.1 12.0 11.0 9.8 9.2 9.2 — — — — — — — — — — — — —

— — — — — — — — — — — — — — — — — — — — 10.3 12.0 11.0 11.2 10.3 8.0 9.3 10.3 9.6 11.1 10.0 — — — —

— — — — — — — — — — — — — — — — — — — — 13.3 12.0 11.8 11.2 13.2 11.3 11.6 13.8 19.9 13.8 14.0 — — — —

Posterior height a b

Height of neural spine a b

The whole vertebra Width across outer edges of postzygapophyses a b

Width across diapophyses a b

3.7 4.8 4.1 6.5 7.5 8.2 9.3 9.6 11.0 11.5 13.9 12.7 13.9 14.2 12.9 14.2 14.2 13.2 13.2 13.3 13.3 13.8 — — — — — — — — — — — — —

13.2 10.8 15.0 15.8 16.4 20.2 22.1 23.3 26.0 27.4 29.2 31.0 33.2 33.7 29.7 27.3 27.9 31.1 32.1 35.1 39.3 44.0 44.2 — — — — — — — — — — — —

6.7 7.3 8.0 8.8 9.0 9.4 10.0 10.7 11.5 12.6 12.8 13.4 14.0 16.5 16.6 17.0 17.5 16.3 15.8 12.0 12.1 11.3 10.0 9.4 — — — — — — — — — — —

5.2 8.3 8.5 9.1 10.0 11.6 12.2 13.7 15.3 15.8 16.3 19.4 21.0 23.0 25.5 31.1 32.3 37.2 38 37.4 32.6 29.5 24.2 21.6 — — — — — — — — — — —

— — — — — — — — — — — — — — — — — — — — 14.5 12.0 11.0 11.9 12.3 14.1 13.8 14.4 15.1 11.8 — — — — —

— — — — — — — — — — — — — — — — — — — — 29.8 31.6 32.3 35.8 36.7 37.1 38.0 — — 40.5 43.4 42.1 40.9 39.2 38.4

— — — — — — — — — — — — — — — — — — — — 16.2 12.7 13.0 12.4 11.3 11.0 — 10.2 9.8 10.1 — — — — —

— — — — — — — — — — — — — — — — — — — — 46.6 41.5 37.9 34.0 30.7 26.6 25.4 — 25.4 24.6 31.5 29.8 29.5 29.5 —

Lower case “a” and “b” refer to exemplars a (PMU 233) and b (PMU 234); “cv”, “d” and “s” refer to cervical, dorsal and sacral vertebrae, respectively.

these coels, which are divided by a horizontal septum, “1h” and “2h” to distinguish them from similarly placed coels in the dorsal series that are divided vertically (see below). The relative sizes and depths of “1h” and “2h” coels vary along the cervical series in accordance with changes to the height and length of the spine. In general, the lower coel tends to be smaller but deeper than the upper one. Beginning with the posterior cervical vertebrae (cervical 12), the EPRL terminates before reaching the prezygapophysis. As a result, the coels are only divided posteriorly and merge into each other anteriorly. However, this appears to be a highly variable feature – for example, a more complete division between the upper and lower coels is still present on the left side of cervical 13. The epipophyses continue into the dorsal series (see below), where they are represented by low rounded projections that appear to be homologous to the triangular “aliform” processes present in Haplocanthosaurus and many macronarians (Upchurch 1998; Upchurch et al. 2004a).

The greatest morphological variation along the cervical series occurs in the structure of the neural spine. The neural spine remains relatively short throughout the cervical series, reaching its greatest height in middle cervical vertebrae and then decreasing slightly towards the cervicodorsal junction (see Table 2). In the most anterior cervical vertebrae, the neural spine is built from robust plate-like SPRLs and SPOLs that converge at a stout summit. Here, the SPRLs extend posteriorly, medially and dorsally, eventually merging into the base of a single plate-like anterior spine margin. The upper part of this anterior plate is developed into a laterally compressed triangular process that projects forwards and overhangs the prezygapophyses in cervical 3 (Fig. 8). The equivalent area in cervical 4 is damaged and the anteriorly projecting triangular process is absent from cervical 5 onwards. The SPOLs of anterior cervical vertebrae converge on the midline at the spine summit. This morphology creates a prominent postspinal cavity between the SPOLs, which opens posteriorly and dorsally above the postzygapophyses.

216


Figure 12 Euhelopus zdanskyi exemplar a (PMU 233). Cervical vertebrae 16 and 17 in right lateral view. See Fig. 7 for abbreviations. Scale bar = 10 cm.

Figure 13 Schematic diagram showing the arrangement of neural arch laminae and pneumatocoels in cervical (right) and dorsal (left) vertebrae.

There is no evidence that the medial surfaces of the SPOLs were excavated by the three coels observed in the axis. In middle cervical vertebrae, the SPRLs are separate ridges that extend posterodorsally onto the anterior face of the spine and merge with each other at the anterior end of the spine summit. These laminae therefore create a triangular prespinal fossa

that is much shallower than the postspinal fossa and floored by the TPRL. The lateral profile of the spines also varies along the cervical series. In anterior cervical vertebrae, the anterior margin of the spine is relatively steep, whereas the posterior margin, formed by the SPOL, is close to horizontal. This impression is enhanced by the presence of the epipophyses, which extend the SPOL backwards in lateral view. The SPOLs become steeper in the middle cervical vertebrae (e.g. cervical 7) and the posterior cervical neural spines have a more symmetrical triangular lateral profile (Figs 10–12). The summit of the neural spine in anterior cervical vertebrae (cervicals 3– 6) is a laterally compressed and anteroposteriorly elongated ridge. Passing along the cervical series posteriorly, this summit region gradually shortens anteroposteriorly and widens transversely. By cervical 8, the neural spine summit overhangs the hollow areas on the lateral surfaces of the spine. The neural spines of cervicals 1–10 are unbifurcated and a faint midline notch first appears in cervical 11. From cervical 12 onwards, the neural spine is shallowly bifurcated (i.e. its depth does not exceed 50 mm). In all bifurcate neural spines, the SPRLs are parallel to one another, extending to the summit of each metapophysis. The base of the notch between the metapophyses develops a slight projection in cervical 15 and, from cervical 16 onwards into the anterior dorsal vertebrae (see below), a finger-like central projection is positioned between the metapophyses. This central process is as tall as the metapophyses themselves in cervical 17 and anterior dorsal vertebrae, so that the spine becomes effectively “trifid” (Fig. 15). This condition is also present in Hudiesaurus, from the Late Jurassic of China (Dong 1997).


217

Figure 14 Camellate pneumaticity in dorsal vertebrae of Euhelopus zdanskyi exemplar a (PMU 233) and exemplar b (PMU 234). Scale bars = 2 cm.

Cervical ribs (Figs 9–12; Supplementary Data Fig. 3)

Figure 15 Euhelopus zdanskyi exemplar a (PMU 233). ‘Trifid’ neural spine in anterior dorsal vertebrae composed of paired metapophyses (mp) and a median tubercle (mt). These correspond to the ‘processus pseudospinosus’ and ‘neuropophysis’, respectively, of Wiman (1929). See Supplementary Data Fig. 1 for comparison. Scale bar = 2 cm.

Below the notch, the wall of bone between the prespinal and postspinal fossae becomes anteroposteriorly thinner in the most posterior cervical vertebrae. Thus, in cervical 16 this wall is relatively thin (a few mm) in its middle and ventral regions. In cervical 17, the anterior face of this wall (i.e. the posterior surface of the prespinal fossa) develops a rugose midline ridge suggesting the attachment of a sheet-like tendon or aponeurosis.

Ribs are not preserved on the axis or cervical 3, but are preserved on cervical 4 and most of the subsequent cervical vertebrae. Their absence on these anterior vertebrae suggests that they were not fused; whereas they are fused to all subsequent cervical vertebrae. The distal ends of the rib shafts are broken in cervical 4, but the majority of other ribs are well preserved and nearly complete. Broken sections in other cervical ribs (e.g. cervical 9) reveal a camellate internal structure. The cervical ribs of Euhelopus resemble those of other sauropods in that they are double-headed, fused to the vertebrae, have short anterior processes and much longer distal shafts. The angle between the tuberculum and capitulum in anterior view is acute, so that the parapophysis–capitulum region is directed laterally and strongly ventrally. As a result, the rib shafts lie well below the ventral surface of each centrum, which represents a derived condition found in neosauropods and closely related forms such as Omeisaurus (Wilson & Sereno 1998). The best preserved middle and posterior cervical tubercula display an autapomorphic zigzag profile in lateral view (see Figs 9 & 11). This morphology is created by posteriorlydirected costal spurs at the point where the diapophysis meets the tuberculum and more ventrally on the anterior margin of the tuberculum (above the anterior process of the rib). In cervical 4, the anterior process of the rib is long and pointed, and it flattens dorsoventrally towards its tip. In

218

Figure 16


Euhelopus zdanskyi exemplar a (PMU 233). Dorsal vertebrae 2–5 in right lateral view. See Fig. 7 for abbreviations. Scale bar = 10 cm.

more posterior cervical vertebrae (e.g. cervical 13), the anterior rib process is prominent, with a subtriangular transverse cross-section and a long, triangular outline in lateral view. The dorsal edge of the triangular transverse cross-section extends posteriorly to the medial margin of the anterior rib process and then merges with the anteroventral edge of the parapophysis. In the rib of cervical 15, the base of the anterior rib process becomes troughlike dorsally because of a ridge on the lateral margin from the tuberculum and similar ridge on the medial side from the capitulum. Towards its tip the process becomes flattened transversely to form a thin plate. Similar ridges also create a trough-like excavation in the base of the distal shaft. The distal shafts are long and very slender and project well beyond the posterior end of the centrum to which the rib attaches, as occurs in most sauropods except diplodocoids (Wilson 2002; Upchurch et al. 2004a; Sereno et al. 2007). Although these distal shafts overlap each other, the arrangement in Euhelopus more closely resembles that of Camarasaurus than that of Mamenchisaurus hochuanensis, which has extremely long distal shafts that form overlapping bundles of three to five ribs (Young & Zhao 1972). Whereas ribs of anterior cervical vertebrae underlap only the vertebra immediately succeeding them, those of posterior cervical vertebrae underlap the succeeding two vertebrae. In anterior cervical vertebrae (e.g. cervicals 4–6), the distal end of the rib shaft develops a very thin medial sheet of bone. Thus, although the shaft narrows dorsoventrally in lateral view, it expands transversely towards its tip. In middle and posterior cervical ribs the distal shafts taper to very narrow, delicate tips. Cervical 16 has a rib resembling those of the preceding vertebrae, whereas in cervical 17 the rib is broader and directed posteroventrally. This change in rib morphology indicates the transition from cervical to dorsal vertebrae (see below).

Dorsal vertebrae (Figs 13–20; Supplementary Data Figs 3, 4) Dorsal vertebrae are defined as having a connection to the sternum via ribs, which enclose the thoracic cavity (Stannius 1846). In sauropods, dorsal ribs are free, double-headed and project ventrally. They differ from cervical ribs, which are fused to the cervical centra and directed parallel to the vertebral axis. As discussed above and in the following description, the cervicodorsal transition is not sharp in Euhelopus, but extends across at least two vertebrae. We identified more “dorsal characteristics” than “cervical characteristics” in presacral 18 and refer to it as a dorsal rather than a cervical vertebra, but there remains ambiguity in the actual position of the cervicodorsal transition. This ambiguity may seem surprising, given that Euhelopus is represented by a complete series of presacral vertebrae, but the absence of transitional cervicodorsal ribs prevents us from using the Stannius (1846) criterion to identify the cervicodorsal transition point. Furthermore, the distinction between cervical and dorsal vertebrae can be difficult to make even in living forms for which soft tissues are known (e.g. Giraffa; Solounias 1999). All dorsal centra are opisthocoelous and all bear large pneumatic fossae (i.e. pleurocoels) that are undivided. The dorsal series displays regional heterogeneity and can be subdivided into “anterior” dorsal vertebrae and “posterior” dorsal vertebrae. Differences in the position and morphology of the centrum, costal articulations, neural spine and external pneumatic structures identify the first four dorsal vertebrae as anterior dorsal vertebrae and the succeeding nine vertebrae as posterior dorsal vertebrae. Anterior and posterior dorsal vertebrae are known from exemplar a; exemplar b consists solely of posterior dorsal vertebrae.

ANTERIOR DORSAL VERTEBRAE (Figs 13, 15, 16; Supplementary Data Figs 3, 4). The first four dorsal


Figure 17 Euhelopus zdanskyi exemplar a (PMU 233). Dorsal vertebra 6 in right lateral view. Dorsal vertebra 7 has been shadowed and a sketch map of vertebral laminae has been overlain upon it. Note the characteristic “K” laminae formed by diapophyseal and parapophyseal laminae on the lateral aspect of the neural arch (see the text and Fig. 18 for explanation). See Fig. 7 for abbreviations. Scale bar = 10 cm.

vertebrae retain the sharply-defined dorsal margin of the pleurocoel present in cervical centra, although the pleurocoels are not divided by an oblique strut. The centra are notably stouter than those of the succeeding posterior dorsal vertebrae. The anterior dorsal parapophyses are located on the centrum rather than the neural arch, and the diapophyses are positioned adjacent to the prezygapophyses (i.e. positioned near the intervertebral foramen). The prezygapophyses are relatively large and the infradiapophyseal lamination is relatively simple. The neural spine is short and does not project beyond the plane of the zygapophyses. It bears a rudimentary bifurcation and the paired metapophyses are flattened dorsally. Dorsal 1 bears characteristics of both the cervical and dorsal regions. Its centrum is as broad as the posterior cervical centra that precede it, but notably shorter anteroposteriorly (Table 2). Like cervical parapophyses, those of dorsal 1 hang below the ventral aspect of the centrum. The parapophyses are not well preserved but they appear to be complete, suggesting that ribs were not fused to them, which is typical of the dorsal series. However, we do not know whether dorsal rib 1 was orientated parallel to the vertebral column (as in the cervical series), perpendicular to it with a contact to the sternum (as in the dorsal series), or was intermediate in morphology. As in the last cervical centrum, the pleurocoel of dorsal 1 is not divided. The centrum bears a ventral hollow with a narrow median strut. The diapophyses, which pro-

219

Figure 18 Euhelopus zdanskyi exemplar a (PMU 233). Close up of neural spine of dorsal vertebra 7 in right lateral view to show vertical division of pneumatic coels (1v, 2v) by the spinodiapophyseal lamina (spdl). See Fig. 7 for other abbreviations. Scale bar = 5 cm.

ject toward the anterior margin of the centrum, are pendant and hang down near the level of the pleurocoel. They bear a flattened lateral surface that is also present on the anterior dorsal vertebrae that follow it. The configuration of diapophyseal laminae changes dramatically over the course of the anterior dorsal vertebrae. In dorsal 1, as in the cervical vertebrae, the diapophysis hangs below the level of the zygapophyses. Consequently, laminae joining the diapophysis and the zygapophyses (PRDL, PODL) have a dorsal component to their orientation, and the PCDL is nearly horizontally orientated. As in the posterior cervical vertebrae, the remnant of the EPRL partially divides the lateral aspect of the metapophysis horizontally into two pneumatic fossae, the upper labelled “1h” and the lower labelled “2h” (see Figs 13, 17 & 18). Pneumatic fossa “3” is present and positioned between the centropostzygapophyseal, posterior centrodiapophyseal and postzygodiapophyseal laminae (see Fig. 13). Pneumatic fossa “1h” is the smallest, due to the relatively short neural spine that rises only slightly above the level of the postzygapophyses. The neural spine is incipiently bifurcated and bears a small median tubercle between its metapophyses. The remaining anterior dorsal vertebrae will be described together. Dorsal centra 2–4 are approximately the same length and only slightly shorter than dorsal 1. They are all notably broader than those of the posterior dorsal vertebrae that follow (see Table 2). Progressing posteriorly from dorsal 1 to dorsal 4, the parapophysis changes in shape and position. In dorsal 1 it is circular and positioned near the ventral centrum, whereas in dorsal 4 it is elliptical and straddles the neurocentral junction. In intervening dorsal centra the parapophysis is of intermediate morphology and position; in dorsal 2 it is circular and positioned at the anteroventral

220

Figure 19


Euhelopus zdanskyi exemplar b (PMU 234). Dorsal vertebrae 5–11 in right lateral view. See Fig. 7 for abbreviations. Scale bar = 10 cm.

corner of the pleurocoel (Fig. 16), and in dorsal 3 it is elliptical and located on the anterodorsal corner of the pleurocoel. In all four anterior dorsal vertebrae, the diapophysis extends forward to the level of the preceding vertebra and projects laterally further than do the cervical or posterior dorsal diapophyses. The neural arch lamination also changes substantially between dorsals 1 and 3. In dorsals 1 and 2, as in the posterior cervical vertebrae, the EPRL is an isolated, horizontally-orientated strut that divides the lateral aspect of the metapophysis into upper and lower pneumatic fossae (Fig. 13; Supplementary Data Fig. 3). This strut disappears in dorsal 3 and is replaced by the spinodiapophyseal lamina (SPDL). There is no vertebrae in which both lamina are present. This vertically orientated strut divides the lateral aspect of the metapophysis into two pneumatic fossae. Thus, a pattern of three pneumatic fossae is retained in this and the remaining dorsal vertebrae, but they are orientated differently and bounded by different laminae (Fig. 13). Whereas in the cervical and anteriormost dorsal neural arches, a horizontal lamina (EPRL) divides the metapophyseal space into upper and lower pneumatic fossae, in all succeeding vertebrae a vertical lamina (SPDL) divides the metapophyseal space into fore and aft pneumatic fossae. The position and bounding laminae of pneumatic fossa 3 does not vary along the presacral vertebral column. The neural spine remains relatively low and incipiently divided in dorsals 2–4 (see Fig. 20). A roughened, flattened region is present on the dorsal surface of the metapophyses, between which emerges an elongate median tubercle. In anterior or posterior view, the neural spine has a “trifid” appearance (see Fig. 19). These anterior dorsal vertebrae may serve as the site of attachment of nuchal ligaments that extend down the neck (Tsuihiji 2004).

POSTERIOR DORSAL VERTEBRAE (Figs 13, 14, 17–20; Supplementary Data Figs 3, 4). The remaining dorsal vertebrae can be readily distinguished from anterior dorsal vertebrae on the basis of many features. The centrum is notably narrower (though not shorter) than those of the preceding verteb-

Figure 20 Schematic diagram showing the arrangement of parapophyseal and diapophyseal laminae of dorsal vertebrae to form “K” laminae (bold lines). See Figs 15 and 17 for comparison. See Fig. 7 for abbreviations.

rae (see Table 2) and the pleurocoel is not sharply bounded dorsally. Unlike the condition in the anterior dorsal vertebrae, the diapophysis does not extend forward level with the intervertebral space, but rather is positioned at mid-centrum. The parapophysis, however, is positioned level with the intervertebral space. A cross-pattern of “K” laminae can be


recognised on the lateral aspect of the neural arch, a feature that may be diagnostic of Euhelopus (Fig. 17). The neural spine is elongate and more deeply bifurcate; the metapophyses are more broadly separated. Exemplar a includes both anterior and posterior dorsal vertebrae, whereas exemplar b and exemplar c consist solely of posterior dorsal vertebrae. The posteriormost preserved portion of exemplar a includes a series of posterior dorsal vertebrae (dorsals 5–8) that is continued in exemplar c (dorsals 8–11). The first preserved vertebra of exemplar b corresponds to presacral 22 (dorsal 4), as suggested by Wiman (1929) in his original description, and the uninterrupted series continues through the sacrum. The centra of dorsals 5 and 6 are more elongate and transversely narrow than the preceding vertebrae, but it is not known whether this trend continues posteriorly, because dorsal 7 is damaged ventrally, and transverse measurements are not available for dorsals 8–11 on exemplar c. In dorsals 5–10, the diapophysis is positioned above mid-centrum or over its posterior half, and the PCDL is orientated subvertically. Along the transition marked by these same dorsal vertebrae, the parapophysis becomes elevated and anteriorly shifted to a position adjacent to the prezygapophysis and it extends into the intervertebral space. The difference in relative anteroposterior positions of the costal articulations must reflect a difference in the orientation and/or shape of the ribs, but only one dorsal rib is preserved (see Fig. 2). Despite its elevation on the neural arch, the parapophysis in posterior dorsal vertebrae is always positioned lower than the diapophysis, and the paradiapophyseal lamina (PPDL) – which first appears in dorsal 5 – is orientated anteroventrally. The PPDL parallels the PRDL and, together with the diapophysis and prezygoparapophyseal lamina (PRPL), they define the boundaries of an elongate coel that is visible in lateral view (see Figs 17 & 19). Both the diapophysis and the parapophysis of posterior dorsal neural arches are supported by laminae that extend anteriorly and posteriorly towards the centrum. These laminae intersect to form a characteristic cross pattern of “K” laminae in dorsal 5 and all more posterior dorsal vertebrae in both exemplar a and exemplar b (see Figs 17 & 19). The identities of the laminae that comprise the “K” laminae is not obvious, but we provide an interpretation in Fig. 20. The vertical portion of the “K” is formed by the PCDL, which can be unambiguously identified on the basis of its connections to the diapophysis and the posterior portion of the centrum. The upper arm of the “K” is most parsimoniously interpreted as the PCPL, on the basis of its clear connection to the parapophysis and its contact to the PCDL near the posterior portion of the centrum. The lower, short arm of the “K” contacts the PCPL posterodorsally and extends anteriorly towards the centrum. Although it does not have a direct connection to the diapophysis, we interpret this to be the ACDL. If this interpretation is correct, it suggests that the ACDL drops out of the dorsal series when the parapophysis is in an intermediate position on the neural arch (replaced by the PPDL and ACPL; Wilson 1999), but then reappears once the parapophysis approaches the level of the prezygapophysis in more posterior dorsal vertebrae. An alternative is to consider the lower, short arm of the “K” to be a novel lamina (e.g. accessory posterior centrodiapophyseal lamina), as Salgado et al. (2005) did in their description of Neuquensaurus. Whilst we admit some uncertainty in our interpretation, we consider it more conservative to identify existing laminae before defining new ones.

221

Figure 21 Euhelopus zdanskyi exemplar a (PMU 233). Dorsal rib in posterior view showing pneumatic coel (co). Scale bar = 5 cm.

The height of the dorsal neural spines increases posteriorly through the posterior dorsal series. The anteroposterior expanse of the neural spine increases abruptly between anterior and posterior dorsal vertebrae, but does not continue increasing posteriorly. The orientation of the SPDL changes between anterior and posterior dorsal vertebrae. Whereas in dorsal 4 the SPDL is forwardly orientated and is conjoined with the PODL near the diapophysis, the SPDL in more posterior dorsal neural arches attaches to the posterior face of the neural spine and is not conjoined with the PODL. This more vertically orientated SPDL divides coels that are subequal in area (coels 1v and 2v). More posterior dorsal neural arches, beginning with dorsal 6, develop triangular lateral processes that hang ventrally. These probably served as attachment sites for epaxial musculature of the neck, as they do in birds (Wedel & Sanders 2002: Table 2), but their functional significance has not been examined.

Dorsal ribs (Fig. 21; Supplementary Data Fig. 3) Only a single dorsal rib, which Wiman (1929:17) identified as the left rib of dorsal 3, was preserved with exemplar a. Wiman’s illustration of that rib (1929: pl. 3, fig. 19) did not draw attention to the large coel that opens on the tuburcular portion of the proximal rib head (Fig. 21). The internal structure of the rib is not known, but pneumaticity extends through approximately two-thirds of the rib length in other titanosauriform sauropods, such as Brachiosaurus (Wilson 2002). Right ribs were preserved in articulation with the last two dorsal vertebrae in exemplar b. These two ribs contact each other and the anterior aspect of the first sacral rib, which is attached to the dorsal surface of the preacetabular process of the ilium (Supplementary Data Fig. 4).

Sacral vertebrae (Figs 22, 23; Supplementary Data Fig. 4) The sacrum has not been completely prepared and remains mounted in the PMU exhibit hall as it was during Wiman’s time (see Fig. 5). Matrix remains between the sacrum and ilium, and only the lateral aspect of the neural spines and

222


Figure 23 Euhelopus zdanskyi exemplar b (PMU 234). Sacral vertebra 6 in left ventrolateral view. Abbreviations: ct, cotyle; il, ilium; is, ischium; pc, pleurocoel; sr, sacral rib. Scale bar = 5 cm. Figure 22 Photograph of PMU technician Nils Hjort preparing the sacrum of Euhelopus in the mid 1920s. This is one of the few known photographs that shows the dorsal aspect of the sacrum, which can no longer be viewed due to the enclosure in which it is stored in (see Fig. 5). Photograph kindly provided by Museum of Evolution, Uppsala University.

diapophyses and the posterior aspect of the last sacra vertebral are visible. Accordingly, our description of sacral morphology is based on those visible parts, figures from Wiman (1929: pl. 4, figs 1–5) and early photographs (Fig. 22). Although the ribs of the 29th and 30th presacral vertebrae indirectly contact the ilium via the rib of the 31st vertebra, we do not consider them to be sacral vertebrae because they do not directly contact the ilium. Only those vertebrae whose ribs contact the sacrum and are thus directly involved in connecting the hindlimb to the axial column are here considered sacral vertebrae. Following this definition (e.g. Romer 1956), Euhelopus has six sacral vertebrae, the first of which is connected to the ilium via an elongate rib that contacts the preacetabular process. A similar condition is present in a specimen referred to Camarasaurus (BYU 17465; Tidwell et al. 2005), in which a sixth sacral vertebra has been added anteriorly. In this latter specimen, the presumptive first sacral rib appears to contact the ilium just anterior to the succeeding rib on the right side; the condition on the left side is more difficult to discern. Most Camarasaurus specimens have five sacral vertebrae and the condition in BYU 17465 was regarded by Tidwell et al. (2005) as agerelated. Euhelopus exemplar b appears to be at an earlier stage of ontogeny, as evidenced by traces of neurocentral sutures visible on all dorsal vertebrae, so the additional sacral vertebra cannot be attributed to old age. The first and second sacral vertebrae of Euhelopus retain the camellate pneumaticity present in the dorsal series; the last sacral appears to lack camellate pneumaticity as the caudal vertebrae presumably would. Nevertheless, the sixth sacral centrum bears a pleurocoel-like depression on its lateral surface and was probably pneumatised, if only partially (Fig. 23). The transition between the dorsal-like pneumaticity of the anterior sacrum and the reduced pneumaticity in the posterior sacrum cannot be assessed in its current state

of preparation. It would be interesting to know whether the intervening vertebrae, which represent primordial sacral or caudosacral vertebrae, resembled the sixth sacral more than the anterior, dorsosacral vertebrae. The presence and extent of the sacricostal yoke could not be assessed due to the matrix filling the pelvic basin. However, it is likely that Euhelopus had a sacricostal yoke that formed part of the articular surface of the acetabulum.

Scapula, coracoid and humerus (Supplementary Data Fig. 5) Young (1935) described a left scapula, coracoid and humerus from what may be the same quarry as exemplar a. The location, size, lack of duplication of these elements and Young’s conviction all suggest that exemplar a and c represent one individual. Unfortunately, the whereabouts of exemplar c is unknown, and we were not able to examine it directly. Our description of these elements herein will be accordingly brief and will focus on salient features gleaned from Young (1935). The scapula and coracoid were articulated, and the humerus was found only 1 metre away. The scapula is an elongate element (1.2 m long) that bears a deep acromion and a narrow blade. The axis of the scapular blade passes through the coracoid articulation but not the glenoid, which appears to face medially (Young 1935: fig. 2). The neck of the blade is quite narrow and it expands dorsally and ventrally. A pronounced muscle scar appears at the ventral margin of the base of the blade. The position of the scar near the glenoid corresponds to the origin of the triceps longus lateralis muscle in extant crocodylians (Meers 2003). The blade is quite thin and its cross-section is nearly flat. This differs from the D-shaped expansion of the scapular blade found in Camarasaurus. There is no asymmetrical dorsal expansion nor thickening of the base of the blade such as that seen in Camarasaurus. The coracoid is about half the dorsoventral height of the base of the scapula. The coracoid portion of the glenoid is smaller than that of the scapula, and it faces laterally (Young 1935: fig. 4).


223

The humerus, like the scapula, is slender and elongate (length = 91 cm). It is strongly expanded proximally but less expanded distally; their maximum transverse diameters are 36.3 cm and 22.5 cm, respectively. At midshaft, the humerus narrows to only 17.5 cm, which is approximately half the breadth of the proximal end. If exemplars a and c pertain to the same animal, as suggested here and elsewhere (Young 1935; Mateer & McIntosh 1985), then a high humerus–femur ratio is implied. However, owing to the incompleteness of proximal and distal ends, the length of the exemplar a femur reported by Wiman (1929; Table 2) is a minimum value. Consequently, the estimate of 0.99 provided by Mateer & McIntosh (1985:129) may be may be too high.

Pelvis (Figs 22, 24, Supplementary data Fig. 4) Both left and right halves of the pelvis are preserved in articulation with the vertebral column in exemplar b (PMU 234; Figs 3, 5 & Supplementary Data Fig. 4). Matrix was not removed from the space enclosed by the pelvis and, consequently, only external features of the pelvis are exposed. Unfortunately, due to the constraints of the enclosure in which it is mounted (see Fig. 5), the dorsal aspect of exemplar b cannot be adequately viewed. As visible in posterior, lateral and dorsal views (Fig. 22, Supplementary Data Fig. 4), the left side of the pelvis has been sheared relative to the right ventrally by 5–12◦ and anteriorly by 7–12◦ . Despite this shearing, the relative orientation of the ilia with respect to the vertebral axis appears to be the same on both sides of the pelvis – the chord passing through the base of the pubic and ischial peduncles forms a 42◦ angle with the vertebral axis. The right and left ilia are well preserved, but both are missing the anterior tip of the preacetabular process, and the left ilium is missing the posterior portion of the blade. The iliac blade is semicircular to crescent-shaped in lateral view. The preacetabular process of the blade extends laterally at a 45◦ angle to the vertebral axis, as in other non-titanosaurian neosauropods (see Wiman 1929: pl. 4, fig. 3). The broken surface of the right ilium exposes large cells that were probably air-filled pneumatic spaces (Fig. 24). Closer examination reveals similar evidence of pneumaticity on the left ilium. Unfortunately, it was not possible to determine the extent of pneumaticity within the ilium or the source of the pneumatisation (i.e. a pneumatic foramen on medial aspect of ilium) because of the inaccessibility of the pelvic basin. Wiman (1929:23) noted this feature, stating “With exception of the acetabular margin, the whole ilium is very cavernous, provided with somewhat larger cavities than the vertebrae. The two other pelvic elements, in contrast, are completely solid.” Camellate pneumaticity extending into the ilium is also known in the titanosaurs Epachthosaurus (Mart´ınez et al. 2004), Lirainosaurus (Sanz et al. 1999) and Sonidosaurus (Xu et al. 2006) and was reported in a sauropod from the Dry Mesa Quarry of the western US (Britt 1993:188). The pubic peduncle projects orthogonally to the vertebral axis and is positioned on the posterior half of the iliac blade; its length is subequal to the maximum height of the iliac blade. The pubic peduncle is more than twice as broad transversely as it is anteroposteriorly. The ischial peduncle is relatively small and does not project much beyond the iliac blade. The right pubis is nearly complete, lacking only part of the iliac peduncle. The left pubis is much more incomplete,

Figure 24 Euhelopus zdanskyi exemplar b (PMU 234). Detail of right iliac blade indicating appendicular pneumaticity in the form of large camellae. Scale bar = 5 cm.

but the proximal portion is well preserved. Like the pubic peduncle of the ilium, the iliac peduncle of the pubis is much broader transversely than anteroposteriorly. As in other sauropods, the pubes meet along an elongate symphysis that is orientated subvertically. The pubic contribution to the acetabulum is smaller than the ischial contribution, in part owing to the enlarged contribution of the pubic peduncle of the ilium. The right ischium is more completely preserved than the left, which lacks a blade. The ischium is shorter than the pubis, as it is in most titanosaurs (Wilson 2002), and the distal blades meet along a narrow edge to form a midline symphysis. The blades are orientated subhorizontally, as in other macronarians and rebbachisaurids (Upchurch 1998; Wilson & Sereno 1998; Wilson 2002; Upchurch et al. 2004a). A small boss is located on the proximolateral surface of the ischium, as in Brachiosaurus (Janensch 1950; MB J3).

Hindlimb (Fig. 25; Supplementary Data Fig. 4) The right hindlimb of exemplar b is nearly complete, lacking only a few pedal phalanges. It is mounted in articulation, which limits observations that can be made on the mutual articular surfaces of the tibia and fibula. The astragalus has not been separated from the tibia, so their articular surfaces likewise cannot be viewed. Femora are present in both exemplar a and exemplar b. The latter is more completely preserved than the former, which lacks a head and tibial condyle. Although no significant differences were observed between the two elements, no autapomorphies were identified that specifically link them.

FEMUR. The femur is fairly straight in anterior view but bears a sharp deflection in its proximal third, as in other

224


Table 3 Measurements (cm) of the tarsal elements of Euhelopus zdanskyi (PMU 234). Element

Dimension

cm

Astragalus

Transverse width Anteroposterior width across lateral surface Height of lateral surface (including ascending process) Maximum diameter Maximum proximodistal thickness

15.2 10.3 9.4

Calcaneum

6.1 4.1

tionship between proximal tibia and fibula is present in other titanosauriforms such as Erketu (IGM 100/1803), Gobititan (You et al. 2003: fig. 3), Tangvayosaurus (J.A.W., pers. obs., 2008), possibly Magyarosaurus (BMNH R 3853), and the titanosaur from Chota Simla, India described by Swinton (1947).

Figure 25 Distal hindlimb of Euhelopus zdanskyi exemplar b (PMU 234) in anterior (left) and oblique proximal (right) views. Abbreviations: ac, anterior crest; cc, cnemial crest; fc, fibular condyle; fe, femur; fi, fibula; tc, tibial crest; ti, tibia. Scale bar = 10 cm.

titanosauriforms (Supplementary Data Fig. 4; Salgado et al. 1997). As in other sauropods, the femoral shaft in Euhelopus is anteroposteriorly compressed. In cross-section its mediolateral axis is 160% its anteroposterior axis, which is not as eccentric as in some titanosaurs (>185%: Wilson & Carrano 1999). The distal condyles of the femur are slightly beveled mediodistally (ca. 4◦ ; Fig. 25), as in Gobititan (You et al. 2003: fig. 2) and some diplodocids (e.g. Apatosaurus; Upchurch et al. 2004b: pl. 9).

CRUS. The tibia and fibula are approximately 68% the length of the femur. The tibia and fibula interlock proximally in a fashion that has not yet been described in other sauropods but appears to be common in titanosauriforms. The cnemial crest of the tibia is orientated laterally, as in other sauropods, and overlaps the anterior aspect of the fibula. The fibula, in turn, has an anterior crest that extends medially into a notch behind the cnemial crest and is sandwiched between it and the body of the tibia (Fig. 25). This mutually overlapping rela-

TARSUS. The astragalus is interlocked with the distal tibia and fibula, and it can only be viewed anteriorly, posteriorly and medially. This element is as broad transversely as is the distal tibia, which is the symplesiomorphic condition for sauropods. The astragalus does not reach the medial margin of the tibia in other Cretaceous Asian titanosauriforms, such as Gobititan (You et al. 2003: fig. 2), Erketu (Ksepka & Norell 2006: fig. 10) and Opisthocoelicaudia (Borsuk-Bialynicka 1977: pl. 14, fig. 2b). In proximal and distal views, the astragalus is subtriangular and tapers to an acute but rounded medial corner. The astragalus also tapers proximodistally towards this corner, so that this part of the element is relatively slender (Table 3). It is not possible to determine the number of ridges and hollows on the posterior surface of the astragalus. The rugose distal surface is convex transversely and strongly rounded anteroposteriorly. As in all sauropods more derived than Vulcanodon (Wilson & Sereno 1998), the anterior surface of the astragalus lacks a fossa containing foramina at the base of the ascending process. In fact, the anterior face of this process and the anterior face of the rest of the bone are continuous with no obvious demarcation between them. The lateral end of the astragalus is shallowly concave and faces laterally and slightly posteriorly. There is no distal lip projecting laterally under the fibula. In the posterodorsal part of this surface, there is a small, deep hollow that does not appear to be the result of damage. As mounted, there is a distinct gap between the distal end of the fibula and the proximal ends of the metatarsals. Wiman (1929:25–26) described only one tarsal element, but a bone he identified as the ungual phalanx of digit III is probably the calcaneum (see Fig. 4; Supplementary data Fig. 4; see also Wilson & Sereno 1998). The calcaneum of Euhelopus is a proximodistally compressed lump of bone that is relatively smooth on its proximal and distal surfaces and rugose on its anterior, posterior, lateral and medial faces (Fig. 26). In proximal view, the calcaneum has a subcircular outline, while in anterior view it is compressed mediolaterally. In these respects, the calcaneum of Euhelopus resembles those of Gobititan (You et al. 2003; P.U., pers. obs., 2007), Erketu (Ksepka & Norell 2006), Camarasaurus and Diplodocus (Bonnan 2000), and differs from the more globular calcanea of Brachiosaurus and Lapparentosaurus, which have a more subrectangular proximal outline (McIntosh 1990). The Euhelopus calcaneum has its greatest


Figure 26 Euhelopus zdanskyi exemplar b (PMU 234). Right calcaneum in lateral (A), medial (B), and anterior (C) views, with proximal towards top. Scale bar = 5 cm.

proximodistal thickness near its lateral margin, so that the proximal articular face slopes slightly mediodistally. One slight anomaly is that the proximal articular surface of the calcaneum in Euhelopus is flat anteroposteriorly and convex transversely and therefore lacks the slight concavity found in the calcanea of most other sauropods (McIntosh 1990; Bonnan 2000). The distal articular surface is strongly convex anteroposteriorly and slightly convex transversely, as is also seen in the calcaneum of Gobititan (P.U., pers. obs., 2007).

PES. The pes includes four metatarsals and several phalanges, none of which were found in articulation and were reconstructed by Wiman (1929:21) in a manner that was “completely arbitrary in several ways” (see Supplementary Data Fig. 4). As discussed below, we suggest several reidentifications and reorientations for pedal elements based on comparisons with articulated sauropod hind feet. In Wiman’s reconstruction and in the skeletal mount, metatarsal I is orientated upside-down (NB: this orientation

225

was corrected by the authors while in Uppsala). This element has the typical short, robust morphology seen in other eusauropods (Upchurch 1998; Wilson & Sereno 1998; Upchurch et al. 2004a). The profile of the proximal end is more subtriangular than D-shaped: there is a long, curving dorsomedial margin, a short, straight ventral margin (facing slightly laterally), and an intermediate length, straight margin facing laterally and a little downwards. Thus there is a more acute ventromedial corner to the proximal end profile than in other sauropods. The proximal articular surface itself is mildly concave over most of its extent, occupying an elliptical area over the central and medial regions of the surface. Towards the ventrolateral corner and ventral margin, this surface becomes mildly convex and slopes strongly distally, so that it does not take part in the true articular surface. The proximal and middle portions of the lateral side of the shaft form the usual triangular striated shallow excavated area seen in other sauropods. This area is bounded dorsally by an acute margin where the dorsal and lateral surfaces meet. At approximately two-thirds of the length of the bone from the proximal end, as the dorsal and ventral surfaces converge towards each other, the anterior tip of the lateral triangular excavation ends in a mild projection. This projection lies on the posterior margin of a shallow, subcircular excavation on the side of the lateral distal condyle. The dorsal surface of the shaft is nearly flat transversely and mildly concave anteroposteriorly (because of the expansion of the proximal end). There is no rugosity on the dorsolateral margins of the shafts of metatarsals I–III. Thus Euhelopus lacks the derived state observed in the pedes of diplodocids (Upchurch 1998; Upchurch et al. 2004a). The stout shaft of the bone is thickest dorsoventrally on its medial side, so that the dorsal surface slopes downwards as it extends laterally. Medially, the side of the shaft is strongly convex dorsoventrally and merges smoothly into both the dorsal and ventral surfaces. The ventral surface is generally mildly convex transversely and concave anteroposteriorly, again because of the expansion of the proximal and distal ends. The distal end is expanded both transversely and dorsoventrally (particularly the medial condyle which is noticeably deeper than the lateral one because of a mild dorsal projection). In Euhelopus, the distolateral condyle of metatarsal I lacks the ventral process seen in diplodocids, Brachiosaurus and Omeisaurus (McIntosh et al. 1992; X. He et al. 1988). The distal end surface is convex dorsoventrally and straight or slightly concave transversely. In dorsal view, the lateral distal condyle projects further distally, so that the bone is longer on its lateral margin than its medial one. There are no distinct selvages (i.e. shelf-like ridges of bone) around the distal condyles, especially the medial one which completely lacks any excavation on its medial surface: instead this area is mildly convex both dorsoventrally and anteroposteriorly. The proximal end of metatarsal II has a subrectangular outline, with the long axis running downwards and a little medially. The dorsal part of this surface is slightly expanded medially and laterally so that these areas overhang the medial and lateral surfaces. The ventral margin is mildly rounded transversely. The proximal articular surface is flat but is slightly convex in places because of irregular projections, particularly on the lateral margin about one-third of the way down from the dorsal margin. Dorsally, the shaft’s surface is flat transversely on the proximal part and becomes very slightly convex distally. There is no distinct transverse

226


Table 4

Measurements (in cm) of the pedal elements of Euhelopus zdanskyi (PMU 234).

Dimension

Mt I

Mt II

Mt III

Mt IV

Ph II.1

Ph III.1

Ph IV.1

Ph I.2

Ph II.3

Length of lateral side Length of medial side Height of proximal end Maximum width of proximal end Minimum transverse width of shaft (viewed dorsally) Transverse width of distal end Maximum height of distal end (medial condyle)

9.8 8.8 7.8 6.7 6.1

12.2 10.8 7.7 5.2 4.3

13.6 12.4 6.6 4.2 2.9

12.1 12.0 6.1 4.5 2.8

4.3 4.5 4.0 6.4 —

4.0 3.9 3.8 5.4 —

2.8 3.3 3.0 5.6 —

12.4 — 7.2 4.7 —

10.9 — 8.7 3.8 —

8.0 4.7

6.4 4.2

5.4 4.1

5.4 3.2

3.8 6.4

3.3 4.7

2.8 5.5

— —

— —

ridge where the proximal and dorsal surfaces meet, nor do the proximal medial and lateral corners form distinct points or projections. The lateral surface has the usual slightly concave triangular area near the proximal end, but this is only weakly developed. The medial surface near the proximal end is also slightly concave, but this is an impression exaggerated by a bulge-like area on the lower part of the medial surface. In transverse cross-section, the shaft has a subcircular outline, being slightly compressed dorsoventrally and forming a sharper angle where the dorsal and lateral margins meet (all other surfaces merge smoothly into each other). The distal end expands both dorsoventrally and transversely. As in metatarsal I the medial condyle is thicker dorsoventrally, whereas the lateral condyle projects more outwardly and distally. Consequently, the element is longer on its lateral side so that the distal end surface lies at an angle to the long axis of the shaft. The distal end surface is strongly convex dorsoventrally. Metatarsal III is placed in the mount as metatarsal IV, but we have reidentified it on the basis of its relative length and robustness: in the eusauropod pes, digit I is the thickest and the metatarsals become narrower in transverse crosssection towards the lateral extreme of the pes (Wilson & Sereno 1998). In many respects its morphology is intermediate between metatarsals II and IV. The long axis of the proximal articular end slopes medially and downwards, but not as strongly as in metatarsal IV. There is a mildly convex area on the dorsal part of the proximal articular surface. The proximal part of the medial surface is generally concave over most of its extent and therefore lacks the bulge-like area of bone near the ventral margin observed in metatarsals II and IV. In transverse cross-section, the shaft is compressed from dorsomedially to ventrolaterally, with an oval to elliptical outline. The shaft is slender and measurements indicate that the minimum shaft widths for metatarsals III and IV are less than 65% of those in metatarsals I and II, which is a derived state characterising eusauropods (Wilson & Sereno 1998; see Table 4). Distally, the articular end surface is squarer than in metatarsal IV with the medial condyle being deeper than the lateral one. The medial condyle also has a flat external surface rather than the more rounded one seen in metatarsal IV. A distinct ridge marks the boundary between distal and ventral surfaces. Metatarsal IV is positioned in the mount as “metatarsal III”, with the proximal and distal ends reversed. We have reidentified this element as metatarsal IV because it is shorter and has a less robust shaft than “metatarsal III” (see above and Table 4). The main difference between metatarsals III and IV is that the long axis of the latter’s proximal end

surface is rotated more strongly relative to that of the distal end. As a result, the long axis of the proximal end surface slopes strongly medially and downwards. Both the proximal and distal surfaces slant relative to the long-axis of the shaft, so that the lateral margins lie more distally than the medial margins. The proximal part of the medial surface bears a convex area near the ventral margin, creating a shallow concavity between it and the dorsomedial margin. In dorsal view, the shaft is narrowest transversely at a point slightly closer to the proximal end than midlength and then widens gradually towards the distal end. The distal articular surface has an elliptical outline, with rounded lateral and medial margins rather than straight edges. There is a more distinct break of slope between the distal and ventral surfaces, but the dorsal and distal surfaces merge smoothly into each other. Three proximal phalanges are preserved, which have been mounted with the metatarsals II–IV. The proximal phalanx of metatarsal I is generally wedge-shaped in sauropods, and none of the preserved phalanges bear this form, indicating phalanx I.1 was not preserved (Mateer & McIntosh 1985). Consequently, the three proximal phalanges are probably correctly assigned to the inner three metatarsals. Phalanx II.1 is a subrectangular element in dorsal view, with the long axis orientated transversely. The proximal end and shaft have an oval outline in transverse cross-section, with a thick, rounded medial side and sharper tapered margin where the dorsal and ventral surfaces meet to form the lateral edge. The proximal articular surface is mildly convex in its medial part and shallowly concave in its lateral part, giving it a slightly sigmoidal profile in dorsal view. This proximal bulge, combined with the expansion of the medial distal condyle, make the bone slightly longer on its medial side than on its lateral one. The ventral surface is very mildly concave both transversely and anteroposteriorly. Distally, the articular surface is strongly convex dorsoventrally and mildly sigmoid transversely because of the expansion of the medial distal condyle. Phalanx III.1 is similar to phalanx II.1, but slightly smaller (see Table 4). The proximal articular surface is very mildly concave in all directions. It is subrectangular in dorsal view with the long-axis running transversely. In transverse cross-section, the shaft is oval and tapers towards a sharper lateral margin. The ventral surface is mildly concave in all directions. Compared to phalanx II.1, the distal end is flatter dorsoventrally and is more distinctly separated from the ventral and dorsal surfaces. This distal articular surface is also more concave transversely. Much of this difference is caused by the distal convexity of the lateral condyle. In distal end view, the articular surface is semicircular in outline, rather than oval, because of the height of the lateral


condyle. Phalanx IV.1 is a small, subrectangular bone that is relatively wide compared to its anteroposterior length. The proximal surface is generally irregular and mildly concave. The outlines of the proximal end and shaft transverse crosssection are not oval because there is a small lateral surface that faces laterally and ventrally, separating the true ventral and dorsal surfaces. Dorsal and lateral surfaces meet at an acute margin, whereas the dorsal and medial surfaces merge smoothly into each other. The ventral surface is mildly concave. Distally, the articular surface is mildly and irregularly convex. Three elements were described by Wiman (1929) as ungual phalanges, and they have been mounted in this position (Supplementary Data Fig. 4). There was no first phalanx preserved with digit 1, and its ungual has therefore been placed in contact with metatarsal I in the mount. The element Wiman (1929) identified as the digit 1 ungual has an unusual appearance, and we cannot be certain that it has been properly identified (see below). The element appears to have been crushed mediolaterally, so that it has an unusually acute ridge extending along its dorsal margin. It is very tall relative to its transverse thickness and there is no sign that the proximal articular surface, which is shallowly convex, slopes at an angle to the long axis of the bone, as it does in neosauropods. Therefore, were this element an ungual, it would have been directed forwards, rather than laterally, a feature that is only retained in basal sauropods. In lateral view, the element has a strongly curved, hook-like profile compared to those of other sauropods. The lateral surface is relatively flat and this surface and the medial surface show no signs of longitudinal nail attachment grooves. The margin where the lateral and ventral surfaces meet forms a sharp ridge. Apart from the general side view profile, the most claw-like aspect of this element is seen on the medial surface. Here, the ventral surface faces downwards and medially, and meets the medial surface at a distinct ridge or break of slope, as in other sauropod unguals. This ventromedial surface is wide at the proximal end but narrows distally to the point where the claw terminates in a very blunt tip. We consider it uncertain or even unlikely that this element is an ungual, but we can forward no better identification for it at this time. The element identified as the ungual of digit two bears the conventional morphology of sauropod unguals. It is larger than the element Wiman (1929) identified as the digit 1 ungual. It is probable that the larger element actually belongs to digit 1, not digit 2. This element is strongly compressed transversely and the proximal articular surface has an oval outline with a rounded, bulging ventral part that narrows dorsally. There is a distinct nail groove that is asymmetrical, positioned higher on the medial side than on the lateral side. The proximal articular surface is subtly divided into upper and lower regions by a transverse ridge. The upper triangular region faces backwards, whereas the lower more rounded region faces backwards and downwards. In dorsal view, the proximal articular surface lies at an angle of approximately 70◦ to the long axis of the claw, so that it faces proximally and a little laterally. This suggests that the claw pointed forwards and laterally when in articulation. The medial surface is smooth and mildly convex both dorsoventrally and anteroposteriorly. On the lateral surface of the ungual, there is a nail attachment groove on the middle and distal regions at approximately midheight. The middle and distal parts of the ventral margin of the medial surface form a distinct, thin

227

ridge that is slightly separated from the rest of the ventral surface. This feature is reminiscent of the more strongly developed ridge in the pedal unguals I–III of Gobititan (P.U., pers. obs., 2007). The element described and mounted as the third ungual is actually the calcaneum, as described above.

Discussion Phylogenetic affinities of Euhelopus Revised scorings for the Wilson (2002) and Upchurch et al. (2004a) matrices are presented in Tables 5 and 6. For the purposes of comparing matrices before and after restudy of Euhelopus, we will track the type of changes to our respective matrices, distinguishing changes between the ambiguous state (“?”) and any unambiguous state (e.g. “0”, “1”) from changes between any two unambiguous states. Sereno (pers. comm., 2008) has called differences in the latter “character state conflict” and differences in the former “character state resolution”, and has used them in a metric to quantify the differences between two matrices. The “Character State Similarity Index” (CSSI; P. Sereno, pers. comm., 2008) ranges between 0 (complete dissimilarity) and 1 (identity) and tallies the total number of character state conflicts (csc) and character state resolutions (csr) relative to the total number of character states (tcs) such that CSSI = (tcs – [csc + 0.5csr])/tcs. In the special case in which a matrix is revised, as is the case here, it is useful to discriminate between two types of character state resolutions. That is, a scoring change from an unambiguous state (e.g. “0” or “1”) to an ambiguous state (i.e. “?”) is “adding ambiguity” whereas a scoring change from the ambiguous state to any unambiguous state is “adding resolution”. “Adding ambiguity” increases missing data, while “adding resolution” reduces missing data. Twenty-two changes were made to the original Wilson (2002) matrix (NB: scoring adjustments made by Wilson [2005c] were not incorporated for the purposes of this reanalysis). Of the 22 changes made, five were substantive changes between unambiguous states and have been noted in the descriptive part of the text. Of the remaining 17 changes, four were “adding resolution” and 13 were “adding ambiguity”. Many of the latter represent a more conservative scoring, rather than a loss of information from the skeleton of Euhelopus. For instance, character 222 coded the size of the proximal condyle of metatarsals I and V proximal condyle relative to the inner metatarsals. Wilson (2002) scored this character as derived on the basis of the size of metatarsal I relative to II–IV, but metatarsal V is not preserved. Despite the likelihood that Euhelopus possessed the derived condition, this character was rescored to be unknown (“?”). Comparison of the Wilson (2002) matrix before and after restudy of Euhelopus indicates a CSSI of 0.9423. Reanalysis of the revised matrix yielded the same topology as Wilson 2002, in which Euhelopus positioned as the sister group of Titanosauria. However, due to changes in characters 81 and 104, support for the node linking this clade dropped from a decay index of 5 to 3. Revision of the scorings of the Upchurch et al. 2004a matrix resulted in 81 changes. Of these changes, the majority

228


Table 5 Scoring changes for the character–taxon matrices of Wilson (2002) and Upchurch et al. (2004a), based on the redescription of Euhelopus zdanskyi presented here. Analysis

Scoring Changes

Wilson (2002)

1(1→0); 4(0→1); 25(?→0); 32(1→?); 34(0→1; 35(0→?); 39(? →1); 41(1→?); 42(1→?); 59(1→?); 66(?→1); 74(0→?); 81(1→0); 100(?→0); 104(1→0); 187(0→?); 188(1→?); 191(0→?); 192(1→?); 222(1→?); 229(1→?); 232(1→?) 5(0→?); 8(0→1); 11(0→1); 20(0→1); 23(1→?); 38(?→1); 39(?→0); 41(0→?); 62–64(0→?); 70(0→?); 78(0→1); 90(0→?); 92(0→1); 94(1→0); 107(0→1); 109(?→0); 113(?→0); 116(0→1); 126(?→0); 127(?→1); 130(?→0); 131(0→1); 134(?→0); 135(1→0); 136(?→1); 138(?→1); 140(?→1); 143(?→1); 144–146(?→9); 148(1→2); 150(1→0); 151(?→0); 153(?→1); 154(?→0); 157(?→1); 158(1→0); 159(?→1); 168(1→?); 201(?→1); 206(0→1); 217(?→0); 218, 219(?→1); 220–222(?→0); 242, 243(1→?); 244(0→?); 245(0→1); 249(?→1); 251(1→?); 253, 254(?→1); 256(?→1); 261–264(?→1); 266, 267(?→1); 268(?→0); 269–273(?→1); 274(1→0); 277(1→0); 279(?→1); 281(?→0); 288(?→0); 289(?→1); 306, 307(1→?)

Upchurch et al. (2004a)

See Table 6 for complete scorings. Wilson (2002) original versus rescored Character State Similarity Index (CSSI) = 0.9423; Upchurch et al. (2004a) original versus rescored CSSI = 0.8414.

(49) added resolution, reducing total missing data. Fifteen scorings added ambiguity, which increased missing data but represent more conservative scoring. Seventeen changes involved the replacement of one known state by another. The CSSI for Upchurch et al. (2004a) original versus rescored is 0.8414. The differences in CSSI reflect the sources of information on Euhelopus morphology available to Wilson (2002) and Upchurch et al. (2004a). Whereas Wilson (2002) used a combination of the published literature and direct personal observation of the Euhelopus material, Upchurch et al. (2004a) relied solely on the published descriptions. Although monographic descriptions can provide a very useful and generally accurate source of character data, direct observation of the relevant specimens always provides a fuller understanding of anatomy and can offer unique insights into phylogenetic relationships. The revised Upchurch et al. (2004a) data matrix was analysed using PAUP 4.0 (Swofford 2002) with the character coding assumptions specified in the original analysis. This produced over 200,000 most parsimonious trees (MPTs) (treelength = 654 steps) before the limit on computer memory meant that the analysis had to be halted. The data matrix was then analysed using the parsimony ratchet programme PAUPMacRat (Sikes & Lewis 2001) in order to determine whether the original heuristic search had become

trapped in a “local” rather than “global” island of maximum parsimony. The shortest PAUPMacRat trees were also 654 steps long, indicating that the original heuristic search had probably discovered a global maximum parsimony island. As expected, the substantial reduction in missing data (–11%), allowed more opportunity for character conflict and, as a consequence, treelength increased 7 steps and tree number increased from 1,056 to more than 200,000. In order to clarify the phylogenetic relationships of Euhelopus, it was necessary to reduce the number of MPTs. Safe taxonomic reduction (Wilkinson 1995) was attempted, but all ingroup taxa have unique combinations of character states and no “safe” a priori deletions of taxa were identified. However, analyses of the original Upchurch et al. (2004a) data matrix indicated that three taxa (i.e. Lapparentosaurus, Nigersaurus and “Pleurocoelus-tex”) can be deleted without affecting the relationships of the remaining ingroup taxa. There is no guarantee that this is also the case for the data matrix that contains the revised Euhelopus scorings. Nevertheless, a priori deletion of these three genera is justified here as a pragmatic step because the focus of the current paper is clarification of the relationships of Euhelopus, not a global reconstruction of sauropod relationships. Reanalysis of the revised data matrix after a priori pruning of Lapparentosaurus, Nigersaurus and “Pleurocoelus-tex” yielded

Table 6 Revised scorings of Euhelopus zdanskyi for the matrices of Wilson (2002) and Upchurch et al. (2004a), based on the redescription presented here Analysis

Scoring

Wilson 2002

01 1 1 0? 1 1 ? ? ? ? ? ? 1 1 1 01 1 ?????????? ? ? ? 1 1 1 ? ? 01

1 ? ? ? ? 1 0? ? ? 1 21 ? 1 01 1 ? 4 ????????10 ? ? 01 1 1 1 1 1 1

? ? ? ? 0? ? ? ? ? 01 1 1 01 1 01 1 11??????11 001 1 01 ? 1 01

? ? 1 1 ? ? ? 01 1 21 01 01 1 1 1 0 001 0000? 1 1 1 1 ? 00? 1 1 1 0

?????????? 1 1 001 ? 03? ? 01 01 ? ? ? ? ? ? 1?11?1?1?1

? ? ? ? 1 001 ? ? 01 ? ? ? ? ? ? ? ? ?????????? 1??0

Upchurch et al. 2004a

1 1 00? 001 1 1 1 ? ? ? ? 1 1 01 ? 1 000001 1 00 ?????????? ????1?1110 ?1?1???11

1 00001 1 01 1 0? ? ? ? 1 1 1 1 1 1 1 1 001 1 1 01 ?????????1 ? 1 1 1 01 01 01

1 0? ? 1 001 1 ? 1 1 1 1 1 1 000? ? 1 1 0001 21 0 1?1??111?? 1 1 1 1 1 1 1 01 1

? ? 1 ? 1 01 1 0? 1 1 1 001 1 1 1 1 001 001 1 01 1 ? ? ? ? ? ? 01 1 0 1 1 1 01 1 01 1 1

? ? ? 0? ? ? ? ? ? ? 1 1 1 1 1 1 1 00 1 1 1 1 ? 00? ? ? 00? ? ? ? ? ? ? ? 01 ? 1 1 1 ? 01 1

? ? ? ? ? ? 1 1 0? 1 1 001 1 021 0 ?????????? ?????????? 1 1 001 ? 1 ? ? ?

See Table 5 for scoring changes.


229

Figure 27 A strict consensus tree of the 97 topologies produced by the a posteriori deletion of Lourinhasaurus and Tehuelchesaurus from 9,894 most parsimonious trees (MPTs) resulting from analysis of the revised Upchurch et al. (2004a) data matrix (see the text for details). Key Chinese taxa are in bold, and some terminal taxa have been combined into suprageneric taxa. Clade names assigned to major sauropod groups are given. The asterisk next to Diplodocoidea indicates that Nemegtosaurus and Quaesitosaurus were included within this clade by Upchurch et al. (2004a).

9,894 MPTs (treelength = 634). The parsimony ratchet was applied once again and the results suggest that the heuristic search found the shortest trees globally. A strict consensus tree of the 9,894 MPTs indicates that Euhelopus and Phuwiangosaurus are basal somphospondylan sauropods that lie just outside of Titanosauria (see Wilson & Upchurch 2003 for clade name definitions). These relationships are shown in Fig. 27, which is a reduced consensus cladogram (Wilkinson 1994) based on the 291 topologies remaining after the a posteriori deletion of Lourinhasaurus and Tehuelchesaurus from the 9,894 MPTs. The support for the position of Euhelopus was examined using bootstrap analysis, decay analysis and constrained topologies. Bootstrap analyses were conducted using 10,000 replicates and the heuristic search in PAUP 4.0. The bootstrap support for the node uniting Euhelopus with Titanosauria is 55%. The decay index of this node is 1. Despite this low value, support at other nodes (i.e. Titanosauriformes, Macronaria) robustly nest Euhelopus within Neosauropoda. A constraint was written in order to force Euhelopus to lie outside Neosauropoda – the position favoured by Upchurch et al. (2004a; Fig. 4). An heuristic search of the data matrix (with the Euhelopus rescorings and the a priori exclusion of Lapparentosaurus, Nigersaurus and “Pleurocoelus-tex”) produced 34 MPTs of length 645, which is 11 steps longer than the most parsimonious solution. A Templeton test compared these constrained MPTs with those produced without a constraint, yielding p-values of 0.056–0.14, just outside the 95% confidence window. (NB: the strict reduced consensus tree [Fig. 27] positions Jobaria outside Neosauropoda, a res-

ult that is in accord with Wilson [2002] but differs from Upchurch et al. [2004a], which identified Jobaria as a basal macronarian. Simply by revising the scorings for Euhelopus, the Upchurch et al. data produces topologies that are more similar to those found by Wilson [2002], not only in terms of the position of Euhelopus but also for Jobaria as well, even though the scorings for the latter remain unchanged.) The results of these tests indicate that the placement of Euhelopus within basal Somphospondyli is only moderately supported. In the case of the bootstrap and decay analyses, it is not clear whether: (1) the position of Euhelopus is unstable; or (2) its position is actually well supported but the bootstrap and decay values are low because other taxa (e.g. Phuwiangosaurus, Austrosaurus) change their relationships relative to Euhelopus with little cost in terms of tree length. The constrained analyses indicate that 11 additional steps are required to force Euhelopus into a position outside Neosauropoda. Although there is a sizable parsimony penalty associated with this topological rearrangement, the Templeton test demonstrates that it cannot be statistically excluded (with 95% confidence) as an explanation of the data.

Euhelopus and the biogeographical history of East Asian sauropods Euhelopus has played an important role in two recent debates concerning the biogeographical history of East Asian sauropods: (1) the existence of a monophyletic Euhelopodidae has been linked to the geographical isolation of East Asia from the Middle Jurassic to the Early Cretaceous (Upchurch 1995,

230


1998); and (2) the placement of Euhelopus at the base of the Titanosauria potentially reinforces recent suggestions that this group originated in Laurasia in general (Weishampel & Jianu 1997) or East Asia in particular (You et al. 2003). A quantitative biogeographical analysis of sauropods, based on their phylogenetic relationships, lies outside the scope of the current study. Nevertheless, resolution of the phylogenetic relationships of Euhelopus presents an opportunity to revisit these debates and address some of the problems that may hinder future progress.

East Asian isolation The distinctive nature of Middle Jurassic to Early Cretaceous Chinese terrestrial faunas (including dinosaurs, pterosaurs and mammals) has led several authors to suggest that endemism occurred as a result of geographical isolation (Milner & Norman 1984; Dong 1992; Russell 1993, 1995; Russell & Zheng 1993; Upchurch 1995; Luo 1999; Barrett et al. 2002). There are at least two geographical barriers that could have isolated East Asia from the rest of Pangaea during the Middle Jurassic: (1) the Turgai (= “Obik” and “Uralian”) epicontinental sea between Europe and central Asia (Milner & Norman 1984; Russell 1993, 1995; Smith et al. 1994; Upchurch 1995; Le Leouff 1997; Upchurch et al. 2002; Scotese 2004; Blakey 2006); and (2) the Mongol-Okhotsk sea between Siberia–Kazakhstan and the Mongolian–Tarim– Junggar blocks (Enkin et al. 1992; Upchurch 1995; Barrett et al. 2002), which closed in the Late Jurassic or Early Cretaceous (Cogné et al. 2005). Most authors believe that the end of the isolation of East Asia occurred in the Early Cretaceous as a result of the establishment of dispersal routes with Europe and/or North America (Weishampel & Bjork 1989; Le Loeuff 1991, 1997; Russell 1993, 1995; Currie 1995; Manabe & Hasegawa 1995; Upchurch 1995; Buffetaut et al. 1997; Norman 1998; Buffetaut & Suteethorn 1999; Barrett et al. 2002; Canudo et al. 2002; Holtz et al. 2004). The preferred date of the end of isolation varies from Late Jurassic to Early Tertiary and depends partly on which of the potential palaeogeographical barriers is favoured and partly on which taxonomic groups are considered (Milner & Norman 1984; Weishampel & Bjork 1989; Russell 1993; Manabe & Hasegawa 1995; Upchurch 1995; Le Loeuff 1997; Barrett et al. 2002; Upchurch et al. 2002). For example, Russell (1993) and Upchurch et al. (2002) argued that isolation ended during the Aptian–Albian, when marine regression caused the Turgai Sea to retreat and allowed iguanodontian dinosaurs and paramacellotid lizards to invade East Asia from Europe. It is also possible that a land connection across the Bering Strait formed at approximately the same time, enabling dispersals between East Asia and western North America (Weishampel & Bjork 1989). Barrett et al. (2002), however, noted the presence of titanosauriform sauropods in East Asia in rocks as old as the Valanginian and, therefore, proposed that isolation ended earlier than previously suspected. Although Barrett et al. (2002) did not specify the palaeogeographical mechanism responsible for this earlier end to isolation, their scenario is more compatible with the closure of the Mongol–Okhotsk sea than it is with the Aptian–Albian marine regression. Sauropods have played an important role in the East Asian isolation hypothesis (EAIH). Russell (1993) proposed that “mamenchisaurs” were restricted to East Asia, whereas

diplodocids were absent from this area but endemic to the rest of Pangaea. This view received support from the cladograms presented by Upchurch (1995, 1998), in which the Middle and Late Jurassic Chinese taxa Shunosaurus, Omeisaurus, Mamenchisaurus and Euhelopus (now thought to be Early Cretaceous in age) formed a monophyletic group of basal eusauropods termed the Euhelopodidae, whereas neosauropod lineages were apparently restricted to the rest of Pangaea during this portion of the Mesozoic. Wilson & Sereno (1998) cast doubt on the monophyly of the Euhelopodidae and suggested that neosauropod lineages had become widespread across Pangaea (including East Asia) prior to the Middle Jurassic and that regional faunal differences were largely caused by differential survival. Furthermore, Rich et al. (1999) noted the morphological similarity of Tehuelchesaurus (from South America) to Omeisaurus, and therefore argued that faunal interchange between East Asia and the rest of Pangaea was possible during the Middle Jurassic (NB: Tehuelchesaurus is now known to have been collected from the Cañadón Calcáreo Formation, which is Late Jurassic in age, much younger than the Middle Jurassic Cañadón Asfalto Formation: Rauhut 2003, 2006). Barrett et al. (2002) reviewed much of the previous literature and identified two competing biogeographical hypotheses to explain the presence of titanosauriform sauropods in the earliest Cretaceous of East Asia: either (1) these taxa invaded East Asia at this time as a result of the end of isolation; or (2) neosauropod lineages were present in East Asia since the Middle Jurassic and their “sudden” appearance in the Early Cretaceous is an artefact of a patchy sampling of the fossil record. Barrett et al. (2002) preferred the first explanation (i.e. EAIH) partly because of the absence of neosauropods in the Jurassic of East Asia and, partly, because the cladogram presented by Upchurch (1998), in which euhelopodids are monophyletic, was the best-supported and most detailed sauropod phylogeny available to them at that time. Thus, the isolation of East Asia from the Middle Jurassic onwards could explain the evolution of the endemic euhelopodid clade, and the end of isolation in the Early Cretaceous might also account for the disappearance of this group and their replacement by titanosauriforms and perhaps other neosauropods (Upchurch 1995, 1998). Since Barrett et al.’s (2002) study, however, new phylogenetic analyses have shifted the balance of evidence so that hypothesis (2), above, may be better supported. Firstly, it is possible that at least one neosauropod lineage was present in East Asia during the Middle Jurassic: Bellusaurus from western China was tentatively identified as a titanosaur by Jacobs et al. (1993), and the cladistic analysis of Upchurch et al. (2004a) placed it as a basal macronarian. However, rejection of the EAIH (i.e. Barrett et al. hypothesis “1”), based on Bellusaurus alone, should be treated with caution: reanalysis of the Upchurch et al. 2004a data matrix presented here results in Bellusaurus being placed outside the Neosauropoda. Secondly, most recent sauropod phylogenies (including the reanalyses here) agree that the Euhelopodidae (sensu Upchurch 1995, 1998) represents a polyphyletic assemblage (Wilson & Sereno 1998; Wilson 2002; Upchurch et al. 2004a; Curry Rogers 2005; Harris 2006), although whether Euhelopus itself should be placed inside or outside of the Neosauropoda has remained controversial until the current study. The breakdown of euhelopodid monophyly has been regarded as undermining the EAIH (Wilson & Sereno 1998;


231

Barrett et al. 2002). It could be argued that Barrett et al.’s (2002) hypothesis (2) should be preferred and the concept of East Asian isolation (at least as an explanation for sauropod biogeographical distributions) should be abandoned. Such a conclusion, however, would be premature because it is possible to reconcile the new interpretations of the relationships of East Asian sauropods with the EAIH. In particular, Wilson (2005a) has noted that all pre-Cretaceous Asian sauropods are non-neosauropods, whereas all Cretaceous Asian sauropods are members of Somphospondyli. Thus, the paraphyletic or polyphyletic assemblage of “euhelopodid” sauropods could represent a relictual fauna that existed in East Asia during the Jurassic, being replaced by more advanced neosauropod lineages that dispersed to this area during the Early Cretaceous. This scenario enables continued support for the EAIH even though the monophyly of the Euhelopodidae (sensu Upchurch 1995) has broken down. Finally, the expectation that East Asian isolation should have resulted in the evolution of a multi-species endemic clade is problematic. Vicariance can be manifested at a variety of taxonomic scales, depending on the amount of time elapsed and the rates of evolutionary change in different lineages. Indeed, the evidence for vicariance can actually be strengthened by re-arrangements of the phylogenetic relationships of taxa that disrupt the monophyly of endemic clades. In order to explain this point more fully, a brief aside concerning area cladograms and vicariance is required (see below).

Vicariance and area cladograms The emergence of a geographical barrier can divide a single species into two isolated populations that eventually diverge to become two daughter species via allopatric speciation. One or both of these daughter species can then radiate within that endemic area to produce an endemic clade. Euhelopodidae has been proposed as one such endemic clade created by the geographical isolation of East Asia. However, phylogenetic biogeographers (e.g. Brooks & McLennan 2002) have argued that the existence of an endemic clade, by itself, provides only relatively weak evidence for vicariance because only the basal node represents a vicariant event (i.e. the separation of the endemic lineage from its sister taxon outside the region of endemism). All other nodes within the endemic clade have resulted from either allopatric speciation within the area of endemism or sympatric speciation. Fortunately, the emplacement of a geographical barrier has the potential to isolate many different species and create multiple endemic clades. Although differences in dispersal capability and evolutionary rate may affect response to the imposed barrier (Simpson 1952), geographical isolation can be expected to affect multiple species. Thus, evidence for vicariance is strengthened when congruent spatiotemporal patterns are found in multiple, independent lineages (Nelson & Platnick 1981; Lieberman 2000; Hunn & Upchurch 2001; Brooks & McLennan 2002). Perhaps counterintuitively, stronger evidence for vicariance can arise when a supposed endemic clade (e.g. Euhelopodidae) is disbanded, and its constituent taxa are redistributed among other clades. This point is illustrated by a hypothetical example. In Figure 28A, the 16 taxa (1–16) form four monophyletic clades (W–Z), each of which is endemic to one of the four geographical areas (A–D). Such a pattern is consistent with a geographical history in which

Figure 28 Two hypothetical scenarios depicting the phylogenetic relationships and biogeographical ranges of taxa 1–16. A–D represent four separate geographical areas that were once in contact and then became isolated from each other in the sequence (A, (B, (C, D))). The separation of A from BCD is event 1 (e1); B from CD is event 2 (e2); and C from D is event 3 (e3). (A) shows the 16 taxa in four monophyletic groups, such that each clade (W–Z) is found in only one of the four areas. (B) Shows the same taxa in a different set of relationships so that, despite the disruption of clades W–Z, there is still evidence for vicariance in the form of the number of nodes consistent with one of the geographical events.

the once continuous area ABCD fragmented in the sequence (A (B (C D))). However, the confinement of each clade to a single area means that the support for vicariance is weak because there is only one cladogenetic event associated with each geographical isolation event (Fig. 28A). Now consider the phylogenetic pattern in Figure 28B, in which the relationships of the 16 taxa have been altered so that the monophyly of the clades W–Z has been disrupted. Despite the breakdown of monophyly with respect to each area, the evidence for vicariance has actually become much stronger, because each of the geographical isolation events is supported by four cladogenetic events in the phylogeny (Fig. 28B). For example, each of the four clades independently supports the hypothesis that the first vicariance event involved the separation of area A from area BCD (event “e1” in Fig. 28B). The hypothetical example outlined above demonstrates that the monophyly of Euhelopodidae can break down without weakening the evidence for vicariance caused by East Asian isolation: indeed, the evidence for vicariance could actually become stronger. Whether the breakdown of Euhelopodidae has weakened or strengthened support for the putative vicariance event cannot be judged merely from qualitative inspections of the alternative phylogenetic positions

232


The biogeographical origin of titanosaurs

Cretaceous (Wilson & Sereno 1998; Day et al. 2002, 2004; Upchurch et al. 2002; Wilson & Upchurch 2003; Canudo & Royo-Torres 2004; Upchurch & Barrett 2005). In this scenario, the origin of Somphospondyli and Titanosauria cannot be traced to vicariance events (at least at the scale of continental fragmentation), although vicariance at a lower taxonomic level (e.g. the genus) could occur. Proponents of this early and global radiation disagree about the impact of Pangaean fragmentation on dinosaur evolution and distribution during the Late Jurassic and Cretaceous. Some workers (e.g. Wilson & Sereno 1998; Sereno 1999) have suggested that subsequent differences in the composition of dinosaur faunas were the result of regional extinctions operating on a once cosmopolitan assemblage. Others, however, have identified vicariance patterns among Cretaceous titanosaurian genera that correspond to the break-up sequence for Pangaea (Upchurch et al. 2002; Canudo & Royo-Torres 2004). In our view, several factors shift the balance of evidence in favour of the biogeographical hypothesis that titanosaurs appeared relatively early (i.e. in or before the Middle Jurassic) and acquired a virtually global distribution prior to Pangaean fragmentation.

Historically, the first discoveries of titanosaur material were made in the southern hemisphere (Wilson & Upchurch 2003). Consequently, many authors suggested that titanosaurs represent a group that originated in Gondwana after it became isolated from Laurasia in the Jurassic or Early Cretaceous (von Huene 1932; Bonaparte 1984, 1999; Bonaparte & Kielan-Jaworowska 1987; Lucas & Hunt 1989; Astibia et al. 1990; Jacobs et al. 1993; Russell 1993; Le Loeuff & Buffetaut 1995; Upchurch 1995; Le Loeuff 1997). Note that this concept originated within the fixist view of continental configuration but was easily adapted to the modern plate tectonic paradigm. Despite the subsequent discovery of titanosaurs in Laurasia (e.g. “Titanosaurus” in Europe and Alamosaurus in North America), this group still seemed to be most abundant and diverse in Gondwana. Thus, titanosaurs have become an iconic example of a Gondwanan radiation and their appearance in Laurasia is usually explained via northward dispersal (e.g. Lucas & Hunt 1989; Le Loeuff & Buffetaut 1995). In recent years, however, the Gondwanan origin of titanosaurs has been challenged on the basis of several lines of evidence. An influx of new discoveries and the results of phylogenetic analyses have identified many northern hemisphere titanosaurs, including some particularly early forms. For example, Day et al. (2002, 2004) described some of the earliest evidence for titanosaurs – wide-gauge trackways from Bathonian age deposits from Ardley (Oxfordshire, England). Furthermore, taxa such as Venenosaurus from North America, “Pelorosaurus becklesii” from Europe and Phuwiangosaurus from Asia, suggest that titanosaurs (or at least basal somphospondylans) were widespread across Laurasia during the Early Cretaceous (Upchurch 1995; Tidwell et al. 2001; Wilson & Upchurch 2003; Upchurch et al. 2004a; Wilson 2005). The reaction to these advances has been varied. Some authors have proposed that titanosaurs originated in Laurasia (Weishampel & Jianu 1997) or in Asia (You et al. 2003). An alternative to these “Laurasian origin” hypotheses is the view that titanosaurs evolved during the Early Jurassic or early Middle Jurassic and dispersed across Pangaea prior to the major fragmentation events in the Late Jurassic and Early

(1) Although the occurrence of many basal titanosaurs (or basal somphospondylans) in the Early Cretaceous of East Asia is intriguing, there are dangers in extrapolating biogeographical inferences directly from this observation. As was shown above, basal phylogenetic positions of taxa in area A can be caused by vicariance in which area A is separated from the rest of the world first. This means that area A (the most basal area in the area cladogram) could either be the “centre of origin” for a clade, or it might simply be the first area to have become isolated after the clade became widespread across ABCD (Croizat et al. 1974). Which scenario we prefer depends on whether the area cladogram for the clade of interest is congruent with a more general pattern indicative of vicariance, which in turn determines the parsimony penalties associated with the competing “vicariance” and “centre of origin + dispersal” interpretations (Brooks & McLennan 2002). In the absence of a quantitative phylogenetic biogeographical analysis, the simple observation that basal members of a clade occupy a particular area should not be used to favour a centre of origin hypothesis over vicariance, or vice versa. (2) It seems probable that titanosaurs had a global distribution in the Early Cretaceous. However, the Early Cretaceous dinosaur fossil record of East Asia (i.e. China) is particularly rich compared to that in Europe and North America. It is, therefore, possible that the accumulation of East Asian forms at the base of the clade is an artefact created by unequal sampling of the fossil record or of taxon selection by phylogeneticists who have favoured more complete specimens. This possibility can be tested by future phylogenetic analyses by targeting the inclusion of Early Cretaceous titanosaurs from outside East Asia. (3) The presence of statistically robust area cladograms (Upchurch et al. 2002), divergence time estimates (Upchurch & Barrett 2005) and direct observations of trackway data (Wilson & Carrano 1999; Day et al. 2002, 2004) all suggest that somphospondylans and perhaps a more derived titanosaur clade had diverged from other

of taxa such as Shunosaurus, Omeisaurus, Mamenchisaurus and Euhelopus proposed by Wilson (2002), Upchurch et al. (2004) and the current study. This is because vicariance patterns are easily obscured by missing data and coherent dispersal events that make it difficult to identify congruence among area relationships (Lieberman 2000; Upchurch & Hunn 2002; Halas et al. 2005). Therefore, the impact of new phylogenetic topologies on the evidence for vicariance should be assessed using biogeographic methods that identify and test congruent area relationships in large and complex data sets and evaluate the probability of obtaining such patterns by chance (Page 1991; Lieberman 2000; Brooks & McLennan 2002; Upchurch & Hunn 2002; Halas et al. 2005). Such an analysis of dinosaurian, or even sauropod, biogeography lies outside the scope of the current study. Consequently, we simply note at this time that the effects of our revision of Euhelopus as a basal somphospondylan on the EAIH will remain unknown until quantitative biogeographical methods are applied in the future.


sauropod lineages by the Bathonian at the latest. Thus, the basal taxa cited by You et al. (2003) occured 20–40 million years after the true origin of the clade and may, therefore, not provide a sound guide to its “centre of origin”. In short, the patchy sampling of the titanosaur fossil record (especially during the Middle and Late Jurassic), uncertainty of ages of several key deposits and lack of resolution for basal titanosaur interrelationships mean that reconstructions of the timing and place of origin of titanosaurs should be treated with great caution. Nevertheless, current evidence clearly demonstrates that the titanosaurian lineage had diverged from other sauropods by the Bathonian; consequently, proposals that titanosaurs originated in either East Asia or Gondwana are inconsistent with our best estimates of their spatiotemporal distribution.

Resurrecting Euhelopodidae? Canudo et al. (2002) described teeth from the Barremian of Spain that possess bosses on the lingual part of each crown. Such lingual crown buttresses are also present in Euhelopus (see above) and have been recognised as an autapomorphy of the genus (Wilson & Sereno 1998: 22). More recently, Buffetaut et al. (2002) and Barrett & Wang (2007) described isolated teeth from the Early Cretaceous of Thailand and China, respectively, several of which also possess these lingual crown buttresses. Finally, Ksepka & Norell (2006) described a partial skeleton of the titanosauriform Erketu from the Early Cretaceous of Mongolia. This material does not include teeth, but the cervical vertebrae of Erketu share a number of derived states with Euhelopus such as extreme elongation of the centrum, “pre-epipophyses” and prong-like epipophyses (see above). The phylogenetic analysis of Ksepka & Norell (2006) placed Erketu within basal Somphospondyli in a trichotomy with Euhelopus and Titanosauria, but they did not take into account the characters mentioned above that could uniquely unite these two East Asian genera. We agree with Canudo et al. (2002) that there may exist a hitherto unsuspected clade of basal titanosauriforms that was widespread across Asia and possibly Europe during the Early Cretaceous (Table 7). In the future, it may be useful to resurrect the term Euhelopodidae as the name for the clade of titanosauriforms closely related to Euhelopus, but we strongly recommend foregoing formalization of this or any other clade name in the absence of a detailed phylogenetic analysis corroborating monophyly of the group and the robustness of the results. In addition, there are a number of newly discovered Asian titanosauriforms (see Table 7) awaiting fuller description that would be critical for inclusion in any phylogenetic assessment of basal titanosauriforms.

Conclusion Continued interest in dinosaur systematics, combined with the mechanistic ease of cladistic methodology (Grant et al. 2003) and an ever-growing store of character datasets, has led to a dramatic increase in the number of hypotheses for dinosaur interrelationships since the first major analyses were published more than 20 years ago (Gauthier 1986; Sereno 1986). More than a score of cladistic analyses have focused on Sauropodomorpha and its subgroups alone (e.g. see

233

Barrett & Batten 2007). This proliferation of analyses has led to many competing hypotheses of interrelationships that conflict, sometimes substantially, with one another – even within the body of work of one author. Transparency is central to cladistic methodology and this attribute allows authors to evaluate previous researchers’ characters and character states, coding strategy, scoring, search methodology and consensus techniques. Critical evaluation of these data is part of the research cycle (Kluge 1991; Jenner 2001), but reassessment of previous work is seldom performed, which can lead to long-standing disagreements about interrelationships of important groups. One potential explanation for this lack of comparative critique may be that there is no methodology for comparisons of competing datasets, although some researchers (e.g. Sereno 2007, pers. comm., 2008) have taken steps in this direction. The phylogenetic affinities of Euhelopus and other Chinese taxa have been the focal disagreement among sauropod systematists for a decade, despite exchanges in the literature (Upchurch 1995, 1998; Wilson & Sereno 1998; Wilson 2002; Upchurch et al. 2002, 2004a). Although this type of exchange is a valuable exercise, it can be time-consuming and allows debates to continue because authors misunderstand each other or get bogged down in semantic debates. An alternative approach, in which disagreeing systematists work together, is rare (e.g. Rieppel & Reisz 1999) but has several advantages. Poorly defined characters and character states can be identified and redefinitions can be found. Character scorings can be discussed and agreed upon, or, if ambiguity remains, agreement to score a state as “?” can be made. Issues regarding the association, articulation and referral of specimens can be jointly reevaluated – enabling more uniform scorings. Finally, more accurate and wide-ranging comparisons between the focal taxa and other forms can be achieved, because workers can combine their knowledge of specimens that perhaps only one of them has examined. The revised Wilson (2002) and Upchurch et al. (2004a) data matrices both support the view that Euhelopus is the sister taxon to Titanosauria. Derived states supporting this position include: single pre- and postspinal laminae on dorsal neural spines; six co-ossified sacral vertebrae; medial deflection of scapular glenoid; medial deflection of proximal part of the femur; and camellate bone tissue structure in presacral vertebrae. Our conclusion that Euhelopus is a basal somphospondylan further disrupts the monophyly of the “Euhelopodidae” and indicates that East Asian sauropods, from the Middle Jurassic to Early Cretaceous, comprised several disparate lineages rather than a single endemic clade. Nevertheless, a period of isolation of East Asia, during the Middle and Late Jurassic, is still feasible. Although there still exists the possibility that Euhelopus or Euhelopus-like taxa were present (but currently unrecorded) in Jurassic sediments in East Asia during its period of geographical isolation, we consider it more likely that Euhelopus was an Early Cretaceous immigrant to East Asia. Although the basal position of Euhelopus with respect to other titanosaurs further undermines previous claims that this clade originated in Gondwana, it would be premature to conclude that the balance of evidence has shifted to a Laurasian or even East Asian centre of origin. The relationships of many other basal titanosaurs and titanosauriforms must be clarified before the impact of East Asian isolation and the break up of Pangaea on sauropod

234

Table 7


Cretaceous Asian sauropod species, listed in order of their stratigraphic appearance.

Taxon

Country

Age

Reference

Material

Nemegtosaurus Nemegt mongoliensis (T) Opisthocoelicaudia Nemegt skarzynskii (T) ‘Nemegtosaurus pachi’ (n.d.) Subashi

Mongolia

Maastrichtian

Nowinski 1971

Skull

Mongolia

Maastrichtian

China

Postcranial skeleton lacking neck Tooth

Quaesitosaurus orientalis (T)

Barungoyot

Mongolia

Erlian

China

Kurzanov & Bannikov 1983 Xu et al. 2006

Skull

Sonidosaurus saihangaobiensis (T) unnamed (?TSF)

?Campanian– Maastrichtian Santonian– Campanian Senonian

Borsuk-Bialynicka 1977 Dong 1977

Kasamatsu

Japan

Santonian

Teeth

‘Antarctosaurus’ jaxarticus’ (n.d.) Unnamed ‘diplodocoid’ Huabeisaurus allocotus (?T)

Dabrazinskaya Svita Dzarakuduk Huiquanpu

Kazakhstan Uzbekistan China

Turonian– Santonian Turonian Late Cretaceous

Tanimoto & Suzuki 1998 Ryabinin 1939

Qingxiusaurus youjiangensis (?T) Borealosaurus wimani (T)

“Redbeds”

China

Late Cretaceous

Sunjiawan

China

Dongyangosaurus sinensis (T)

Fangyan

China

Early Late Cretaceous Early Late Cretaceous

Gobititan shenzhouensis (TSF) unnamed Tangvayosaurus hoffeti (TSF) Jianshanosaurus lixianensis (?T)

Xinminbao Group

China

Albian

You et al. 2003

Jiufotang Grès Supérieurs Jinhua

China Laos China

Albian Aptian–Albian Aptian–Albian

Wang et al. 1998 Allain et al. 1999 Tang et al. 2001b

Ilek

Russia

Aptian–Albian

Averianov et al. 2002

‘Ultrasaurus tabriensis’ (n.d.) Gugyedong

South Korea

Aptian

Chiayusaurus lacustris (?TSF) unnamed (TSF) unnamed ‘brachiosaurid’ (?TSF) Dongbeititan dongi (TSF) Pukyongosaurus millenniumi (TSF)

Xinminbao Group Xinminbao Group Jinju

China China South Korea

Kim 1983; Lee et al. 1997 Barremian–Aptian Bohlin 1953 Barremian–Aptian Dong 1997 Barremian–Aptian Lim et al. 2001

Yixian Hasandong

China South Korea

Barremian Barremian

Wang et al. 2007 Dong et al. 2001

‘Chiayusaurus asianensis’ (n.d.) unnamed ‘titanosaurid’ (indet.) unnamed ‘camarasaurid’ (indet.) Euhelopus zdanskyi (TSF) unnamed (TSF)

Hasandong

South Korea

Barremian

Lee et al. 1997, 2001

Partial skeleton 7 partial cervical vertebrae, dorsal centrum, ribs, chevrons Tooth

Hasandong

South Korea

Barremian

Lee et al. 1997, 2001

Tooth

Hasandong

South Korea

Barremian

Lee et al. 1997, 2001

Tooth

Mengyin Kuwajima

China Japan

Wiman 1929 Barrett et al. 2002

2 partial skeletons 9 teeth

unnamed ‘brachiosaurid’ (?TSF) Toba sauropod (?TSF)

Kitadani

Japan

Azuma 2003

Teeth

Matsuo Group

Japan

‘Asiatosaurus mongoliensis’ (n.d.) Mongolosaurus haplodon (?T) Erketu ellisoni (TSF)

Oshih

Mongolia

?Neocomian ?Barriasian– ?Hauterivian Hauterivian– Barremian Valangian– Barremian Early Cretaceous

Tomida & Tsumura 2006 Osborn 1924b

2 partial caudal vertebrae, 2 partial limb bones 2 teeth

On Gong

Mongolia

Early Cretaceous

Gilmore 1933

Bor Gové

Mongolia

Early Cretaceous

Ksepka & Norell 2006

Tooth, basisphenoid, cervical vertebrae 1–3 Anterior cervical series, sternal plate, distal hindlimb

Shestakovo sauropod (TSF)

Formation

Partial skeleton

?

Sues & Averianov 2004 Teeth, postcranial bones Pang & Cheng 2000 2 teeth, nearly complete postcranium Mo et al. 2008 Anterior caudal neural spine, sternal plate, humerus You et al. 2004 Caudal vertebra L¨ u et al. 2008

10 dorsal vertebrae, sacrum, first 2 caudal vertebrae, dorsal ribs, ilia, pubis, ischium Caudal vertebrae, distal hindlimb ? Partial skeleton 5 dorsal and 3 caudal vertebrae, partial scapula, pubis, ischium and femur Teeth, mid-caudal vertebrae, pes Partial humerus Tooth Cervical vertebra Tooth


Table 7

235

Continued

Taxon

Formation

Country

Age

Reference

Material

Phuwiangosaurus sirindhornae (T) Fusuisaurus zhaoi (TSF)

Sao Khua

Thailand

Early Cretaceous

Martin et al. 1994

Multiple skeletons

Napai

China

Early Cretaceous

Mo et al. 2006

Huanghetitan luijiaxiaensis (TSF)

Hekou Group

China

Early Cretaceous

You et al. 2006

‘Asiatosaurus kwanshiensis’ (n.d.) Jiutaisaurus xidiensis (n.d.)

Napan

China

Cretaceous

Hou et al. 1975

Anterior caudal vertebrae, dorsal ribs, ilium, pubis, distal femur Sacrum, 2 caudal vertebrae, fragmentary cervical ribs, partial chevron, scapula, coracoid Tooth, partial vertebra, ribs

Quantou

China

Cretaceous

W.-H. Wu et al. 2006

18 anterior caudal vertebrae & chevrons

Youngest species are at the top and oldest species are at the bottom; Early and Late Cretaceous species are separated by solid line. Parenthetical abbreviations after species name indicate phylogenetic affinities: indet., indeterminate sauropod; n.d., nomen dubium; T, Titanosauria; TSF, Titanosauriformes.

evolutionary history can be fully assessed. We anticipate that the relationships of Euhelopus relative to other basal somphospondylans and titanosaurs will become more fully resolved in the near future, when newly discovered taxa are described in more detail and systematic effort is directed towards the interrelationships of Titanosauriformes.

Acknowledgements We thank S. Stuenes and J. Peel for access to the collections of the PMU and for allowing us to publish archival photographs (Figs 5 & 26). We thank B. Miljour for drafting Figure 1 and for her assistance with the Supplementary Figures. P. Barrett kindly provided information on microwear in isolated Euhelopus teeth, as well as other useful information. N. Insel translated the original description of Euhelopus by Wiman (1929), which is available on the Polyglot Paleontologist website. We are grateful to P. Barrett, P. Mannion and S. Upchurch for their help with data collection and photos. S. Upchurch also provided insightful observations on the “K” lamina. We thank P. Sereno for allowing us to use the Character State Similarity Index defined in his in review manuscript. M. D’Emic, T. Ikejiri, P. Mannion, L. Salgado, J. Whitlock and an anonymous referee provided careful reviews of the manuscript. J.A.W.’s research was supported by National Science Foundation grant DEB-0640434 and a Woodrow Wilson National Fellowship Foundation Career Enhancement Fellowship for Junior Faculty.

References Allain, R., Tarquet, P., Battail, B., Dejax, J., Richir, P., Véran, M., Limon-Duparcmeur, F., Vacant, R., Mateus, O., Sayarath, P., Khenthavong, B. & Phouyavong, S. 1999. Un nouveau genere de dinosaure sauropode de la formation des Grés supérieurs (Aptien– Albien) du Laos. Comptes Rendus de l’Academie des Sciences Paris, Sciences de la Terre et des Planétes 329: 609–616. Alexander, J. H. 1857. Universal dictionary of weights and measures, ancient and modern; reduced to the standards of the United States of America. W. Minifie, Baltimore, 158 pp. Andrews, R. C. 1932. The New Conquest of Central Asia. A Narrative of the Explorations of the Central Asiatic Expeditions in Mongolia and

China, 1921–1930. The American Museum of Natural History, New York, 678 pp. Astibia, H., Buffetaut, E., Buscalioni, A. D., Cappetta, H., Corral, C., Estes, R., Garcia-Garmilla, F., Jaeger, J. J., Jimenez-Fuentes, E., Le Loeuff, J., Mazin, J. M., Orue-Etxebarria, X., Pereda-Suberbiola, J., Powell, J. E., Rage, J.-C., Rodriguez-Lazaro, J., Sanz, J. L. & Tong, H. 1990. The fossil vertebrates of Lano (Basque Country, Spain); new evidence on the composition and affinities of the Late Cretaceous continental faunas of Europe. Terra Nova 2: 460–466. Averianov, A. O. & Skutschas, P. 2000. A eutherian mammal from the Early Cretaceous of Russia and biostratigraphy of the Asian Early Cretaceous vertebrate assemblages. Lethaia 33: 330–340. —, Voronkevich, A. V., Maschenko, E. N., Leshchinskiy, S. V. & Faygertz, A. V. 2002. A sauropod foot from the Early Cretaceous of Western Siberia, Russia. Acta Palaeontologica Polonica 47: 117–124. Azuma, Y. 2003. Early Cretaceous vertebrate remains from Katsuyama City, Fukui Prefecture, Japan. Memoir of the Fukui Prefectural Dinosaur Museum 2: 17–21. Barrett, P. M. & Batten, D. J. 2007. Evolution and Palaeobiology of Early Sauropodomorph Dinosaurs. Special Papers in Palaeontology 77: 1–288. — & Wang, X.-L. 2007. Basal titanosauriform (Dinosauria, Sauropoda) teeth from the Lower Cretaceous Yixian Formation of Liaoning Province, China. Palaeoworld 16: 265–271. —, Hasegawa, Y., Manabe, M., Isaji, S. & Matsouka, H. 2002. Sauropod dinosaurs from the Lower Cretaceous of eastern Asia: taxonomic and biogeographical implications. Palaeontology 45: 1197–1217. Benson, R. 2008. New information on Stokesosaurus, a tyrannosauroid (Dinosauria: Theropoda) from North America and the United Kingdom. Journal of Vertebrate Paleontology 28: 732–750. Blakey, R. C. 2006. Global paleogeographic views of earth history – late Precambrian to Recent. Available at: http://jan.ucc.nau.edu/∼rcb/ globaltext2.html. Bohlin, B. 1953. Reports from the Scientific Expedition to the NorthWestern Provinces of China Under Leadership of Dr. Sven Hedin. VI. Vertebrate Palaeontology 6: Fossil Reptiles from Mongolia and Kansu. Statens Etnografiska Museum, Stockholm, 113 pp. Bonnan, M. 2000. The presence of a calcaneum in a diplodocid sauropod. Journal of Vertebrate Paleontology 20: 317–323. Bonaparte, J. F. 1984. Late Cretaceous faunal interchange of terrestrial vertebrates between the Americas. Pp. 19–24 in Reif, W. E. & Westphal, F. (eds) Third Symposium on Mesozoic Terrestrial Ecosystems, Short Papers. Attempto Verlag, Tübingen. — 1986. The early radiation and phylogenetic relationships of sauropod dinosaurs, based on vertebral anatomy. Pp. 247–258 in Padian, K. (ed.) The Beginning of the Age of Dinosaurs. Cambridge University Press, Cambridge.

236


— 1999. Tetrapod faunas from South America and India: a palaeobiogeographic interpretation. Proceedings of the Indian National Science Academy 65 A: 427–437. — & Kielan-Jaworowska, Z. 1987. Late Cretaceous dinosaur and mammal faunas of Laurasia and Gondwana. Pp. 22–29 in Currie, P. & Koster, E. H. (eds) Fourth Symposium on Mesozoic Terrestrial Ecosystems, Short Papers. Tyrell Museum of Paleontology, Drumheller. Borsuk-Bialynicka, M. 1977. A new camarasaurid sauropod Opisthocoelicaudia skarzynskii gen. n., sp. n. from the Upper Cretaceous of Mongolia. Palaeontologica Polonica 37: 6–63. Britt, B. B. 1993. Pneumatic postcranial bones in dinosaurs and other archosaurs. Unpublished Ph.D. Dissertation: University of Calgary, Alberta. Brooks, D. R. & McLennan, D. A. 2002. The Nature of Diversity: an Evolutionary Voyage of Discovery. University of Chicago Press, Chicago, 676 pp. Buffetaut, E. & Suteethorn, V. 1999. The dinosaur fauna of the Sao Khua Formation of Thailand and the beginning of the Cretaceous radiation of dinosaurs in Asia. Palaeogeography, Palaeoclimatology, Palaeoecology 150: 13–23. Buffetaut, E., Suteethorn, V., Tong, H.-Y., Chaimanee, Y. & Khansubha, S. 1997. New dinosaur discoveries in the Jurassic and Cretaceous of northeastern Thailand. Pp. 177–187 in Dheeradilok, P. (ed.) International Conference on Stratigraphy and Tectonic Evolution of Southeast Asia and the South Pacific, Department of Mineral Resources, Bangkok. Buffetaut, E., Suteethorn, V., Le Loeuff, J., Cuny, G., Tong, H. & Khansubha, S. 2002. A review of the sauropod dinosaurs of Thailand. Pp. 95–101 in Mantajit, N. (ed.) The Symposium on Geology of Thailand. Department of Mineral Resources, Bangkok. Calvo, J. O. & Salgado, L. 1995. Rebbachisaurus tessonei sp. nov. a new Sauropoda from the Albian–Cenomanian of Argentina; new evidence on the origin of the Diplodocidae. Gaia 11: 13–33. Canudo, J. I. & Royo-Torres, R. 2004. El fémur saurópodo (Dinosauria) de Peñaroya de Tastavins (Teruel, Aptiense inferior, Cretácico inferior): Implicaciones paleobiogeográficas para los Titanosauriformes. Geo-Temas 6: 19–21. ˜ —, Ruiz-Omenaca, J. I., Barco, J. L. & Royo-Torres, R. 2002. ¿Saurópodos asiáticos en el Barremiense inferior (Cretácico Inferior) de España? Ameghiniana 39: 443–452. Chen, P., Li, W., Chen, J., Ye, C., Wang, Z., Shen, Y. & Sun, D. 1982. Stratigraphical classification of Jurassic and Cretaceous in China. Scientia Sinica (series B) 25: 1227–1248. Cogné, J.-P., Kravchinsky, V. A., Halim, N. & Hankard, F. 2005. Late Jurassic–Early Cretaceous closure of the Mongol–Okhotsk Ocean demonstrated by new Mesozoic palaeomagnetic results from the TransBa¨ıkal area (SE Siberia). Geophysical Journal International 163: 813– 832. Croizat, L., Nelson, G. & Rosen, D. E. 1974. Centers of origin and related concepts. Systematic Zoology 23: 265–287. Currie, P. J. 1995. Wandering dragons, the dinosaurs of Canada and China. Pp.101 in Sun, A. & Wang, Y. (eds) Sixth Symposium on Terrestrial Mesozoic Ecosystems and Biotas, Short Papers. China Ocean Press, Beijing. Curry Rogers, K. A. 2005. Titanosauria: a phylogenetic overview. Pp. 50–103 in Curry Rogers, K. A. & Wilson, J. A. (eds) The Sauropods: Evolution and Paleobiology. University of California Press, Berkeley and Los Angeles. Day, J. J., Upchurch, P., Norman, D. B., Gale, A. S. & Powell, H. P. 2002. Sauropod trackways, evolution and behavior. Science 296: 1659. —, Norman, D. B., Gale, A. S., Upchurch, P. & Powell, H. P. 2004. New Middle Jurassic dinosaur trackways from Oxfordshire, U.K. Palaeontology 47: 319–348. Depéret, C. 1896. Note sur les Dinosauriens Sauropodes & Théropodes du Crétacé Supérieur de Madagascar. Bulletin de la Société Géologique de France 24: 176–194. Dong, Z.-M. 1977. [On the dinosaurian remains from Turpan, Xinjiang]. Vertebrata Palasiatica 15: 59–65. [In Chinese with English abstract] — 1992. Dinosaurs of China. China Ocean Press, Beijing, 188 pp.

— 1995. The dinosaur complexes of China and their biochronology. Pp. 91–96 in Sun, A. & Wang, Y. (eds) Sixth Symposium of Mesozoic Terrestrial Ecosystems and Biota, Short Papers. China Ocean Press, Beijing. — 1997. On the sauropods from the Mazongshan area, Gansu Province, China. Pp. 19–23 in Dong, Z. (ed.) Sino–Japanese Silk Road Dinosaur Expedition. China Ocean Press, Beijing. — 1998. On Asiatic sauropods and their phylogenetic relationships. Pp. 79–88 in Galopim, A. M. C., Andrade, A. M. & Silva, C. M. (eds) I Encontro Internacional sobre Paleobiologia dos Dinossáurios. Lisbon. —, Zhou, S. & Zhang, Y. 1983. The dinosaurian remains from Sichuan Basin, China. Palaeontologia Sinica (New Series C) 23: 1–145. —, Paik, I. S. & Kim, H. J. 2001. A preliminary report on a sauropod from the Hasandong Formation (Lower Cretaceous), Korea. Pp. 41–53 in Deng, T. & Wang, Y. (eds) Proceedings of the Eighth Annual Meeting of the Chinese Society of Vertebrate Paleontology. China Ocean Press, Beijing. Enkin, R. J., Zhentu, Y., Yan, C. & Courtillot, V. 1992. Paleomagnetic constraints on the geodynamic history of the major blocks of China from the Permian to the present. Journal of Geophysical Research 97: 13953–13989. Falconer, H. 1868. Notes on fossil remains found in the Valley Indus below Attok, and at Jubbulpoor. Pp. 414–419 in Murchinson, C. (ed.) Palaeontological Memoirs and Notes of the Late Hugh Falconer. Robert Hardwicke, London. Gauthier, J. A. 1986. Saurischian monophyly and the origin of birds. Pp. 1–55 in Padian, K. (ed.) The Origin of Birds and the Evolution of Flight. Memoirs of the California Academy of Sciences 8. California Academy of Sciences, San Francisco. Gilmore, C. W. 1933. Two new dinosaurian reptiles from Mongolia with notes on some fragmentary specimens. American Museum Novitates 679: 1–20. — 1936. Osteology of Apatosaurus with special reference to specimens in the Carnegie Museum. Memoirs of the Carnegie Museum 11: 175–300. Granger, W. & Gregory, W. K. 1923. Protoceratops andrewsi, a preceratopsian dinosaur from Mongolia. American Museum Novitates 72: 1–9. Grant, T., Faivovich, J. & Pol, D. 2003. The perils of ‘point-and-click’ systematics. Cladistics 19: 276–285. Halas, D., Zampar, D. & Brooks, D. R. 2005. A historical biogeographical protocol for studying biotic diversification by taxon pulses. Journal of Biogeography 32: 249–260. Harris, J. D. 2006. The significance of Suuwassea emiliae (Dinosauria: Sauropoda) for flagellicaudatan interrelationships and evolution. Journal of Systematic Palaeontology 4: 185–198. He, H.-Y., Wang, X.-L., Zhou, Z.-H., Wang, F., Boven, A., Shi, G.-H. & Zhu, R.-X. 2004. Timing of the Jiufotang Formation (Jehol Group) in Liaoning, northeastern China, and its implications. Geophysical Research Letters 31, L12605. He, X., Li, K. & Cai, K. 1988. [The Middle Jurassic Dinosaur Fauna from Dashanpu, Zigong, Sichuan, Vol. IV. Sauropod Dinosaurs (2), Omeisaurus Tianfuensis]. Sichuan Scientific and Technological Publishing House, Chengdu, 144 pp. [In Chinese with English abstract]. Holtz Jr., T. R., Chapman, R. E. & Lamanna, M. C. 2004. Mesozoic biogeography of Dinosauria. Pp. 627–642 in Weishampel, D. B., Dodson, P. & Osmólska, H. (eds) The Dinosauria, 2nd Edition. University of California Press, Berkeley. Hou, L.-H., Yeh, H.-K. & Zhao, X.-J. 1975. [Fossil reptiles from Fusui, Kwangshi]. Vertebrata PalAsiatica 13: 24–33. [In Chinese with English abstract]. Huene, F. von 1932. Die fossil Reptil-Ordnung Saurischia, ihre Entwicklung und Geschichte. Monographen zur Geologie und Palaeontologie 4: 1–361. — 1956. Paläontologie und Phylogenie der niederen Tetrapoden. VEB Gustav Fisher Verlag, Jena. Hunn, C. A. & Upchurch, P. 2001. The importance of time/space in diagnosing the causality of phylogenetic events: towards a ‘chronobiogeographical’ paradigm? Systematic Biology 50: 391–407.


Ikejiri, T., Tidwell, V. & Trexler, D. L. 2005. New adult specimens of Camarasaurus lentus highlight ontogenetic variation within the species. Pp. 154–186 in Tidwell, V. & Carpenter, K. (eds) Thunderlizards: the Sauropodomorph Dinosaurs. Indiana University Press, Bloomington. Jacobs, L. L., Winkler, D. A., Downs, W. R. & Gomani, E. M. 1993. New material of an Early Cretaceous titanosaurid sauropod dinosaur from Malawi. Palaeontology 36: 523–534. Janensch, W. 1929. Material und Formengehalt der Sauropoden in der Ausbeute der Tendaguru Expedition. Palaeontographica (Supplement 7) 2: 1–34. — 1935–1936. Die Schädel der Sauropoden Brachiosaurus, Barosaurus, und Dicraeosaurus aus den Tendaguru-Schichten Deutsch-Ostafrikas. Palaeontographica (Supplement 7) 3: 147–298. — 1950. Die Wirbelsäule von Brachiosaurus brancai. Palaeontographica (Supplement 7) 2: 27–93. Jenner, R. A. 2001. Bilateral phylogeny and uncritical recycling of morphological data sets. Systematic Biology 50: 730–742. Kim, H. M. 1983. Cretaceous dinosaurs from Korea. Journal of the Geological Society of Korea 19: 115–122. Kluge, A. G. 1991. Boine snake phylogeny and research cycles. Miscellaneous Publications of the Museum of Zoology, University of Michigan 178: 1–58. Ksepka, D. T. & Norell, M. A. 2006. Erketu ellisoni, a long-necked sauropod from Bor Guvé (Dornogov Aimag, Mongolia). American Museum Novitates 3508: 1–16. Kurzanov, S. M. & Bannikov, A. F. 1983. A new sauropod from the Upper Cretaceous of Mongolia. Paleontological Journal 2: 90–96. Lapparent, A. F. & Lavocat, R. 1955. Dinosauriens. Pp. 785–962 in Piveteau, J. (ed.) Traité de Paléontologie. Masson et Cie , Paris. Lee, Y.-N., Yang, S.-Y. & Park, E.-J. 1997. Sauropod dinosaur remains from the Gyeongsang Supergroup, Korea. Palaeontological Society of Korea, Special Publication 2: 103–114. —, Yu, K.-M. & Wood, C. B. 2001. A review of the vertebrate faunas from the Gyeongsang Supergroup (Cretaceous) in South Korea. Palaeogeography, Palaeoclimatology, Palaeoecology 165: 357–373. Leidy, J. 1858. Hadrosaurus foulkii, a new saurian from the Cretaceous of New Jersey. Proceedings of the Academy of Natural Sciences of Philadelphia 8: 72–73. Le Loeuff, J. 1991. The Campano–Maastrichtian vertebrate faunas from southern Europe and their relationships with other faunas in the world: paleobiogeographical implications. Cretaceous Research 12: 93–114. — 1997. Biogeography. Pp. 51–56 in Currie, P. J. & Padian, K. (eds) Encyclopedia of Dinosaurs. Academic Press Inc., San Diego. — & Buffetaut, E. 1995. The evolution of Late Cretaceous non-marine vertebrate faunas in Europe. Pp. 181–184 in Sun, A. & Wang, Y. (eds) Sixth Symposium on Terrestrial Mesozoic Ecosystems and Biotas, Short Papers. China Ocean Press, Beijing. Lieberman, B. S. 2000. Paleobiogeography: using fossils to study global change, plate tectonics and evolution. Kluver Academic/Plenum Publishers, New York, 208 pp. Lim, J.-D., Martin, L. D. & Baek, K.-S. 2001. The first discovery of a brachiosaurid from the Asian continent. Naturwissenschaften 88: 82–84. ¨ J., Azuma, Y., Chen, R., Zheng, W. & Jin, X. 2008. A new tiLu, tanosauriform sauropod from the early Late Cretaceous of Dongyang Zhejiang Province. Acta Geologica Sinica 82: 225–235. Luo, Z. 1999. A refugium for relicts. Nature 400: 23–25. Lucas, S. G. & Hunt, A. P. 1989. Alamosaurus and the sauropod hiatus in the Cretaceous of the North American western interior. Geological Society of America Special Paper 238: 75–85. Lydekker, R. 1893. The dinosaurs of Patagonia. Anales del Museo de La Plata 2: 1–14. Madsen Jr., J. H., McIntosh, J. S. & Berman, D. S. 1995. Skull and atlas– axis complex of the Upper Jurassic sauropod Camarasaurus Cope (Reptilia: Saurischia). Bulletin of the Carnegie Museum of Natural History 31: 1–115. Manabe, M. & Hasegawa, Y. 1995. Diapsid fauna and its palaeobiogeographical implication; the Neocomian section of the Tetori Group,

237

Japan. Pp. 179 in Sun, A. & Wang, Y. (eds) Sixth Symposium on Mesozoic Terrestrial Ecosystems and Biota, Short Papers. China Ocean Press, Beijing. Mantell, G. A. 1825. Notice on the Iguanodon, a newly discovered fossil reptile, from the sandstone of Tilgate Forest, in Sussex. Philosophical Transactions of the Royal Society of London 115: 179–186. Marsh, O. C. 1878. Principal characters of American Jurassic dinosaurs. Pt. I. American Journal of Science 16: 411–416. — 1882. Classification of the Dinosauria. American Journal of Science 23: 81–86. — 1895. On the affinities and classification of the dinosaurian reptiles. American Journal of Science (Series 3) 50: 413–423. Martin, V., Buffetaut, E. & Suteethorn, V. 1994. A new genus of sauropod dinosaur from the Sao Khua Formation (Late Jurassic or Early Cretaceous) of northeastern Thailand. Comptes Rendus de l’Academie des Sciences Paris (série II) 319: 1085–1092. Martin-Rolland, V. 1999. Les sauropodes chinois. Revue Paléobiologie, Geneve 18: 287–315. Mart´ınez, R. D., Giménez, O., Rodr´ıguez, J., Luna, M. & Lamanna, M. C. 2004. An articulated specimen of the basal Titanosaurian (Dinosauria: Sauropoda) Epachthosaurus sciuttoi from the early Late Cretaceous Bajo Barreal formation of Chubut Province, Argentina. Journal of Vertebrate Paleontology 24: 107–120. Mateer, N. J. & Lucas, S. G. 1985. Swedish vertebrate palaeontology in China: a history of the Lagrelius collection. Bulletin of the Geological Institutions of the University of Uppsala, new series 11: 1–24. — & McIntosh, J. S. 1985. A new reconstruction of the skull of Euhelopus zdanskyi (Saurischia:Sauropoda). Bulletin of the Geological Institutions of the University of Uppsala, new series 11: 25– 132. Mayor, A. 2000. The First Fossil Hunters. Princeton University Press, Princeton, 361 pp. McIntosh, J. S. 1990. Sauropoda. Pp. 345–401 in Weishampel, D. B., Dodson, P. & Osmólska, H. (eds) The Dinosauria. University of California Press, Berkeley. —, Coombs Jr., W. P. & Russell, D. A. 1992. A new diplodocid sauropod (Dinosauria) from Wyoming, U.S.A. Journal of Vertebrate Paleontology 12: 158–167. Meers, M. B. 2003. Crocodylian forelimb musculature and its relevance to Archosauria. The Anatomical Record Part A 274A: 891–916. Milner, A. R. & Norman, D. B. 1984. The biogeography of advanced ornithopod dinosaurs (Archosauria: Ornithischia) – a cladistic-vicariance model. Pp. 145–150 in Reif, W.-E. & Westphal, F. (eds) Third Symposium on Mesozoic Terrestrial; Ecosystems, Short Papers. Attempto Verlag, Tübingen. Mo, J., Wang, W., Huang, Z., Huang, X. & Xing, X. 2006. A basal titanosauriform from the Early Cretaceous of Guangxi, China. Acta Geologica Sinica 80: 486–489. —, Huang, C.-L., Zhao, Z.-R., Wang, W. & Xing, X. 2008. [A new titanosaur (Dinosauria: Sauropoda) from the Late Cretaceous of Guangxi, China]. Vertebrata PalAsiatica 46: 147–156 [In Chinese with English summary]. Needham, J. 1956. Science and Civilisation in China. Volume 2: History of Scientific Thought. Cambridge University Press, Cambridge, 697 pp. Nelson, G. & Platnick, N. 1981. Systematics and Biogeography: Cladistics and Vicariance. Columbia University Press, New York, 567 pp. Norman, D. B. 1998. On Asian ornithopods (Dinosauria: Ornithischia). 3. A new species of iguanodontid dinosaur. Zoological Journal of the Linnean Society of London 122: 291–348. Nowinski, A. 1971. Nemegtosaurus mongoliensis n. gen., s. sp. (Sauropoda) from the Uppermost Cretaceous of Mongolia. Palaeontologia Polonica 25: 57–81. Osborn, H. F. 1923. Two Lower Cretaceous dinosaurs from Mongolia. American Museum Novitates 93: 1–3. — 1924a. Psittcosaurus and Protiguanodon: two Lower Cretaceous iguanodonts from Mongolia. American Museum Novitates 127: 1–16. — 1924b. Sauropoda and Theropoda of the Lower Cretaceous of Mongolia. American Museum Novitates 128: 1–7.

238


Ostrom, J. H. & McIntosh, J. S. 1966. Marsh’s Dinosaurs. Yale University Press, New Haven, 388 pp. Ouyang, H. & Ye, Y. 2002. [The first mamenchisaurian skeleton with complete skull Mamenchisaurus youngi]. Sichuan Scientific and Technology Publications, Sichuan, 111 pp. [In Chinese with English abstract]. Owen, R. 1841. A description of a portion of the skeleton of Cetiosaurus, a gigantic extinct saurian occurring in the Oolitic Formation of different parts of England. Proceedings of the Geological Society of London 3: 457–462. Page, R. D. M. 1991. Random dendrograms and null hypotheses in cladistic biogeography. Systematic Zoology 40: 54–62. Pang, Q. & Cheng, Z. 2000. A new family of sauropod dinosaur from the Upper Cretaceous of Tianzhen, Shanxi Province, China. Acta Geologica Sinica 74: 117–125. Plot, R. 1677. The Natural History of Oxford-shire, being an essay towards the Natural History of England. The Theater, Oxford, 358 pp. Rauhut, O. W. M. 2003. A dentary of Patagosaurus (Sauropoda) from the Middle Jurassic of Patagonia. Ameghiniana 40: 425–432. — 2006. A brachiosaurid sauropod from the Late Jurassic Cañadón Calcáreo Formation of Chubut, Argentina. Fossil Record 9: 226–237. ´ Rich, T. H., Vickers-Rich, P., Gimenez, O., Cuneo, R., Puerta, P. & Vacca, R. 1999. A new sauropod dinosaur from Chubut Province, Argentina. Pp. 61–84 in Tomida, Y., Rich, T. H. & Vickers-Rich, P. (eds) Proceedings of the Second Gondwana Dinosaur Symposium. National Science Museum Monographs, Tokyo. Rieppel, O. & Reisz, R. R. 1999. The origin and early evolution of turtles. Annual Review of Ecology and Systematics 30: 1–22. Romer, A. S. 1956. Osteology of the Reptiles. University of Chicago Press, Chicago, 772 pp. ˜ Ruiz-Omenaca, J. I., Canudo, J. I. & Cuenca Bescós, G. 1997. Primera evidencia de un a´ rea de alimentación de dinosaurios herbivoros en el cretácico inferior de España (Teruel). Monografias de la Academia de Ciencias Exactas F´ısicas Qu´ımicas y Naturales de Zaragoza 10: 1–48. Russell, D. A. 1993. The role of central Asia in dinosaurian biogeography. Canadian Journal of Earth Science 30: 2002–2012. — 1995. China and the lost worlds of the dinosaurian era. Historical Biology 10: 3–12. — & Zheng, Z. 1993. A large mamenchisaurid from the Junggar Basin, Xinjiang, People’s Republic of China. Canadian Journal of Earth Sciences 30: 2082–2095. Ryabinin, A. N. 1939. [Some results of the studies of the Upper Cretaceous dinosaurian fauna from the vicinity of the Station Sary-Agach, south Kazakhstan.] Problemy Palaeontologii 4: 175–184. [In Russian]. Salgado, L., Coria, R. & Calvo, J. O. 1997. Evolution of titanosaurid sauropods. I: phylogenetic analysis based on the postcranial evidence. Ameghiniana 34: 3–32. —, Apestegu´ıa, S. & Heredia, S. 2005. A new specimen of Neuquensaurus australis, a Late Cretaceous saltasaurine titanosaur from north Patagonia. Journal of Vertebrate Paleontology 25: 623– 634. —, Carvalho, I. S. & Garrido, A. C. 2006. Zapalasaurus bonapartei, un nuevo dinosaurio saurópodo de La Formación La Amarga (Cretácico Inferior), noroeste de Patagonia, Provincia de Neuquén, Argentina. Geobios 39: 695–707. Sanz, J. L., Powell, J. E., Loeuff, J. L., Martinez, R. & Pereda Suberbiola, X. 1999. Sauropod remains from the Upper Cretaceous of Laño (northcentral Spain). Titanosaur phylogenetic relationships. ´ Estudios del Museo de Ciencias Naturales de Alava 14: 235–255. Säve-Söderbergh, G. 1946. Carl Wiman 1867 10/3 – 1944 15/6 [obituary]. Avtryck ur Geologiska Föreningens i Stockholm Förhandlingar 68: 393–404. Scotese, C. R. 2004. Cenozoic and Mesozoic paleogeography: changing terrestrial pathways. Pp. 9–26 in Lomolino, M. V. & Heaney, L. R. (eds) Frontiers of Biogeography. Sinauer Associates, Sunderland, Massachusetts. Seeley, H. G. 1887. The classification of the Dinosauria. Geological Magazine 3: 562. Sereno, P. C. 1986. Phylogeny of the bird-hipped dinosaurs (order Ornithischia). National Geographic Research 2: 234–256.

— 1999. Evolution of dinosaurs. Science 284: 2137–2147. — 2007. Logical basis for morphological characters in phylogenetics. Cladistics 23: 1–23. —, Wilson, J. A., Witmer, L. M., Whitlock, J. A., Maga, A., Ide, O. & Rowe, T. A. 2007. Structural extremes in a Cretaceous dinosaur. PLoS ONE 2: e1230. doi:10.1371/journal.pone.0001230. Sikes, D. S. & Lewis, P. O. 2001. PAUPRat: a tool to implement Parsimony Ratchet searches using PAUP. Department of Ecology and Evolutionary Biology, University of Connecticut, Storrs. Simpson, G. G. 1952. Probabilities of dispersal in geologic time. Bulletin of the American Museum of Natural History 99: 163–176. Smith, A. G., Smith, D. G. & Funnell, B. M. 1994. Atlas of Mesozoic and Cenozoic coastlines. Cambridge University Press, Cambridge, 102 pp. Solounias, N. 1999. The remarkable anatomy of the giraffe’s neck. Journal of Zoology 247: 257–268. Stannius, H. 1846. Lehrbuch der vergleichenden Anatomie der Wirbelthiere. Verlag von Veit & Comp., Berlin, 482 pp. Steel, R. 1970. Saurischia. Handbuch der Paläoherpetologie 13: 1–88. Sues, H.-D. & Averianov, A. O. 2004. Dinosaurs from the Upper Cretaceous (Turonian) of Dzharakuduk, Kyzlkum Desert, Uzbekistan. Journal of Vertebrate Paleontology 24: 119A–120A. Swinton, W. E. 1947. New discoveries of Titanosaurus indicus Lyd. Annals and Magazine of Natural History (series 11) 14: 112–123. Swofford, D. L. 2002. PAUP: Phylogenetic Analysis Using Parsimony version 4.10b10. Sinauer Associates, Sunderland, Massachusetts. Tang, F., Jin, X., Kang, X. & Zhang, G. 2001a. [Omeisaurus maoianus a complete Sauropoda from Jingyan, Sichuan]. Research works of Natural History Museum of Zhejiang 10: 1–127. [In Chinese with English summary]. —, Kang, X.-M., Jin, X.-S., Feng, W. & Wu, W.-T. 2001b. A new sauropod dinosaur of Cretaceous from Jiangshan, Zheijiang Province. Vertebrata PalAsiatica 39: 272–281. Tanimoto, M. & Suzuki, C. 1998. Wear facet of teeth in the Nemegtosauridae – a new knowlege of Suzuki’s sauropod specimen from Iwaki. Research Reports Iwaki Natural History 9: 1–3. Tidwell, V., Carpenter, K. & Meyer, S.. 2001. New titanosauriform (Sauropoda) from the Poison Strip Member of the Cedar Mountain Formation (Lower Cretaceous), Utah. Pp. 139–165 in Tanke, D. & Carpenter, K. (eds) Mesozoic Vertebrate Life: New Research Inspired by the Paleontology of Phillip J. Currie. Indiana University Press, Bloomington. —, Stadtman, K. L. & Shaw, A. 2005. Age-related characteristics found in a partial pelvis of Camarasaurus. Pp. 180–186 in Tidwell, V. & Carpenter, K. (eds) Thunder-Lizards: The Sauropodomorph Dinosaurs. Indiana University Press, Bloomington. Tomida, Y. & Tsumura, Y. 2006. A partial skeleton of titanosaurian sauropod from the Early Cretaceous of Toba City, central Japan. Journal of the Paleontological Society of Korea 22: 217–238. Tsuihiji, T. 2004. The ligament system in the neck of Rhea americana and its implication for the bifurcated neural spines of sauropod dinosaurs. Journal of Vertebrate Paleontology 24: 165–172. Upchurch, P. 1995. Evolutionary history of sauropod dinosaurs. Philosophical Transactions of the Royal Society of London B 349: 365–390. — 1998. The phylogenetic relationships of sauropod dinosaurs. Zoological Journal of the Linnean Society 124: 43–103. — 1999. The phylogenetic relationships of the Nemegtosauridae. Journal of Vertebrate Paleontology 19: 106–125. — & Barrett, P. M. 2000. The evolution of sauropod feeding. Pp. 79–122 in Sues, H.-D. (ed.) Evolution of Feeding in Terrestrial Vertebrates. Cambridge University Press, Cambridge. — & — 2005. Phylogenetic and taxic perspectives on sauropod diversity. Pp. 104–124 in Curry Rogers, K. A. & Wilson, J. A. (eds) The Sauropods: Evolution and Paleobiology. University of California Press, Berkeley and Los Angeles. — & Hunn, C. A. 2002 ‘Time’: the neglected dimension in cladistic biogeography? Geobios 24, 277–286. — & Martin, J. 2002. The Rutland Cetiosaurus: the anatomy and relationships of a Middle Jurassic British sauropod dinosaur. Palaeontology 45: 1049–1074.


—, Hunn, C. A. & Norman, D. B. 2002. An analysis of dinosaurian biogeography: evidence for the existence of vicariance and dispersal patterns caused by geological events. Proceedings of the Royal Society of London B 269: 613–621. —, Barrett, P. M. & Dodson, P. 2004a. Sauropoda. Pp. 259–324 in Weishampel, D. B., Dodson, P. & Osmólska, H. (eds) The Dinosauria. 2nd Edition. University of California Press, Berkeley. —, Tomida, Y. & Barrett, P. M. 2004b. A new specimen of Apatosaurus ajax (Sauropoda: Diplodocidae) from the Morrison Formation (Upper Jurassic) of Wyoming, USA. National Science Museum Monographs 26: 1–108. —, Barrett, P. M., Zhao, X. & Xing, X. 2007. A re-evaluation of Chinshakiangosaurus chunghoensis Ye vide Dong 1992 (Dinosauria, Sauropodomorpha): implications for cranial evolution in basal sauropod dinosaurs. Geological Magazine 144: 247–262. Wagler, J. 1832. Neue Sippen und Gattungen der Vögel. Isis 11: 1224. Wang, X., Wang, Y., Wang, Y., Xu, X., Tang, Z., Zhang, F. & Hu, Y. 1998. [Stratigraphic sequence and vertebrate-bearing beds of the lower part of the Yixian Formation in Sihetun and neighboring area, western Liaoning, China. Vertebrata PalAsiatica 36: 81–101. [In Chinese with English abstract]. —, You, H., Meng, Q., Gao, C., Cheng, X. & Liu, J. 2007. Dongbeititan dongi, the first sauropod dinosaur from the Lower Cretaceous Jehol Group of western Liaoning Province, China. Acta Geologica Sinica 81: 911–916. Wedel, M. J., Cifelli, R. L. & Sanders, R. K. 2000. Sauroposeidon proteles, a new sauropod from the Early Cretaceous of Oklahoma. Journal of Vertebrate Paleontology 20: 109–114. Wedel, M. J. & Sanders, R. K. 2002. Osteological correlates of cervical musculature in Aves and Sauropoda (Dinosauria: Saurischia), with comments on the cervical ribs of Apatosaurus. PaleoBios 22: 1–6. Weishampel, D. B. 1990. Dinosaurian distributions. Pp. 63–189 in Weishampel, D. B., Dodson, P. & Osmólska, H. (eds) The Dinosauria. University of California Press, Berkeley. — & Bjork, P. R. 1989. The first indisputable remains of Iguanodon (Ornithischia: Ornithopoda) from North America: Iguanodon lakotaensis, sp. nov. Journal of Vertebrate Paleontology 9: 56–66. — & Jianu, C.-M. 1997. The importance of phylogeny in paleogeographic analyses, with examples from North-American hadrosaurids and European titanosaurids. Sargetia, Scientia Naturae 17: 261–279. —, Barrett, P. M., Coria, R. A., Le Loeuff, J., Xing, X., Sahni, A., Gomani, E. M. & Noto, C. R. 2004. Dinosaurian distributions. Pp. 517–606 in Weishampel, D. B., Dodson, P. & Osmólska, H. (eds) The Dinosauria, 2nd Edition. University of California Press, Berkeley. Wilkinson, M. 1994. Common cladistic information and its consensus representation: reduced Adams and reduced cladistic consensus trees and profiles. Systematic Biology 43: 343–368. — 1995. Coping with abundant missing entries in phylogenetic inference using parsimony. Systematic Biology 44: 501–514. Wilson, J. A. 1999. A nomenclature for vertebral laminae in sauropods and other saurischian dinosaurs. Journal of Vertebrate Paleontology 19: 639–653. — 2002. Sauropod dinosaur phylogeny: critique and cladistic analysis. Zoological Journal of the Linnean Society 136: 217–276. — 2005a. Redescription of the Mongolian sauropod Nemegtosaurus mongoliensis Nowinski (Dinosauria: Saurischia) and comments on Late Cretaceous sauropod diversity. Journal of Systematic Palaeontology 3: 283–318. — 2005b. Morphological patterns in sauropod evolution. Pp. 15–49 in Curry Rogers, K. A. & Wilson, J. A. (eds) The Sauropods: Evolution and Paleobiology. University of California Press, Berkeley.

239

— 2005c. Integrating ichnofossil and body fossil records to estimate locomotor posture and spatiotemporal distribution of early sauropod dinosaurs: a stratocladistic approach. Paleobiology 31: 400–423. — 2006. Anatomical nomenclature of fossil vertebrates: standardized terms or lingua franca? Journal of Vertebrate Paleontology 26: 511– 518. — & Carrano, M. T. 1999. Titanosaurs and the origin of “wide-gauge” trackways: a biomechanical and systematic perspective on sauropod locomotion. Paleobiology 25: 252–267. — & Sereno, P. C. 1998. Early evolution and higher-level phylogeny of sauropod dinosaurs. Journal of Vertebrate Paleontology 18: 1–68. — & Upchurch, P. 2003. A revision of Titanosaurus Lydekker (Dinosauria – Sauropoda), the first dinosaur genus with a ‘Gondwanan’ distribution. Journal of Systematic Palaeontology 1: 125–160. —, Sereno, P. C., Srivastava, S., Bhatt, D. K., Khosla, A. & Sahni, A. 2003. A new abelisaurid (Dinosauria, Theropoda) from the Lameta Formation (Cretaceous, Maastrichtian) of India. Contributions from the Museum of Paleontology, The University of Michigan 31: 1–42. Wiman, C. 1929. Die Kreide-Dinosaurier aus Shantung. Palaeontologia Sinica (series C) 6: 1–67. Wu, W.-H., Dong, Z.-M., Sun, Y.-W., Li, C.-T. & Li, T. 2006. A new sauropod dinosaur from the Cretaceous of Jiutai, Jilin, China. Global Geology 25: 6–8. Wu, X.-C., Brinkman, D. B. & Lu, J.-C. 1994. A new species of Shantungosuchus from the Lower Cretaceous of Inner Mongolia (China), with comments on S. chuhsienensis Young, 1961 and the phylogenetic position of the genus. Journal of Vertebrate Paleontology 14: 210– 229 Xu, X., Zhang, X., Tan, Q., Zhao, X. & Tan, L. 2006. A new titanosaurian sauropod from Late Cretaceous of Nei Mongol, China. Acta Geologica Sinica 80: 20–26. Yates, A. & Kitching, J. W. 2003. The earliest known sauropod dinosaur and the first steps towards sauropod evolution. Proceedings of the Royal Society of London B 270: 1753–1758. You, H., Tang, F. & Luo, Z. 2003. A new basal titanosaur (Dinosauria: Sauropoda) from the Early Cretaceous of China. Acta Geologica Sinica 77: 424–429. —, Ji, Q., Lamanna, M. C., Li, J. & Li, Y. 2004. A titanosaurian sauropod dinosaur with opisthocoelous caudal vertebrae from the Early Late Cretaceous of Liaoning Provice, China. Acta Geologica Sinica 78: 907–911. —, Li, D., Zhou, L. & Ji, Q. 2006. Huanghetitan luijiaxiaensis, a new sauropod dinosaur from the Lower Cretaceous Hekou Group of Lanzhou Basin, Gansu Province, China. Geological Review 52: 668–674. Young, C. C. 1935. Dinosaurian remains from Mengyin, Shantung. Bulletin of the Geological Society of China 15: 519–533. — 1937. A new dinosaurian from Sinkiang. Palaeontologica Sinica (series C) 2: 1–25. — 1939. On a new Sauropoda, with notes on other fragmentary reptiles from Szechuan. Bulletin of the Geological Society of China 19: 279– 315. — 1954. On a new sauropod from Yiping, Szechuan, China. Acta Scientia Sinica 3: 491–504. — 1958. New sauropods from China. Vertebrata PalAsiatica 2: 1–27. — & Zhao, X.-J. 1972. [Mamenchisaurus hochuanensis, sp. nov.]. Academia Sinica 8: 1–30. [In Chinese with English abstract]. Zhang, Y. 1988. [The Middle Jurassic dinosaur fauna from Dashanpu, Zigong, Sichuan]. Journal of the Chengdu College of Geology 3: 1– 87. [In Chinese with English abstract]. Zhou, Z., Barrett, P. M. & Hilton, J. 2003. An exceptionally preserved Lower Cretaceous ecosystem. Nature 421: 807–814.