Anatomical study of two previously undescribed specimens of Clevosaurus hudsoni (Lepidosauria: Rhynchocephalia) from Cromhall Quarry, UK, aided by computed tomography, yields additional information on the skeleton and hitherto undescribed bones

Anatomical study of two previously undescribed specimens of Clevosaurus hudsoni (Lepidosauria:... Abstract We investigate two well-preserved and previously undescribed specimens of Clevosaurus hudsoni from a Late Triassic fissure deposit at Cromhall Quarry, SW Britain. For the first time computed tomography (CT) scans of British Triassic fissure specimens have been successfully digitally segmented. Visualisation software was used to isolate bone from matrix and to separate individual bones from each other, revealing hidden cranial and postcranial elements. The CT data, together with stereoscopic microscope analysis, have enabled a full evaluation of the specimens including previously poorly known or undescribed elements of the type species of the clevosaur clade. We present detailed descriptions of the cervical vertebrae including the atlas-axis complex. Little studied bones such as the gastralia and epipodials are detailed here and a gap in the lower temporal bar is confirmed. Sclerotic ossicles are presented for the first time for C. hudsoni. A fully fused scapulocoracoid and unfused astragalus and calcaneum provide new insights into clevosaur ontogeny. The CT scans provide key information on post mortem movement and taphonomy of the specimen, revealing fragmentation of part of the skull by the right arm, which has been thrust into the right side of the skull displacing both cranial and jaw bones. Clevosaurus, Bristol fissure, rhynchocephalian, sphenodontian, computed tomography, anatomy, Triassic INTRODUCTION During the Late Triassic and Early Jurassic, SW Britain in the region of Bristol and South Wales was inhabited by a diverse vertebrate fauna, living on an archipelago of Carboniferous limestone and sandstone palaeo-islands, positioned ~30° north of the equator (Robinson, 1957; Marshall & Whiteside, 1980; Fraser, 1994; Evans & Kermack, 1994; Whiteside & Marshall, 2008; van den Berg et al., 2012; Whiteside et al., 2016). Vertical joints and fractures on the limestone surface, which developed into a karst topography of fissures, dolines, swallow holes and caverns, were subsequently infilled by a range of lithologies, from breccias, conglomerates and marls to recrystallized limestones, shales, siltstones and calcareous sandstones (Whiteside et al., 2016). These fissure deposits are host to well-documented vertebrate faunas including fish, archosaurs, lepidosaurs and mammals. Early rhynchocephalians are prominent among the diverse tetrapod assemblage, occurring in several of the fissure deposits from numerous quarries around the Bristol area. The type species of Clevosaurus, C. hudsoniSwinton, 1939, the subject of this paper, was the first Triassic terrestrial lepidosaur to be described from the region. It was discovered at Cromhall Quarry (then called Slickstones) by F.G. Hudson, initially reported by Swinton (1939) and later described by Robinson (1973) and Fraser (1988). Subsequently, there have been discoveries of Clevosaurus in other Bristol region quarries (Marshall & Whiteside, 1980; Fraser, 1994; Whiteside & Marshall, 2008; Foffa et al., 2014; Klein et al., 2015) and from South Wales (Fraser, 1994; Säilä, 2005; Whiteside & Marshall, 2008). Other species of Clevosaurus are described from Late Triassic and Early Jurassic deposits in the eastern United States, Canada, South Africa, Brazil, Luxembourg, Belgium and China and recent cladistical work has identified a Clevosauridae clade (Hsiou, de Franca & Ferigolo, 2015). Robinson (1973) gave a partial description of Clevosaurus as ‘Glevosaurus’ in a mistaken view that the name attribution (Clevum = Gloucester) had been wrongly spelled, but C or G is appropriate. The first detailed study of C. hudsoni by Fraser (1988) focussed on an extensive collection of individual bones and articulated specimens, all from Cromhall Quarry. He provided an osteological description of the cranial and postcranial elements and, based on limb bone ratios and comparisons with limb proportions in Sphenodon punctatusGray, 1842 (the only extant rhynchocephalian), reconstructed C. hudsoni as a 250-mm-long, agile, quadrupedal, lizard-like reptile. In this paper, we present two previously undescribed specimens of C. hudsoni from Cromhall Quarry, collected by the University College team led by Pamela L. Robinson in 1953–1955 (unpublished field notes of Robinson held in the NHMUK). The first specimen (NHMUK PV R36832) is an incomplete, partially articulated, but well-preserved skeleton of C. hudsoni. The second specimen, NMHUK PV R36846, is a mainly complete, articulated hind limb of the same species with part of one digit and the shaft and proximal part of the femur missing. The aim of this paper is to provide a detailed description of the two specimens, focussing on the bones that are poorly known or are unknown for C. hudsoni and clevosaurs in general. In addition to stereoscopic microscope analysis, we digitally segmented computed tomography (CT) scans using visualization software, to reveal hidden cranial and postcranial elements. This is the first time that scanning of vertebrates from the UK Triassic fissure deposits has been successfully attempted; previously, a CT scan of PantydracoGalton, Yates & Kermack, 2007, from Pant-y-ffynnon was deemed unsatisfactory (S. Chapman, personal communication, 2016). The CT surface models we generated, together with microscope examination, enable a fuller evaluation and description of C. hudsoni, yielding important new information. Institutional abbreviations and specimens From NHMUK, Natural History Museum, London, UK, C. hudsoni NHMUK PV R36832 and PV R36846; S. punctatus NHMUK 1861; Oct 1928; 65.5.43; 1972.1499; 1985 1212; 97.2.6.10. From UMZC, University Museum of Zoology, Cambridge, United Kingdom, S. punctatus R2614 and R2587. Geological setting Cromhall Quarry (Ordnance Survey Grid Reference ST 704916), the most northerly of the Late Triassic–Early Jurassic fissure localities (Fig. 1A), lies 20 km northeast of Bristol City. Cromhall Village is southwest of the quarry, with Charfield due east. Cromhall Quarry, originally known as Slickstones or Woodend Quarry (Hiscock, 2009), was first worked for building stone in the late 19th century and, although no longer active, is still registered as a working quarry. Quarrying has exposed Carboniferous limestones (and dolomites) of the Black Rock Group, Gully Oolite and Clifton Down Formation (Fig. 1B). These are underlain by limestones and shales of the Avon Group as well as the Tintern Sandstone Formation (Old Red Sandstone). The Carboniferous limestones are well jointed, with a dominant N-S joint direction (Walkden & Fraser, 1993). Walkden & Fraser (1993) suggest that most of the Cromhall Quarry fissure fills are equivalent to the Late Triassic Mercia Mudstone Group sediments but Whiteside & Marshall (2008) and Whiteside et al. (2016) regard the infillings as occurring at, and post, the initiation of the Rhaetian transgression in Penarth Group times. Figure 1. View largeDownload slide A, palaeogeographical map showing the principal Late Triassic/Early Jurassic tetrapod-bearing fissure deposits near Bristol, with current coastline superimposed (modified from Whiteside & Marshall, 2008). B, simplified geology map of the area around Cromhall Quarry. Mercia Mudstone Group is Triassic, Old Red Sandstone is Devonian and the other labelled strata are Carboniferous (geology map and legend derived from BGS website Open Science data, last accessed 29 October 2017). Figure 1. View largeDownload slide A, palaeogeographical map showing the principal Late Triassic/Early Jurassic tetrapod-bearing fissure deposits near Bristol, with current coastline superimposed (modified from Whiteside & Marshall, 2008). B, simplified geology map of the area around Cromhall Quarry. Mercia Mudstone Group is Triassic, Old Red Sandstone is Devonian and the other labelled strata are Carboniferous (geology map and legend derived from BGS website Open Science data, last accessed 29 October 2017). Various lithologies are present within the fissure deposits at Cromhall Quarry, which consist predominantly of red or green mudstones, or marls, interbedded with re-cemented limestone debris, in places dolomitized and often with reworked detrital crinoid ossicles (Fraser, 1982; Fraser & Walkden, 1983; Walkden & Fraser, 1993). White and yellow quartzose and calcareous sandstones and polymictic conglomerates also occur (Fraser, 1985; Walkden & Fraser, 1993; Whiteside et al., 2016). The shape and character of the fissures in the quarry vary, reflecting both variation in host rock morphology and mode of formation (Whiteside et al., 2016). Walkden & Fraser (1993) sub-divided the Bristol fissure systems into two broad types – those that were tectonic in origin and those that were karstic, concluding that ‘in many cases, tectonic features were exploited by karstic dissolution’ (Walkden & Fraser, 1993: 571). The fissures at Cromhall Quarry are regarded as karstic features, which initiated along joint systems in the limestone or exploited tectonic fractures and subsequently expanded by solution, in either sub-aerial or phreatic conditions (Walkden & Fraser, 1993; Whiteside & Marshall, 2008; Whiteside et al., 2016). It is clear from Robinson’s (1957) diagrams and a cave featured in Whiteside et al. (2016) that caverns, not just expanded dolines, formed a short distance below the limestone surface in substantial parts of the fissure system. Furthermore, red sediments tend to underlie the green lithologies. Cavern formation close to the limestone surface indicates a high water table, which provides the phreatic conditions required (Whiteside & Marshall, 2008; Whiteside et al., 2016). This would have most probably occurred when the sea level was high, during the Rhaetian transgression. The first mentions of the Robinson collection of reptile fossils at Slickstones (Cromhall Quarry) are Robinson, Kermack & Joysey (1952) and Robinson (1955). Our C. hudsoni fossils were collected in 1953–1955 from cavern sediments now quarried away, in locality B, Cromhall Quarry (Robinson, NHMUK unpublished notes; Fig. 2A, B). Robinson’s localities B and C (Fig. 2A) lay between sites 1 and 2 of Fraser (1988). In her notes, Robinson described the host rock as a ‘foss. red clay’ (Robinson, NHMUK unpublished notes; Fig. 2C), which contrasts to the buff-coloured matrix at fissure site 1, the probable source of the C. hudsoni specimens studied by Fraser (1988). Walkden & Fraser (1993) recorded C. hudsoni from ‘fenestral limestone’ in their site 1 and the ‘slot fissures’ which they concluded were Rhaetian. However, the location of our specimens suggests a different interpretation of their proposed chronolithological sequence, as the red matrix occurs substantially below the fenestral limestone. Walkden & Fraser (1993) suggested that these red lithologies are Norian and Robinson (1971) assigned them specifically to late Norian. However, we consider that probably all C. hudsoni fossils are Rhaetian. Recent research on the conchostracans identified as Euestheria brodieanaJones, 1862, collected by Pamela Robinson and Tom Fry from the same fissure location (below ‘B’ in Fig. 2A) and the same red marl as C. hudsoni in Cromhall Quarry, has demonstrated that the stratum is late Rhaetian, equivalent to the Cotham Member, Lilstock Formation (Morton et al., 2017). Figure 2. View largeDownload slide A, sketch of the original fissure locality, no longer in existence (modified after Robinson, 1957). B, photograph of the quarry from 1954; the ‘pendulum’ has already been removed (P.L. Robinson, unpublished notebook, NHMUK). C, copy of the original field sketch of fossil locality B by P.L. Robinson, redrawn with minor modification for clarity by the authors (after P.L. Robinson, unpublished notebook, NHMUK). Figure 2. View largeDownload slide A, sketch of the original fissure locality, no longer in existence (modified after Robinson, 1957). B, photograph of the quarry from 1954; the ‘pendulum’ has already been removed (P.L. Robinson, unpublished notebook, NHMUK). C, copy of the original field sketch of fossil locality B by P.L. Robinson, redrawn with minor modification for clarity by the authors (after P.L. Robinson, unpublished notebook, NHMUK). Early interpretations of the Cromhall Quarry fissures as ‘upland’ underground water courses or cave systems, with narrow passages forming an opening to the surface (Robinson, 1957; Halstead & Nicholl, 1971), have been superseded by evidence from the western wall of Cromhall Quarry, revealing separate openings for each of the seven fissures exposed there. Each fissure is thought to represent either a separately filled sinkhole or doline (Fraser, 1985), several dolines that coalesced, or a cavity of undetermined geometry (Walkden & Fraser, 1993), which were probably united at depth (Fraser, 1985; Walkden & Fraser, 1993). Whiteside et al. (2016) note that such caverns probably occurred in the ‘fresh/saline water-mixing zone of freshwater lenses on small limestone palaeo-islands’ (Whiteside & Marshall, 2008; Whiteside et al., 2016: 264), as observed at Tytherington Quarry. These fissures are, therefore, in marginal marine locations rather than in the uplands. The seven principal fissure sites at Cromhall Quarry have yielded at least 14 tetrapod genera, identified from tens of thousands of fossils (Fraser, 1985, 1986, 1994; Walkden & Fraser, 1993; Whiteside et al., 2016). Rhynchocephalians dominate the faunal assemblage at Cromhall Quarry; PlanocephalosaurusFraser, 1982, DiphydontosaurusWhiteside, 1986, and Clevosaurus being the most abundant (Fraser & Walkden, 1983; Fraser, 1994). Additionally, archosauromorph and procolophonid fossils are present (Fraser, 1982; Fraser & Walkden, 1983), in addition to marine and non-marine fish (Walkden & Fraser, 1993;,Fraser, 1994; Whiteside & Marshall, 2008; Whiteside et al., 2016). Mammaliamorphs are presumed absent in the Cromhall Quarry fauna (Whiteside et al., 2016). Individual bones transported from disarticulated skeletons form the majority of the fossils. Fraser (1985) attributed the high degree of rounding and polishing of some bones to attrition, as disarticulated elements were ‘swept along with terrestrial debris into watercourses and thence into the sinkhole systems’ (Fraser, 1985: 286). However, at Cromhall Quarry, complete or partially articulated specimens are recorded (e.g. Fraser, 1988; Whiteside & Marshall, 2008). The specimens of C. hudsoni under investigation here are partial skeletons, fully or mostly articulated. Given the articulation evident in many parts of the fossils, it is likely there was minimal transportation of these skeletons prior to fossilization. Whiteside et al. (2016) consider that dissociation of some of the cranial and postcranial elements of NHMUK PV R36832 occurred in situ, probably by bacterial action in subaqueous conditions. MATERIAL AND METHODS The fossil material used in this study is comprised of a partially complete skeleton (NHMUK PV R36832) and partially complete hind limb (NMHUK PV R36846) of C. hudsoni in a matrix of red marl. The specimens were extensively prepared at University College London and the NHMUK and no further excavation of the material was required. The specimens were observed under a stereoscopic microscope. We took the photographs with a Canon EOS 70D, using proprietary Canon EOS DIGITAL software, version 29.1A (Digital Photo Professional 3.14.40; EOS Utility 2.14.10). Photographs were then aligned and stacked using Adobe Photoshop CS6. The background was removed and the figures prepared using Autodesk AutoCAD LT 2016. To supplement detailed microscopic examination of the fossils, we reconstructed cranial and postcranial portions of C. hudsoni, using CT scans and visualization software. CT is increasingly being used as a non-destructive, investigative tool (Cunningham et al., 2014) and applications include reconstruction of fossil material for anatomical description and visualization of internal anatomy (e.g. Lautenschlager et al., 2014; Bever et al., 2015; Porro, Rayfield & Clack, 2015a, b). Both specimens were CT scanned at µ-VIS X-Ray Imaging Centre, Faculty of Engineering and the Environment, University of Southampton. µ-CT images of NHMUK PV R36832 (scan ref. 1077) were obtained using a custom built, dual source 225/450 kV walk-in room (Nikon Metrology, UK). µ-CT images of NMHUK PV R36846 (scan ref. 1234) were obtained using a XT H225 L micro-focus CT system (Nikon Metrology). Both scans were acquired with a micro-focus 225 kV source, fitted with a tungsten reflection target, together with a Perkin Elmer XRD 1621 detector. The scan settings for NMHUK PV R36846 were: 160 kVp, 44 µA, 177 ms exposure, 3142 projections acquired during a full 360° rotation, using an average of 64 frames per projection. The source to object distance was set at 180 mm, with a source to detector distance of 702 mm, resulting in a 51 µm reconstructed voxel resolution. We used 1-mm Cu filtration, together with the beryllium window that forms part of the target housing. The specimen was mounted within a Perspex tube and held by phenolic foam to prevent movement. A higher resolution ‘local’ scan was subsequently conducted on the skull region, using the same tube voltage and projection settings, but with the source to object distance adjusted such that the reconstructed voxel resolution was close to 30 µm. The scan settings for NMHUK PV R36846 were: 80 kVp, 93 µA, 500 ms exposure, 5001 projections acquired during a full 360° rotation using an average of 16 frames per projection. The source to object distance was set at 92 mm with a source to detector distance of 802 mm, resulting in a 23-µm reconstructed voxel resolution. There was no filtration of the hind limb scans, other than the beryllium window. We mounted the hind limb specimen within a carbon fibre reinforced plastic tube (inner diameter 2 mm, outer diameter 4 mm). Reconstructions of all three scans used a filtered back projection algorithm, implemented within CTPro and CTAgent software packages (Nikon Metrology). The CT scans were processed using Avizo 9.01 (FEI Visualisation Sciences Group) 3D visualization software. The X-ray attenuation properties of the fossil bone and rock matrix are very similar, with relatively poor contrast, so automatic density thresholding of the data was not possible. Therefore, the CT scans were interpreted using manual segmentation. Assessing the data slice by slice, interpolation was carried out across a maximum of three slices (but frequently every slice was segmented). In this manner, bones were separated from the matrix and from each other. In some portions of the CT scans, segmentation was not possible and in other instances portions of bone were segmented but, perhaps as a result of bone damage or poor contrast, only partial elements could be discerned. Where segmentation was achieved, we assigned individual bones to different fields within the segmentation editor and 3D surface models were created of each field. Anatomical descriptions of all visible bones derive from examination of the specimens, by eye and under the microscope. Surface models created from the CT scans provide additional information on non-visible parts of the bones and elements fully concealed within the matrix. Abbreviations Abbreviations used in the Figures 2–19: add, adductor; amp, amphicoelous; ant, anterior; antlat, anterolateral; ar, articular; arch, arch; ast, astragalus; at, atlas; ax, axis; b, bone; ba, basal; bi, bicondylar; bl, blade; bo, basioccipital; br, branchial; bu, bulge; c, cervical; cal, calcaneum; cap, capitellum; car, carpal; ccv, concavity; cd, coronoid; ce, centrum; cen, centrale; cir, circular; cl, clavicle; co, condyle; com, complex; con, contact; cond, condyloid; cor, coracoid; cr, crest; cvx, convexity; d, dentary; de, dental; del, deltopectoral; di, distal; do, dorsal; dp, depression; E, East; ect, ectopterygoid; ectp, ectepicondyle (ectepicondylar); entp, entepicondyle (entepicondylar); epi, epipterygoid; exo, exoccipital; ext, extension; f, frontal; fa, facet; fac, facial; fe, femur; fen, fenestra; fi, fissure; fib, fibula; fl, flange; fo, foramen; fos, fossa; fu, furrow; fx, flexure; gas, gastralium; gr, groove; hu, humerus; hy, hyoid; hyp, hypapophysis; i, ischium; ic, intercentrum; icl, interclavicle; j, jugal; jt, joint; l, left; lat, lateral; m, medial; maf, magnum foramen; md, mandibular; me, metotic; metac, metacarpal; metat, metatarsal; mk, meckelian; mn, mental; mx, maxilla; n, nasal; ne, neural; no, notch; o, orbit; oc, occipital; od, odontode; op, opisthotic; oss, ossified; p, parietal; pal, palatine; par, paroccipital; pf, postfrontal; phx, phalanx; pif, pineal foramen; pl, plate; po, postorbital; pos, posterior; poslat, posterolateral; poz, postzygapophysis; pr, process; prf, prefrontal; pro, prootic; prz, prezygapophysis; pt, pterygoid; px, proximal; q, quadrate; r, right; ra, radius; rad, radiale; rar, retroarticular; ri, ridge; rw, row; s, secondary; sc, scapulocoracoid; sca, scapula; sco, sclerotic ossicle; sk, skirt; so, supraoccipital; sp, stapes; spi, spine; spl, splint; sq, squamosal; st, supratemporal; sty, styloid; su, suture; sub, sub; sul, sulcus; suplab, supralabial; sur, surangular; syn, synapophysis; tab, table; tar, tarsal; th, thickening; thy, thyroid; tib, tibia; to, tooth; tra, transverse; tro, trochlea; tu, tubercle; ug, ungual; ul, ulna; uln, ulnare; v, vertebra; ve, ventral; w, wear. SYSTEMATIC PALAEONTOLOGY Class: Reptilia Subclass: Diapsida Superorder: Lepidosauria Duméril & Bibron, 1839 (sensu Evans, 1984) Order: Rhynchocephalia Gunther, 1867 (sensu Gauthier, Estes & de Queiroz, 1988) Suborder: Sphenodontia Williston, 1925 (sensu Benton, 1985) Family: Clevosauridae Bonaparte & Sues, 2006 (sensu Hsiou et al., 2015) Genus: Clevosaurus Swinton, 1939 Type species: Clevosaurus hudsoni Swinton, 1939 Included species: Clevosaurus minor Fraser, 1988; C. latidens* Fraser, 1993; C. bairdi Sues, Shubin & Olsen, 1994; C. wangi Wu, 1994, C. mcgilli Wu, 1994 and C. petilus Wu, 1994; C. convallis Säilä, 2005; C. brasiliensis Bonaparte & Sues, 2006 and C. sectumsemper Klein et al., 2015. Remarks: The most recent cladistic analysis by Hsiou et al. (2015) resolved a Clevosauridae clade with the following apomorphic features: antorbital region forming one-quarter of the skull length (reversed to between one-third to one-quarter in C. brasiliensis, C. wangi and C. petilus); a narrow and elongated dorsal process of the jugal; palatine teeth forming a single row, plus one isolated tooth. Hsiou et al. (2015) define the clade as ‘all taxa more closely related to Clevosaurus than to Sphenodon’ (Hsiou et al., 2015: 4). *Clevosaurus latidens is positioned outside of Clevosauridae in this analysis. Characters that occur in clevosaurs, but are not restricted to the genus include: a lateral forked flange of the premaxilla preventing contact between the maxilla and the external naris [horizontal posterior flange not present in C. convallis (Säilä, 2005)]; a dorsally expanded lateral process of the premaxilla; suborbital fenestra bounded solely by the ectopterygoid and palatine; a high, steeply inclined coronoid process of the dentary; flanged teeth; a broad maxillary-jugal contact (Säilä, 2005; Bonaparte & Sues, 2006; Jones, 2006; Hsiou et al., 2015; Klein et al., 2015). Diagnosis: Based on Swinton (1939), Robinson (1973) and Fraser (1988), specimen NHMUK PV R36832 can be diagnosed as C. hudsoni. The diagnosis is based on the following principal features, identified in the specimen: Acrodont dentition. The maxilla bears four large additional teeth that increase in size posteriorly; the teeth are conical in form and have postero-lingual flanges. Three smaller conical teeth occur posteriorly to that additional set on the maxilla. The dentary has four large, conical, additional teeth which increase in size posteriorly; these additional teeth have anterolateral flanges. Teeth are evident on the pterygoid and palatine. Two rows of teeth are present on the pterygoid. A single lateral row of large teeth is present on the palatine but because of incomplete preservation of the medial region of the palatine, the small isolated tooth medially offset from the lateral row, typical of clevosaurs, is not recorded. Incomplete lower temporal bar. There is a gap between the posterior process of the jugal and the quadrate on this specimen. Fraser (1988) notes that in some specimens of C. hudsoni, weak contact is made between the jugal and quadratojugal. There is no contact on this specimen. A flattened, plate-like quadrate present. Large pineal foramen. Postorbital triangular in shape. Supratemporal present. Specimen NMHUK PV R36846 is designated the hind limb of C. hudsoni based on its great similarity to C. hudsoni, figured and described by Fraser (1988: fig. 35). Identifiable features on this specimen include: Size of tibia falls within size range of elements measured by Fraser (1988). Hooked fifth metatarsal. Fifth tarsal fused to fifth metatarsal to form a single tarsometatarsal. A single astragalocalcaneum, formed by fusion of ankle bones, is considered diagnostic of clevosaurs but on this specimen the elements appear to be separate (see description of NMHUK PV R36846 below). RESULTS Anatomical description of NHMUK PV R36832 NHMUK PV R36832 (Fig. 3A, B) preserves many of the skull bones and some postcranial elements, many in close association and some partly articulated. Most of the bones on the left side of the skull have been preserved; frontal, parietal, prefrontal, jugal, postfrontal, postorbital, quadrate are present, together with probable pterygoid, palatine and ectopterygoid (Fig. 3C–F; Supporting Information, Appendices S1, S2). However, excepting the frontal, parietal, postfrontal and partial supratemporal, many bones on the right side are absent. The premaxillae are also missing. The (probable) left nasal is broken, as are the anterior regions of both the dentary and maxilla. The articular complex is present, but there is no evidence of an unfused angular, which projects anteriorly, ventral to the dentary in the drawing by Fraser (1988). Bones of the braincase are visible out of position, between the left parietal and left postorbital (Fig. 3C, D; Supporting Information, Appendices S1, S2). Sclerotic ossicles are preserved in the left orbit (Fig. 3E, F). Figure 3. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. A, entire specimen. B, diagrammatic representation of the specimen with key showing location of views in (C–F) and in Figure 4. Skull bones in (C, D) dorsolateral view and (E, F) left lateral view (surface models are in artificial colour in all figures). Figure 3. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. A, entire specimen. B, diagrammatic representation of the specimen with key showing location of views in (C–F) and in Figure 4. Skull bones in (C, D) dorsolateral view and (E, F) left lateral view (surface models are in artificial colour in all figures). The specimen also comprises articulated cervical vertebrae (Fig. 4; Supporting Information, Appendices S3, S4), including the atlas/axis complex, together with the left scapulocoracoid, humerus, radius, ulna and some carpals and metacarpals. There is additionally a probable dorsal vertebra, a right ischium (Fig. 3A, B), gastralia, chevrons, ribs and numerous partially broken bones and fragments, some of which cannot be accurately identified. We describe the skeleton below and have provided longer accounts of poorly known or unknown bones. Figure 4. View largeDownload slide Postcranial bones of Clevosaurus hudsoni specimen NHMUK PV R36832. (A) photograph and (B) surface model (refer to Fig. 3B for location on specimen). Figure 4. View largeDownload slide Postcranial bones of Clevosaurus hudsoni specimen NHMUK PV R36832. (A) photograph and (B) surface model (refer to Fig. 3B for location on specimen). Tooth bearing marginal bones Left maxilla: The anterior region is missing, but the mid- and posterior region of the left maxilla is distinctive of C. hudsoni with four prominent flanged acrodont additional (sensuRobinson, 1976) teeth, increasing in size posteriorly, followed by three simple uniform subconical teeth (a feature of sphenodontians). The bone has a central depression and anteriorly there are three supralabial foramina, where the superior alveolar nerve and maxillary artery probably exited. The anterodorsal nasal facet and anterior edge of the facial process is obscured by both matrix and a (probable) broken portion of the maxilla. The long suborbital margin downturns sharply posteriorly (Fig. 5A, B) and the maxilla extends further with a lateral projection, terminating at a broad tip. Notches and irregularities along this margin are likely to be artefacts of fossil preservation or preparation. The CT scan reveals facets on the medial surface for each of the adjacent elements: jugal, prefrontal, palatine and ectopterygoid (Fig. 5B). Figure 5. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left maxilla in (A) lateral and (B) medial views. Parietals in (C) dorsal and (D) ventral views. Frontals in (E) dorsal and (F) ventral views. Figure 5. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left maxilla in (A) lateral and (B) medial views. Parietals in (C) dorsal and (D) ventral views. Frontals in (E) dorsal and (F) ventral views. The dentition on NHMUK PV R36832 is fully acrodont, characteristic of the Sphenodontia, with the teeth fused centrally about the crown of the jaw. Hatchling dentition is absent on the specimen, possibly due to incomplete preservation, but perhaps indicating an adult individual (Robinson, 1973; Fraser, 1988). The first additional tooth is small, with a rounded tip and the other three teeth are broadly conical with flattened tips. The enamel of these teeth has vertical striations. A ridge of secondary bone is present above the tooth row (Fig. 5A), but it is unlike some (aged?) clevosaur individuals where the teeth are worn and obliterated by the secondary bone forming a single cutting surface (Fraser, 1988). A noticeable posteromedial flange is present on the second, third and fourth teeth. In addition, there is a ridge on the medial surface possibly indicating the dorsal extent of occlusion of the dentary dentition but wear facets are not sufficiently clear in the CT scan. Dermal roofing bones The paired parietals (surrounding a large pineal foramen) and frontals (Fig. 5C–F) are typical of C. hudsoni. The parietals contacted the frontals in an asymmetric interdigitating suture. Also asymmetrical is the complex medial suture between the frontals. The frontals are thickened ventrally where they form the upper margins of the orbits. A supratemporal, a bone that was previously noted in C. hudsoni by Robinson (1973), is identified by its contact with the right parietal and by facets suggesting a tongue-in-groove connection (Fig. 5C). The anterior of both frontals is missing, as is therefore any contact with the nasals. We have, however, identified a probable left nasal that has a W-shaped frontal facet (Fig. 6A, B), similar to that described for C. wangi, from China (Wu, 1994) but not recorded previously in C. hudsoni. The microscope photograph and CT scan reveal a prefrontal facet on the nasal. Figure 6. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left nasal in (A) dorsolateral and (B) ventromedial views. C, right postfrontal in dorsal view. D, articulation between left postfrontal and left postorbital in anterolateral view. Left postorbital in (E) lateral and (F) medial views. Left prefrontal in (G) dorsolateral and (H) posterolateral views. Left jugal in (I) lateral and (J) medial views. Figure 6. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left nasal in (A) dorsolateral and (B) ventromedial views. C, right postfrontal in dorsal view. D, articulation between left postfrontal and left postorbital in anterolateral view. Left postorbital in (E) lateral and (F) medial views. Left prefrontal in (G) dorsolateral and (H) posterolateral views. Left jugal in (I) lateral and (J) medial views. Circumorbital bones The right postfrontal is complete, in life position, and exhibits the facets and overlap with the frontal and the parietal (Fig. 6C); the facet that underlay the postorbital is also clear. The left postfrontal, while remaining in substantial contact with the extensive overlapping facet of the postorbital (Fig. 6D), has separated and displaced ventrally from the roofing elements. The left postorbital is large, triangular and is thickened at the orbital margin (Fig. 6E). The CT images show the overlapping medial facets on the bone for the postfrontal dorsally, jugal anteroventrally and squamosal posteroventrally (Fig. 6F). The left prefrontal has rotated out of position so that the contact with the nasal and frontal is not preserved. However, part of the maxillary facet is visible, as is the posterolateral bulge (Fig. 6G). The rugose keel of the orbital margin is discernible on the CT scan (Fig. 6H). The triradiate left jugal (Fig. 6I), displays a rounded posterior process, with a slight lateral bulge, but no evidence of a facet for quadratojugal or squamosal. This confirms the description of Robinson (1973) that at least some adult specimens of C. hudsoni do not have a complete lower temporal bar. Images from the CT scan reveal maxillary and postorbital facets on the lateral surface (Fig. 6J), which underlay the respective bones. A facet for the ectopterygoid is present on the medial surface of the anterior process. Palatoquadrate and cheek bones The left quadrate comprises a single broad, blade-like bone with a well-defined, posterior keel, dorsoventrally aligned (Fig. 7A, B). This keel thickens ventrally to form a pronounced anterolaterally directed condyle. A thin, slightly convex, blade-like plate projects from the keel anteromedially. This plate includes the margin of a probable quadratojugal foramen (sensuRobinson, 1973), as it disappears into the matrix. In contrast to the description of a very thin quadratojugal fused to the quadrate by Fraser (1988), there is no evidence of a separate quadratojugal or a clear suture between the bones. Therefore, it is unclear whether the condyle is formed from the quadrate alone, as suggested by Fraser (1988), or includes a quadratojugal, as described by Robinson (1973). A pterygoid facet is evident on the lateral surface of this plate and a squamosal facet is just discernible dorsally, although this is obscured by matrix and is not clear on the CT scan. The quadrate does not contact the posterior process of the jugal. Figure 7. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left quadrate in (A) lateral and (B) medial views. Palate region in (C), (D) ventral view (refer to Fig. 3A for location of view). E, left lateral view of specimen between maxilla and dentary, showing bones out of position (refer to Fig. 3A for location of view). Possible left pterygoid in (F), (G) lateral view. Figure 7. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left quadrate in (A) lateral and (B) medial views. Palate region in (C), (D) ventral view (refer to Fig. 3A for location of view). E, left lateral view of specimen between maxilla and dentary, showing bones out of position (refer to Fig. 3A for location of view). Possible left pterygoid in (F), (G) lateral view. Palatal complex The underside of specimen NHMUK PV R36832 reveals a ventral view of the palatal area (Fig. 7C) which, although damaged and infilled in part with matrix, does show some details. There are two rows of small teeth running approximately parallel to each other, with the medial set extending more anteriorly indicating that they belong to the right pterygoid. There appears to be at least seven teeth in the lateral set and at least eight in the medial set that are discernible on the CT scan (Fig. 7D; Supporting Information, Appendix S2). The pterygoid plate projects anteriorly towards a fragment of bone that may represent a point of contact with the right vomer. There is no sign of the vomers but they are thin, fragile elements (Fraser, 1988) and thus rarely preserved. The posterior process of the right pterygoid is tentatively identified, based on CT evidence of connection with the anterior pterygoid plate. A few small, conical teeth, revealed by the CT scan (Fig. 7D) in a medial position to the right pterygoid probably represent teeth of the left pterygoid. A fragment of the left pterygoid plate appears to be preserved on the specimen, but it is difficult to separate bone from matrix in this part of the CT scan. Two stout conical teeth, preserved on a broken but robust bone positioned laterally to the anterior plate of the right pterygoid (Fig. 7D), are thought to belong to the right palatine. Part of the left palatine with several robust teeth is preserved out of position, embedded in matrix between the left maxilla and left dentary (Fig. 7E; Supporting Information, Appendix S1). Anterior to this, a slender bone with medially flattened teeth possibly represents a broken portion of the anterior left maxilla. In a more posterior position between the maxilla and dentary, there is a flexed, curved solid bone (Fig. 7E), which we identify as an ectopterygoid. There is uncertainty about the diagnosis of the bone pictured in Figure 7F, which is evidently out of alignment with the surrounding cranial elements. We considered the possibility that this may be part of a left squamosal. However, based on its position on the specimen and on information from the CT scan, we consider it more likely to be the quadrate flange of the pterygoid. It is comparable to the pterygoid of Sphenodon, illustrated by Jones et al. (2011: fig. 36), with quadrate facets on the plate-like element of the bone and the projection (or ramus), possibly articulating with the squamosal. The CT scan reveals two processes within the matrix (Fig. 7G). These projections are likely to be flanges of the pterygoid for articulation with the epipterygoid (posteriorly) and ectopterygoid (anteriorly). Braincase bones A good portion of the supraoccipital, which forms the roof to the braincase (Fig. 8A–D), is visible. There is a central shallow dorsal crest running anteroposteriorly. Posteriorly, the supraoccipital defines the dorsal surface of the foramen magnum; the anteriormost region is embedded in matrix and is not discernible on the CT scan. Contacts with the other braincase elements are generally not preserved but the supraoccipital appears to be fused with the left prootic and there is a facet for the right exoccipital. The right prootic is poorly preserved, but a depression lying posterolaterally to the right side of the supraoccipital possibly marks the contact between these two bones. Figure 8. View largeDownload slide Photograph and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Braincase bones in (A, B) posterodorsal, (C) right lateral and (D) anteroventral views. Figure 8. View largeDownload slide Photograph and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Braincase bones in (A, B) posterodorsal, (C) right lateral and (D) anteroventral views. The crescent-shaped occipital condyle, posteriorly positioned on the basioccipital, forms the ventral surface of the magnum foramen (Fig. 8A, B). Posterolateral projections of the occipital condyle meet the left and right exoccipitals, which form the lateral walls of the magnum foramen. Although a possible suture is present, they appear fused on the right-hand side, a condition considered rare in Clevosaurus by Fraser (1988). However, an overlap and probable suture are visible at the basioccipital contact with the left exoccipital. This feature is also observed in a specimen of Sphenodon (ref. NHMUK 1861). Lateral facets are present on the exoccipitals, which probably contacted the opisthotic in life but are separated by a narrow matrix infill on the specimen. Similarly, as a result of post mortem rotational movement, the contact between the exoccipitals and the supraoccipital is not in life position. However, a convolute margin is present on the dorsal surface of the right exoccipital, matching one on the supraoccipital (Fig. 8A). Anterior to the occipital condyle, the left and right basal tubercles (Fig. 8B) are fused to, and project posterolaterally from, the basioccipital. The metotic fissure, a prominent depression, lies at the base of the tubercles and there is a probable parasphenoid anteriorly, but its identification cannot be confirmed because little of the bone is visible on the surface and it is unidentifiable on the CT scan. The left opisthotic, forming the posterodorsal region of the braincase and displaying the paroccipital process, is well preserved. It is probable that the left opisthotic and anteriorly positioned prootic are fused. The paroccipital process projects posterolaterally, separated by a ridge from the anteromedial part of the left opisthotic (Fig. 8A). Lateral to this ridge and posterior to the opisthotic, there is a notable circular depression with two lateral processes, probably a muscle attachment site. A similar feature is visible on the opisthotic of Sphenodon (ref NHMUK 97.2.6.10; 1972.1499). Lateral to this, the left paroccipital process projects in a curved sweep anterolaterally, with a concave ventral surface. Medially, a tapering flange projects from the ventral surface of the paroccipital process; this is the ‘descending process’ of Fraser (1988: 140). This flange meets the basioccipital anterior to the left basal tubercle, at the metotic fissure; the medial surface probably formed a border to the fenestra ovalis. The right opisthotic, preserved without a paroccipital process, is only partially visible and is just discernible on the CT scan (Fig. 8B). The scan reveals the curved anterodorsal morphology of the right prootic (Fig. 8C) but, although the right opisthotic and prootic appear fused, the contacts between the prootic and surrounding bones cannot be confirmed. Lower Jaw Left dentary: The left dentary, like the maxilla, is missing the anteriormost region. The dentary is a robust bone (Fig. 9A–D), with a pronounced lateral ridge marking the extensive development of secondary bone, indicative of an adult individual (Fraser, 1988). Below this ridge, there are three mental foramina (Fig. 9A), where the inferior alveolar nerve exited, as described for other clevosaurs (Sues et al., 1994). The dentary dentition is fully acrodont, including four additional teeth (sensuRobinson, 1976) which increase in size posteriorly. Each of the four teeth has a shallow anterolateral flange, extending basally; this is most pronounced on the second and third teeth. A steeper, shorter posterior flange is present on the first and second teeth. The first tooth is considerably smaller than the others and the tip is quite rounded. The base of the tooth is well defined and there are prominent posterior wear facets caused by occlusion on this and the second tooth (Fig. 9B), that match exactly in size with the maxillary dentition. The second tooth is conical in form. The shallowly sloping anterior flange extends to meet the posterior flange of the first tooth, and its base forms a prominent ridge. The third tooth has a rounded base and a large, broad, rounded tip. The shallow anterior flange curves dorsally before meeting the posterior flange of the second tooth. A posterior wear facet is visible. The tip of the fourth tooth is obscured by matrix, but the CT scan reveals that it is clearly the largest (Fig. 9C). It has a rounded appearance, with a rounded shallow base and a broad posterior wear facet. The flange is notably reduced. The CT scan shows wear facets on the lateral surface of the dentary, posterior to the acrodont tooth row, caused by occlusion with the posterior three uniform teeth of the maxilla. Figure 9. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left dentary in (A, C) lateral and (D) medial views. B, partial left dentary in lateral view showing wear facets. E, left articular complex in dorsal view. F, left surangular in lateral view. G, left articular in dorsal view. H, right articular in lateral view. Figure 9. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left dentary in (A, C) lateral and (D) medial views. B, partial left dentary in lateral view showing wear facets. E, left articular complex in dorsal view. F, left surangular in lateral view. G, left articular in dorsal view. H, right articular in lateral view. Wear facets are also evident on the medial surface of the left dentary (Fig. 9D), indicative of the occluding palatine teeth. The meckelian groove is present on the medial surface, below the sub-dental ridge, and is obscured anteriorly by bone ventral to the first additional tooth. Although the anterior tip is missing, there are no indications of hatchling and successive teeth such as are found in some (smaller) specimens of C. hudsoni (Fraser, 1988). The full extent of the depth of the dentary is clear on the CT scan (Fig. 9C, D). The dorsal region of the coronoid process curves medially and there is a slight lateral thickening at its base. We find no evidence from the CT scan of a separate coronoid bone. The mandibular foramen, which accommodated the inferior alveolar nerve, is a pronounced feature on the posterior base of the coronoid process, with an opposing foramen visible on the surangular portion of the articular complex (Fig. 9E, F), which has become detached from the dentary. The surangular forms the lateral surface dorsal to the posterior process of the dentary. Posterior part of lower jaw: The left articular complex of articular, prearticular and surangular (Fig. 9E) thickens posteriorly, terminating in a retroarticular process. The curved articular complex is partly fused. The anterior portion of the surangular is fused to the articular; the mandibular foramen is present at the boundary between the two bones. Posterior to the foramen, a suture is discernible between the ventral surface of the surangular and the dorsal surface of the articular (Fig. 9F). The left prearticular, which forms the ventral surface of the articular complex, appears to have dissociated from the left articular, with matrix filling the gap (Fig. 9E). Posterior to the surangular, the articular complex curves initially posterolaterally and then in a broad sweep medially. The articular forms a medially directed ridge, which passes up posterodorsally to the retroarticular process, the tip of which is obscured by matrix and may be incomplete. Two well-defined sulci are present, one lateral and the other medial (Fig. 9E, G). These articular fossae, separated by a well-defined ridge, articulated with the condyle of the quadrate. The right articular is revealed by CT scan (Supporting Information, Appendices S1, S2), wholly embedded in matrix. The surface model, generated from the scan data (Fig. 9H), exhibits the distinctive retroarticular process and the articular fossae, separated by a distinctive ridge. Additional cranial bones hidden within the matrix of NHMUK PV R36832 Several bones embedded in the matrix of NHMUK PV R36832 have been revealed by the CT scans (Supporting Information, Appendices S1, S2). None of the bones articulate with those around them and all are thought to be out of position, probably the result of sedimentary slumping. A number of slender rod-like bones are present, which are tentatively identified based on their morphology (Fig. 10A, B) and relative positions in the rock matrix. We consider a slender bone, which has a twist in the shaft and evidence of expanded ends, to be an epipterygoid. A number of rod-like bones, which are flattened at one end and taper to a circular termination at the other, are probably branchial bones of the of the hyoid arch. These bones are usually cartilaginous, but may be ossified in Sphenodon (Romer, 1956). Two probable stapes are also present. Figure 10. View largeDownload slide Surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. A, anterolateral view of skull showing surfaces of slender bones hidden within the matrix (large cranial elements shown for relative position). B, posteroventral view of (A). Figure 10. View largeDownload slide Surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. A, anterolateral view of skull showing surfaces of slender bones hidden within the matrix (large cranial elements shown for relative position). B, posteroventral view of (A). Postcranial bones Left humerus: The humerus is a large, relatively well-preserved bone (Fig. 11A), although there is damage to the surface at both proximal and distal ends. Only the anterior surface of the bone is exposed, but the CT scan reveals much of the complete fossil (Fig. 11B, C) including the ectepicondyle and entepicondyle. Also revealed is the ectepicondylar foramen, through which passed the radial nerve and blood vessels (Fraser & Walkden, 1984). Three faces are formed at the proximal end, separated by distinct ridges, which, together with a tubercle on the ventral edge, would have formed points of attachment for the pectoral muscles (Fraser, 1988). The axial twist along the shaft results in a 90° angle between the proximal and distal heads. Figure 11. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left humerus in (A) anterodorsal, (B) dorsal and (C) ventral views. Left ulna in (D) dorsolateral, (E) anterior and (F) posterior views. G, left distal forelimb bones in lateral view. H, isolated distal left forelimb bones in lateral view. Figure 11. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left humerus in (A) anterodorsal, (B) dorsal and (C) ventral views. Left ulna in (D) dorsolateral, (E) anterior and (F) posterior views. G, left distal forelimb bones in lateral view. H, isolated distal left forelimb bones in lateral view. Left ulna: The ulna is robust, with expanded proximal and distal heads and a slender shaft (Fig. 11D–F). Although the proximal head does not articulate with the humerus, it is in close proximity to where the trochlea would lie. There is no evidence of an olecranon, probably because of breakage. A broad shallow groove just below the proximal head is identified as the distal end of the trochlear notch. The distal head of the ulna is anteroposteriorly flattened and ventrally there is a pronounced coronoid process for muscle attachment. Left carpus and manus: Several distal elements of the forelimb are present on the specimen and, although some lie in close proximity to the distal head of the radius, they are separated and, in part, quite displaced (Fig. 11G). The elements include three carpals and three metacarpals/phalanges. An ungual phalanx displays a concavity in the proximal head, which articulated with the preceding element. One of the preserved carpals, notably thin, flattened and angular, is identified as the radiale. A second smaller, more rounded, faceted bone is probably a centrale and a third bone, multifaceted, angular and solid may be the ulnare but, given its small size, it is considered more likely to be the fourth distal carpal. Although the bones were compared with those of Sphenodon (ref NHMUK Oct 1928), ascribing the metacarpals/phalanges and ungual phalanx to a particular digit is difficult. However, the position and size of the metacarpals indicates they are from either digit one or two. Two further manus bones (Fig. 11H) lie some distance from the left forelimb epipodials, close to the left dentary. The larger of the two bones is likely to be a metacarpal, rather than a metatarsal, based on its preservation near the skull. The bone has a distinct fossa at the proximal end, probably the site of muscle or tendon attachment. A distal or ungual phalanx, separate from the metacarpal/tarsal is probably from a different digit. Right ischium: The ischium is the only part of the pelvic girdle that is found on the surface of the specimen. By comparison with Sphenodon (ref. NHMUK Oct 1928), the bone is likely to be the right ischium, with the lateral view exposed on the surface of the rock (Fig. 12A). It is a thin but broad plate-like bone, roughly hatchet-shaped, that narrows to a thickened neck and flares anterodorsally to where it would have contacted the pubis and ilium. The posteroventral margin of the ischium is arc-like, more curved than the rhomboid form illustrated by Fraser (1988: fig. 31). There is a thickened lateral keel terminating in a posterior process, with a pronounced tubercle positioned immediately posterior to the neck, for attachment of tail musculature ligaments and tendons (Fraser, 1988). The CT scan shows the medial side of the ischium to be thickened below the neck (Fig. 12B). A pubis and a fragment of bone, possibly the ilium, are also revealed by the scan, but the bones do not lie in their life positions. Full segmentation of the pubis was not achieved, but it is clearly the anteroventral element of the pelvis. The short contact surface with the ischium is clear, as is part of the margin of both acetabulum and thyroid fenestra. The ventral portion of the bone is not detectable. The assigned ilium is based on its proximity to the other two bones. Other skeletal elements Some bones of the C. hudsoni pectoral girdle have not been previously described in detail, because the material available was incomplete or not sufficiently clear. For comparative purposes, we reference part of the postcranial skeleton for Planocephalosaurus robinsonaeFraser, 1982, by Fraser & Walkden (1984) and specimens of Sphenodon from the NHMUK. The interclavicle is a T-shaped bone, with the clavicle articulation occurring on the anteroventral surface (Fraser, 1988). Our interclavicle specimen, although partially embedded in matrix (Fig. 12C), is similar, but much larger than the interclavicle of P. robinsonae (Fraser & Walkden, 1984: fig. 12a; plate 53, fig. 16). The clavicle facet and dorsal curvature of the bone are visible at the proximal end. The anterior crossbar is partly exposed but the T-junction between anterior and posterior elements of the bone is hidden beneath the left clavicle in matrix. Both clavicles, slender bones which curved dorsally, have become detached (Fig. 12C). The left clavicle has a dorsal notch where it articulated with the scapulocoracoid and the right bone displays the interclavicle facet. Figure 12. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. A, right ischium in lateral view. B, pelvic girdle bones as preserved in the specimen, out of position (right ischium in medial view). C, postcranial bones at or near the pectoral region, left lateral view. Figure 12. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. A, right ischium in lateral view. B, pelvic girdle bones as preserved in the specimen, out of position (right ischium in medial view). C, postcranial bones at or near the pectoral region, left lateral view. There are numerous bone fragments on the surface and, while some of these may be fragments of ribs or other elements, we describe only those that are recognizable. A few scattered ribs on NHMUK PV R36832 (Fig. 12C) include several short bicipital elements, closely associated with the anterior cervical vertebrae and two larger sub-bicipital ribs, lying close to the posterior cervicals. Two long, narrow, rod-like dorsal ribs are positioned in a posterior position to the humerus and scapulocoracoid. Bones previously poorly known or unknown Sclerotic ossicles: Sclerotic ossicles are found in the matrix of the left orbit (Fig. 13). The bones are very thin and are poorly defined, some being angular and others more rounded. They are flattened with a low, slightly concave profile. In life, we expect the sclerotic plates to have overlapped in the manner of Sphenodon (Underwood, 1970), where some ossicles are overlapped by either or both neighbours and others overlap both neighbours. A complete overlap is not observed on NHMUK PV R36832, but post mortem slippage of bones has undoubtedly occurred. These bones have not been described previously for C. hudsoni, but Sues et al. (1994: 331) suggest that two ‘featureless bony platelets’ in the left orbit of a C. bairdi fossil from the McCoy Brook Formation, Nova Scotia, are sclerotic ossicles. Figure 13. View largeDownload slide Photographs of Clevosaurus hudsoni specimen NHMUK PV R36832. A, partial left lateral view of the skull, showing the position of sclerotic ossicles in the orbital area. B, enlarged view of preserved sclerotic ossicles. Figure 13. View largeDownload slide Photographs of Clevosaurus hudsoni specimen NHMUK PV R36832. A, partial left lateral view of the skull, showing the position of sclerotic ossicles in the orbital area. B, enlarged view of preserved sclerotic ossicles. Sphenodon is known to have 16 (at most seventeen) sclerotic ossicles, which overlap corneally, have a distinct waist orbitally and form a flattish scleral ring (Underwood, 1970). Transposing simple ratios of ossicle size and ring circumference in Sphenodon to NHMUK PV R36832 produces a similar number for C. hudsoni but, admittedly, there are few plates preserved, with little articulation, so this is a broad estimate. Cervical vertebrae: Without a complete, articulated specimen, Fraser (1988) was unable to be sure of the number of cervical vertebrae found in C. hudsoni but suggested that there are eight, based on comparisons with P. robinsonae (estimated from dissociated specimens; Fraser & Walkden, 1984), and Sphenodon (Romer, 1956; Hoffstetter & Gasc, 1969). However, Hoffstetter & Gasc (1969) noted that the Jurassic rhynchocephalian genera have seven cervical vertebrae. The CT scan of our specimen reveals greater detail than previously described (Fig. 14A–H), confirming that eight cervical vertebrae are present in C. hudsoni. The elements that comprise cervical vertebrae 1 and 2, the atlas/axis complex (Fig. 14B, D–H; Supporting Information, Appendix S3), while retaining their essential symmetry, have become slightly displaced. The atlas comprises left and right neural arches, and the first two intercentra, which lie either side of the centrum (odontoid). The elements comprising the atlas are separate, but remain in close proximity. It is probably that, as is the case for at least some Sphenodon individuals (Romer, 1956; Hoffstetter & Gasc, 1969), the atlas bones may have been separate in life. Romer (1956) records the atlas centrum of the adult Sphenodon fusing with the following intercentrum and partially fusing with the axis centrum. Hoffstetter & Gasc (1969), however, suggest that the second intercentrum is fused to the axis centrum, which is lengthened anteriorly by fusion with the atlas centrum, a relationship also described for Sphenodon by Jones et al. (2009). NHMUK PV R36832 appears to align most closely with the observation of Romer (1956), in that the second intercentrum is fused to the centrum of the atlas (Fig. 14D); neither element, however, is fused to the axis. The axis is distinguished from posterior cervical vertebrae by the neural spine, which is enlarged for head support (Romer, 1956). Additional intercentra are present between the axis and third cervical vertebra, and between the third and fourth vertebrae. Romer (1956) describes a ventral hypapophysis on the intercentra of Sphenodon, which Hoffstetter & Gasc (1969) consider to be poorly differentiated on the cervical vertebrae of rhynchocephalians. The intercentra on this specimen are ventrally convex and, while not very well defined, appear to have a rudimentary hypapophysis (Fig. 14D). A proatlas was not identified on NHMUK PV36832. Figure 14. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. A, cervical vertebrae in lateral view. B, disarticulated atlas bones in lateral view. C, cervical vertebrae in lateral view. D, atlas and axis in lateral view. Atlas and axis in (E) anterior, (F) dorsal, (G) posterior and (H) ventral views. Figure 14. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. A, cervical vertebrae in lateral view. B, disarticulated atlas bones in lateral view. C, cervical vertebrae in lateral view. D, atlas and axis in lateral view. Atlas and axis in (E) anterior, (F) dorsal, (G) posterior and (H) ventral views. Ten vertebrae were identified on the CT scan (Fig. 15A). There is a notable difference in morphology between those anteriorly positioned and the more posterior vertebrae. Vertebrae 1–8 are cervical vertebrae whereas 9 and 10 are dorsal vertebrae. The CT scan enables the vertebrae to be viewed in all orientations (Supporting Information, Appendix S4), revealing the complete morphology. The neural arches have been preserved intact and fully attached to the centrum. The cervical vertebrae are distinctively hourglass shaped, notochordal and amphicoelous (Fig. 15B–F). As noted by Fraser (1988), the diapophysis and parapophysis have fused forming a diagonal anteroventral-posterodorsal synapophysis. The two dorsal vertebrae, each with an amphicoelous notochordal centrum and a synapophysis, are similar in form to the cervical vertebrae but are wider and have a shorter neural spine (Fig. 15G–J). Another vertebra, separate from the articulated region (refer to Fig. 3A, B), is identified as dorsal because of its position relative to the rest of the skeleton. The lack of a ‘transverse process’ indicates that it is not from the caudal region. Figure 15. View largeDownload slide Photograph and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. A, cervical and dorsal vertebrae in posterodorsal view. B, fourth cervical vertebra in lateral view. Third cervical vertebra in (C) anterior, (D) dorsal, (E) lateral and (F) posterior views. Second dorsal vertebra in (G) anterior, (H) dorsal, (I) lateral and (J) posterior views. Figure 15. View largeDownload slide Photograph and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. A, cervical and dorsal vertebrae in posterodorsal view. B, fourth cervical vertebra in lateral view. Third cervical vertebra in (C) anterior, (D) dorsal, (E) lateral and (F) posterior views. Second dorsal vertebra in (G) anterior, (H) dorsal, (I) lateral and (J) posterior views. Scapulocoracoid: The scapulocoracoid is a single fused bone, comprised of the dorsal scapula blade and ventral coracoid (Fig. 16A–C). There is no evidence, from either the actual specimen or CT scan, of a suture between the two bones. The tubercle, for attachment of the triceps tendon (Fraser, 1988), is present at the end of the anterodorsal-posteroventral ridge. The glenoid is a prominent feature on the scapulocoracoid, flanked by two ridge-like processes, with the coracoid foramen positioned anteriorly. Notably, the coracoid foramen in this specimen is positioned centrally between the processes (Fig. 16A), directly behind the glenoid, unlike that illustrated by Fraser (1988: fig. 28), where the foramen is offset dorsally. Figure 16. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left scapulocoracoid in (A, B) lateral and (C) medial views. Left radius in (D) dorsolateral, (E) anterior and (F) posterior views. Figure 16. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left scapulocoracoid in (A, B) lateral and (C) medial views. Left radius in (D) dorsolateral, (E) anterior and (F) posterior views. Left radius: Fraser (1988) suggests that the radius of C. hudsoni is rarely preserved due to its delicate structure, but our specimen is quite solid, despite a crosswise kink midway along the shaft of the bone (Fig. 16D–F). From a narrow shaft, the proximal and distal heads broaden slightly. The proximal head has a sub-oval cross-section, with a flattened posterior surface and a prominent lateral groove, where the radius articulated with the ulna. A terminal depression in the proximal head, the ‘shallow oval concavity’ of Fraser (1988: 151), articulated with the capitellum of the humerus. The blunt distal head of the radius exhibits anteroposterior flattening, similar to the ulna. The CT scan shows the distal styloid process, where ligaments and tendons were attached. Gastralia: Gastralia are the bony rods or belly ribs, ventrally positioned on the skeleton. They are generally regarded as strengthening elements in the ventral abdominal wall, protecting the abdomen which, in C. hudsoni lies close to the ground, owing to a sprawling gait (Fraser, 1988). Recent studies on the function of gastralia in other taxa, however, suggest an additional role in respiration (e.g. Carrier & Farmer, 2000; Claessens, 2004), which may also be the case in rhynchocephalians. Our specimen has 15 observable rows of gastralia, 13 of which have retained the apex of the V, or chevron (Fig. 17). Based on an illustration of the gastralia of Sphenodon by Romer (1956: fig. 202) and our observations of Sphenodon specimens, the smallest gastralium is interpreted as being the most posterior/caudal. This posteriormost gastralium has the shortest medial chevron with the most acute angle. The length of each successive anterior chevron increases across the set. The angle of the medial chevron is obtuse mid-region (the widest angle is approximately centrally positioned), becoming successively more acute anteriorly. The gastralia are thickest at the apex of the chevron and thin laterally; the elements are segmented, not all apices retain their associated lateral splints and some lateral elements have no apex. The medial chevron of the first gastralium (caudal) lacks additional lateral splints, but this may be an artefact of fossilization. Lateral segments become increasingly slender as the gastralia lengthen; more caudally positioned lateral splints are as thick as the medial chevrons and those in an anterior position are long and delicate. Overlap between the segments is not uniform, varyingly positioned on either anterior or posterior side of the medial gastralium (Fig. 17C). The chevrons appear for the most part to be fused, although one gastralium has an overlap at the apex which may represent a break or indicate separate elements (Fig. 17B). A similar chevron was observed on a specimen of Sphenodon (ref. NHMUK Oct 1928). Figure 17. View largeDownload slide Photographs of Clevosaurus hudsoni specimen NHMUK PV R36832. A, gastralia in ventral view. B, enlarged view of apex of medial splint. C, enlarged view of overlap between medial and lateral splints. Figure 17. View largeDownload slide Photographs of Clevosaurus hudsoni specimen NHMUK PV R36832. A, gastralia in ventral view. B, enlarged view of apex of medial splint. C, enlarged view of overlap between medial and lateral splints. Fraser (1988), describing each gastralium as two lateral splints joined by a medial apex (chevron), recorded 25 rows of gastralia for C. hudsoni. This is the same number as Sphenodon, but there does not appear to be that many rows on the articulated specimen on which he based his description (Fraser, 1988: plate 2, fig. 4). It is difficult to be sure of the total number, as many of the gastralia illustrated by Fraser (1988) are broken and out of position. However, of those present, there are only ten or 11 apical chevrons. Furthermore, we consider that the overall shape of the gastralia structure of our specimen to be similar to Sphenodon, indicating that a total of 15 or 16 rows is likely. This smaller number seems reasonable as C. hudsoni is estimated to have reached a maximum length of 25 cm (Fraser, 1988) compared to the much longer 60 cm of the extant Sphenodon (Benton, 2015). Anatomical description of NHMUK PV R36846 Apart from the missing mid-region and proximal part of a broken femur and broken fourth digit, the left hind limb specimen of C. hudsoni (NHMUK PV R36846) is almost complete and well preserved (Fig. 18A). The morphology we describe is based on microscope examination and surface models generated from the CT scan (Figs 18, 19; Supporting Information, Appendix S5). The astragalus and calcaneum are considered separately. Figure 18. View largeDownload slide Photograph and surface models of Clevosaurus hudsoni left hind limb specimen NHMUK PV R36846. Specimen in (A, B) dorsal, (C) ventral and (D) posterolateral views. Figure 18. View largeDownload slide Photograph and surface models of Clevosaurus hudsoni left hind limb specimen NHMUK PV R36846. Specimen in (A, B) dorsal, (C) ventral and (D) posterolateral views. Femur and epipodials The tibia and fibula are intact and the distal condylar portion of the femur is present (Fig. 18A–D). The curve on the end of femoral shaft is less pronounced on this specimen than that illustrated by Fraser (1988: fig. 33). The lateral and medial tibial condyles, separated by a furrow, are well displayed on the ventral surface of the femur (Fig. 19A). Tibia: The tibia is a robust bone with a near cylindrical slender shaft, expanding to sub-rounded proximal and distal heads (Fig. 19B–D). The proximal end displays the femoral condyles and a sulcus that would have accommodated the medial surface of the head of the fibula. The shaft of the tibia is slightly anteroposteriorly flattened and broadly concave towards the fibula along its length. The distal head is concave on the ventromedial surface, for articulation with the ‘astragalocalcaneum’. Figure 19. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36846. A, distal head of broken femur in ventrolateral view. Left tibia in (B) anterior, (C) lateral and (D) posterior views. Left fibula in (E) anterior and (F) posterior views. G, astragalus, calcaneum and tarsal bones in dorsal view. H, digit i in dorsolateral view. I, digit i phalanx and ungual in dorsolateral view. Astragalus and calcaneum in (J) dorsal and (K) ventral views. Figure 19. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36846. A, distal head of broken femur in ventrolateral view. Left tibia in (B) anterior, (C) lateral and (D) posterior views. Left fibula in (E) anterior and (F) posterior views. G, astragalus, calcaneum and tarsal bones in dorsal view. H, digit i in dorsolateral view. I, digit i phalanx and ungual in dorsolateral view. Astragalus and calcaneum in (J) dorsal and (K) ventral views. Fibula: The fibula is a very slender bone (Fig. 19E, F) that appears to have a slight axial twist on the (anteroposteriorly) flattened shaft, but this possibly reflects breakage. The shaft broadens slightly at both proximal and distal heads. Articulation facets for adjacent bones are poorly preserved. Tarsus and pes Three tarsals are present between the astragalus and calcaneum and the metatarsals (Fig. 19G). There is no evidence of a first digit tarsal and the second almost merges with the third, occurring as a ventromedial attachment with a small dorsolateral projection. These second and third tarsals are small, well-rounded bones with articulation facets. The fourth tarsal, larger and with a prominent deep foramen on the dorsal surface, is more polygonal in shape, displaying a curved convex lateral facet for articulation with the medial surfaces of the astragalus and calcaneum. A distal facet articulates with the tarsometatarsal (fifth metatarsal), which has the distinctive lepidosaurian hooked morphology (Fig. 19G). The robust tarsometatarsal has a depressed medial facet to articulate with the adjacent fourth metatarsal and proximal facets for articulation with the calcaneum and fourth tarsus. The metatarsals and phalanges vary in size, but are somewhat similar morphologically; the metatarsals have a slender shaft, dorsoventrally flattened, with expanded proximal and broad, faceted distal heads. Metatarsal 1 (Fig. 19H) illustrates the typically bicondylar convex distal head that accommodated the concave head of the phalanx. The phalanges have a more circular shaft, notably faceted bicondylar distal heads, and proximal heads which are rounded and concave. The distal heads have a convex tip, which articulated with the concave head of the next phalanx. The ungual phalanges are mediolaterally compressed and have a strongly concave proximal surface that articulated with the penultimate phalanx (Fig. 19H). A groove is visible on the lateral and medial surfaces of each ungual phalanx (Fig. 19I). These are likely to have been attachment points for tendons providing surface traction for the claw. The phalangeal formula for NHMUK PV R36846 is uncertain because the fourth digit is broken beyond the second phalanx. However, based on the digits preserved, the hind limb phalangeal formula for C. hudsoni is 2: 3: 4: _: 4. In the absence of a complete hind limb, Fraser (1988) assumed that the phalangeal formula for C. hudsoni would be the same as that of Sphenodon (2: 3: 4: 5: 4), which is accepted here. Bones previously poorly known or unknown Astragalus and calcaneum: Clevosaurs typically possess an astragalocalcaneum; a single fused proximal tarsal element of the medial astragalus and the lateral calcaneum. However, on NHMUK PV R36846, the astragalocalcaneum appears unfused (Figs 18A, 19J, K). There is a clear separation between the astragalus and the calcaneum and, while this may reflect breakage along a line of weakness defined by the suture between the two elements, followed by medioventral rotation of the calcaneum, it is more likely that the two bones were never fused. Matrix infill between the two elements may give the appearance of fusion at their lateral contact, but the ventral view facilitated by the CT scan (Fig. 19K) delineates clearly the outline of each bone. Fraser (1988) describes a fused astragalocalcaneum in C. hudsoni and Sues et al. (1994) suggest that the poorly preserved astragalus and calcaneum on their specimen of C. bairdi appears to be fused. Additionally, Klein et al. (2015) identify a single element in C. sectumsemper. Thus, the probable unfused ‘astragalocalcaneum’ of our specimen is unusual among clevosaurs, but visual and CT scan evidence suggests that the separation between the astragalus and the calcaneum is more than just a break. The astragalus has a concave lateral facet that would have articulated with a medial convexity on the calcaneum (Fig. 19K). The margins of both bones are well defined and, while the edge may have undergone some attrition during preservation, if there had been a break along a suture line some degree of irregularity might be expected. Sues et al. (1994) do not provide an image of the astragalocalcaneum for C. bairdi, but Fraser (1988: fig. 35) and Klein et al. (2015: fig. 6K, L) provide an illustration and photograph, respectively. The line of suture between the two bones is not delineated by Fraser (1988). A vague suture is identifiable on both anterior and posterior surfaces of the C. sectumsemper specimen (Klein et al., 2015), but it is not a deep feature and does not appear to present a possible line of weakness. The astragalus and calcaneum preserved on specimen NMHUK PV R36846 are robust bones. If these elements were at one time a single unit, in a manner comparable to C. sectumsemper, then force would have been required to separate them. While the ‘astragalocalcaneum’ might be considered to be in a vulnerable position on the hind limb of Clevosaurus, given the exceptional preservation of this specimen and the fact that the bones are mostly undamaged, it is more probably that the bones were never fused. It is notable that although one NHMUK Sphenodon specimen (ref NHMUK 65.5.43) showed no evidence of a suture between the astragalus and the calcaneum, two other specimens, one adult and one probable sub-adult, had obvious sutures (ref. NHMUK 1861; Oct 1928). The astragalus has broadly concave proximal facets for articulation with the sub-rounded distal heads of both tibia and fibula (Fig. 19J). Facets are also present on the distal surfaces of both the astragalus and calcaneum for articulation with the tarsals. Both bones are flattened dorsoventrally and have broad concave anterior surfaces. DISCUSSION Taphonomy of NHMUK PV R36832 and NHMUK PV R36846 The positioning of the bones on NHMUK PV R36832 indicates partial dissociation in situ. The left-hand side of the skull is much more intact but many neighbouring bones have dissociated, some slightly and others to a greater degree. Some bones are broken but others, especially the more robust, are largely complete. There are good examples of near complete articulation or connection in the paired frontals and parietals, the cervical vertebrae and the left scapulocoracoid, humerus and forearm bones. It provides evidence for the hypothesis that the accumulations of numerous isolated bones in the British Triassic fissures commonly occurred through the dissociation of articulated corpses that washed into the fissure and were preserved by rapid, bacterial decomposition in a subaqueous, or at least moist, environment (Whiteside & Marshall, 2008; Whiteside et al., 2016). The lack of any bite marks on this fossil also contrasts with the predator accumulation hypothesis of Evans & Kermack (1994). The more fragmentary right-hand side of the skull is explained by the CT scan, which reveals the presence of an entire right humerus (Supporting Information, Appendices S1, S2) with smaller bones, the right ulna and radius, positioned next to the distal end of the humerus. These images indicate that the right arm was pushed through the skull by post mortem sedimentary slumping, causing the displacement of head bones. The movements of matrix in the fissure that resulted in this partial fragmentation may relate to compression of the sediments, caused by new material falling onto the decaying skeleton in an aqueous environment. However, although the original fieldnotes of P.L. Robinson do not have precise descriptions of the strata (due to the quarrying operations), the matrix movement could relate to seismic movements, such as those resulting in the slumping structures found in the Late Triassic in the UK (Gallois, 2009). NHMUK PV R36846 is a very well-preserved hind limb fossil, with bones in life or almost life positions, strengthening the view that entire carcasses were regularly washed into the fissures. What is not clear is whether some individuals were alive or recently killed in the waters that washed in. It is conceivable that this individual was freshly killed just before or at burial. Whatever the circumstances, the soft tissue would have rapidly decomposed and in other cases further flows would have dissociated and deposited the isolated bones used to describe the fauna at, for example, Cromhall Quarry (Fraser & Walkden, 1983) or Tytherington Quarry (Whiteside & Marshall, 2008). Comparison with other clevosaurs Based on recent cladistic analysis (Hsiou et al., 2015), the Clevosauridae clade currently comprises C. bairdi from the Early Jurassic of Nova Scotia, Canada (Sues et al., 1994), C. wangi., C. petilus and C. mcgilli from the Early Jurassic of Yunnan Province, China (Wu, 1994), C. brasiliensis from the Late Triassic of Rio Grande do Sul, Brazil (Bonaparte & Sues, 2006), C. convallis from the Early Jurassic of South Wales (Säilä, 2005) and C. hudsoni and C. minor from the Late Triassic of SW England (Fraser, 1988). Clevosaurus sectumsemper had not been described at the time of the analysis, but Klein et al. (2015) considered it a distinct taxon from the other SW UK clevosaurs; future cladistic analysis may confirm the validity of this suggestion. Clevosaurus bairdi, C. brasiliensis and the three Chinese clevosaurs, C. wangi, C. petilus and C. mcgilli, have each been diagnosed from partial or deformed skulls, while the UK species (excepting C. hudsoni) have been described primarily from jaw bones. Jaw bones have been used here to compare NHMUK PV R36832 with other clevosaurs. A fully acrodont dentition is shared by all clevosaurs, together with the presence of secondary dentine. Evidence of tooth wear, indicative of orthal occlusion between maxilla and dentary is common. Significant differences occur, however, in the number of additional teeth between species and the presence or absence of prominent tooth flanges (Table 1). Interestingly, a separate coronoid bone attached to the coronoid process of the dentary has been identified in some but not all clevosaurs; notably it has not been recorded in C. hudsoni. Table 1. Differences in dentition and dentary morphology among clevosaurs Species Location Reference Additional teeth Prominent flange on teeth Coronoid process Clevosaurus bairdi Nova Scotia, Canada Sues et al. (1994) 5–6 On 3 additional teeth only CB Clevosaurus brasiliensis Rio Grande Do Sul, Brazil Bonaparte & Sues (2006) 2–3 Not conspicuous NC Clevosaurus wangi Yunnan Province, China Wu (1994); Jones (2006) 5* Yes; small CB Clevosaurus petilus Yunnan Province, China Wu (1994); Jones (2006) 5* Small CB Clevosaurus mcgilli Yunnan Province, China Wu (1994); Jones (2006) 5* Yes; small CB Clevosaurus convallis Wales, UK Säilä (2005) 6† Yes N Clevosaurus sectumsemper Bristol, UK Klein et al. (2015) 4 Yes NC Clevosaurus hudsoni Bristol, UK Fraser (1988) 4 Yes NC Clevosaurus minor Bristol, UK Fraser (1988) 4 Yes N NHMUK PV R36832 Bristol, UK This study 4 Yes NC Species Location Reference Additional teeth Prominent flange on teeth Coronoid process Clevosaurus bairdi Nova Scotia, Canada Sues et al. (1994) 5–6 On 3 additional teeth only CB Clevosaurus brasiliensis Rio Grande Do Sul, Brazil Bonaparte & Sues (2006) 2–3 Not conspicuous NC Clevosaurus wangi Yunnan Province, China Wu (1994); Jones (2006) 5* Yes; small CB Clevosaurus petilus Yunnan Province, China Wu (1994); Jones (2006) 5* Small CB Clevosaurus mcgilli Yunnan Province, China Wu (1994); Jones (2006) 5* Yes; small CB Clevosaurus convallis Wales, UK Säilä (2005) 6† Yes N Clevosaurus sectumsemper Bristol, UK Klein et al. (2015) 4 Yes NC Clevosaurus hudsoni Bristol, UK Fraser (1988) 4 Yes NC Clevosaurus minor Bristol, UK Fraser (1988) 4 Yes N NHMUK PV R36832 Bristol, UK This study 4 Yes NC ‘Additional teeth’ refer to the sequence of large teeth that follow the hatchling dentition on the dentary and maxilla; they do not include the small subconical teeth that follow on many rhynchocephalian maxillae. CB, separate coronoid bone present on coronoid process of dentary; NC, separate coronoid bone not recorded; N, coronoid process not preserved. *Number of additional teeth based on maximum recorded by Jones (2006). †Clevosaurus convallis has six large additional dentary teeth followed by one or two smaller teeth (Säilä, 2005). View Large Table 1. Differences in dentition and dentary morphology among clevosaurs Species Location Reference Additional teeth Prominent flange on teeth Coronoid process Clevosaurus bairdi Nova Scotia, Canada Sues et al. (1994) 5–6 On 3 additional teeth only CB Clevosaurus brasiliensis Rio Grande Do Sul, Brazil Bonaparte & Sues (2006) 2–3 Not conspicuous NC Clevosaurus wangi Yunnan Province, China Wu (1994); Jones (2006) 5* Yes; small CB Clevosaurus petilus Yunnan Province, China Wu (1994); Jones (2006) 5* Small CB Clevosaurus mcgilli Yunnan Province, China Wu (1994); Jones (2006) 5* Yes; small CB Clevosaurus convallis Wales, UK Säilä (2005) 6† Yes N Clevosaurus sectumsemper Bristol, UK Klein et al. (2015) 4 Yes NC Clevosaurus hudsoni Bristol, UK Fraser (1988) 4 Yes NC Clevosaurus minor Bristol, UK Fraser (1988) 4 Yes N NHMUK PV R36832 Bristol, UK This study 4 Yes NC Species Location Reference Additional teeth Prominent flange on teeth Coronoid process Clevosaurus bairdi Nova Scotia, Canada Sues et al. (1994) 5–6 On 3 additional teeth only CB Clevosaurus brasiliensis Rio Grande Do Sul, Brazil Bonaparte & Sues (2006) 2–3 Not conspicuous NC Clevosaurus wangi Yunnan Province, China Wu (1994); Jones (2006) 5* Yes; small CB Clevosaurus petilus Yunnan Province, China Wu (1994); Jones (2006) 5* Small CB Clevosaurus mcgilli Yunnan Province, China Wu (1994); Jones (2006) 5* Yes; small CB Clevosaurus convallis Wales, UK Säilä (2005) 6† Yes N Clevosaurus sectumsemper Bristol, UK Klein et al. (2015) 4 Yes NC Clevosaurus hudsoni Bristol, UK Fraser (1988) 4 Yes NC Clevosaurus minor Bristol, UK Fraser (1988) 4 Yes N NHMUK PV R36832 Bristol, UK This study 4 Yes NC ‘Additional teeth’ refer to the sequence of large teeth that follow the hatchling dentition on the dentary and maxilla; they do not include the small subconical teeth that follow on many rhynchocephalian maxillae. CB, separate coronoid bone present on coronoid process of dentary; NC, separate coronoid bone not recorded; N, coronoid process not preserved. *Number of additional teeth based on maximum recorded by Jones (2006). †Clevosaurus convallis has six large additional dentary teeth followed by one or two smaller teeth (Säilä, 2005). View Large The distinctive ‘boomerang-shaped’ maxilla of C. bairdi (Sues et al., 1994), which has flanges on only three of the additional teeth, is markedly different morphologically to that of C. hudsoni including NHMUK PV R63832. The maxillary dentition of C. brasiliensis is variously described as comprising two large flanged teeth, followed by one or two conical teeth or ‘a large conical anterior tooth, followed by several small teeth and in turn by two large ones’ (Bonaparte & Sues, 2006: 919). Hsiou et al. (2015) confirm that the large flanges seen on the additional teeth of C. hudsoni (and observed on NHMUK PV R36832) are not seen on the holotype of C. brasiliensis. The fossil material used by Wu (1994) to assign the three Chinese species was reassessed by Jones (2006), who concluded that as a result of poor preservation, the diagnostic features of the specimens could not be adequately assessed to erect three new taxa and assigned Clevosaurus sp. to all three. In contrast to Wu (1994), Jones (2006) reports that although their dentary and maxillary teeth ‘may possess small flanges’ (Jones, 2006: 558), they are not as extensive as those of C. hudsoni (and also NHMUK PV R36832). However, Hsiou et al. (2015) considered that the three Chinese taxa each possessed unique characters, warranting their inclusion in the cladistic analysis. Clevosaurus convallis possesses six large additional dentary teeth, which generally increase in size posteriorly, but the sixth tooth of this series is smaller than the fifth (Säilä, 2005). The posteromedial flanges on the additional maxillary teeth and anterior flanges on the additional dentary teeth of C. convallis (Säilä, 2005) are features shared with C. hudsoni (Fraser, 1988) including NHMUK PV R36832. However, Säilä (2005) indicates that the anterolateral flanges of the additional teeth on the dentary of C. convallis are not as long as those on C. hudsoni and, on some specimens, there is a complete loss of tooth overlap, with the flanges worn by occlusion with the maxilla; the maxillary flanges of C. convallis are thought to be comparable in size to C. hudsoni. Clevosaurus sectumsemper possesses the same number of additional teeth as C. hudsoni and NHMUK PV R36832, which increase in size posteriorly. However, the tooth bases on the dentary of C. sectumsemper are described by Klein et al. (2015) as more ventrally positioned posteriorly than for C. hudsoni and they further describe pronounced gaps between the four additional teeth, with no overlap of the anterolateral flanges. Therefore, although they share some features, the marginal dentition of NHMUK PV R36832 differs significantly from that of both C. convallis and C. sectumsemper, but it is identical to C. hudsoni. Clevosaurus minor closely resembles C. hudsoni and NHMUK PV R36832 in the shape and configuration of the maxillary and dentary dentition (Fraser, 1988) and in possessing the same number of additional teeth. These teeth are flanged and increase in size posteriorly, and there are two or three small, subconical teeth on the maxilla, posterior to the additional set. However, Fraser & Walkden (1983) and Fraser (1988) observe a considerable difference in tooth wear and the growth of secondary dentine on similarly sized specimens of C. minor and C. hudsoni, the former exhibiting high degrees of wear in the largest specimens examined, and the latter showing signs of immaturity, often without the full complement of teeth. There are also differences between the palatines of the two species (Fraser & Walkden, 1983; Fraser, 1988). Despite the possibility of differences in diet producing such variance, C. minor is not considered to be a juvenile of C. hudsoni (Fraser & Walkden, 1983) and sexual dimorphism is also discounted as a possible explanation by Fraser (1988). Furthermore, C. minor and C. hudsoni rarely occur within the same fissure deposit at Cromhall Quarry (Fraser & Walkden, 1983; Fraser, 1988; Walkden & Fraser, 1993), strengthening the notion that they are separate taxa. Based on the evidence, Fraser & Walkden (1983) and Fraser (1988) concluded that C. minor and C. hudsoni are distinct species. NHMUK PV R36832 was found between fissure sites 1 and 2, a location where C. minor is either absent or very rare (Walkden & Fraser, 1993). This fact, combined with the differences in tooth wear and morphology of the palatine, along with the relative sizes of C. minor and NHMUK PV R36832, indicate that NHMUK PV R36832 is not a specimen of C. minor. It is evident from the posterior wear facets on NHMUK PV R36832 that the maxillary teeth occluded on the lateral side and between the dentary teeth, with each maxillary tooth contacting the flanges between the dentary teeth. That NHMUK PV36832 is an adult specimen is demonstrated by the size of the teeth, the deep wear facets and the secondary dentition. However, the bases to the teeth on both maxilla and dentary are clearly visible and there is no evidence of a single cutting edge of bone, described by Robinson (1973) and Fraser (1988) for some of the most mature (or aged) individuals, where the maxillary teeth have either been worn by use, or are almost totally obscured by the ventral growth of secondary dentine. Similarly, although the flanges between the teeth on the dentary give the impression of a continuous cutting surface, the tips of the additional teeth on NHMUK PV R36832 stand proud and project dorsally. It is, therefore, most probable that NHMUK PV R36832 was an adult, but not an aged, individual. Differing patterns of tooth wear occur in similar sized clevosaurs and comparable wear patterns have been observed in specimens of different size. This variability led Fraser (1988), Säilä (2005) and Klein et al. (2015) to conclude that differences in diet could be responsible for variations in dental wear patterns in C. minor, C. convallis and C. sectumsemper. Säilä (2005) concluded that sexual dimorphism could not be ruled out as a potential cause of variation in tooth wear, but this was not investigated further and had previously been discounted for C. minor, by Fraser (1988), who considered variation in diet more plausible. A diet of arthropods versus one of shelled molluscs or small tetrapods was suggested by Klein et al. (2015) to account for tooth wear variation in C. sectumsemper. Fraser (1988) suggested that juvenile clevosaurs probably fed on small insects and soft bodied invertebrates and postulated that the continuous cutting edge of the jaw and edentulous regions of the mature animals supported facultative herbivory (Fraser & Walkden, 1983). However, transversely elongated teeth are more indicative of herbivory and we suggest C. hudsoni with cutting blades running sub-parallel to the long axis of the jaw was not primarily herbivorous, but rather that the dentition was adapted for faunivory. Preservation and ontogeny The preservation of NHMUK PV R36832 and NHMUK PV R36846 has enabled assessment of bones that have been described rarely, or not at all, for clevosaurs (Table 2). Finding sclerotic ossicles is unusual, as is the preservation of fragile bones such as the stapes and hyoid. Although cervical vertebrae are occasionally preserved, fossilization of an articulated sequence including the atlas/axis complex is rare, as is the presence of the gastralia. The scapulocoracoid and the astragalus and calcaneum are particularly significant for the insights they may provide on ontogeny and/or variation in clevosaurs. The fused scapulocoracoid of NHMUK PV R36832 accords with the observation of Fraser (1988) for C. hudsoni that there were no growth stages in his collection where the scapula and coracoid bones were separated by a complete suture. Klein et al. (2015), however, believed that the isolated coracoid of C. sectumsemper showed no evidence of having broken from a fused scapulocoracoid, concluding that juveniles may have possessed separate elements that fused in adulthood. It is notable that a scapulocoracoid suture is visible on both a sub-adult and an adult Sphenodon (ref. NHMUK 1861; 1985 1212) but not on all specimens. Table 2. Notable morphological features described as part of this study for specimens NHMUK PV R36832 and NHMUK PV R36846 and their occurrence in other known clevosaurs Species Location Reference Sclerotic ossicles Stapes/hyoid Atlas/axis centra Gastralia Scapulocoracoid Astragalocalcaneum Clevosaurus bairdi Nova Scotia, Canada Sues et al. (1994) Y N N N N F Clevosaurus brasiliensis Rio Grande Do Sul, Brazil Bonaparte & Sues (2006) N N N N N N Clevosaurus wangi Yunnan Province, China Wu (1994); Jones (2006) N Y U N N N Clevosaurus petilus Yunnan Province, China Wu (1994); Jones (2006) N Y N N N N Clevosaurus mcgilli Yunnan Province, China Wu (1994); Jones (2006) N N N N N N Clevosaurus convallis Wales, UK Säilä (2005) N N N N N N Clevosaurus sectumsemper Bristol, UK Klein et al. (2015) N N N N NF F Clevosaurus hudsoni Bristol, UK Fraser (1988) N N N Y (25) F F Clevosaurus minor Bristol, UK Fraser (1988) N N N N N N NHMUK PV R36832 Bristol, UK This study Y Y NF Y (15) F N NHMUK PV R36846 Bristol, UK This study NA NA NA NA NA NF Species Location Reference Sclerotic ossicles Stapes/hyoid Atlas/axis centra Gastralia Scapulocoracoid Astragalocalcaneum Clevosaurus bairdi Nova Scotia, Canada Sues et al. (1994) Y N N N N F Clevosaurus brasiliensis Rio Grande Do Sul, Brazil Bonaparte & Sues (2006) N N N N N N Clevosaurus wangi Yunnan Province, China Wu (1994); Jones (2006) N Y U N N N Clevosaurus petilus Yunnan Province, China Wu (1994); Jones (2006) N Y N N N N Clevosaurus mcgilli Yunnan Province, China Wu (1994); Jones (2006) N N N N N N Clevosaurus convallis Wales, UK Säilä (2005) N N N N N N Clevosaurus sectumsemper Bristol, UK Klein et al. (2015) N N N N NF F Clevosaurus hudsoni Bristol, UK Fraser (1988) N N N Y (25) F F Clevosaurus minor Bristol, UK Fraser (1988) N N N N N N NHMUK PV R36832 Bristol, UK This study Y Y NF Y (15) F N NHMUK PV R36846 Bristol, UK This study NA NA NA NA NA NF Number in parentheses in ‘Gastralia’ column indicates number of gastralia recorded. F, element preserved and bones fused; N, element not preserved; NA, bone region not present in specimen; NF, element preserved and bones not fused; U, undetermined; Y, element preserved. View Large Table 2. Notable morphological features described as part of this study for specimens NHMUK PV R36832 and NHMUK PV R36846 and their occurrence in other known clevosaurs Species Location Reference Sclerotic ossicles Stapes/hyoid Atlas/axis centra Gastralia Scapulocoracoid Astragalocalcaneum Clevosaurus bairdi Nova Scotia, Canada Sues et al. (1994) Y N N N N F Clevosaurus brasiliensis Rio Grande Do Sul, Brazil Bonaparte & Sues (2006) N N N N N N Clevosaurus wangi Yunnan Province, China Wu (1994); Jones (2006) N Y U N N N Clevosaurus petilus Yunnan Province, China Wu (1994); Jones (2006) N Y N N N N Clevosaurus mcgilli Yunnan Province, China Wu (1994); Jones (2006) N N N N N N Clevosaurus convallis Wales, UK Säilä (2005) N N N N N N Clevosaurus sectumsemper Bristol, UK Klein et al. (2015) N N N N NF F Clevosaurus hudsoni Bristol, UK Fraser (1988) N N N Y (25) F F Clevosaurus minor Bristol, UK Fraser (1988) N N N N N N NHMUK PV R36832 Bristol, UK This study Y Y NF Y (15) F N NHMUK PV R36846 Bristol, UK This study NA NA NA NA NA NF Species Location Reference Sclerotic ossicles Stapes/hyoid Atlas/axis centra Gastralia Scapulocoracoid Astragalocalcaneum Clevosaurus bairdi Nova Scotia, Canada Sues et al. (1994) Y N N N N F Clevosaurus brasiliensis Rio Grande Do Sul, Brazil Bonaparte & Sues (2006) N N N N N N Clevosaurus wangi Yunnan Province, China Wu (1994); Jones (2006) N Y U N N N Clevosaurus petilus Yunnan Province, China Wu (1994); Jones (2006) N Y N N N N Clevosaurus mcgilli Yunnan Province, China Wu (1994); Jones (2006) N N N N N N Clevosaurus convallis Wales, UK Säilä (2005) N N N N N N Clevosaurus sectumsemper Bristol, UK Klein et al. (2015) N N N N NF F Clevosaurus hudsoni Bristol, UK Fraser (1988) N N N Y (25) F F Clevosaurus minor Bristol, UK Fraser (1988) N N N N N N NHMUK PV R36832 Bristol, UK This study Y Y NF Y (15) F N NHMUK PV R36846 Bristol, UK This study NA NA NA NA NA NF Number in parentheses in ‘Gastralia’ column indicates number of gastralia recorded. F, element preserved and bones fused; N, element not preserved; NA, bone region not present in specimen; NF, element preserved and bones not fused; U, undetermined; Y, element preserved. View Large The separate astragalus and calcaneum on NMHUK PV R36846 contrasts with a fused astragalocalcaneum described previously for C. hudsoni (Fraser, 1988), C. sectumsemper (Klein et al., 2015) and C. bairdi (Sues et al., 1994). NMHUK PV R36846 is an isolated hindlimb, so it is not possible to assess the animal’s age directly. However, we can compare our specimen with those of Fraser (1988), who took measurements of fore and hind limb bones from fully ossified individuals. The tibia on NHMUK PV R36846 falls within the measured lengths recorded (Fraser, 1988), lying at the lower end of the range, but is longer than the shortest specimen. This may support the notion that our specimen derived from a young adult and that subsequent ontogenetic fusion of the astragalus and calcaneum may have occurred, had the individual survived to full maturity. However, it is also plausible that the specimen derives from the smaller gender of a dimorphic species. As described above, we have found individual variation in the fused/sutured/unsutured condition of ‘scapulocoracoids’ and ‘astragalocalcanea’, irrespective of animal size, in a range of Sphenodon specimens in the NHMUK. It is, therefore, possible that these bones are isolated components in juveniles that fuse in particular adults, with some or no trace of a suture. Our findings are important, as a fused astragalocalcaneum is considered an apomorphy common to Rhynchocephalia and Squamata (e.g. Evans, 2003). The fully fused astragalocalcaneum condition described for C. hudsoni (Fraser, 1988) may simply indicate a mature adult or a variable character in the species. Correspondingly, the assertion by Klein et al. (2015) that the scapulocoracoid of C. sectumsemper was unfused in juveniles and fused in adults may also apply to C. hudsoni, contraFraser (1988), and to clevosaurs in general. CONCLUSION Two previously undescribed specimens of C. hudsoni, NHMUK PV R36832 and NMHUK PV R36846, from a Late Triassic fissure infill at Cromhall Quarry, Gloucestershire, have been investigated using stereoscopic microscopic analysis and CT. A detailed description of the visible bones was augmented by the creation of 3D digital reconstructions from CT data, providing additional information on material that was either partially, or fully, embedded in matrix. It is the first time that successful CT scans have been obtained from the British Triassic fissure deposits. The possibility of the CT scans revealing detailed information on the braincase bones was hindered by the similarity in X-ray attenuation properties of the fossilized bone and the rock matrix. Despite this, 3D reconstructions have been generated for numerous bones from both specimens, including bones from the right side of NHMUK PV R36832 that are not visible and were presumed lost. Key elements preserved in these specimens have not been previously described. This paper provides new information on the cervical vertebrae including the atlas-axis complex, the pectoral girdle, the fore limb, hind limb, the gastralia as well as aspects of the skull such as the sclerotic ossicles of C. hudsoni. We have been able to significantly add and amend information on the type species of the clevosaur clade. SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article at the publisher’s website: Appendix S1. Animation of Avizo surface model of NHMUK PV R36832: cranial lateral aspect. Appendix S2. Animation of Avizo surface model of NHMUK PV R36832: cranial dorsoventral aspect. Appendix S3. Animation of Avizo surface model of NHMUK PV R36832: atlas/axis complex. Appendix S4. Animation of Avizo surface model of NHMUK PV R36832: cervical/dorsal vertebrae. Appendix S5. Animation of Avizo surface model of NHMUK PV R36846: left hindlimb. ACKNOWLEDGEMENTS We thank Sandra Chapman and Patrick Campbell (NHMUK) for arranging the loans of Clevosaurus and facilitating our visit to investigate Sphenodon skeletons, respectively. Timothy Smithson and Jason Head (UMZC) are thanked for the loan of Sphenodon skeletons. We are very grateful to Mark Mavrogordato and Kathryn Rankin for carrying out the CT scans at µ-VIS X-Ray Imaging Centre, Faculty of Engineering and the Environment, University of Southampton. Mark Mavrogordato is additionally thanked for help in drafting the methodology of a CT scan. We greatly appreciate the support given by Simon Chen and the Cromhall Diving Centre who have been exceptionally helpful in permitting access to Cromhall Quarry. We thank two anonymous reviewers for their constructive comments. REFERENCES Benton MJ . 1985 . Classification and phylogeny of the diapsid reptiles . Zoological Journal of the Linnean Society 84 : 97 – 164 . Google Scholar CrossRef Search ADS Benton MJ . 2015 . Vertebrate palaeontology, 4th edn . London : Wiley-Blackwell . Bever GS , Lyson TR , Field DJ , Bhullar BA . 2015 . Evolutionary origin of the turtle skull . Nature 525 : 239 – 242 . Google Scholar CrossRef Search ADS PubMed Bonaparte JF , Sues HD . 2006 . A new species of Clevosaurus (Lepidosauria: Rhynchocephalia) from the Upper Triassic of Rio Grande do Sul, Brazil . Palaeontology 49 : 917 – 923 . Google Scholar CrossRef Search ADS Carrier DR , Farmer CG . 2000 . The evolution of pelvic aspiration in archosaurs . Paleobiology 26 : 271 – 293 . Google Scholar CrossRef Search ADS Claessens LPAM . 2004 . Dinosaur gastralia; origin, morphology, and function . Journal of Vertebrate Paleontology 24 : 89 – 106 . Google Scholar CrossRef Search ADS Cunningham JA , Rahman IA , Lautenschlager S , Rayfield EJ , Donoghue PC . 2014 . A virtual world of paleontology . Trends in Ecology & Evolution 29 : 347 – 357 . Google Scholar CrossRef Search ADS PubMed Evans SE . 1984 . The classification of the Lepidosauria . Zoological Journal of the Linnean Society 82 : 87 – 100 . Google Scholar CrossRef Search ADS Evans SE . 2003 . At the feet of the dinosaurs: the early history and radiation of lizards . Biological Reviews 78 : 513 – 551 . Google Scholar CrossRef Search ADS PubMed Evans SE , Kermack KA . 1994 . Assemblages of small tetrapods from the Early Jurassic of Britain . In: Fraser NC , Sues HD , eds. In the shadow of the dinosaurs: early Mesozoic tetrapods . New York : Cambridge University Press , 271 – 282 . Foffa D , Whiteside DI , Viegas PA , Benton MJ . 2014 . Vertebrates from the Late Triassic Thecodontosaurus-bearing rocks of Durdham Down, Clifton (Bristol, UK) . Proceedings of the Geologists’ Association 125 : 317 – 328 . Google Scholar CrossRef Search ADS Fraser NC . 1982 . A new rhynchocephalian from the British Upper Triassic . Palaeontology 25 : 709 – 725 . Fraser NC . 1985 . Vertebrate faunas from Mesozoic fissure deposits of South West Britain . Modern Geology 9 : 273 – 300 . Fraser NC . 1986 . New Triassic sphenodontids from south-west England and a review of their classification . Palaeontology 29 : 165 – 186 . Fraser NC . 1988 . The osteology and relationships of Clevosaurus (Reptilia: Sphenodontida) . Philosophical Transactions of the Royal Society B 321 : 125 – 178 . Google Scholar CrossRef Search ADS Fraser NC . 1993 . A new sphenodontian from the early Mesozoic of England and North America: implications for correlating early Mesozoic continental deposits . In: Lucas SG , Morales M , eds. The nonmarine Triassic . Albuquerque : Museum of Natural History and Science , 135 – 139 . Fraser NC . 1994 . Assemblages of small tetrapods from British Late Triassic fissure deposits . In: Fraser NC , Sues HD , eds. In the shadow of the dinosaurs: early Mesozoic tetrapods . New York : Cambridge University Press , 214 – 226 . Fraser NC , Walkden GM . 1983 . The ecology of a Late Triassic reptile assemblage from Gloucestershire, England . Palaeogeography, Palaeoclimatology, Palaeoecology 42 : 341 – 365 . Google Scholar CrossRef Search ADS Fraser NC , Walkden GM . 1984 . The postcranial skeleton of the Upper Triassic sphenodontid Planocephalosaurus robinsonae . Palaeontology 27 : 575 – 595 . Gallois RW . 2009 . The lithostratigraphy of the Penarth Group (Late Triassic) of the Severn Estuary area . Geoscience in South-West England 12 : 71 – 84 . Galton PM , Yates AM , Kermack D . 2007 . Pantydraco n. gen. for Thecodontosaurus caducus YATES, 2003, a basal sauropodomorph dinosaur from the Upper Triassic or Lower Jurassic of South Wales, UK . Neues Jahrbuch für Geologie und Paläontologie – Abhandlungen 243 : 119 – 125 . Google Scholar CrossRef Search ADS Gauthier JA , Estes R , de Queiroz K . 1988 . A phylogenetic analysis of Lepidosauromorpha . In: Estes R , Pregill G , eds. Phylogenetic relationships of the lizard families . Stanford : Stanford University Press , 15 – 98 . Gray JE . 1842 . Description of two hitherto unrecorded species of reptiles from New Zealand; presented to the British Museum by Dr. Dieffenbach . In: Gray JE , ed. The zoological miscellany, Vol. 2 . London : Treuttel, Würtz & Co ., 72 . Halstead LB , Nicholl PG . 1971 . Fossilized caves of Mendip . Studies in Speleology 2 : 93 – 102 . Hiscock C . 2009 . Slickstones Quarry, Cromhall – SSSI & RIGS . Outcrop – The Newsletter of the Avon RIGS Group 24 : 1 – 3 . Hoffstetter R , Gasc JP . 1969 . Vertebrae and ribs of modern reptiles . In: Gans C , Parsons TS , eds. Biology of the Reptilia Volume 1, Morphology A . New York : Academic Press , 201 – 302 . Hsiou AS , De Franca MAG , Ferigolo J . 2015 . New data on the Clevosaurus (Sphenodontia: Clevosauridae) from the Upper Triassic of southern Brazil . PLoS ONE 10 : e0137523 . Google Scholar CrossRef Search ADS PubMed Jones MEH . 2006 . The Early Jurassic clevosaurs from China (Diapsida: Lepidosauria) . New Mexico Museum Natural History Science Bulletin 37 : 548 – 562 . Jones MEH , Curtis N , Fagan MJ , O’Higgins P , Evans SE . 2011 . Hard tissue anatomy of the cranial joints in Sphenodon (Rhynchocephalia): sutures, kinesis, and skull mechanics . Palaeontologia Electronica 14 : 1 –92. Jones MEH , Curtis N , O’Higgins P , Fagan M , Evans SE . 2009 . The head and neck muscles associated with feeding in Sphenodon (Reptilia: Lepidosauria: Rhynchocephalia) . Palaeontologia Electronica 12 : 1–56 . Jones TR . 1862 . A monograph of the fossil Estheriae . Monograph of the Palaeontographical Society 14 : 1 – 134 . Klein CG , Whiteside DI , Selles de Lucas V , Viegas PA , Benton MJ . 2015 . A distinctive Late Triassic microvertebrate fissure fauna and a new species of Clevosaurus (Lepidosauria: Rhynchocephalia) from Woodleaze Quarry, Gloucestershire, UK . Proceedings of the Geologists’ Association 126 : 402 – 416 . Google Scholar CrossRef Search ADS Lautenschlager S , Witmer LM , Altangerel P , Zanno LE , Rayfield EJ . 2014 . Cranial anatomy of Erlikosaurus andrewsi (Dinosauria, Therizinosauria): new insights based on digital reconstruction . Journal of Vertebrate Paleontology 34 : 1263 – 1291 . Google Scholar CrossRef Search ADS Marshall JEA , Whiteside DI . 1980 . Marine influence in the Triassic “uplands” . Nature 287 : 627 – 628 . Google Scholar CrossRef Search ADS Morton JD , Whiteside DI , Hethke M , Benton MJ . 2017 . Biostratigraphy and geometric morphometrics of conchostracans (Crustacea, Branchiopoda) from the Late Triassic fissure deposits of Cromhall Quarry, UK . Palaeontology 60 : 349 – 374 . Google Scholar CrossRef Search ADS Porro LB , Rayfield EJ , Clack JA . 2015a . Descriptive anatomy and three-dimensional reconstruction of the skull of the early tetrapod Acanthostega gunnari Jarvik, 1952 . PLoS ONE 10 : e0118882 . Google Scholar CrossRef Search ADS Porro LB , Rayfield EJ , Clack JA . 2015b . Computed tomography, anatomical description and three-dimensional reconstruction of the lower jaw of Eusthenopteron foordi Whiteaves, 1881 from the Upper Devonian of Canada . Palaeontology 58 : 1 – 17 . Google Scholar CrossRef Search ADS Robinson PL . 1955 . Exhibition of specimens from Slickstones Quarry, Gloucestershire . Proceedings of the Geological Society, London 1527 : 113 – 115 . Robinson PL . 1957 . The Mesozoic fissures of the Bristol Channel area and their vertebrate faunas . Zoological Journal of the Linnean Society 43 : 260 – 282 . Google Scholar CrossRef Search ADS Robinson PL . 1971 . A problem of faunal replacement on Permo-Triassic continents . Palaeontology 14 : 131 – 153 . Robinson PL . 1973 . A problematic reptile from the British Upper Trias . Journal of the Geological Society, London 129 : 457 – 479 . Google Scholar CrossRef Search ADS Robinson PL . 1976 . How Sphenodon and Uromastyx grow their teeth and use them . In: Bellairs AD’A , Cox CB , eds. Morphology and biology of reptiles . London : Academic Press, Linnean Society Symposium Series , 43 – 67 . Robinson PL , Kermack KA , Joysey KA . 1952 . Exhibition of specimens from Slickstones Quarry, Gloucestershire . Proceedings of the Geological Society, London 1485 : 86 – 87 . Romer AS . 1956 . Osteology of the reptiles . Chicago : University of Chicago Press . Säilä LK . 2005 . A new species of the sphenodontian reptile Clevosaurus from the Lower Jurassic of South Wales . Palaeontology 48 : 817 – 831 . Google Scholar CrossRef Search ADS Sues HD , Shubin NH , Olsen PE . 1994 . A new sphenodontian (Lepidosauria: Rhynchocephalia) from the McCoy Brook Formation (Lower Jurassic) of Nova Scotia, Canada . Journal of Vertebrate Paleontology 14 : 327 – 340 . Google Scholar CrossRef Search ADS Swinton WE . 1939 . A new Triassic rhynchocephalian from Gloucestershire . Annals and Magazine of Natural History Series 11 4 : 591 – 594 . Google Scholar CrossRef Search ADS Underwood G . 1970 . The eye . In: Gans C , Parsons TS , eds. Biology of the Reptilia volume 2, morphology B . New York : Academic Press , 1 – 97 . van den Berg T , Whiteside DI , Viegas PA , Schouten R , Benton MJ . 2012 . The Late Triassic microvertebrate fauna of Tytherington, UK . Proceedings of the Geologists’ Association 123 : 638 – 648 . Google Scholar CrossRef Search ADS Walkden GM , Fraser NC . 1993 . Late Triassic fissure sediments and vertebrate faunas: environmental change and faunal succession at Cromhall, South West Britain . Modern Geology 18 : 511 – 535 . Whiteside DI . 1986 . The head skeleton of the Rhaetian sphenodontid Diphydontosaurus avonis gen. et. sp. nov. and the modernizing of a living fossil . Philosophical Transactions of the Royal Society London B 312 : 379 – 430 . Google Scholar CrossRef Search ADS Whiteside DI , Duffin CJ , Gill PG , Marshall JEA , Benton MJ . 2016 . The Late Triassic and Early Jurassic fissure faunas from Bristol and South Wales: stratigraphy and setting . Palaeontologia Polonica 67 : 257 – 287 . Whiteside DI , Marshall JEA . 2008 . The age, fauna and palaeoenvironment of the Late Triassic fissure deposits of Tytherington, South Gloucestershire, UK . Geological Magazine 145 : 102 – 147 . Google Scholar CrossRef Search ADS Wu XC . 1994 . Late Triassic-Early Jurassic sphenodontians from China and the phylogeny of the Sphenodontia . In: Fraser NC , Sues HD , eds. In the shadow of the dinosaurs: early Mesozoic tetrapods . New York : Cambridge University Press , 38 – 69 . © 2017 The Linnean Society of London, Zoological Journal of the Linnean Society This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Zoological Journal of the Linnean Society Oxford University Press

Anatomical study of two previously undescribed specimens of Clevosaurus hudsoni (Lepidosauria: Rhynchocephalia) from Cromhall Quarry, UK, aided by computed tomography, yields additional information on the skeleton and hitherto undescribed bones

Loading next page...
 
/lp/ou_press/anatomical-study-of-two-previously-undescribed-specimens-of-MQSW7QACCF
Publisher
The Linnean Society of London
Copyright
© 2017 The Linnean Society of London, Zoological Journal of the Linnean Society
ISSN
0024-4082
eISSN
1096-3642
D.O.I.
10.1093/zoolinnean/zlx087
Publisher site
See Article on Publisher Site

Abstract

Abstract We investigate two well-preserved and previously undescribed specimens of Clevosaurus hudsoni from a Late Triassic fissure deposit at Cromhall Quarry, SW Britain. For the first time computed tomography (CT) scans of British Triassic fissure specimens have been successfully digitally segmented. Visualisation software was used to isolate bone from matrix and to separate individual bones from each other, revealing hidden cranial and postcranial elements. The CT data, together with stereoscopic microscope analysis, have enabled a full evaluation of the specimens including previously poorly known or undescribed elements of the type species of the clevosaur clade. We present detailed descriptions of the cervical vertebrae including the atlas-axis complex. Little studied bones such as the gastralia and epipodials are detailed here and a gap in the lower temporal bar is confirmed. Sclerotic ossicles are presented for the first time for C. hudsoni. A fully fused scapulocoracoid and unfused astragalus and calcaneum provide new insights into clevosaur ontogeny. The CT scans provide key information on post mortem movement and taphonomy of the specimen, revealing fragmentation of part of the skull by the right arm, which has been thrust into the right side of the skull displacing both cranial and jaw bones. Clevosaurus, Bristol fissure, rhynchocephalian, sphenodontian, computed tomography, anatomy, Triassic INTRODUCTION During the Late Triassic and Early Jurassic, SW Britain in the region of Bristol and South Wales was inhabited by a diverse vertebrate fauna, living on an archipelago of Carboniferous limestone and sandstone palaeo-islands, positioned ~30° north of the equator (Robinson, 1957; Marshall & Whiteside, 1980; Fraser, 1994; Evans & Kermack, 1994; Whiteside & Marshall, 2008; van den Berg et al., 2012; Whiteside et al., 2016). Vertical joints and fractures on the limestone surface, which developed into a karst topography of fissures, dolines, swallow holes and caverns, were subsequently infilled by a range of lithologies, from breccias, conglomerates and marls to recrystallized limestones, shales, siltstones and calcareous sandstones (Whiteside et al., 2016). These fissure deposits are host to well-documented vertebrate faunas including fish, archosaurs, lepidosaurs and mammals. Early rhynchocephalians are prominent among the diverse tetrapod assemblage, occurring in several of the fissure deposits from numerous quarries around the Bristol area. The type species of Clevosaurus, C. hudsoniSwinton, 1939, the subject of this paper, was the first Triassic terrestrial lepidosaur to be described from the region. It was discovered at Cromhall Quarry (then called Slickstones) by F.G. Hudson, initially reported by Swinton (1939) and later described by Robinson (1973) and Fraser (1988). Subsequently, there have been discoveries of Clevosaurus in other Bristol region quarries (Marshall & Whiteside, 1980; Fraser, 1994; Whiteside & Marshall, 2008; Foffa et al., 2014; Klein et al., 2015) and from South Wales (Fraser, 1994; Säilä, 2005; Whiteside & Marshall, 2008). Other species of Clevosaurus are described from Late Triassic and Early Jurassic deposits in the eastern United States, Canada, South Africa, Brazil, Luxembourg, Belgium and China and recent cladistical work has identified a Clevosauridae clade (Hsiou, de Franca & Ferigolo, 2015). Robinson (1973) gave a partial description of Clevosaurus as ‘Glevosaurus’ in a mistaken view that the name attribution (Clevum = Gloucester) had been wrongly spelled, but C or G is appropriate. The first detailed study of C. hudsoni by Fraser (1988) focussed on an extensive collection of individual bones and articulated specimens, all from Cromhall Quarry. He provided an osteological description of the cranial and postcranial elements and, based on limb bone ratios and comparisons with limb proportions in Sphenodon punctatusGray, 1842 (the only extant rhynchocephalian), reconstructed C. hudsoni as a 250-mm-long, agile, quadrupedal, lizard-like reptile. In this paper, we present two previously undescribed specimens of C. hudsoni from Cromhall Quarry, collected by the University College team led by Pamela L. Robinson in 1953–1955 (unpublished field notes of Robinson held in the NHMUK). The first specimen (NHMUK PV R36832) is an incomplete, partially articulated, but well-preserved skeleton of C. hudsoni. The second specimen, NMHUK PV R36846, is a mainly complete, articulated hind limb of the same species with part of one digit and the shaft and proximal part of the femur missing. The aim of this paper is to provide a detailed description of the two specimens, focussing on the bones that are poorly known or are unknown for C. hudsoni and clevosaurs in general. In addition to stereoscopic microscope analysis, we digitally segmented computed tomography (CT) scans using visualization software, to reveal hidden cranial and postcranial elements. This is the first time that scanning of vertebrates from the UK Triassic fissure deposits has been successfully attempted; previously, a CT scan of PantydracoGalton, Yates & Kermack, 2007, from Pant-y-ffynnon was deemed unsatisfactory (S. Chapman, personal communication, 2016). The CT surface models we generated, together with microscope examination, enable a fuller evaluation and description of C. hudsoni, yielding important new information. Institutional abbreviations and specimens From NHMUK, Natural History Museum, London, UK, C. hudsoni NHMUK PV R36832 and PV R36846; S. punctatus NHMUK 1861; Oct 1928; 65.5.43; 1972.1499; 1985 1212; 97.2.6.10. From UMZC, University Museum of Zoology, Cambridge, United Kingdom, S. punctatus R2614 and R2587. Geological setting Cromhall Quarry (Ordnance Survey Grid Reference ST 704916), the most northerly of the Late Triassic–Early Jurassic fissure localities (Fig. 1A), lies 20 km northeast of Bristol City. Cromhall Village is southwest of the quarry, with Charfield due east. Cromhall Quarry, originally known as Slickstones or Woodend Quarry (Hiscock, 2009), was first worked for building stone in the late 19th century and, although no longer active, is still registered as a working quarry. Quarrying has exposed Carboniferous limestones (and dolomites) of the Black Rock Group, Gully Oolite and Clifton Down Formation (Fig. 1B). These are underlain by limestones and shales of the Avon Group as well as the Tintern Sandstone Formation (Old Red Sandstone). The Carboniferous limestones are well jointed, with a dominant N-S joint direction (Walkden & Fraser, 1993). Walkden & Fraser (1993) suggest that most of the Cromhall Quarry fissure fills are equivalent to the Late Triassic Mercia Mudstone Group sediments but Whiteside & Marshall (2008) and Whiteside et al. (2016) regard the infillings as occurring at, and post, the initiation of the Rhaetian transgression in Penarth Group times. Figure 1. View largeDownload slide A, palaeogeographical map showing the principal Late Triassic/Early Jurassic tetrapod-bearing fissure deposits near Bristol, with current coastline superimposed (modified from Whiteside & Marshall, 2008). B, simplified geology map of the area around Cromhall Quarry. Mercia Mudstone Group is Triassic, Old Red Sandstone is Devonian and the other labelled strata are Carboniferous (geology map and legend derived from BGS website Open Science data, last accessed 29 October 2017). Figure 1. View largeDownload slide A, palaeogeographical map showing the principal Late Triassic/Early Jurassic tetrapod-bearing fissure deposits near Bristol, with current coastline superimposed (modified from Whiteside & Marshall, 2008). B, simplified geology map of the area around Cromhall Quarry. Mercia Mudstone Group is Triassic, Old Red Sandstone is Devonian and the other labelled strata are Carboniferous (geology map and legend derived from BGS website Open Science data, last accessed 29 October 2017). Various lithologies are present within the fissure deposits at Cromhall Quarry, which consist predominantly of red or green mudstones, or marls, interbedded with re-cemented limestone debris, in places dolomitized and often with reworked detrital crinoid ossicles (Fraser, 1982; Fraser & Walkden, 1983; Walkden & Fraser, 1993). White and yellow quartzose and calcareous sandstones and polymictic conglomerates also occur (Fraser, 1985; Walkden & Fraser, 1993; Whiteside et al., 2016). The shape and character of the fissures in the quarry vary, reflecting both variation in host rock morphology and mode of formation (Whiteside et al., 2016). Walkden & Fraser (1993) sub-divided the Bristol fissure systems into two broad types – those that were tectonic in origin and those that were karstic, concluding that ‘in many cases, tectonic features were exploited by karstic dissolution’ (Walkden & Fraser, 1993: 571). The fissures at Cromhall Quarry are regarded as karstic features, which initiated along joint systems in the limestone or exploited tectonic fractures and subsequently expanded by solution, in either sub-aerial or phreatic conditions (Walkden & Fraser, 1993; Whiteside & Marshall, 2008; Whiteside et al., 2016). It is clear from Robinson’s (1957) diagrams and a cave featured in Whiteside et al. (2016) that caverns, not just expanded dolines, formed a short distance below the limestone surface in substantial parts of the fissure system. Furthermore, red sediments tend to underlie the green lithologies. Cavern formation close to the limestone surface indicates a high water table, which provides the phreatic conditions required (Whiteside & Marshall, 2008; Whiteside et al., 2016). This would have most probably occurred when the sea level was high, during the Rhaetian transgression. The first mentions of the Robinson collection of reptile fossils at Slickstones (Cromhall Quarry) are Robinson, Kermack & Joysey (1952) and Robinson (1955). Our C. hudsoni fossils were collected in 1953–1955 from cavern sediments now quarried away, in locality B, Cromhall Quarry (Robinson, NHMUK unpublished notes; Fig. 2A, B). Robinson’s localities B and C (Fig. 2A) lay between sites 1 and 2 of Fraser (1988). In her notes, Robinson described the host rock as a ‘foss. red clay’ (Robinson, NHMUK unpublished notes; Fig. 2C), which contrasts to the buff-coloured matrix at fissure site 1, the probable source of the C. hudsoni specimens studied by Fraser (1988). Walkden & Fraser (1993) recorded C. hudsoni from ‘fenestral limestone’ in their site 1 and the ‘slot fissures’ which they concluded were Rhaetian. However, the location of our specimens suggests a different interpretation of their proposed chronolithological sequence, as the red matrix occurs substantially below the fenestral limestone. Walkden & Fraser (1993) suggested that these red lithologies are Norian and Robinson (1971) assigned them specifically to late Norian. However, we consider that probably all C. hudsoni fossils are Rhaetian. Recent research on the conchostracans identified as Euestheria brodieanaJones, 1862, collected by Pamela Robinson and Tom Fry from the same fissure location (below ‘B’ in Fig. 2A) and the same red marl as C. hudsoni in Cromhall Quarry, has demonstrated that the stratum is late Rhaetian, equivalent to the Cotham Member, Lilstock Formation (Morton et al., 2017). Figure 2. View largeDownload slide A, sketch of the original fissure locality, no longer in existence (modified after Robinson, 1957). B, photograph of the quarry from 1954; the ‘pendulum’ has already been removed (P.L. Robinson, unpublished notebook, NHMUK). C, copy of the original field sketch of fossil locality B by P.L. Robinson, redrawn with minor modification for clarity by the authors (after P.L. Robinson, unpublished notebook, NHMUK). Figure 2. View largeDownload slide A, sketch of the original fissure locality, no longer in existence (modified after Robinson, 1957). B, photograph of the quarry from 1954; the ‘pendulum’ has already been removed (P.L. Robinson, unpublished notebook, NHMUK). C, copy of the original field sketch of fossil locality B by P.L. Robinson, redrawn with minor modification for clarity by the authors (after P.L. Robinson, unpublished notebook, NHMUK). Early interpretations of the Cromhall Quarry fissures as ‘upland’ underground water courses or cave systems, with narrow passages forming an opening to the surface (Robinson, 1957; Halstead & Nicholl, 1971), have been superseded by evidence from the western wall of Cromhall Quarry, revealing separate openings for each of the seven fissures exposed there. Each fissure is thought to represent either a separately filled sinkhole or doline (Fraser, 1985), several dolines that coalesced, or a cavity of undetermined geometry (Walkden & Fraser, 1993), which were probably united at depth (Fraser, 1985; Walkden & Fraser, 1993). Whiteside et al. (2016) note that such caverns probably occurred in the ‘fresh/saline water-mixing zone of freshwater lenses on small limestone palaeo-islands’ (Whiteside & Marshall, 2008; Whiteside et al., 2016: 264), as observed at Tytherington Quarry. These fissures are, therefore, in marginal marine locations rather than in the uplands. The seven principal fissure sites at Cromhall Quarry have yielded at least 14 tetrapod genera, identified from tens of thousands of fossils (Fraser, 1985, 1986, 1994; Walkden & Fraser, 1993; Whiteside et al., 2016). Rhynchocephalians dominate the faunal assemblage at Cromhall Quarry; PlanocephalosaurusFraser, 1982, DiphydontosaurusWhiteside, 1986, and Clevosaurus being the most abundant (Fraser & Walkden, 1983; Fraser, 1994). Additionally, archosauromorph and procolophonid fossils are present (Fraser, 1982; Fraser & Walkden, 1983), in addition to marine and non-marine fish (Walkden & Fraser, 1993;,Fraser, 1994; Whiteside & Marshall, 2008; Whiteside et al., 2016). Mammaliamorphs are presumed absent in the Cromhall Quarry fauna (Whiteside et al., 2016). Individual bones transported from disarticulated skeletons form the majority of the fossils. Fraser (1985) attributed the high degree of rounding and polishing of some bones to attrition, as disarticulated elements were ‘swept along with terrestrial debris into watercourses and thence into the sinkhole systems’ (Fraser, 1985: 286). However, at Cromhall Quarry, complete or partially articulated specimens are recorded (e.g. Fraser, 1988; Whiteside & Marshall, 2008). The specimens of C. hudsoni under investigation here are partial skeletons, fully or mostly articulated. Given the articulation evident in many parts of the fossils, it is likely there was minimal transportation of these skeletons prior to fossilization. Whiteside et al. (2016) consider that dissociation of some of the cranial and postcranial elements of NHMUK PV R36832 occurred in situ, probably by bacterial action in subaqueous conditions. MATERIAL AND METHODS The fossil material used in this study is comprised of a partially complete skeleton (NHMUK PV R36832) and partially complete hind limb (NMHUK PV R36846) of C. hudsoni in a matrix of red marl. The specimens were extensively prepared at University College London and the NHMUK and no further excavation of the material was required. The specimens were observed under a stereoscopic microscope. We took the photographs with a Canon EOS 70D, using proprietary Canon EOS DIGITAL software, version 29.1A (Digital Photo Professional 3.14.40; EOS Utility 2.14.10). Photographs were then aligned and stacked using Adobe Photoshop CS6. The background was removed and the figures prepared using Autodesk AutoCAD LT 2016. To supplement detailed microscopic examination of the fossils, we reconstructed cranial and postcranial portions of C. hudsoni, using CT scans and visualization software. CT is increasingly being used as a non-destructive, investigative tool (Cunningham et al., 2014) and applications include reconstruction of fossil material for anatomical description and visualization of internal anatomy (e.g. Lautenschlager et al., 2014; Bever et al., 2015; Porro, Rayfield & Clack, 2015a, b). Both specimens were CT scanned at µ-VIS X-Ray Imaging Centre, Faculty of Engineering and the Environment, University of Southampton. µ-CT images of NHMUK PV R36832 (scan ref. 1077) were obtained using a custom built, dual source 225/450 kV walk-in room (Nikon Metrology, UK). µ-CT images of NMHUK PV R36846 (scan ref. 1234) were obtained using a XT H225 L micro-focus CT system (Nikon Metrology). Both scans were acquired with a micro-focus 225 kV source, fitted with a tungsten reflection target, together with a Perkin Elmer XRD 1621 detector. The scan settings for NMHUK PV R36846 were: 160 kVp, 44 µA, 177 ms exposure, 3142 projections acquired during a full 360° rotation, using an average of 64 frames per projection. The source to object distance was set at 180 mm, with a source to detector distance of 702 mm, resulting in a 51 µm reconstructed voxel resolution. We used 1-mm Cu filtration, together with the beryllium window that forms part of the target housing. The specimen was mounted within a Perspex tube and held by phenolic foam to prevent movement. A higher resolution ‘local’ scan was subsequently conducted on the skull region, using the same tube voltage and projection settings, but with the source to object distance adjusted such that the reconstructed voxel resolution was close to 30 µm. The scan settings for NMHUK PV R36846 were: 80 kVp, 93 µA, 500 ms exposure, 5001 projections acquired during a full 360° rotation using an average of 16 frames per projection. The source to object distance was set at 92 mm with a source to detector distance of 802 mm, resulting in a 23-µm reconstructed voxel resolution. There was no filtration of the hind limb scans, other than the beryllium window. We mounted the hind limb specimen within a carbon fibre reinforced plastic tube (inner diameter 2 mm, outer diameter 4 mm). Reconstructions of all three scans used a filtered back projection algorithm, implemented within CTPro and CTAgent software packages (Nikon Metrology). The CT scans were processed using Avizo 9.01 (FEI Visualisation Sciences Group) 3D visualization software. The X-ray attenuation properties of the fossil bone and rock matrix are very similar, with relatively poor contrast, so automatic density thresholding of the data was not possible. Therefore, the CT scans were interpreted using manual segmentation. Assessing the data slice by slice, interpolation was carried out across a maximum of three slices (but frequently every slice was segmented). In this manner, bones were separated from the matrix and from each other. In some portions of the CT scans, segmentation was not possible and in other instances portions of bone were segmented but, perhaps as a result of bone damage or poor contrast, only partial elements could be discerned. Where segmentation was achieved, we assigned individual bones to different fields within the segmentation editor and 3D surface models were created of each field. Anatomical descriptions of all visible bones derive from examination of the specimens, by eye and under the microscope. Surface models created from the CT scans provide additional information on non-visible parts of the bones and elements fully concealed within the matrix. Abbreviations Abbreviations used in the Figures 2–19: add, adductor; amp, amphicoelous; ant, anterior; antlat, anterolateral; ar, articular; arch, arch; ast, astragalus; at, atlas; ax, axis; b, bone; ba, basal; bi, bicondylar; bl, blade; bo, basioccipital; br, branchial; bu, bulge; c, cervical; cal, calcaneum; cap, capitellum; car, carpal; ccv, concavity; cd, coronoid; ce, centrum; cen, centrale; cir, circular; cl, clavicle; co, condyle; com, complex; con, contact; cond, condyloid; cor, coracoid; cr, crest; cvx, convexity; d, dentary; de, dental; del, deltopectoral; di, distal; do, dorsal; dp, depression; E, East; ect, ectopterygoid; ectp, ectepicondyle (ectepicondylar); entp, entepicondyle (entepicondylar); epi, epipterygoid; exo, exoccipital; ext, extension; f, frontal; fa, facet; fac, facial; fe, femur; fen, fenestra; fi, fissure; fib, fibula; fl, flange; fo, foramen; fos, fossa; fu, furrow; fx, flexure; gas, gastralium; gr, groove; hu, humerus; hy, hyoid; hyp, hypapophysis; i, ischium; ic, intercentrum; icl, interclavicle; j, jugal; jt, joint; l, left; lat, lateral; m, medial; maf, magnum foramen; md, mandibular; me, metotic; metac, metacarpal; metat, metatarsal; mk, meckelian; mn, mental; mx, maxilla; n, nasal; ne, neural; no, notch; o, orbit; oc, occipital; od, odontode; op, opisthotic; oss, ossified; p, parietal; pal, palatine; par, paroccipital; pf, postfrontal; phx, phalanx; pif, pineal foramen; pl, plate; po, postorbital; pos, posterior; poslat, posterolateral; poz, postzygapophysis; pr, process; prf, prefrontal; pro, prootic; prz, prezygapophysis; pt, pterygoid; px, proximal; q, quadrate; r, right; ra, radius; rad, radiale; rar, retroarticular; ri, ridge; rw, row; s, secondary; sc, scapulocoracoid; sca, scapula; sco, sclerotic ossicle; sk, skirt; so, supraoccipital; sp, stapes; spi, spine; spl, splint; sq, squamosal; st, supratemporal; sty, styloid; su, suture; sub, sub; sul, sulcus; suplab, supralabial; sur, surangular; syn, synapophysis; tab, table; tar, tarsal; th, thickening; thy, thyroid; tib, tibia; to, tooth; tra, transverse; tro, trochlea; tu, tubercle; ug, ungual; ul, ulna; uln, ulnare; v, vertebra; ve, ventral; w, wear. SYSTEMATIC PALAEONTOLOGY Class: Reptilia Subclass: Diapsida Superorder: Lepidosauria Duméril & Bibron, 1839 (sensu Evans, 1984) Order: Rhynchocephalia Gunther, 1867 (sensu Gauthier, Estes & de Queiroz, 1988) Suborder: Sphenodontia Williston, 1925 (sensu Benton, 1985) Family: Clevosauridae Bonaparte & Sues, 2006 (sensu Hsiou et al., 2015) Genus: Clevosaurus Swinton, 1939 Type species: Clevosaurus hudsoni Swinton, 1939 Included species: Clevosaurus minor Fraser, 1988; C. latidens* Fraser, 1993; C. bairdi Sues, Shubin & Olsen, 1994; C. wangi Wu, 1994, C. mcgilli Wu, 1994 and C. petilus Wu, 1994; C. convallis Säilä, 2005; C. brasiliensis Bonaparte & Sues, 2006 and C. sectumsemper Klein et al., 2015. Remarks: The most recent cladistic analysis by Hsiou et al. (2015) resolved a Clevosauridae clade with the following apomorphic features: antorbital region forming one-quarter of the skull length (reversed to between one-third to one-quarter in C. brasiliensis, C. wangi and C. petilus); a narrow and elongated dorsal process of the jugal; palatine teeth forming a single row, plus one isolated tooth. Hsiou et al. (2015) define the clade as ‘all taxa more closely related to Clevosaurus than to Sphenodon’ (Hsiou et al., 2015: 4). *Clevosaurus latidens is positioned outside of Clevosauridae in this analysis. Characters that occur in clevosaurs, but are not restricted to the genus include: a lateral forked flange of the premaxilla preventing contact between the maxilla and the external naris [horizontal posterior flange not present in C. convallis (Säilä, 2005)]; a dorsally expanded lateral process of the premaxilla; suborbital fenestra bounded solely by the ectopterygoid and palatine; a high, steeply inclined coronoid process of the dentary; flanged teeth; a broad maxillary-jugal contact (Säilä, 2005; Bonaparte & Sues, 2006; Jones, 2006; Hsiou et al., 2015; Klein et al., 2015). Diagnosis: Based on Swinton (1939), Robinson (1973) and Fraser (1988), specimen NHMUK PV R36832 can be diagnosed as C. hudsoni. The diagnosis is based on the following principal features, identified in the specimen: Acrodont dentition. The maxilla bears four large additional teeth that increase in size posteriorly; the teeth are conical in form and have postero-lingual flanges. Three smaller conical teeth occur posteriorly to that additional set on the maxilla. The dentary has four large, conical, additional teeth which increase in size posteriorly; these additional teeth have anterolateral flanges. Teeth are evident on the pterygoid and palatine. Two rows of teeth are present on the pterygoid. A single lateral row of large teeth is present on the palatine but because of incomplete preservation of the medial region of the palatine, the small isolated tooth medially offset from the lateral row, typical of clevosaurs, is not recorded. Incomplete lower temporal bar. There is a gap between the posterior process of the jugal and the quadrate on this specimen. Fraser (1988) notes that in some specimens of C. hudsoni, weak contact is made between the jugal and quadratojugal. There is no contact on this specimen. A flattened, plate-like quadrate present. Large pineal foramen. Postorbital triangular in shape. Supratemporal present. Specimen NMHUK PV R36846 is designated the hind limb of C. hudsoni based on its great similarity to C. hudsoni, figured and described by Fraser (1988: fig. 35). Identifiable features on this specimen include: Size of tibia falls within size range of elements measured by Fraser (1988). Hooked fifth metatarsal. Fifth tarsal fused to fifth metatarsal to form a single tarsometatarsal. A single astragalocalcaneum, formed by fusion of ankle bones, is considered diagnostic of clevosaurs but on this specimen the elements appear to be separate (see description of NMHUK PV R36846 below). RESULTS Anatomical description of NHMUK PV R36832 NHMUK PV R36832 (Fig. 3A, B) preserves many of the skull bones and some postcranial elements, many in close association and some partly articulated. Most of the bones on the left side of the skull have been preserved; frontal, parietal, prefrontal, jugal, postfrontal, postorbital, quadrate are present, together with probable pterygoid, palatine and ectopterygoid (Fig. 3C–F; Supporting Information, Appendices S1, S2). However, excepting the frontal, parietal, postfrontal and partial supratemporal, many bones on the right side are absent. The premaxillae are also missing. The (probable) left nasal is broken, as are the anterior regions of both the dentary and maxilla. The articular complex is present, but there is no evidence of an unfused angular, which projects anteriorly, ventral to the dentary in the drawing by Fraser (1988). Bones of the braincase are visible out of position, between the left parietal and left postorbital (Fig. 3C, D; Supporting Information, Appendices S1, S2). Sclerotic ossicles are preserved in the left orbit (Fig. 3E, F). Figure 3. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. A, entire specimen. B, diagrammatic representation of the specimen with key showing location of views in (C–F) and in Figure 4. Skull bones in (C, D) dorsolateral view and (E, F) left lateral view (surface models are in artificial colour in all figures). Figure 3. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. A, entire specimen. B, diagrammatic representation of the specimen with key showing location of views in (C–F) and in Figure 4. Skull bones in (C, D) dorsolateral view and (E, F) left lateral view (surface models are in artificial colour in all figures). The specimen also comprises articulated cervical vertebrae (Fig. 4; Supporting Information, Appendices S3, S4), including the atlas/axis complex, together with the left scapulocoracoid, humerus, radius, ulna and some carpals and metacarpals. There is additionally a probable dorsal vertebra, a right ischium (Fig. 3A, B), gastralia, chevrons, ribs and numerous partially broken bones and fragments, some of which cannot be accurately identified. We describe the skeleton below and have provided longer accounts of poorly known or unknown bones. Figure 4. View largeDownload slide Postcranial bones of Clevosaurus hudsoni specimen NHMUK PV R36832. (A) photograph and (B) surface model (refer to Fig. 3B for location on specimen). Figure 4. View largeDownload slide Postcranial bones of Clevosaurus hudsoni specimen NHMUK PV R36832. (A) photograph and (B) surface model (refer to Fig. 3B for location on specimen). Tooth bearing marginal bones Left maxilla: The anterior region is missing, but the mid- and posterior region of the left maxilla is distinctive of C. hudsoni with four prominent flanged acrodont additional (sensuRobinson, 1976) teeth, increasing in size posteriorly, followed by three simple uniform subconical teeth (a feature of sphenodontians). The bone has a central depression and anteriorly there are three supralabial foramina, where the superior alveolar nerve and maxillary artery probably exited. The anterodorsal nasal facet and anterior edge of the facial process is obscured by both matrix and a (probable) broken portion of the maxilla. The long suborbital margin downturns sharply posteriorly (Fig. 5A, B) and the maxilla extends further with a lateral projection, terminating at a broad tip. Notches and irregularities along this margin are likely to be artefacts of fossil preservation or preparation. The CT scan reveals facets on the medial surface for each of the adjacent elements: jugal, prefrontal, palatine and ectopterygoid (Fig. 5B). Figure 5. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left maxilla in (A) lateral and (B) medial views. Parietals in (C) dorsal and (D) ventral views. Frontals in (E) dorsal and (F) ventral views. Figure 5. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left maxilla in (A) lateral and (B) medial views. Parietals in (C) dorsal and (D) ventral views. Frontals in (E) dorsal and (F) ventral views. The dentition on NHMUK PV R36832 is fully acrodont, characteristic of the Sphenodontia, with the teeth fused centrally about the crown of the jaw. Hatchling dentition is absent on the specimen, possibly due to incomplete preservation, but perhaps indicating an adult individual (Robinson, 1973; Fraser, 1988). The first additional tooth is small, with a rounded tip and the other three teeth are broadly conical with flattened tips. The enamel of these teeth has vertical striations. A ridge of secondary bone is present above the tooth row (Fig. 5A), but it is unlike some (aged?) clevosaur individuals where the teeth are worn and obliterated by the secondary bone forming a single cutting surface (Fraser, 1988). A noticeable posteromedial flange is present on the second, third and fourth teeth. In addition, there is a ridge on the medial surface possibly indicating the dorsal extent of occlusion of the dentary dentition but wear facets are not sufficiently clear in the CT scan. Dermal roofing bones The paired parietals (surrounding a large pineal foramen) and frontals (Fig. 5C–F) are typical of C. hudsoni. The parietals contacted the frontals in an asymmetric interdigitating suture. Also asymmetrical is the complex medial suture between the frontals. The frontals are thickened ventrally where they form the upper margins of the orbits. A supratemporal, a bone that was previously noted in C. hudsoni by Robinson (1973), is identified by its contact with the right parietal and by facets suggesting a tongue-in-groove connection (Fig. 5C). The anterior of both frontals is missing, as is therefore any contact with the nasals. We have, however, identified a probable left nasal that has a W-shaped frontal facet (Fig. 6A, B), similar to that described for C. wangi, from China (Wu, 1994) but not recorded previously in C. hudsoni. The microscope photograph and CT scan reveal a prefrontal facet on the nasal. Figure 6. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left nasal in (A) dorsolateral and (B) ventromedial views. C, right postfrontal in dorsal view. D, articulation between left postfrontal and left postorbital in anterolateral view. Left postorbital in (E) lateral and (F) medial views. Left prefrontal in (G) dorsolateral and (H) posterolateral views. Left jugal in (I) lateral and (J) medial views. Figure 6. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left nasal in (A) dorsolateral and (B) ventromedial views. C, right postfrontal in dorsal view. D, articulation between left postfrontal and left postorbital in anterolateral view. Left postorbital in (E) lateral and (F) medial views. Left prefrontal in (G) dorsolateral and (H) posterolateral views. Left jugal in (I) lateral and (J) medial views. Circumorbital bones The right postfrontal is complete, in life position, and exhibits the facets and overlap with the frontal and the parietal (Fig. 6C); the facet that underlay the postorbital is also clear. The left postfrontal, while remaining in substantial contact with the extensive overlapping facet of the postorbital (Fig. 6D), has separated and displaced ventrally from the roofing elements. The left postorbital is large, triangular and is thickened at the orbital margin (Fig. 6E). The CT images show the overlapping medial facets on the bone for the postfrontal dorsally, jugal anteroventrally and squamosal posteroventrally (Fig. 6F). The left prefrontal has rotated out of position so that the contact with the nasal and frontal is not preserved. However, part of the maxillary facet is visible, as is the posterolateral bulge (Fig. 6G). The rugose keel of the orbital margin is discernible on the CT scan (Fig. 6H). The triradiate left jugal (Fig. 6I), displays a rounded posterior process, with a slight lateral bulge, but no evidence of a facet for quadratojugal or squamosal. This confirms the description of Robinson (1973) that at least some adult specimens of C. hudsoni do not have a complete lower temporal bar. Images from the CT scan reveal maxillary and postorbital facets on the lateral surface (Fig. 6J), which underlay the respective bones. A facet for the ectopterygoid is present on the medial surface of the anterior process. Palatoquadrate and cheek bones The left quadrate comprises a single broad, blade-like bone with a well-defined, posterior keel, dorsoventrally aligned (Fig. 7A, B). This keel thickens ventrally to form a pronounced anterolaterally directed condyle. A thin, slightly convex, blade-like plate projects from the keel anteromedially. This plate includes the margin of a probable quadratojugal foramen (sensuRobinson, 1973), as it disappears into the matrix. In contrast to the description of a very thin quadratojugal fused to the quadrate by Fraser (1988), there is no evidence of a separate quadratojugal or a clear suture between the bones. Therefore, it is unclear whether the condyle is formed from the quadrate alone, as suggested by Fraser (1988), or includes a quadratojugal, as described by Robinson (1973). A pterygoid facet is evident on the lateral surface of this plate and a squamosal facet is just discernible dorsally, although this is obscured by matrix and is not clear on the CT scan. The quadrate does not contact the posterior process of the jugal. Figure 7. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left quadrate in (A) lateral and (B) medial views. Palate region in (C), (D) ventral view (refer to Fig. 3A for location of view). E, left lateral view of specimen between maxilla and dentary, showing bones out of position (refer to Fig. 3A for location of view). Possible left pterygoid in (F), (G) lateral view. Figure 7. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left quadrate in (A) lateral and (B) medial views. Palate region in (C), (D) ventral view (refer to Fig. 3A for location of view). E, left lateral view of specimen between maxilla and dentary, showing bones out of position (refer to Fig. 3A for location of view). Possible left pterygoid in (F), (G) lateral view. Palatal complex The underside of specimen NHMUK PV R36832 reveals a ventral view of the palatal area (Fig. 7C) which, although damaged and infilled in part with matrix, does show some details. There are two rows of small teeth running approximately parallel to each other, with the medial set extending more anteriorly indicating that they belong to the right pterygoid. There appears to be at least seven teeth in the lateral set and at least eight in the medial set that are discernible on the CT scan (Fig. 7D; Supporting Information, Appendix S2). The pterygoid plate projects anteriorly towards a fragment of bone that may represent a point of contact with the right vomer. There is no sign of the vomers but they are thin, fragile elements (Fraser, 1988) and thus rarely preserved. The posterior process of the right pterygoid is tentatively identified, based on CT evidence of connection with the anterior pterygoid plate. A few small, conical teeth, revealed by the CT scan (Fig. 7D) in a medial position to the right pterygoid probably represent teeth of the left pterygoid. A fragment of the left pterygoid plate appears to be preserved on the specimen, but it is difficult to separate bone from matrix in this part of the CT scan. Two stout conical teeth, preserved on a broken but robust bone positioned laterally to the anterior plate of the right pterygoid (Fig. 7D), are thought to belong to the right palatine. Part of the left palatine with several robust teeth is preserved out of position, embedded in matrix between the left maxilla and left dentary (Fig. 7E; Supporting Information, Appendix S1). Anterior to this, a slender bone with medially flattened teeth possibly represents a broken portion of the anterior left maxilla. In a more posterior position between the maxilla and dentary, there is a flexed, curved solid bone (Fig. 7E), which we identify as an ectopterygoid. There is uncertainty about the diagnosis of the bone pictured in Figure 7F, which is evidently out of alignment with the surrounding cranial elements. We considered the possibility that this may be part of a left squamosal. However, based on its position on the specimen and on information from the CT scan, we consider it more likely to be the quadrate flange of the pterygoid. It is comparable to the pterygoid of Sphenodon, illustrated by Jones et al. (2011: fig. 36), with quadrate facets on the plate-like element of the bone and the projection (or ramus), possibly articulating with the squamosal. The CT scan reveals two processes within the matrix (Fig. 7G). These projections are likely to be flanges of the pterygoid for articulation with the epipterygoid (posteriorly) and ectopterygoid (anteriorly). Braincase bones A good portion of the supraoccipital, which forms the roof to the braincase (Fig. 8A–D), is visible. There is a central shallow dorsal crest running anteroposteriorly. Posteriorly, the supraoccipital defines the dorsal surface of the foramen magnum; the anteriormost region is embedded in matrix and is not discernible on the CT scan. Contacts with the other braincase elements are generally not preserved but the supraoccipital appears to be fused with the left prootic and there is a facet for the right exoccipital. The right prootic is poorly preserved, but a depression lying posterolaterally to the right side of the supraoccipital possibly marks the contact between these two bones. Figure 8. View largeDownload slide Photograph and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Braincase bones in (A, B) posterodorsal, (C) right lateral and (D) anteroventral views. Figure 8. View largeDownload slide Photograph and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Braincase bones in (A, B) posterodorsal, (C) right lateral and (D) anteroventral views. The crescent-shaped occipital condyle, posteriorly positioned on the basioccipital, forms the ventral surface of the magnum foramen (Fig. 8A, B). Posterolateral projections of the occipital condyle meet the left and right exoccipitals, which form the lateral walls of the magnum foramen. Although a possible suture is present, they appear fused on the right-hand side, a condition considered rare in Clevosaurus by Fraser (1988). However, an overlap and probable suture are visible at the basioccipital contact with the left exoccipital. This feature is also observed in a specimen of Sphenodon (ref. NHMUK 1861). Lateral facets are present on the exoccipitals, which probably contacted the opisthotic in life but are separated by a narrow matrix infill on the specimen. Similarly, as a result of post mortem rotational movement, the contact between the exoccipitals and the supraoccipital is not in life position. However, a convolute margin is present on the dorsal surface of the right exoccipital, matching one on the supraoccipital (Fig. 8A). Anterior to the occipital condyle, the left and right basal tubercles (Fig. 8B) are fused to, and project posterolaterally from, the basioccipital. The metotic fissure, a prominent depression, lies at the base of the tubercles and there is a probable parasphenoid anteriorly, but its identification cannot be confirmed because little of the bone is visible on the surface and it is unidentifiable on the CT scan. The left opisthotic, forming the posterodorsal region of the braincase and displaying the paroccipital process, is well preserved. It is probable that the left opisthotic and anteriorly positioned prootic are fused. The paroccipital process projects posterolaterally, separated by a ridge from the anteromedial part of the left opisthotic (Fig. 8A). Lateral to this ridge and posterior to the opisthotic, there is a notable circular depression with two lateral processes, probably a muscle attachment site. A similar feature is visible on the opisthotic of Sphenodon (ref NHMUK 97.2.6.10; 1972.1499). Lateral to this, the left paroccipital process projects in a curved sweep anterolaterally, with a concave ventral surface. Medially, a tapering flange projects from the ventral surface of the paroccipital process; this is the ‘descending process’ of Fraser (1988: 140). This flange meets the basioccipital anterior to the left basal tubercle, at the metotic fissure; the medial surface probably formed a border to the fenestra ovalis. The right opisthotic, preserved without a paroccipital process, is only partially visible and is just discernible on the CT scan (Fig. 8B). The scan reveals the curved anterodorsal morphology of the right prootic (Fig. 8C) but, although the right opisthotic and prootic appear fused, the contacts between the prootic and surrounding bones cannot be confirmed. Lower Jaw Left dentary: The left dentary, like the maxilla, is missing the anteriormost region. The dentary is a robust bone (Fig. 9A–D), with a pronounced lateral ridge marking the extensive development of secondary bone, indicative of an adult individual (Fraser, 1988). Below this ridge, there are three mental foramina (Fig. 9A), where the inferior alveolar nerve exited, as described for other clevosaurs (Sues et al., 1994). The dentary dentition is fully acrodont, including four additional teeth (sensuRobinson, 1976) which increase in size posteriorly. Each of the four teeth has a shallow anterolateral flange, extending basally; this is most pronounced on the second and third teeth. A steeper, shorter posterior flange is present on the first and second teeth. The first tooth is considerably smaller than the others and the tip is quite rounded. The base of the tooth is well defined and there are prominent posterior wear facets caused by occlusion on this and the second tooth (Fig. 9B), that match exactly in size with the maxillary dentition. The second tooth is conical in form. The shallowly sloping anterior flange extends to meet the posterior flange of the first tooth, and its base forms a prominent ridge. The third tooth has a rounded base and a large, broad, rounded tip. The shallow anterior flange curves dorsally before meeting the posterior flange of the second tooth. A posterior wear facet is visible. The tip of the fourth tooth is obscured by matrix, but the CT scan reveals that it is clearly the largest (Fig. 9C). It has a rounded appearance, with a rounded shallow base and a broad posterior wear facet. The flange is notably reduced. The CT scan shows wear facets on the lateral surface of the dentary, posterior to the acrodont tooth row, caused by occlusion with the posterior three uniform teeth of the maxilla. Figure 9. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left dentary in (A, C) lateral and (D) medial views. B, partial left dentary in lateral view showing wear facets. E, left articular complex in dorsal view. F, left surangular in lateral view. G, left articular in dorsal view. H, right articular in lateral view. Figure 9. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left dentary in (A, C) lateral and (D) medial views. B, partial left dentary in lateral view showing wear facets. E, left articular complex in dorsal view. F, left surangular in lateral view. G, left articular in dorsal view. H, right articular in lateral view. Wear facets are also evident on the medial surface of the left dentary (Fig. 9D), indicative of the occluding palatine teeth. The meckelian groove is present on the medial surface, below the sub-dental ridge, and is obscured anteriorly by bone ventral to the first additional tooth. Although the anterior tip is missing, there are no indications of hatchling and successive teeth such as are found in some (smaller) specimens of C. hudsoni (Fraser, 1988). The full extent of the depth of the dentary is clear on the CT scan (Fig. 9C, D). The dorsal region of the coronoid process curves medially and there is a slight lateral thickening at its base. We find no evidence from the CT scan of a separate coronoid bone. The mandibular foramen, which accommodated the inferior alveolar nerve, is a pronounced feature on the posterior base of the coronoid process, with an opposing foramen visible on the surangular portion of the articular complex (Fig. 9E, F), which has become detached from the dentary. The surangular forms the lateral surface dorsal to the posterior process of the dentary. Posterior part of lower jaw: The left articular complex of articular, prearticular and surangular (Fig. 9E) thickens posteriorly, terminating in a retroarticular process. The curved articular complex is partly fused. The anterior portion of the surangular is fused to the articular; the mandibular foramen is present at the boundary between the two bones. Posterior to the foramen, a suture is discernible between the ventral surface of the surangular and the dorsal surface of the articular (Fig. 9F). The left prearticular, which forms the ventral surface of the articular complex, appears to have dissociated from the left articular, with matrix filling the gap (Fig. 9E). Posterior to the surangular, the articular complex curves initially posterolaterally and then in a broad sweep medially. The articular forms a medially directed ridge, which passes up posterodorsally to the retroarticular process, the tip of which is obscured by matrix and may be incomplete. Two well-defined sulci are present, one lateral and the other medial (Fig. 9E, G). These articular fossae, separated by a well-defined ridge, articulated with the condyle of the quadrate. The right articular is revealed by CT scan (Supporting Information, Appendices S1, S2), wholly embedded in matrix. The surface model, generated from the scan data (Fig. 9H), exhibits the distinctive retroarticular process and the articular fossae, separated by a distinctive ridge. Additional cranial bones hidden within the matrix of NHMUK PV R36832 Several bones embedded in the matrix of NHMUK PV R36832 have been revealed by the CT scans (Supporting Information, Appendices S1, S2). None of the bones articulate with those around them and all are thought to be out of position, probably the result of sedimentary slumping. A number of slender rod-like bones are present, which are tentatively identified based on their morphology (Fig. 10A, B) and relative positions in the rock matrix. We consider a slender bone, which has a twist in the shaft and evidence of expanded ends, to be an epipterygoid. A number of rod-like bones, which are flattened at one end and taper to a circular termination at the other, are probably branchial bones of the of the hyoid arch. These bones are usually cartilaginous, but may be ossified in Sphenodon (Romer, 1956). Two probable stapes are also present. Figure 10. View largeDownload slide Surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. A, anterolateral view of skull showing surfaces of slender bones hidden within the matrix (large cranial elements shown for relative position). B, posteroventral view of (A). Figure 10. View largeDownload slide Surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. A, anterolateral view of skull showing surfaces of slender bones hidden within the matrix (large cranial elements shown for relative position). B, posteroventral view of (A). Postcranial bones Left humerus: The humerus is a large, relatively well-preserved bone (Fig. 11A), although there is damage to the surface at both proximal and distal ends. Only the anterior surface of the bone is exposed, but the CT scan reveals much of the complete fossil (Fig. 11B, C) including the ectepicondyle and entepicondyle. Also revealed is the ectepicondylar foramen, through which passed the radial nerve and blood vessels (Fraser & Walkden, 1984). Three faces are formed at the proximal end, separated by distinct ridges, which, together with a tubercle on the ventral edge, would have formed points of attachment for the pectoral muscles (Fraser, 1988). The axial twist along the shaft results in a 90° angle between the proximal and distal heads. Figure 11. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left humerus in (A) anterodorsal, (B) dorsal and (C) ventral views. Left ulna in (D) dorsolateral, (E) anterior and (F) posterior views. G, left distal forelimb bones in lateral view. H, isolated distal left forelimb bones in lateral view. Figure 11. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left humerus in (A) anterodorsal, (B) dorsal and (C) ventral views. Left ulna in (D) dorsolateral, (E) anterior and (F) posterior views. G, left distal forelimb bones in lateral view. H, isolated distal left forelimb bones in lateral view. Left ulna: The ulna is robust, with expanded proximal and distal heads and a slender shaft (Fig. 11D–F). Although the proximal head does not articulate with the humerus, it is in close proximity to where the trochlea would lie. There is no evidence of an olecranon, probably because of breakage. A broad shallow groove just below the proximal head is identified as the distal end of the trochlear notch. The distal head of the ulna is anteroposteriorly flattened and ventrally there is a pronounced coronoid process for muscle attachment. Left carpus and manus: Several distal elements of the forelimb are present on the specimen and, although some lie in close proximity to the distal head of the radius, they are separated and, in part, quite displaced (Fig. 11G). The elements include three carpals and three metacarpals/phalanges. An ungual phalanx displays a concavity in the proximal head, which articulated with the preceding element. One of the preserved carpals, notably thin, flattened and angular, is identified as the radiale. A second smaller, more rounded, faceted bone is probably a centrale and a third bone, multifaceted, angular and solid may be the ulnare but, given its small size, it is considered more likely to be the fourth distal carpal. Although the bones were compared with those of Sphenodon (ref NHMUK Oct 1928), ascribing the metacarpals/phalanges and ungual phalanx to a particular digit is difficult. However, the position and size of the metacarpals indicates they are from either digit one or two. Two further manus bones (Fig. 11H) lie some distance from the left forelimb epipodials, close to the left dentary. The larger of the two bones is likely to be a metacarpal, rather than a metatarsal, based on its preservation near the skull. The bone has a distinct fossa at the proximal end, probably the site of muscle or tendon attachment. A distal or ungual phalanx, separate from the metacarpal/tarsal is probably from a different digit. Right ischium: The ischium is the only part of the pelvic girdle that is found on the surface of the specimen. By comparison with Sphenodon (ref. NHMUK Oct 1928), the bone is likely to be the right ischium, with the lateral view exposed on the surface of the rock (Fig. 12A). It is a thin but broad plate-like bone, roughly hatchet-shaped, that narrows to a thickened neck and flares anterodorsally to where it would have contacted the pubis and ilium. The posteroventral margin of the ischium is arc-like, more curved than the rhomboid form illustrated by Fraser (1988: fig. 31). There is a thickened lateral keel terminating in a posterior process, with a pronounced tubercle positioned immediately posterior to the neck, for attachment of tail musculature ligaments and tendons (Fraser, 1988). The CT scan shows the medial side of the ischium to be thickened below the neck (Fig. 12B). A pubis and a fragment of bone, possibly the ilium, are also revealed by the scan, but the bones do not lie in their life positions. Full segmentation of the pubis was not achieved, but it is clearly the anteroventral element of the pelvis. The short contact surface with the ischium is clear, as is part of the margin of both acetabulum and thyroid fenestra. The ventral portion of the bone is not detectable. The assigned ilium is based on its proximity to the other two bones. Other skeletal elements Some bones of the C. hudsoni pectoral girdle have not been previously described in detail, because the material available was incomplete or not sufficiently clear. For comparative purposes, we reference part of the postcranial skeleton for Planocephalosaurus robinsonaeFraser, 1982, by Fraser & Walkden (1984) and specimens of Sphenodon from the NHMUK. The interclavicle is a T-shaped bone, with the clavicle articulation occurring on the anteroventral surface (Fraser, 1988). Our interclavicle specimen, although partially embedded in matrix (Fig. 12C), is similar, but much larger than the interclavicle of P. robinsonae (Fraser & Walkden, 1984: fig. 12a; plate 53, fig. 16). The clavicle facet and dorsal curvature of the bone are visible at the proximal end. The anterior crossbar is partly exposed but the T-junction between anterior and posterior elements of the bone is hidden beneath the left clavicle in matrix. Both clavicles, slender bones which curved dorsally, have become detached (Fig. 12C). The left clavicle has a dorsal notch where it articulated with the scapulocoracoid and the right bone displays the interclavicle facet. Figure 12. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. A, right ischium in lateral view. B, pelvic girdle bones as preserved in the specimen, out of position (right ischium in medial view). C, postcranial bones at or near the pectoral region, left lateral view. Figure 12. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. A, right ischium in lateral view. B, pelvic girdle bones as preserved in the specimen, out of position (right ischium in medial view). C, postcranial bones at or near the pectoral region, left lateral view. There are numerous bone fragments on the surface and, while some of these may be fragments of ribs or other elements, we describe only those that are recognizable. A few scattered ribs on NHMUK PV R36832 (Fig. 12C) include several short bicipital elements, closely associated with the anterior cervical vertebrae and two larger sub-bicipital ribs, lying close to the posterior cervicals. Two long, narrow, rod-like dorsal ribs are positioned in a posterior position to the humerus and scapulocoracoid. Bones previously poorly known or unknown Sclerotic ossicles: Sclerotic ossicles are found in the matrix of the left orbit (Fig. 13). The bones are very thin and are poorly defined, some being angular and others more rounded. They are flattened with a low, slightly concave profile. In life, we expect the sclerotic plates to have overlapped in the manner of Sphenodon (Underwood, 1970), where some ossicles are overlapped by either or both neighbours and others overlap both neighbours. A complete overlap is not observed on NHMUK PV R36832, but post mortem slippage of bones has undoubtedly occurred. These bones have not been described previously for C. hudsoni, but Sues et al. (1994: 331) suggest that two ‘featureless bony platelets’ in the left orbit of a C. bairdi fossil from the McCoy Brook Formation, Nova Scotia, are sclerotic ossicles. Figure 13. View largeDownload slide Photographs of Clevosaurus hudsoni specimen NHMUK PV R36832. A, partial left lateral view of the skull, showing the position of sclerotic ossicles in the orbital area. B, enlarged view of preserved sclerotic ossicles. Figure 13. View largeDownload slide Photographs of Clevosaurus hudsoni specimen NHMUK PV R36832. A, partial left lateral view of the skull, showing the position of sclerotic ossicles in the orbital area. B, enlarged view of preserved sclerotic ossicles. Sphenodon is known to have 16 (at most seventeen) sclerotic ossicles, which overlap corneally, have a distinct waist orbitally and form a flattish scleral ring (Underwood, 1970). Transposing simple ratios of ossicle size and ring circumference in Sphenodon to NHMUK PV R36832 produces a similar number for C. hudsoni but, admittedly, there are few plates preserved, with little articulation, so this is a broad estimate. Cervical vertebrae: Without a complete, articulated specimen, Fraser (1988) was unable to be sure of the number of cervical vertebrae found in C. hudsoni but suggested that there are eight, based on comparisons with P. robinsonae (estimated from dissociated specimens; Fraser & Walkden, 1984), and Sphenodon (Romer, 1956; Hoffstetter & Gasc, 1969). However, Hoffstetter & Gasc (1969) noted that the Jurassic rhynchocephalian genera have seven cervical vertebrae. The CT scan of our specimen reveals greater detail than previously described (Fig. 14A–H), confirming that eight cervical vertebrae are present in C. hudsoni. The elements that comprise cervical vertebrae 1 and 2, the atlas/axis complex (Fig. 14B, D–H; Supporting Information, Appendix S3), while retaining their essential symmetry, have become slightly displaced. The atlas comprises left and right neural arches, and the first two intercentra, which lie either side of the centrum (odontoid). The elements comprising the atlas are separate, but remain in close proximity. It is probably that, as is the case for at least some Sphenodon individuals (Romer, 1956; Hoffstetter & Gasc, 1969), the atlas bones may have been separate in life. Romer (1956) records the atlas centrum of the adult Sphenodon fusing with the following intercentrum and partially fusing with the axis centrum. Hoffstetter & Gasc (1969), however, suggest that the second intercentrum is fused to the axis centrum, which is lengthened anteriorly by fusion with the atlas centrum, a relationship also described for Sphenodon by Jones et al. (2009). NHMUK PV R36832 appears to align most closely with the observation of Romer (1956), in that the second intercentrum is fused to the centrum of the atlas (Fig. 14D); neither element, however, is fused to the axis. The axis is distinguished from posterior cervical vertebrae by the neural spine, which is enlarged for head support (Romer, 1956). Additional intercentra are present between the axis and third cervical vertebra, and between the third and fourth vertebrae. Romer (1956) describes a ventral hypapophysis on the intercentra of Sphenodon, which Hoffstetter & Gasc (1969) consider to be poorly differentiated on the cervical vertebrae of rhynchocephalians. The intercentra on this specimen are ventrally convex and, while not very well defined, appear to have a rudimentary hypapophysis (Fig. 14D). A proatlas was not identified on NHMUK PV36832. Figure 14. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. A, cervical vertebrae in lateral view. B, disarticulated atlas bones in lateral view. C, cervical vertebrae in lateral view. D, atlas and axis in lateral view. Atlas and axis in (E) anterior, (F) dorsal, (G) posterior and (H) ventral views. Figure 14. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. A, cervical vertebrae in lateral view. B, disarticulated atlas bones in lateral view. C, cervical vertebrae in lateral view. D, atlas and axis in lateral view. Atlas and axis in (E) anterior, (F) dorsal, (G) posterior and (H) ventral views. Ten vertebrae were identified on the CT scan (Fig. 15A). There is a notable difference in morphology between those anteriorly positioned and the more posterior vertebrae. Vertebrae 1–8 are cervical vertebrae whereas 9 and 10 are dorsal vertebrae. The CT scan enables the vertebrae to be viewed in all orientations (Supporting Information, Appendix S4), revealing the complete morphology. The neural arches have been preserved intact and fully attached to the centrum. The cervical vertebrae are distinctively hourglass shaped, notochordal and amphicoelous (Fig. 15B–F). As noted by Fraser (1988), the diapophysis and parapophysis have fused forming a diagonal anteroventral-posterodorsal synapophysis. The two dorsal vertebrae, each with an amphicoelous notochordal centrum and a synapophysis, are similar in form to the cervical vertebrae but are wider and have a shorter neural spine (Fig. 15G–J). Another vertebra, separate from the articulated region (refer to Fig. 3A, B), is identified as dorsal because of its position relative to the rest of the skeleton. The lack of a ‘transverse process’ indicates that it is not from the caudal region. Figure 15. View largeDownload slide Photograph and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. A, cervical and dorsal vertebrae in posterodorsal view. B, fourth cervical vertebra in lateral view. Third cervical vertebra in (C) anterior, (D) dorsal, (E) lateral and (F) posterior views. Second dorsal vertebra in (G) anterior, (H) dorsal, (I) lateral and (J) posterior views. Figure 15. View largeDownload slide Photograph and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. A, cervical and dorsal vertebrae in posterodorsal view. B, fourth cervical vertebra in lateral view. Third cervical vertebra in (C) anterior, (D) dorsal, (E) lateral and (F) posterior views. Second dorsal vertebra in (G) anterior, (H) dorsal, (I) lateral and (J) posterior views. Scapulocoracoid: The scapulocoracoid is a single fused bone, comprised of the dorsal scapula blade and ventral coracoid (Fig. 16A–C). There is no evidence, from either the actual specimen or CT scan, of a suture between the two bones. The tubercle, for attachment of the triceps tendon (Fraser, 1988), is present at the end of the anterodorsal-posteroventral ridge. The glenoid is a prominent feature on the scapulocoracoid, flanked by two ridge-like processes, with the coracoid foramen positioned anteriorly. Notably, the coracoid foramen in this specimen is positioned centrally between the processes (Fig. 16A), directly behind the glenoid, unlike that illustrated by Fraser (1988: fig. 28), where the foramen is offset dorsally. Figure 16. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left scapulocoracoid in (A, B) lateral and (C) medial views. Left radius in (D) dorsolateral, (E) anterior and (F) posterior views. Figure 16. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36832. Left scapulocoracoid in (A, B) lateral and (C) medial views. Left radius in (D) dorsolateral, (E) anterior and (F) posterior views. Left radius: Fraser (1988) suggests that the radius of C. hudsoni is rarely preserved due to its delicate structure, but our specimen is quite solid, despite a crosswise kink midway along the shaft of the bone (Fig. 16D–F). From a narrow shaft, the proximal and distal heads broaden slightly. The proximal head has a sub-oval cross-section, with a flattened posterior surface and a prominent lateral groove, where the radius articulated with the ulna. A terminal depression in the proximal head, the ‘shallow oval concavity’ of Fraser (1988: 151), articulated with the capitellum of the humerus. The blunt distal head of the radius exhibits anteroposterior flattening, similar to the ulna. The CT scan shows the distal styloid process, where ligaments and tendons were attached. Gastralia: Gastralia are the bony rods or belly ribs, ventrally positioned on the skeleton. They are generally regarded as strengthening elements in the ventral abdominal wall, protecting the abdomen which, in C. hudsoni lies close to the ground, owing to a sprawling gait (Fraser, 1988). Recent studies on the function of gastralia in other taxa, however, suggest an additional role in respiration (e.g. Carrier & Farmer, 2000; Claessens, 2004), which may also be the case in rhynchocephalians. Our specimen has 15 observable rows of gastralia, 13 of which have retained the apex of the V, or chevron (Fig. 17). Based on an illustration of the gastralia of Sphenodon by Romer (1956: fig. 202) and our observations of Sphenodon specimens, the smallest gastralium is interpreted as being the most posterior/caudal. This posteriormost gastralium has the shortest medial chevron with the most acute angle. The length of each successive anterior chevron increases across the set. The angle of the medial chevron is obtuse mid-region (the widest angle is approximately centrally positioned), becoming successively more acute anteriorly. The gastralia are thickest at the apex of the chevron and thin laterally; the elements are segmented, not all apices retain their associated lateral splints and some lateral elements have no apex. The medial chevron of the first gastralium (caudal) lacks additional lateral splints, but this may be an artefact of fossilization. Lateral segments become increasingly slender as the gastralia lengthen; more caudally positioned lateral splints are as thick as the medial chevrons and those in an anterior position are long and delicate. Overlap between the segments is not uniform, varyingly positioned on either anterior or posterior side of the medial gastralium (Fig. 17C). The chevrons appear for the most part to be fused, although one gastralium has an overlap at the apex which may represent a break or indicate separate elements (Fig. 17B). A similar chevron was observed on a specimen of Sphenodon (ref. NHMUK Oct 1928). Figure 17. View largeDownload slide Photographs of Clevosaurus hudsoni specimen NHMUK PV R36832. A, gastralia in ventral view. B, enlarged view of apex of medial splint. C, enlarged view of overlap between medial and lateral splints. Figure 17. View largeDownload slide Photographs of Clevosaurus hudsoni specimen NHMUK PV R36832. A, gastralia in ventral view. B, enlarged view of apex of medial splint. C, enlarged view of overlap between medial and lateral splints. Fraser (1988), describing each gastralium as two lateral splints joined by a medial apex (chevron), recorded 25 rows of gastralia for C. hudsoni. This is the same number as Sphenodon, but there does not appear to be that many rows on the articulated specimen on which he based his description (Fraser, 1988: plate 2, fig. 4). It is difficult to be sure of the total number, as many of the gastralia illustrated by Fraser (1988) are broken and out of position. However, of those present, there are only ten or 11 apical chevrons. Furthermore, we consider that the overall shape of the gastralia structure of our specimen to be similar to Sphenodon, indicating that a total of 15 or 16 rows is likely. This smaller number seems reasonable as C. hudsoni is estimated to have reached a maximum length of 25 cm (Fraser, 1988) compared to the much longer 60 cm of the extant Sphenodon (Benton, 2015). Anatomical description of NHMUK PV R36846 Apart from the missing mid-region and proximal part of a broken femur and broken fourth digit, the left hind limb specimen of C. hudsoni (NHMUK PV R36846) is almost complete and well preserved (Fig. 18A). The morphology we describe is based on microscope examination and surface models generated from the CT scan (Figs 18, 19; Supporting Information, Appendix S5). The astragalus and calcaneum are considered separately. Figure 18. View largeDownload slide Photograph and surface models of Clevosaurus hudsoni left hind limb specimen NHMUK PV R36846. Specimen in (A, B) dorsal, (C) ventral and (D) posterolateral views. Figure 18. View largeDownload slide Photograph and surface models of Clevosaurus hudsoni left hind limb specimen NHMUK PV R36846. Specimen in (A, B) dorsal, (C) ventral and (D) posterolateral views. Femur and epipodials The tibia and fibula are intact and the distal condylar portion of the femur is present (Fig. 18A–D). The curve on the end of femoral shaft is less pronounced on this specimen than that illustrated by Fraser (1988: fig. 33). The lateral and medial tibial condyles, separated by a furrow, are well displayed on the ventral surface of the femur (Fig. 19A). Tibia: The tibia is a robust bone with a near cylindrical slender shaft, expanding to sub-rounded proximal and distal heads (Fig. 19B–D). The proximal end displays the femoral condyles and a sulcus that would have accommodated the medial surface of the head of the fibula. The shaft of the tibia is slightly anteroposteriorly flattened and broadly concave towards the fibula along its length. The distal head is concave on the ventromedial surface, for articulation with the ‘astragalocalcaneum’. Figure 19. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36846. A, distal head of broken femur in ventrolateral view. Left tibia in (B) anterior, (C) lateral and (D) posterior views. Left fibula in (E) anterior and (F) posterior views. G, astragalus, calcaneum and tarsal bones in dorsal view. H, digit i in dorsolateral view. I, digit i phalanx and ungual in dorsolateral view. Astragalus and calcaneum in (J) dorsal and (K) ventral views. Figure 19. View largeDownload slide Photographs and surface models of Clevosaurus hudsoni specimen NHMUK PV R36846. A, distal head of broken femur in ventrolateral view. Left tibia in (B) anterior, (C) lateral and (D) posterior views. Left fibula in (E) anterior and (F) posterior views. G, astragalus, calcaneum and tarsal bones in dorsal view. H, digit i in dorsolateral view. I, digit i phalanx and ungual in dorsolateral view. Astragalus and calcaneum in (J) dorsal and (K) ventral views. Fibula: The fibula is a very slender bone (Fig. 19E, F) that appears to have a slight axial twist on the (anteroposteriorly) flattened shaft, but this possibly reflects breakage. The shaft broadens slightly at both proximal and distal heads. Articulation facets for adjacent bones are poorly preserved. Tarsus and pes Three tarsals are present between the astragalus and calcaneum and the metatarsals (Fig. 19G). There is no evidence of a first digit tarsal and the second almost merges with the third, occurring as a ventromedial attachment with a small dorsolateral projection. These second and third tarsals are small, well-rounded bones with articulation facets. The fourth tarsal, larger and with a prominent deep foramen on the dorsal surface, is more polygonal in shape, displaying a curved convex lateral facet for articulation with the medial surfaces of the astragalus and calcaneum. A distal facet articulates with the tarsometatarsal (fifth metatarsal), which has the distinctive lepidosaurian hooked morphology (Fig. 19G). The robust tarsometatarsal has a depressed medial facet to articulate with the adjacent fourth metatarsal and proximal facets for articulation with the calcaneum and fourth tarsus. The metatarsals and phalanges vary in size, but are somewhat similar morphologically; the metatarsals have a slender shaft, dorsoventrally flattened, with expanded proximal and broad, faceted distal heads. Metatarsal 1 (Fig. 19H) illustrates the typically bicondylar convex distal head that accommodated the concave head of the phalanx. The phalanges have a more circular shaft, notably faceted bicondylar distal heads, and proximal heads which are rounded and concave. The distal heads have a convex tip, which articulated with the concave head of the next phalanx. The ungual phalanges are mediolaterally compressed and have a strongly concave proximal surface that articulated with the penultimate phalanx (Fig. 19H). A groove is visible on the lateral and medial surfaces of each ungual phalanx (Fig. 19I). These are likely to have been attachment points for tendons providing surface traction for the claw. The phalangeal formula for NHMUK PV R36846 is uncertain because the fourth digit is broken beyond the second phalanx. However, based on the digits preserved, the hind limb phalangeal formula for C. hudsoni is 2: 3: 4: _: 4. In the absence of a complete hind limb, Fraser (1988) assumed that the phalangeal formula for C. hudsoni would be the same as that of Sphenodon (2: 3: 4: 5: 4), which is accepted here. Bones previously poorly known or unknown Astragalus and calcaneum: Clevosaurs typically possess an astragalocalcaneum; a single fused proximal tarsal element of the medial astragalus and the lateral calcaneum. However, on NHMUK PV R36846, the astragalocalcaneum appears unfused (Figs 18A, 19J, K). There is a clear separation between the astragalus and the calcaneum and, while this may reflect breakage along a line of weakness defined by the suture between the two elements, followed by medioventral rotation of the calcaneum, it is more likely that the two bones were never fused. Matrix infill between the two elements may give the appearance of fusion at their lateral contact, but the ventral view facilitated by the CT scan (Fig. 19K) delineates clearly the outline of each bone. Fraser (1988) describes a fused astragalocalcaneum in C. hudsoni and Sues et al. (1994) suggest that the poorly preserved astragalus and calcaneum on their specimen of C. bairdi appears to be fused. Additionally, Klein et al. (2015) identify a single element in C. sectumsemper. Thus, the probable unfused ‘astragalocalcaneum’ of our specimen is unusual among clevosaurs, but visual and CT scan evidence suggests that the separation between the astragalus and the calcaneum is more than just a break. The astragalus has a concave lateral facet that would have articulated with a medial convexity on the calcaneum (Fig. 19K). The margins of both bones are well defined and, while the edge may have undergone some attrition during preservation, if there had been a break along a suture line some degree of irregularity might be expected. Sues et al. (1994) do not provide an image of the astragalocalcaneum for C. bairdi, but Fraser (1988: fig. 35) and Klein et al. (2015: fig. 6K, L) provide an illustration and photograph, respectively. The line of suture between the two bones is not delineated by Fraser (1988). A vague suture is identifiable on both anterior and posterior surfaces of the C. sectumsemper specimen (Klein et al., 2015), but it is not a deep feature and does not appear to present a possible line of weakness. The astragalus and calcaneum preserved on specimen NMHUK PV R36846 are robust bones. If these elements were at one time a single unit, in a manner comparable to C. sectumsemper, then force would have been required to separate them. While the ‘astragalocalcaneum’ might be considered to be in a vulnerable position on the hind limb of Clevosaurus, given the exceptional preservation of this specimen and the fact that the bones are mostly undamaged, it is more probably that the bones were never fused. It is notable that although one NHMUK Sphenodon specimen (ref NHMUK 65.5.43) showed no evidence of a suture between the astragalus and the calcaneum, two other specimens, one adult and one probable sub-adult, had obvious sutures (ref. NHMUK 1861; Oct 1928). The astragalus has broadly concave proximal facets for articulation with the sub-rounded distal heads of both tibia and fibula (Fig. 19J). Facets are also present on the distal surfaces of both the astragalus and calcaneum for articulation with the tarsals. Both bones are flattened dorsoventrally and have broad concave anterior surfaces. DISCUSSION Taphonomy of NHMUK PV R36832 and NHMUK PV R36846 The positioning of the bones on NHMUK PV R36832 indicates partial dissociation in situ. The left-hand side of the skull is much more intact but many neighbouring bones have dissociated, some slightly and others to a greater degree. Some bones are broken but others, especially the more robust, are largely complete. There are good examples of near complete articulation or connection in the paired frontals and parietals, the cervical vertebrae and the left scapulocoracoid, humerus and forearm bones. It provides evidence for the hypothesis that the accumulations of numerous isolated bones in the British Triassic fissures commonly occurred through the dissociation of articulated corpses that washed into the fissure and were preserved by rapid, bacterial decomposition in a subaqueous, or at least moist, environment (Whiteside & Marshall, 2008; Whiteside et al., 2016). The lack of any bite marks on this fossil also contrasts with the predator accumulation hypothesis of Evans & Kermack (1994). The more fragmentary right-hand side of the skull is explained by the CT scan, which reveals the presence of an entire right humerus (Supporting Information, Appendices S1, S2) with smaller bones, the right ulna and radius, positioned next to the distal end of the humerus. These images indicate that the right arm was pushed through the skull by post mortem sedimentary slumping, causing the displacement of head bones. The movements of matrix in the fissure that resulted in this partial fragmentation may relate to compression of the sediments, caused by new material falling onto the decaying skeleton in an aqueous environment. However, although the original fieldnotes of P.L. Robinson do not have precise descriptions of the strata (due to the quarrying operations), the matrix movement could relate to seismic movements, such as those resulting in the slumping structures found in the Late Triassic in the UK (Gallois, 2009). NHMUK PV R36846 is a very well-preserved hind limb fossil, with bones in life or almost life positions, strengthening the view that entire carcasses were regularly washed into the fissures. What is not clear is whether some individuals were alive or recently killed in the waters that washed in. It is conceivable that this individual was freshly killed just before or at burial. Whatever the circumstances, the soft tissue would have rapidly decomposed and in other cases further flows would have dissociated and deposited the isolated bones used to describe the fauna at, for example, Cromhall Quarry (Fraser & Walkden, 1983) or Tytherington Quarry (Whiteside & Marshall, 2008). Comparison with other clevosaurs Based on recent cladistic analysis (Hsiou et al., 2015), the Clevosauridae clade currently comprises C. bairdi from the Early Jurassic of Nova Scotia, Canada (Sues et al., 1994), C. wangi., C. petilus and C. mcgilli from the Early Jurassic of Yunnan Province, China (Wu, 1994), C. brasiliensis from the Late Triassic of Rio Grande do Sul, Brazil (Bonaparte & Sues, 2006), C. convallis from the Early Jurassic of South Wales (Säilä, 2005) and C. hudsoni and C. minor from the Late Triassic of SW England (Fraser, 1988). Clevosaurus sectumsemper had not been described at the time of the analysis, but Klein et al. (2015) considered it a distinct taxon from the other SW UK clevosaurs; future cladistic analysis may confirm the validity of this suggestion. Clevosaurus bairdi, C. brasiliensis and the three Chinese clevosaurs, C. wangi, C. petilus and C. mcgilli, have each been diagnosed from partial or deformed skulls, while the UK species (excepting C. hudsoni) have been described primarily from jaw bones. Jaw bones have been used here to compare NHMUK PV R36832 with other clevosaurs. A fully acrodont dentition is shared by all clevosaurs, together with the presence of secondary dentine. Evidence of tooth wear, indicative of orthal occlusion between maxilla and dentary is common. Significant differences occur, however, in the number of additional teeth between species and the presence or absence of prominent tooth flanges (Table 1). Interestingly, a separate coronoid bone attached to the coronoid process of the dentary has been identified in some but not all clevosaurs; notably it has not been recorded in C. hudsoni. Table 1. Differences in dentition and dentary morphology among clevosaurs Species Location Reference Additional teeth Prominent flange on teeth Coronoid process Clevosaurus bairdi Nova Scotia, Canada Sues et al. (1994) 5–6 On 3 additional teeth only CB Clevosaurus brasiliensis Rio Grande Do Sul, Brazil Bonaparte & Sues (2006) 2–3 Not conspicuous NC Clevosaurus wangi Yunnan Province, China Wu (1994); Jones (2006) 5* Yes; small CB Clevosaurus petilus Yunnan Province, China Wu (1994); Jones (2006) 5* Small CB Clevosaurus mcgilli Yunnan Province, China Wu (1994); Jones (2006) 5* Yes; small CB Clevosaurus convallis Wales, UK Säilä (2005) 6† Yes N Clevosaurus sectumsemper Bristol, UK Klein et al. (2015) 4 Yes NC Clevosaurus hudsoni Bristol, UK Fraser (1988) 4 Yes NC Clevosaurus minor Bristol, UK Fraser (1988) 4 Yes N NHMUK PV R36832 Bristol, UK This study 4 Yes NC Species Location Reference Additional teeth Prominent flange on teeth Coronoid process Clevosaurus bairdi Nova Scotia, Canada Sues et al. (1994) 5–6 On 3 additional teeth only CB Clevosaurus brasiliensis Rio Grande Do Sul, Brazil Bonaparte & Sues (2006) 2–3 Not conspicuous NC Clevosaurus wangi Yunnan Province, China Wu (1994); Jones (2006) 5* Yes; small CB Clevosaurus petilus Yunnan Province, China Wu (1994); Jones (2006) 5* Small CB Clevosaurus mcgilli Yunnan Province, China Wu (1994); Jones (2006) 5* Yes; small CB Clevosaurus convallis Wales, UK Säilä (2005) 6† Yes N Clevosaurus sectumsemper Bristol, UK Klein et al. (2015) 4 Yes NC Clevosaurus hudsoni Bristol, UK Fraser (1988) 4 Yes NC Clevosaurus minor Bristol, UK Fraser (1988) 4 Yes N NHMUK PV R36832 Bristol, UK This study 4 Yes NC ‘Additional teeth’ refer to the sequence of large teeth that follow the hatchling dentition on the dentary and maxilla; they do not include the small subconical teeth that follow on many rhynchocephalian maxillae. CB, separate coronoid bone present on coronoid process of dentary; NC, separate coronoid bone not recorded; N, coronoid process not preserved. *Number of additional teeth based on maximum recorded by Jones (2006). †Clevosaurus convallis has six large additional dentary teeth followed by one or two smaller teeth (Säilä, 2005). View Large Table 1. Differences in dentition and dentary morphology among clevosaurs Species Location Reference Additional teeth Prominent flange on teeth Coronoid process Clevosaurus bairdi Nova Scotia, Canada Sues et al. (1994) 5–6 On 3 additional teeth only CB Clevosaurus brasiliensis Rio Grande Do Sul, Brazil Bonaparte & Sues (2006) 2–3 Not conspicuous NC Clevosaurus wangi Yunnan Province, China Wu (1994); Jones (2006) 5* Yes; small CB Clevosaurus petilus Yunnan Province, China Wu (1994); Jones (2006) 5* Small CB Clevosaurus mcgilli Yunnan Province, China Wu (1994); Jones (2006) 5* Yes; small CB Clevosaurus convallis Wales, UK Säilä (2005) 6† Yes N Clevosaurus sectumsemper Bristol, UK Klein et al. (2015) 4 Yes NC Clevosaurus hudsoni Bristol, UK Fraser (1988) 4 Yes NC Clevosaurus minor Bristol, UK Fraser (1988) 4 Yes N NHMUK PV R36832 Bristol, UK This study 4 Yes NC Species Location Reference Additional teeth Prominent flange on teeth Coronoid process Clevosaurus bairdi Nova Scotia, Canada Sues et al. (1994) 5–6 On 3 additional teeth only CB Clevosaurus brasiliensis Rio Grande Do Sul, Brazil Bonaparte & Sues (2006) 2–3 Not conspicuous NC Clevosaurus wangi Yunnan Province, China Wu (1994); Jones (2006) 5* Yes; small CB Clevosaurus petilus Yunnan Province, China Wu (1994); Jones (2006) 5* Small CB Clevosaurus mcgilli Yunnan Province, China Wu (1994); Jones (2006) 5* Yes; small CB Clevosaurus convallis Wales, UK Säilä (2005) 6† Yes N Clevosaurus sectumsemper Bristol, UK Klein et al. (2015) 4 Yes NC Clevosaurus hudsoni Bristol, UK Fraser (1988) 4 Yes NC Clevosaurus minor Bristol, UK Fraser (1988) 4 Yes N NHMUK PV R36832 Bristol, UK This study 4 Yes NC ‘Additional teeth’ refer to the sequence of large teeth that follow the hatchling dentition on the dentary and maxilla; they do not include the small subconical teeth that follow on many rhynchocephalian maxillae. CB, separate coronoid bone present on coronoid process of dentary; NC, separate coronoid bone not recorded; N, coronoid process not preserved. *Number of additional teeth based on maximum recorded by Jones (2006). †Clevosaurus convallis has six large additional dentary teeth followed by one or two smaller teeth (Säilä, 2005). View Large The distinctive ‘boomerang-shaped’ maxilla of C. bairdi (Sues et al., 1994), which has flanges on only three of the additional teeth, is markedly different morphologically to that of C. hudsoni including NHMUK PV R63832. The maxillary dentition of C. brasiliensis is variously described as comprising two large flanged teeth, followed by one or two conical teeth or ‘a large conical anterior tooth, followed by several small teeth and in turn by two large ones’ (Bonaparte & Sues, 2006: 919). Hsiou et al. (2015) confirm that the large flanges seen on the additional teeth of C. hudsoni (and observed on NHMUK PV R36832) are not seen on the holotype of C. brasiliensis. The fossil material used by Wu (1994) to assign the three Chinese species was reassessed by Jones (2006), who concluded that as a result of poor preservation, the diagnostic features of the specimens could not be adequately assessed to erect three new taxa and assigned Clevosaurus sp. to all three. In contrast to Wu (1994), Jones (2006) reports that although their dentary and maxillary teeth ‘may possess small flanges’ (Jones, 2006: 558), they are not as extensive as those of C. hudsoni (and also NHMUK PV R36832). However, Hsiou et al. (2015) considered that the three Chinese taxa each possessed unique characters, warranting their inclusion in the cladistic analysis. Clevosaurus convallis possesses six large additional dentary teeth, which generally increase in size posteriorly, but the sixth tooth of this series is smaller than the fifth (Säilä, 2005). The posteromedial flanges on the additional maxillary teeth and anterior flanges on the additional dentary teeth of C. convallis (Säilä, 2005) are features shared with C. hudsoni (Fraser, 1988) including NHMUK PV R36832. However, Säilä (2005) indicates that the anterolateral flanges of the additional teeth on the dentary of C. convallis are not as long as those on C. hudsoni and, on some specimens, there is a complete loss of tooth overlap, with the flanges worn by occlusion with the maxilla; the maxillary flanges of C. convallis are thought to be comparable in size to C. hudsoni. Clevosaurus sectumsemper possesses the same number of additional teeth as C. hudsoni and NHMUK PV R36832, which increase in size posteriorly. However, the tooth bases on the dentary of C. sectumsemper are described by Klein et al. (2015) as more ventrally positioned posteriorly than for C. hudsoni and they further describe pronounced gaps between the four additional teeth, with no overlap of the anterolateral flanges. Therefore, although they share some features, the marginal dentition of NHMUK PV R36832 differs significantly from that of both C. convallis and C. sectumsemper, but it is identical to C. hudsoni. Clevosaurus minor closely resembles C. hudsoni and NHMUK PV R36832 in the shape and configuration of the maxillary and dentary dentition (Fraser, 1988) and in possessing the same number of additional teeth. These teeth are flanged and increase in size posteriorly, and there are two or three small, subconical teeth on the maxilla, posterior to the additional set. However, Fraser & Walkden (1983) and Fraser (1988) observe a considerable difference in tooth wear and the growth of secondary dentine on similarly sized specimens of C. minor and C. hudsoni, the former exhibiting high degrees of wear in the largest specimens examined, and the latter showing signs of immaturity, often without the full complement of teeth. There are also differences between the palatines of the two species (Fraser & Walkden, 1983; Fraser, 1988). Despite the possibility of differences in diet producing such variance, C. minor is not considered to be a juvenile of C. hudsoni (Fraser & Walkden, 1983) and sexual dimorphism is also discounted as a possible explanation by Fraser (1988). Furthermore, C. minor and C. hudsoni rarely occur within the same fissure deposit at Cromhall Quarry (Fraser & Walkden, 1983; Fraser, 1988; Walkden & Fraser, 1993), strengthening the notion that they are separate taxa. Based on the evidence, Fraser & Walkden (1983) and Fraser (1988) concluded that C. minor and C. hudsoni are distinct species. NHMUK PV R36832 was found between fissure sites 1 and 2, a location where C. minor is either absent or very rare (Walkden & Fraser, 1993). This fact, combined with the differences in tooth wear and morphology of the palatine, along with the relative sizes of C. minor and NHMUK PV R36832, indicate that NHMUK PV R36832 is not a specimen of C. minor. It is evident from the posterior wear facets on NHMUK PV R36832 that the maxillary teeth occluded on the lateral side and between the dentary teeth, with each maxillary tooth contacting the flanges between the dentary teeth. That NHMUK PV36832 is an adult specimen is demonstrated by the size of the teeth, the deep wear facets and the secondary dentition. However, the bases to the teeth on both maxilla and dentary are clearly visible and there is no evidence of a single cutting edge of bone, described by Robinson (1973) and Fraser (1988) for some of the most mature (or aged) individuals, where the maxillary teeth have either been worn by use, or are almost totally obscured by the ventral growth of secondary dentine. Similarly, although the flanges between the teeth on the dentary give the impression of a continuous cutting surface, the tips of the additional teeth on NHMUK PV R36832 stand proud and project dorsally. It is, therefore, most probable that NHMUK PV R36832 was an adult, but not an aged, individual. Differing patterns of tooth wear occur in similar sized clevosaurs and comparable wear patterns have been observed in specimens of different size. This variability led Fraser (1988), Säilä (2005) and Klein et al. (2015) to conclude that differences in diet could be responsible for variations in dental wear patterns in C. minor, C. convallis and C. sectumsemper. Säilä (2005) concluded that sexual dimorphism could not be ruled out as a potential cause of variation in tooth wear, but this was not investigated further and had previously been discounted for C. minor, by Fraser (1988), who considered variation in diet more plausible. A diet of arthropods versus one of shelled molluscs or small tetrapods was suggested by Klein et al. (2015) to account for tooth wear variation in C. sectumsemper. Fraser (1988) suggested that juvenile clevosaurs probably fed on small insects and soft bodied invertebrates and postulated that the continuous cutting edge of the jaw and edentulous regions of the mature animals supported facultative herbivory (Fraser & Walkden, 1983). However, transversely elongated teeth are more indicative of herbivory and we suggest C. hudsoni with cutting blades running sub-parallel to the long axis of the jaw was not primarily herbivorous, but rather that the dentition was adapted for faunivory. Preservation and ontogeny The preservation of NHMUK PV R36832 and NHMUK PV R36846 has enabled assessment of bones that have been described rarely, or not at all, for clevosaurs (Table 2). Finding sclerotic ossicles is unusual, as is the preservation of fragile bones such as the stapes and hyoid. Although cervical vertebrae are occasionally preserved, fossilization of an articulated sequence including the atlas/axis complex is rare, as is the presence of the gastralia. The scapulocoracoid and the astragalus and calcaneum are particularly significant for the insights they may provide on ontogeny and/or variation in clevosaurs. The fused scapulocoracoid of NHMUK PV R36832 accords with the observation of Fraser (1988) for C. hudsoni that there were no growth stages in his collection where the scapula and coracoid bones were separated by a complete suture. Klein et al. (2015), however, believed that the isolated coracoid of C. sectumsemper showed no evidence of having broken from a fused scapulocoracoid, concluding that juveniles may have possessed separate elements that fused in adulthood. It is notable that a scapulocoracoid suture is visible on both a sub-adult and an adult Sphenodon (ref. NHMUK 1861; 1985 1212) but not on all specimens. Table 2. Notable morphological features described as part of this study for specimens NHMUK PV R36832 and NHMUK PV R36846 and their occurrence in other known clevosaurs Species Location Reference Sclerotic ossicles Stapes/hyoid Atlas/axis centra Gastralia Scapulocoracoid Astragalocalcaneum Clevosaurus bairdi Nova Scotia, Canada Sues et al. (1994) Y N N N N F Clevosaurus brasiliensis Rio Grande Do Sul, Brazil Bonaparte & Sues (2006) N N N N N N Clevosaurus wangi Yunnan Province, China Wu (1994); Jones (2006) N Y U N N N Clevosaurus petilus Yunnan Province, China Wu (1994); Jones (2006) N Y N N N N Clevosaurus mcgilli Yunnan Province, China Wu (1994); Jones (2006) N N N N N N Clevosaurus convallis Wales, UK Säilä (2005) N N N N N N Clevosaurus sectumsemper Bristol, UK Klein et al. (2015) N N N N NF F Clevosaurus hudsoni Bristol, UK Fraser (1988) N N N Y (25) F F Clevosaurus minor Bristol, UK Fraser (1988) N N N N N N NHMUK PV R36832 Bristol, UK This study Y Y NF Y (15) F N NHMUK PV R36846 Bristol, UK This study NA NA NA NA NA NF Species Location Reference Sclerotic ossicles Stapes/hyoid Atlas/axis centra Gastralia Scapulocoracoid Astragalocalcaneum Clevosaurus bairdi Nova Scotia, Canada Sues et al. (1994) Y N N N N F Clevosaurus brasiliensis Rio Grande Do Sul, Brazil Bonaparte & Sues (2006) N N N N N N Clevosaurus wangi Yunnan Province, China Wu (1994); Jones (2006) N Y U N N N Clevosaurus petilus Yunnan Province, China Wu (1994); Jones (2006) N Y N N N N Clevosaurus mcgilli Yunnan Province, China Wu (1994); Jones (2006) N N N N N N Clevosaurus convallis Wales, UK Säilä (2005) N N N N N N Clevosaurus sectumsemper Bristol, UK Klein et al. (2015) N N N N NF F Clevosaurus hudsoni Bristol, UK Fraser (1988) N N N Y (25) F F Clevosaurus minor Bristol, UK Fraser (1988) N N N N N N NHMUK PV R36832 Bristol, UK This study Y Y NF Y (15) F N NHMUK PV R36846 Bristol, UK This study NA NA NA NA NA NF Number in parentheses in ‘Gastralia’ column indicates number of gastralia recorded. F, element preserved and bones fused; N, element not preserved; NA, bone region not present in specimen; NF, element preserved and bones not fused; U, undetermined; Y, element preserved. View Large Table 2. Notable morphological features described as part of this study for specimens NHMUK PV R36832 and NHMUK PV R36846 and their occurrence in other known clevosaurs Species Location Reference Sclerotic ossicles Stapes/hyoid Atlas/axis centra Gastralia Scapulocoracoid Astragalocalcaneum Clevosaurus bairdi Nova Scotia, Canada Sues et al. (1994) Y N N N N F Clevosaurus brasiliensis Rio Grande Do Sul, Brazil Bonaparte & Sues (2006) N N N N N N Clevosaurus wangi Yunnan Province, China Wu (1994); Jones (2006) N Y U N N N Clevosaurus petilus Yunnan Province, China Wu (1994); Jones (2006) N Y N N N N Clevosaurus mcgilli Yunnan Province, China Wu (1994); Jones (2006) N N N N N N Clevosaurus convallis Wales, UK Säilä (2005) N N N N N N Clevosaurus sectumsemper Bristol, UK Klein et al. (2015) N N N N NF F Clevosaurus hudsoni Bristol, UK Fraser (1988) N N N Y (25) F F Clevosaurus minor Bristol, UK Fraser (1988) N N N N N N NHMUK PV R36832 Bristol, UK This study Y Y NF Y (15) F N NHMUK PV R36846 Bristol, UK This study NA NA NA NA NA NF Species Location Reference Sclerotic ossicles Stapes/hyoid Atlas/axis centra Gastralia Scapulocoracoid Astragalocalcaneum Clevosaurus bairdi Nova Scotia, Canada Sues et al. (1994) Y N N N N F Clevosaurus brasiliensis Rio Grande Do Sul, Brazil Bonaparte & Sues (2006) N N N N N N Clevosaurus wangi Yunnan Province, China Wu (1994); Jones (2006) N Y U N N N Clevosaurus petilus Yunnan Province, China Wu (1994); Jones (2006) N Y N N N N Clevosaurus mcgilli Yunnan Province, China Wu (1994); Jones (2006) N N N N N N Clevosaurus convallis Wales, UK Säilä (2005) N N N N N N Clevosaurus sectumsemper Bristol, UK Klein et al. (2015) N N N N NF F Clevosaurus hudsoni Bristol, UK Fraser (1988) N N N Y (25) F F Clevosaurus minor Bristol, UK Fraser (1988) N N N N N N NHMUK PV R36832 Bristol, UK This study Y Y NF Y (15) F N NHMUK PV R36846 Bristol, UK This study NA NA NA NA NA NF Number in parentheses in ‘Gastralia’ column indicates number of gastralia recorded. F, element preserved and bones fused; N, element not preserved; NA, bone region not present in specimen; NF, element preserved and bones not fused; U, undetermined; Y, element preserved. View Large The separate astragalus and calcaneum on NMHUK PV R36846 contrasts with a fused astragalocalcaneum described previously for C. hudsoni (Fraser, 1988), C. sectumsemper (Klein et al., 2015) and C. bairdi (Sues et al., 1994). NMHUK PV R36846 is an isolated hindlimb, so it is not possible to assess the animal’s age directly. However, we can compare our specimen with those of Fraser (1988), who took measurements of fore and hind limb bones from fully ossified individuals. The tibia on NHMUK PV R36846 falls within the measured lengths recorded (Fraser, 1988), lying at the lower end of the range, but is longer than the shortest specimen. This may support the notion that our specimen derived from a young adult and that subsequent ontogenetic fusion of the astragalus and calcaneum may have occurred, had the individual survived to full maturity. However, it is also plausible that the specimen derives from the smaller gender of a dimorphic species. As described above, we have found individual variation in the fused/sutured/unsutured condition of ‘scapulocoracoids’ and ‘astragalocalcanea’, irrespective of animal size, in a range of Sphenodon specimens in the NHMUK. It is, therefore, possible that these bones are isolated components in juveniles that fuse in particular adults, with some or no trace of a suture. Our findings are important, as a fused astragalocalcaneum is considered an apomorphy common to Rhynchocephalia and Squamata (e.g. Evans, 2003). The fully fused astragalocalcaneum condition described for C. hudsoni (Fraser, 1988) may simply indicate a mature adult or a variable character in the species. Correspondingly, the assertion by Klein et al. (2015) that the scapulocoracoid of C. sectumsemper was unfused in juveniles and fused in adults may also apply to C. hudsoni, contraFraser (1988), and to clevosaurs in general. CONCLUSION Two previously undescribed specimens of C. hudsoni, NHMUK PV R36832 and NMHUK PV R36846, from a Late Triassic fissure infill at Cromhall Quarry, Gloucestershire, have been investigated using stereoscopic microscopic analysis and CT. A detailed description of the visible bones was augmented by the creation of 3D digital reconstructions from CT data, providing additional information on material that was either partially, or fully, embedded in matrix. It is the first time that successful CT scans have been obtained from the British Triassic fissure deposits. The possibility of the CT scans revealing detailed information on the braincase bones was hindered by the similarity in X-ray attenuation properties of the fossilized bone and the rock matrix. Despite this, 3D reconstructions have been generated for numerous bones from both specimens, including bones from the right side of NHMUK PV R36832 that are not visible and were presumed lost. Key elements preserved in these specimens have not been previously described. This paper provides new information on the cervical vertebrae including the atlas-axis complex, the pectoral girdle, the fore limb, hind limb, the gastralia as well as aspects of the skull such as the sclerotic ossicles of C. hudsoni. We have been able to significantly add and amend information on the type species of the clevosaur clade. SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article at the publisher’s website: Appendix S1. Animation of Avizo surface model of NHMUK PV R36832: cranial lateral aspect. Appendix S2. Animation of Avizo surface model of NHMUK PV R36832: cranial dorsoventral aspect. Appendix S3. Animation of Avizo surface model of NHMUK PV R36832: atlas/axis complex. Appendix S4. Animation of Avizo surface model of NHMUK PV R36832: cervical/dorsal vertebrae. Appendix S5. Animation of Avizo surface model of NHMUK PV R36846: left hindlimb. ACKNOWLEDGEMENTS We thank Sandra Chapman and Patrick Campbell (NHMUK) for arranging the loans of Clevosaurus and facilitating our visit to investigate Sphenodon skeletons, respectively. Timothy Smithson and Jason Head (UMZC) are thanked for the loan of Sphenodon skeletons. We are very grateful to Mark Mavrogordato and Kathryn Rankin for carrying out the CT scans at µ-VIS X-Ray Imaging Centre, Faculty of Engineering and the Environment, University of Southampton. Mark Mavrogordato is additionally thanked for help in drafting the methodology of a CT scan. We greatly appreciate the support given by Simon Chen and the Cromhall Diving Centre who have been exceptionally helpful in permitting access to Cromhall Quarry. We thank two anonymous reviewers for their constructive comments. REFERENCES Benton MJ . 1985 . Classification and phylogeny of the diapsid reptiles . Zoological Journal of the Linnean Society 84 : 97 – 164 . Google Scholar CrossRef Search ADS Benton MJ . 2015 . Vertebrate palaeontology, 4th edn . London : Wiley-Blackwell . Bever GS , Lyson TR , Field DJ , Bhullar BA . 2015 . Evolutionary origin of the turtle skull . Nature 525 : 239 – 242 . Google Scholar CrossRef Search ADS PubMed Bonaparte JF , Sues HD . 2006 . A new species of Clevosaurus (Lepidosauria: Rhynchocephalia) from the Upper Triassic of Rio Grande do Sul, Brazil . Palaeontology 49 : 917 – 923 . Google Scholar CrossRef Search ADS Carrier DR , Farmer CG . 2000 . The evolution of pelvic aspiration in archosaurs . Paleobiology 26 : 271 – 293 . Google Scholar CrossRef Search ADS Claessens LPAM . 2004 . Dinosaur gastralia; origin, morphology, and function . Journal of Vertebrate Paleontology 24 : 89 – 106 . Google Scholar CrossRef Search ADS Cunningham JA , Rahman IA , Lautenschlager S , Rayfield EJ , Donoghue PC . 2014 . A virtual world of paleontology . Trends in Ecology & Evolution 29 : 347 – 357 . Google Scholar CrossRef Search ADS PubMed Evans SE . 1984 . The classification of the Lepidosauria . Zoological Journal of the Linnean Society 82 : 87 – 100 . Google Scholar CrossRef Search ADS Evans SE . 2003 . At the feet of the dinosaurs: the early history and radiation of lizards . Biological Reviews 78 : 513 – 551 . Google Scholar CrossRef Search ADS PubMed Evans SE , Kermack KA . 1994 . Assemblages of small tetrapods from the Early Jurassic of Britain . In: Fraser NC , Sues HD , eds. In the shadow of the dinosaurs: early Mesozoic tetrapods . New York : Cambridge University Press , 271 – 282 . Foffa D , Whiteside DI , Viegas PA , Benton MJ . 2014 . Vertebrates from the Late Triassic Thecodontosaurus-bearing rocks of Durdham Down, Clifton (Bristol, UK) . Proceedings of the Geologists’ Association 125 : 317 – 328 . Google Scholar CrossRef Search ADS Fraser NC . 1982 . A new rhynchocephalian from the British Upper Triassic . Palaeontology 25 : 709 – 725 . Fraser NC . 1985 . Vertebrate faunas from Mesozoic fissure deposits of South West Britain . Modern Geology 9 : 273 – 300 . Fraser NC . 1986 . New Triassic sphenodontids from south-west England and a review of their classification . Palaeontology 29 : 165 – 186 . Fraser NC . 1988 . The osteology and relationships of Clevosaurus (Reptilia: Sphenodontida) . Philosophical Transactions of the Royal Society B 321 : 125 – 178 . Google Scholar CrossRef Search ADS Fraser NC . 1993 . A new sphenodontian from the early Mesozoic of England and North America: implications for correlating early Mesozoic continental deposits . In: Lucas SG , Morales M , eds. The nonmarine Triassic . Albuquerque : Museum of Natural History and Science , 135 – 139 . Fraser NC . 1994 . Assemblages of small tetrapods from British Late Triassic fissure deposits . In: Fraser NC , Sues HD , eds. In the shadow of the dinosaurs: early Mesozoic tetrapods . New York : Cambridge University Press , 214 – 226 . Fraser NC , Walkden GM . 1983 . The ecology of a Late Triassic reptile assemblage from Gloucestershire, England . Palaeogeography, Palaeoclimatology, Palaeoecology 42 : 341 – 365 . Google Scholar CrossRef Search ADS Fraser NC , Walkden GM . 1984 . The postcranial skeleton of the Upper Triassic sphenodontid Planocephalosaurus robinsonae . Palaeontology 27 : 575 – 595 . Gallois RW . 2009 . The lithostratigraphy of the Penarth Group (Late Triassic) of the Severn Estuary area . Geoscience in South-West England 12 : 71 – 84 . Galton PM , Yates AM , Kermack D . 2007 . Pantydraco n. gen. for Thecodontosaurus caducus YATES, 2003, a basal sauropodomorph dinosaur from the Upper Triassic or Lower Jurassic of South Wales, UK . Neues Jahrbuch für Geologie und Paläontologie – Abhandlungen 243 : 119 – 125 . Google Scholar CrossRef Search ADS Gauthier JA , Estes R , de Queiroz K . 1988 . A phylogenetic analysis of Lepidosauromorpha . In: Estes R , Pregill G , eds. Phylogenetic relationships of the lizard families . Stanford : Stanford University Press , 15 – 98 . Gray JE . 1842 . Description of two hitherto unrecorded species of reptiles from New Zealand; presented to the British Museum by Dr. Dieffenbach . In: Gray JE , ed. The zoological miscellany, Vol. 2 . London : Treuttel, Würtz & Co ., 72 . Halstead LB , Nicholl PG . 1971 . Fossilized caves of Mendip . Studies in Speleology 2 : 93 – 102 . Hiscock C . 2009 . Slickstones Quarry, Cromhall – SSSI & RIGS . Outcrop – The Newsletter of the Avon RIGS Group 24 : 1 – 3 . Hoffstetter R , Gasc JP . 1969 . Vertebrae and ribs of modern reptiles . In: Gans C , Parsons TS , eds. Biology of the Reptilia Volume 1, Morphology A . New York : Academic Press , 201 – 302 . Hsiou AS , De Franca MAG , Ferigolo J . 2015 . New data on the Clevosaurus (Sphenodontia: Clevosauridae) from the Upper Triassic of southern Brazil . PLoS ONE 10 : e0137523 . Google Scholar CrossRef Search ADS PubMed Jones MEH . 2006 . The Early Jurassic clevosaurs from China (Diapsida: Lepidosauria) . New Mexico Museum Natural History Science Bulletin 37 : 548 – 562 . Jones MEH , Curtis N , Fagan MJ , O’Higgins P , Evans SE . 2011 . Hard tissue anatomy of the cranial joints in Sphenodon (Rhynchocephalia): sutures, kinesis, and skull mechanics . Palaeontologia Electronica 14 : 1 –92. Jones MEH , Curtis N , O’Higgins P , Fagan M , Evans SE . 2009 . The head and neck muscles associated with feeding in Sphenodon (Reptilia: Lepidosauria: Rhynchocephalia) . Palaeontologia Electronica 12 : 1–56 . Jones TR . 1862 . A monograph of the fossil Estheriae . Monograph of the Palaeontographical Society 14 : 1 – 134 . Klein CG , Whiteside DI , Selles de Lucas V , Viegas PA , Benton MJ . 2015 . A distinctive Late Triassic microvertebrate fissure fauna and a new species of Clevosaurus (Lepidosauria: Rhynchocephalia) from Woodleaze Quarry, Gloucestershire, UK . Proceedings of the Geologists’ Association 126 : 402 – 416 . Google Scholar CrossRef Search ADS Lautenschlager S , Witmer LM , Altangerel P , Zanno LE , Rayfield EJ . 2014 . Cranial anatomy of Erlikosaurus andrewsi (Dinosauria, Therizinosauria): new insights based on digital reconstruction . Journal of Vertebrate Paleontology 34 : 1263 – 1291 . Google Scholar CrossRef Search ADS Marshall JEA , Whiteside DI . 1980 . Marine influence in the Triassic “uplands” . Nature 287 : 627 – 628 . Google Scholar CrossRef Search ADS Morton JD , Whiteside DI , Hethke M , Benton MJ . 2017 . Biostratigraphy and geometric morphometrics of conchostracans (Crustacea, Branchiopoda) from the Late Triassic fissure deposits of Cromhall Quarry, UK . Palaeontology 60 : 349 – 374 . Google Scholar CrossRef Search ADS Porro LB , Rayfield EJ , Clack JA . 2015a . Descriptive anatomy and three-dimensional reconstruction of the skull of the early tetrapod Acanthostega gunnari Jarvik, 1952 . PLoS ONE 10 : e0118882 . Google Scholar CrossRef Search ADS Porro LB , Rayfield EJ , Clack JA . 2015b . Computed tomography, anatomical description and three-dimensional reconstruction of the lower jaw of Eusthenopteron foordi Whiteaves, 1881 from the Upper Devonian of Canada . Palaeontology 58 : 1 – 17 . Google Scholar CrossRef Search ADS Robinson PL . 1955 . Exhibition of specimens from Slickstones Quarry, Gloucestershire . Proceedings of the Geological Society, London 1527 : 113 – 115 . Robinson PL . 1957 . The Mesozoic fissures of the Bristol Channel area and their vertebrate faunas . Zoological Journal of the Linnean Society 43 : 260 – 282 . Google Scholar CrossRef Search ADS Robinson PL . 1971 . A problem of faunal replacement on Permo-Triassic continents . Palaeontology 14 : 131 – 153 . Robinson PL . 1973 . A problematic reptile from the British Upper Trias . Journal of the Geological Society, London 129 : 457 – 479 . Google Scholar CrossRef Search ADS Robinson PL . 1976 . How Sphenodon and Uromastyx grow their teeth and use them . In: Bellairs AD’A , Cox CB , eds. Morphology and biology of reptiles . London : Academic Press, Linnean Society Symposium Series , 43 – 67 . Robinson PL , Kermack KA , Joysey KA . 1952 . Exhibition of specimens from Slickstones Quarry, Gloucestershire . Proceedings of the Geological Society, London 1485 : 86 – 87 . Romer AS . 1956 . Osteology of the reptiles . Chicago : University of Chicago Press . Säilä LK . 2005 . A new species of the sphenodontian reptile Clevosaurus from the Lower Jurassic of South Wales . Palaeontology 48 : 817 – 831 . Google Scholar CrossRef Search ADS Sues HD , Shubin NH , Olsen PE . 1994 . A new sphenodontian (Lepidosauria: Rhynchocephalia) from the McCoy Brook Formation (Lower Jurassic) of Nova Scotia, Canada . Journal of Vertebrate Paleontology 14 : 327 – 340 . Google Scholar CrossRef Search ADS Swinton WE . 1939 . A new Triassic rhynchocephalian from Gloucestershire . Annals and Magazine of Natural History Series 11 4 : 591 – 594 . Google Scholar CrossRef Search ADS Underwood G . 1970 . The eye . In: Gans C , Parsons TS , eds. Biology of the Reptilia volume 2, morphology B . New York : Academic Press , 1 – 97 . van den Berg T , Whiteside DI , Viegas PA , Schouten R , Benton MJ . 2012 . The Late Triassic microvertebrate fauna of Tytherington, UK . Proceedings of the Geologists’ Association 123 : 638 – 648 . Google Scholar CrossRef Search ADS Walkden GM , Fraser NC . 1993 . Late Triassic fissure sediments and vertebrate faunas: environmental change and faunal succession at Cromhall, South West Britain . Modern Geology 18 : 511 – 535 . Whiteside DI . 1986 . The head skeleton of the Rhaetian sphenodontid Diphydontosaurus avonis gen. et. sp. nov. and the modernizing of a living fossil . Philosophical Transactions of the Royal Society London B 312 : 379 – 430 . Google Scholar CrossRef Search ADS Whiteside DI , Duffin CJ , Gill PG , Marshall JEA , Benton MJ . 2016 . The Late Triassic and Early Jurassic fissure faunas from Bristol and South Wales: stratigraphy and setting . Palaeontologia Polonica 67 : 257 – 287 . Whiteside DI , Marshall JEA . 2008 . The age, fauna and palaeoenvironment of the Late Triassic fissure deposits of Tytherington, South Gloucestershire, UK . Geological Magazine 145 : 102 – 147 . Google Scholar CrossRef Search ADS Wu XC . 1994 . Late Triassic-Early Jurassic sphenodontians from China and the phylogeny of the Sphenodontia . In: Fraser NC , Sues HD , eds. In the shadow of the dinosaurs: early Mesozoic tetrapods . New York : Cambridge University Press , 38 – 69 . © 2017 The Linnean Society of London, Zoological Journal of the Linnean Society This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Zoological Journal of the Linnean SocietyOxford University Press

Published: Dec 15, 2017

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off