During the Triassic Period, 200–251 Ma, many tetrapod clades first appeared and began to diversify, including dinosaurs, mammals, lissamphibians, pterosaurs, and turtles. Among the clades undergoing diversification were the lepidosauromorphs, especially the rhynchocephalians. Indeed, sphenodontians were actually the most common lepidosaurs during much of the early Mesozoic, and the most extensive records of early sphenodontian evolution have, historically, come from fissure fills of Late Triassic–Early Jurassic age in the UK (e.g. Evans 1984, 2003; Evans and Kermack 1994; Fraser 1994; Säilä 2005) and mainland Europe (Sigogneau-Russell et al. 1994), and, to a lesser extent, stratified deposits of Early Jurassic age in China (Wu 1994; Jones 2006a). Other Upper Triassic records of sphenodontian diversification are restricted to a few relatively complete skulls and a variety of more fragmentary fossils. Thus, relatively complete Upper Triassic sphenodontian skulls are known from the Lossiemouth Sandstone of Scotland (Huene 1910, 1912; Fraser and Benton 1989), the German Keuper (Jaekel 1911; Huene 1929; Fraser and Benton 1989), the Newark Supergroup in eastern North America (Sues and Baird 1993; Sues et al. 1994) and the Caturrita Formation in Brazil (Bonaparte and Sues 2006). More fragmentary fossils from various sites in the Chinle Group of the western USA (Murry 1986, 1987, 1989a, b; Fraser 1993; Harris et al. 1999; Heckert 2004), the Calcare di Zorzino in Italy (Renesto 1995), and the Elliot Formation in South Africa (Sues and Reisz 1995) complete the known Triassic record of sphenodontians. To date, the only verified report of a Triassic squamate is that of Tikiguana estesi Datta and Ray from the Tiki Formation in India (Datta and Ray 2006): all other purported Triassic squamate records have been disproven (Evans 2003). Here we report on the most complete sphenodontian recovered from the Upper Triassic Chinle Group in the American West, describing new morphological features and briefly commenting on their significance within the broader context of early sphenodontian evolution and biostratigraphy.
Abstract: We document here a new taxon of sphenodontian, Whitakersaurus bermani gen. et sp. nov., that is also the most complete sphenodontian fossil from the Upper Triassic Chinle Group in the south-western USA and the first Chinle sphenodontian represented by more than a single fragmentary dentulous element. The holotype was recovered during preparation of block C-8-82 from the famous Coelophysis (Whitaker) quarry at Ghost Ranch, New Mexico, and is the most complete small vertebrate recovered from the quarry. Detailed lithostratigraphy and geologic mapping demonstrate that the Whitaker quarry is in the Rock Point Formation of the Chinle Group, so Whitakersaurus is the first sphenodontian reported from this unit. Records of the phytosaur Redondasaurus at the quarry and elsewhere in the Chinle Group demonstrate that the quarry, and thus Whitakersaurus, is of Apachean (late Norian–Rhaetian) age. The sphenodontian specimen consists of incomplete left and right dentaries, a partial left? maxilla?, and impressions of a probable palatal element, all of which preserve multiple teeth. Whitakersaurus is distinct from other sphenodontians in possessing a unique combination of the following features: marginal dentition pleurodont anteriorly and posteriorly acrodont; pronounced heterodonty in dentary, with as many as 15 smaller, peg-like teeth anteriorly and several larger, posterior teeth that are conical and striated; faint radial ornamentation of posterior tooth crowns; presence of c. 19 dentary teeth; and absence of a distinct flange on posterior teeth. Numerous other details distinguish it from both more primitive and more derived taxa. Whitakersaurus, therefore, helps to document further mosaic evolution and an extensive diversification event of sphenodontians during Triassic time. Although sphenodontian taxa are relatively easily recognized, widely distributed, and common small- or microvertebrate fossils, the long stratigraphic ranges of taxa known from multiple specimens hinders their utility as index fossils with which to correlate strata across Pangaea.
Institutional abbreviations: CMNH, Carnegie Museum of Natural History, Pittsburg; MNA, Museum of Northern Arizona, Flagstaff; NMMNH, New Mexico Museum of Natural History and Science, Albuquerque.
Material and Methods
The fossil we describe here, NMMNH P-43125, is a fragmentary skull and incomplete lower jaws recovered during preparation of the NMMNH’s block from the famous Coelophysis quarry at Ghost Ranch, New Mexico (Text-fig. 1A). This large block (C-8-82; c. 2 m long, 1.4 m wide, and 1.2 m thick for an estimated mass of 5440 kg) was excavated by the CMNH in 1982 (Text-fig. 1B). From our communications with personnel at other institutions, it is clear that most blocks from the site excavated by the CMNH were carved out and transported right side up, and preparation commenced from the top down. However, we chose to rotate this block, approaching the principal fossil horizon from the underside (see Pierce et al. 2004 for a brief description of the technique), providing direct access to the densest, highest quality bonebed (see taphonomy section) and, we believe, directly facilitating recovery of the specimen described here.
In this paper we refer to the tooth-bearing bones, and the dentition as a whole, using standard anatomical terms for so-called ‘lower’ vertebrates (e.g. anterior/posterior, dorsal/ventral, lateral/medial). When referring to specific teeth we use dental terminology similar to that advocated by Smith and Dodson (2003) (e.g. mesial/distal, basal/apical, labial/lingual). When there is some chance of ambiguity, we clarify (e.g. use of rostral/caudal rather than anterior/posterior for bones associated with the mesial/distal dentition). The taxonomic headings of the systematic palaeontology section follow Evans et al. (2001) and Bonaparte and Sues (2006).
For purposes of illustration the specimen was photographed under both a light microscope and a scanning electron microscope (SEM). The light microscope is an Olympus SZX12 with dedicated QColor 5-megapixel digital camera utilizing QCapture software version 2.70. The specimen was originally photographed uncoated on a JEOL-JSM5800 SEM housed at the Institute of Meteoritics at the UNM Department of Earth and Planetary Sciences (micrographs in Pl. 1, figs 1, 3–6). However, better results were achieved after coating the specimen and using a Quanta 200 ESEM running XT microscope server imaging software housed at the College of Arts and Sciences microscopy facility at Appalachian State University (all other SEM images).
Stratigraphy and age
The famous Coelophysis (or Whitaker, after its discoverer) quarry at Ghost Ranch, north of Abiquiu, New Mexico, is situated high in the local Upper Triassic stratigraphic section (Text-fig. 1C). The quarry is located in strata mapped by Koning et al. (2006) as the Rock Point Formation, and stratigraphic work demonstrates that the quarry is c. 30 m above the contact of the Rock Point Formation with the underlying Painted Desert Member of the Petrified Forest Formation and 35 m below the Middle Jurassic Entrada Sandstone (Lucas and Hunt 1992; Hunt and Lucas 1993a–c; Sullivan et al. 1996; Lucas et al. 2003, 2005; Lucas and Tanner 2007). The Rock Point Formation vertebrate fauna at the Coelophysis quarry is distinct from that of underlying units and is considered to be Apachean (late Norian–Rhaetian) in age on the non-marine Triassic tetrapod timescale of Lucas (1998). Because of the long history of mistakenly attributing a lower stratigraphic position, and thus an erroneously old age to the quarry (e.g. Colbert 1989, 1990; Dubiel 1989a; Rowe and Gauthier 1990), we briefly review the stratigraphy of the Rock Point Formation, the fauna of the Coelophysis quarry and its age in the following paragraphs.
The stratigraphic position of the Coelophysis quarry has been variously identified as the upper Petrified Forest Member (= Painted Desert Member of the Petrified Forest Formation in our usage) (e.g. Colbert 1989, 1990; Schwartz and Gillette 1994), the ‘siltstone member’ (Stewart et al. 1972; O’Sullivan 1974, 1977; Nesbitt, 2007), Chinle Formation undifferentiated (e.g. Nesbitt and Norell 2005; Nesbitt et al. 2006), the Owl Rock Member (= Formation of our usage) (Dubiel 1989a; Goldstein et al. 1996), and the Rock Point Formation (= ‘Member’ of previous usage) (e.g. Lucas and Hunt 1992; Hunt and Lucas 1993a–c; Long and Murry 1995; Sullivan et al. 1996; Harris and Downs 2002; Tykoski and Rowe, 2004). We follow Lucas (1993, 1997) and correlate strata identified by Stewart et al. (1972) as the ‘siltstone member’ as part of the Rock Point Formation, as did Stewart et al.
Smith et al. (1961) provided the first reasonably detailed geological maps of the Ghost Ranch area, mapping the Ghost Ranch quadrangle at a scale of 1:24,000. Smith et al. (1961) mapped all upper Chinle sediments as the ‘upper shale member’ of the Chinle Formation, in large part because their map was somewhat generalized and relied on topographic expression to differentiate many units, thereby missing the relatively subtle contact of fine-grained, non-bentonitic Rock Point Formation sediments on fine-grained, bentonitic, slope-forming Painted Desert Member sediments. Koning et al. (2006), employing a more modern understanding of non-marine stratigraphy and sedimentology, provided a considerably more detailed map that accurately placed the Coelophysis quarry in the Rock Point Formation.
Assignment of the strata that contain the Ghost Ranch Coelophysis quarry to the Rock Point Formation is well supported because: (1) these strata closely resemble the type Rock Point strata in the Four Corners region (Harshbarger et al. 1957; Lucas et al. 1997); (2) Rock Point strata in the Chama Basin are at the top of the Chinle Group, as they are elsewhere (e.g. Stewart et al. 1972; Lucas 1993); and (3) Rock Point strata at Ghost Ranch yield Apachean-age vertebrate fossils (Lucas and Hunt 1993; Lucas et al. 2003, 2005). Stewart et al. (1972) recognized this correlation, equating the ‘siltstone member’ (our Rock Point Formation) to strata they assigned to the Rock Point Formation (Member) elsewhere.
Indeed, the Rock Point strata at Ghost Ranch are demonstrably different from underlying strata of the Painted Desert Member. They are dominantly siltstones and fine-grained sandstones lacking bentonitic (smectite/illite clay minerals). These strata are massive or else repetitively bedded in sets typically ranging from 1 to 5 m thick and weather to shades of orange (Koning et al. 2006). In contrast, strata of the Painted Desert Member are dominantly bentonitic mudrocks deposited as thick (3–10 m) tabular beds with occasional channel deposits of ripple-laminated to planar-laminated sandstone with minor intraformational conglomerate and, occasionally, well-developed palaeosols that, collectively, weather to shades of red and brown (Lucas 1993; Lucas et al. 2003, 2005; Koning et al. 2006).
Dubiel (1989a) and Goldstein et al. (1996; see also coauthored abstracts with Goldstein by Hargrave et al. 1996 and Trinh et al. 1996) assigned upper Chinle strata in the Ghost Ranch area to the Owl Rock ‘Member’ of the Chinle Formation based on sedimentological models (e.g. Dubiel 1989a, b), postulating that these strata are a lake margin facies of that unit. Lucas and Hunt (1992, p. 158) presented a detailed refutation of Dubiel’s assignment, and subsequent work on Owl Rock sedimentology (Tanner 2000) has rejected the sedimentological model of Owl Rock deposition advocated by Dubiel (1989a, b). Furthermore, Lucas et al. (2003, 2005) have demonstrated that the correlations of Goldstein and colleagues are contrary to all available outcrop data.
Coelophysis is by far the most abundant vertebrate known from the Ghost Ranch quarry, but by no means is it the only taxon known from the site. Other vertebrates reported from the Coelophysis quarry include an indeterminate chondostrean, redfieldiid and coelacanthid osteichthyans, a drepanosaurid, the phytosaur Redondasaurus bermani Hunt and Lucas, the sphenosuchian Hesperosuchus, the enigmatic archosaur Vancleavea or a closely allied form, the rauisuchian Postosuchus, and the shuvosaurid Shuvosaurus (=Effigia) okeeffeae Nesbitt and Norell (Schaeffer 1967; Colbert 1989; Hunt and Lucas 1993a–c; Long and Murry 1995; Clark et al. 2000; Harris and Downs 2002; Lucas et al. 2003; Nesbitt and Norell 2005; Lucas et al. 2007; Nesbitt 2007).
Of the vertebrate taxa known from the site, R. bermani is the most age diagnostic: records of Redondasaurus we recognize are restricted to uppermost Triassic rocks, including the Redonda Formation in east-central New Mexico, the Rock Point Formation, and the basal Wingate Sandstone in Utah (Hunt and Lucas 1993c; Lucas et al. 1997; Heckert et al. 2001; Hungerbühler 2002; Lucas and Tanner 2007). While some have questioned the validity of this taxon (e.g. Long and Murry 1995), others (e.g. Hungerbühler 2002) recognize it as a valid genus, which, combined with its stratigraphic distribution, validates its utility as an index taxon of the Apachean land-vertebrate faunachron (Lucas, 1998). Unfortunately, all other vertebrates from the Coelophysis quarry are relatively non-diagnostic from a biostratigraphic perspective, as all of the fish as well as Hesperosuchus, Postosuchus, and Vancleavea have long stratigraphic ranges (Huber et al. 1993; Long and Murry 1995; Lucas 1997; Hunt et al. 2002, 2005). Coelophysoid theropods are extremely difficult to identify at the genus level and have a stratigraphic ‘career’ that ranges from the Carnian to the Hettangian (early Late Triassic–Early Jurassic) (Heckert and Lucas 2000). To date, Shuvosaurus okeeffae has only been found at the Coelophysis quarry, and, thus, is not of biostratigraphic utility, and other records of Shuvosaurus, S. inexpectatus, are from strata of older, Revueltian, age in West Texas (Lucas et al. 2007).
Turning the block over prior to preparation permitted careful microstratigraphic study of bedding under the main bonebed. The block shows two fining-upward sequences above a basal sandy siltstone (Text-fig. 1C). The base of sequence 1 contains the ostracode Darwinula and the conchostracan Lioestheria, and probably indicates a topographic low that was the site of an ephemeral pond. The basal coarse silt and very fine sand of sequence 1 fines up to very fine silt with clay lenses (units 2–5). Sequence 2 (units 6–7) fines up from a basal matrix-supported sandy conglomerate to very fine silt. Beds 4–7 contain numerous elongate rip-up mud clasts that are closely aligned and trend WNW–ESE. Trends were measured on 44 rip-ups; 90 per cent were aligned to within 20 degrees. The dinosaur skeletons show general alignment with the mud clasts. In the uppermost bed, above the main bonebed, bones are less common, somewhat disarticulated and not as well preserved. This probably indicates partial reworking of this layer. Overall, the block stratigraphy indicates at least two flood events of increasing energy (Rinehart et al. 2004b).
To date, the fauna from block C-8-82 is relatively representative of the quarry as a whole. It comprises the invertebrates mentioned previously and the redfieldiid fish Synorichthys, the coelacanth fish Chinlea, the archosauromorph Vancleavea, the sphenodontian described here, a juvenile phytosaur, possibly pertaining to Redondasaurus gregorii, and the theropod dinosaur Coelophysis bauri. Additionally, an indeterminate redfieldiid-like fish, several isolated enigmatic scutes, and teeth of an archosauriform (possibly a sphenosuchian or ornithischian dinosaur) are present. In general, the fossil material coarsens up throughout the block. Sequence 1 contains (in ascending order) invertebrates, scattered fish scales and bones, articulated fish, small juvenile dinosaurs, and Vancleavea. Sequence 2 yielded the sphenodontian, the phytosaur, and larger juvenile and adult dinosaurs (Rinehart et al. 2004a).
Completely articulated skeletons of Coelophysis are the norm in the block. Given the excellent preservation and complete lack of evidence of scavenging or weathering, we believe that the most parsimonious taphonomic interpretation is that a large congregation of Coelophysis, along with several other animals on a river floodplain, were overcome by a sequence of flood events, washed into a topographic low, and immediately buried.
The sphenodontian fossil, NMMNH P-43125, was discovered in the main dinosaur bone layer of the block, c. 300 mm below the top (star, Text-fig. 1B–C) and within 2 cm of the westernmost edge. Within this area of the block the matrix is finely brecciated. Nearby there is a small fault with several cm of offset (bone sequences, such as articulated vertebral series, wrap up and around the throw of the fault) that has caused some disruption of the sediments. Additionally, desiccation during the 25 years since excavation has caused the rock to shrink away from the field jacket and to crumble into the resulting void. It is unknown how much of the sphenodontian was originally fossilized, but these factors probably account for the fact that so little of it has been recovered from a block that contains so many complete skeletal fossils.
DIAPSIDA Osborn, 1903
LEPIDOSAURIA Haeckel, 1866
RHYNCHOCEPHALIA Günter, 1867
SPHENODONTIA Williston, 1925
WHITAKERSAURUS gen. nov.
Derivation of name. After the late George O. Whitaker, field assistant to Edwin H. Colbert and the discoverer of the Coelophysis quarry at Ghost Ranch, and Latin, saurus, lizard.
Type and only known species. Whitakersaurus bermani.
Occurrence. Known only from the type locality, the Coelophysis (Whitaker) quarry, Upper Triassic of north-central New Mexico, USA.
Diagnosis. As for the type and only species, Whitakersaurus bermani (see below).
Whitakersaurus bermani sp. nov.
Text-figures 2–3, Plates 1–2
Derivation of name. After David S. Berman, palaeontologist at the Carnegie Museum whose field work in the early 1980s resulted in excavation of numerous blocks from the Coelophysis quarry, including the block from which the holotype was obtained.
Holotype. NMMNH P-43125, an associated maxillary fragment and incomplete lower jaws on two small blocks.
Locality. The Coelophysis (Whitaker) quarry, as above. The NMMNH locality number for this site is L-3115, but many other institutions also house fossils of associated vertebrates from this site (see Colbert 1989).
Horizon and age. Rock Point Formation (Upper Triassic: Norian–Rhaetian). The tetrapod fauna of the Rock Point Formation is of Apachean age, which corresponds to a late Norian–Rhaetian (Late–latest Triassic) age on marine timescales (Lucas 1998; Lucas and Tanner 2007).
Diagnosis. A genus of sphenodontian lepidosaur distinguished from Gephyrosaurus (as well as most squamates) by its reduced tooth count (less than 20 dentary teeth) consisting of anteriorly pleurodont and posteriorly acrodont dentition and pronounced heterodonty, with posterior teeth significantly larger than the anterior teeth; ridged posterior teeth that are more ornamented than those of Diphydontosaurus; also distinct from Diphydontosaurus in that the teeth are taller, more conical, and possess a radial ornamentation; distinguished from both Diphydontosaurus and Gephyrosaurus in lacking the normal alternating replacement pattern for pleurodont dentitions described by Edmunds (1960); also distinguished from Gephyrosaurus by its much lower tooth count (<20 as opposed to c. 40); distinct from Planocephalosaurus and other, more derived sphenodontians (e.g. Clevosauridae, Sphenodontinae, Eilinodontinae, and Opisthodontia sensuApesteguía and Novas 2003) by presence of at least 19 dentary teeth and the lack of a distinct flange on the posterior dentition. Whitakersaurus differs from Paleollanosaurus in lacking tiny posterior teeth. The larger teeth of Whitakersaurus also exhibit more texture than those of Paleollanosaurus, are not as closely spaced, and are more conical and less recurved. The pronounced size increase of the posterior additional teeth differs from those of Opisthias, and Whitakersaurus contains many more teeth than Polysphenodon.
Description. NMMNH P-43125 is preserved on two small (c. 1.5-cm-square) part-counterpart blocks of matrix, which we designate block A (larger part) and block B (smaller counterpart) (Text-fig. 2). Block A (Text-fig. 2C–D) contains fragments of three elements and impressions of a fourth. Bones preserved on block A include a left? maxilla? fragment, an incomplete left dentary, and part of the right dentary. Bone impressions preserved on block A include most of the remainder of the right dentary and a tooth-bearing element that we tentatively interpret as a palatal? element. Block B (Text-fig. 2A–B) preserves the majority of the right dentary as a counterpart to the impressions in block A and some bone impressions that probably do not pertain to a sphenodontian, as they are much larger than the elements described here. Block A is c. 15 mm long parallel to the lower jaw elements, as much as 12 mm wide perpendicular to them, and as much as 7 mm thick. Block B is much smaller: c. 10 mm long, 7 mm wide and 4 mm thick. We are certain that the holotype documented here represents a single individual because the two blocks (A and B) are part and counterpart and all of the specimens are preserved in close association (the lower jaws are in near articulation). Given the close association of these elements as well as the rarity of small vertebrates from this site, we are confident that all the rhynchocephalian fossils we describe here pertain to a single individual.
On block A, the left dentary is best preserved in occlusal view, although much of the lateral surface is visible, as is a small portion of the anteriormost medial surface near the symphysis (Text-fig. 2C–D; Pl. 1 figs 2–5; Pl. 2 figs 2–6). The coronoid process appears to have been lost, but the posteriormost portion of the preserved bone may preserve the articulation with the surangular. As preserved, the specimen is c. 11 mm long, and c. 19 tooth positions are visible. Complete or nearly complete teeth occupy 13 of the 19 preserved positions. Four broken teeth are evident: these are in tooth positions 1 and 4–6 (Text-fig. 3B; Pl. 1, fig. 3). Position 10 is broken at or below the tooth line, and matrix and breakage similarly obscures a possible tooth position in the vicinity of the fourteenth preserved tooth position. The first nine tooth positions are pleurodont, with the tooth roots exposed on the lingual side. All teeth posterior to the tenth tooth position are solidly fused to the jaw, exhibiting the acrodont dentition typical of most derived sphenodontians. The dentary is broken between tooth positions 18 and 19, and the dorsal surface is broken posterior to that position, so we are not certain that the dentition consisted of only 19 positions, although the right dentary appears to confirm this. A thin line extending dorsoventrally from behind the nineteenth position could be a suture with another element, probably the surangular (Text-fig. 2D).
The dentary teeth are markedly heterodont, with smaller, peg-like teeth in the anteriormost 14 positions and larger, laterally compressed, conical teeth in positions 15–19 (Text-fig. 2D; Pl. 1, fig. 2; Pl. 2, figs 2–6). The anterior teeth gradually increase in size from the symphysis posteriorly, with teeth in positions 11–14 at least twice the diameter and height of most anterior teeth. However, the teeth in positions 17–19 (Pl. 1, figs 2, 4; Text-fig. 3C) are more than twice the size of the teeth in positions 11–14 (Pl. 1, figs 3, 5; Text-fig. 3B), and the change in shape is abrupt and evident by tooth position 15 (Pl. 1, fig. 2). The anterior teeth appear proportionately tall (2–3 times basoapically tall relative to mesiodistal length or labiolingual width), whereas the posterior teeth have mesiodistally elongate bases and, therefore, are no more than c. 1.5 times as tall as mesiodistally long. The anterior teeth preserve little, if any, ornamentation, whereas the posterior teeth are striated vertically, with the numerous fine striations radiating basally from the apex of the tooth. For example, the largest and best preserved of these teeth, tooth 18, has a mesial, a posterior, and two faint labial ridges (Pl. 1, fig. 4; Pl. 2, fig. 6). There is a slight tendency for the ridges to develop to the extent that a Planocephalosaurus-like groove is visible just lingual to the distal margin of teeth 16 and 18.
The preserved maxilla fragment is less than 2 mm long and preserves all or part of four relatively large teeth, presumably from the posterior dentition (Text-figs 2C–D, 3D; Pl. 1, fig. 6; Pl. 2, fig. 1). We describe this element as if it were a left maxilla visible in lateral view. We do this in part because the impressions of the opposite side of the bone show a relatively smooth surface lacking nutrient foramina or other texture, which is otherwise common on the external (lateral) surface of reptilian bones. The teeth are too large and the element too rod-like (not platy) to consider this a palatal element. Enough bone is preserved to demonstrate acrodont implantation, and the preserved bone and associated impression give some indication of the thickness of the bone (Text-figs 2D, 3D), although little else can be said regarding the bone itself. The preserved teeth are closely spaced, but only the first preserved tooth position contains a complete tooth; the teeth in the other three positions are broken. The teeth are conical and preserve striations radiating basally from the tip (Text-fig. 3D; Pl. 2, fig. 1). Five ridges are evident on the exposed side of the most complete tooth, one each along the mesial and distal surfaces and three on the presumed labial surface. At least one of the three bifurcates near its basal terminus. The complete tooth is very slightly taller than mesiodistally long, laterally compressed, and appears slightly recurved towards the others, supporting our interpretation that this is the anteriormost preserved tooth on the fragment. The second tooth appears similar, and the third and fourth teeth are so close together that their bases are nearly coalesced.
An additional fragment on block A consists of a thin rod of bone and associated impressions, the latter of which indicate the presence of six of the smaller, peg-like teeth (Text-fig. 2C–D). This element is thin and platy, with low, conical teeth. We interpret this as a probable palatal element, presumably either the palatine or the pterygoid. The tooth crowns, while tiny, are nearly as tall as the element is thick. The element thus resembles the pterygoid of Clevosaurus (see Fraser, 1988, fig. 14), although, because only a single tooth row is evident, the palatine cannot be ruled out, even though it tends to possess larger teeth. The teeth are numerous and more regularly arranged relative to those of the vomer in Clevosaurus (Fraser, 1988, fig. 14) or Planocephalosaurus (Fraser, 1982, fig. 1b).
Block B includes a right dentary and the impression of part of another bone that is significantly larger than any associated with the sphenodontian, so it probably pertains to another taxon, presumably one of the larger archosaurs that are abundant at the quarry (Text-fig. 2A–B). The dentary is preserved in lateral aspect, although some of the teeth, especially the anterior ones, are clearly visible in occlusal view, and the ventral margin of the bone is also visible. The posterior portion of the tooth row is broken and displaced ventrally. Total length as preserved is c. 9 mm. Ventral to the tooth row several nutrient foramina are evident, particularly beneath the second, fourth, and seventh tooth positions. At a minimum, there are 16 preserved tooth positions, although breakage on either side of tooth position 7 could conceivably mask the presence of as many as two more positions (not included in tooth counts reported here). All positions are occupied by teeth, and only in the first, thirteenth and sixteenth preserved positions are the teeth broken. As in the left dentary in block A, the teeth are strongly heterodont, with the first 12 preserved positions occupied by small, relatively tall, peg-like teeth. Much larger, more mesial–distally elongate teeth that were presumably similar to the posterior teeth of the left dentary occupied positions 14 and 16, and larger, conical, striated teeth occupy positions 17 and 18 (Text-figs 2A, 3A). Given the preservation of the specimen, it is extremely difficult to assess tooth implantation, but the first nine tooth positions are exposed much more deeply on the lingual side, and we interpret these as pleurodont. The more posterior teeth are clearly fused to the dentary but are only visible in labial to obliquely occlusal view, and we interpret these as acrodont.
Comparisons of tooth counts suggest that no more than 1–2 positions are missing from the right dentary. Presumably the missing position(s) are from its anteriormost part or the damaged area in the vicinity of the seventh preserved tooth position. The posteriormost four teeth on the right dentary form an alternating pattern of a smaller, then larger, tooth, with both large teeth having a much smaller tooth anterior to them. These teeth probably represent the ‘additional teeth’ of sphenodontian dentitions described by Robinson (1976), and P-43125 therefore exhibits relatively few (as few as 3, as many as 5) ‘additional teeth’.
Discussion. The small size and numerous tiny teeth of P-43125 suggest that it may pertain to a juvenile, or even hatchling, individual, although it is not markedly smaller than many already described Late Triassic and Early Jurassic sphenodontians. Modern Sphenodon hatchlings possess a dentition of numerous small teeth termed the ‘hatchling dentition’ by Robinson (1976). The dentary dentition in P-43125 differs from the hatchling dentition of Sphenodon because, except in the four posteriormost teeth described above, it lacks the alternating pattern of large and small teeth described by Robinson (1976). Another possibility is that the small size of P-43125 may indicate that it is a juvenile but that the tooth replacement scheme documented by Robinson (1976) had not developed in this particular lineage, although it has been reported in other Triassic and Jurassic taxa (e.g. Fraser 1986, 1988; Reynoso 2003).
It remains possible that the anterior dentition of P-43125 is pleurodont, as the lingual surface of the tooth row is visible at a level below that of the labial surface in both dentaries. This area is very difficult to observe, however, as it is only somewhat visible at a high angle to the bone, and too little bone is exposed to evaluate, for example, the size, shape, and position of the Meckelian groove (Text-fig. 3B; Pl. 1, fig. 3).
Comparisons. Because many early sphenodontians are known from abundant, if disarticulated to disassociated, material from rich fissure fills (e.g. Evans 1980; Fraser 1982, 1986; Whiteside 1986; Säilä 2005) and similar deposits (Evans et al. 2001), or from remarkably complete skulls (e.g. Sues et al. 1994; Bonaparte and Sues 2006), we choose to provide more detailed comparisons with these taxa to justify the erection of a new genus for the Coelophysis quarry material. As in the diagnosis, we attempt to discuss the taxa in largely systematic fashion, from more primitive rhynchocephalians (e.g. Gephyrosaurus) to more derived sphenodontians and then superficially similar, but distantly related, taxa such as acrodont squamates. We follow Evans (2003) and Apesteguía and Novas (2003) for systematic placement, with more fragmentary taxa that have not been analyzed with cladistic methods interspersed at our interpretation of their approximate phylogenetic position (Text-fig. 4).
The dentition of Whitakersaurus differs from that of Diphydontosaurus, in which the majority of the anterior teeth are distinctly pleurodont (Whiteside 1986). Approximately half (nine of 20) tooth positions of Whitakersaurus are pleurodont, with the more posterior teeth acrodont. Whitakersaurus also possesses fewer dentary teeth (20) than Diphydontosaurus (24) (Whiteside 1986). The specimen we describe is comparable in size to the Diphydontosaurus skeleton from the Calcare di Zorzino described by Renesto (1995). Unlike the Italian specimen, however, P-43125 has fewer pleurodont teeth anteriorly, and the degree of heterodonty described here is greater than the ‘gradual increase in size posteriorly’ described by Renesto (1995, p. 151). Additionally, the left dentary we describe is less robust and tapers to a thin symphysis, unlike the wider symphysis described by Renesto (1995).
Both Gephyrosaurus (Evans 1980) and Diphydontosaurus (Whiteside 1986) apparently have the ‘typical’ tooth replacement scheme documented for non-acrodont reptiles by Edmund (1960). In this scheme, alternating tooth positions are replaced in waves that pass anteriorly through the marginal dentition. No such replacement pattern is evident in Whitakersaurus.
NMMNH P-43125 is distinct from problematic, fragmentary taxa described from the Placerias quarry in the Upper Triassic Bluewater Creek Formation of east-central Arizona by Murry (1987). Specifically, it is much smaller than the putative ‘eolacertilian’ illustrated by Murry (1987, fig. 9). Similarly, it has a more homodont, and generally smaller, dentition than the putative kuehneosaurid Murry (1987, figs 7–8) described, most of which he felt exhibited ‘subpleurodont,’ not acrodont, implantation.
Harris et al. (1999) provided a preliminary redescription of acrodont tooth fragments from the Placerias quarry assigned to the Sphenodontia by Murry (1987), assigning four morphotypes (A–D) to the various dentigerous fragments. Morphotype A teeth are strongly ridged and resemble those of Planocephalosaurus and, thus, are not relevant to discussion here. The ridges on teeth of this morphotype are much stronger (better developed) than those seen on NMMNH P-43125. Morphotype B teeth are somewhat similar to those of Whitakersaurus in that they are ‘parabolic in side view, moderately laterally compressed, and lack surface ornamentation but possess a sagittal ridge’ (Harris et al. 1999, p. 49A). Morphotype C teeth are tall and conical, with radial ridges, but unlike the tall, conical posterior teeth documented here, are uncompressed laterally. The teeth of Whitakersaurus are neither as tall nor as cylindrical as the morphotype D teeth of Harris et al. (1999). Thus, although there are some similarities between Whitakersaurus and the teeth from the Placerias quarry, no synapomorphy or autapomorphy is evident to refer definitively any of the latter to the former.
In many ways, the specimen described here is similar to Planocephalosaurus, as both taxa possess teeth with distinctive radial ridges and have a similar tooth count (Fraser 1982). Erection of a new genus for Whitakersaurus is warranted, however, based in large part on the following differences from Planocephalosaurus: (1) only some of the teeth in Whitakersaurus are ridged, whereas all, or nearly all, those of Planocephalosaurus are ridged (Fraser 1982); (2) the ridges on Whitakersaurus teeth are comparatively less well developed (less pronounced) than those of Planocephalosaurus; (3) the teeth of Whitakersaurus are not recurved to the extent seen in the additional teeth of Planocephalosaurus; (4) Whitakersaurus lacks the single, extremely large, posteriormost additional tooth that characterizes the dentition of most species of Planocephalosaurus; and (5) Whitakersaurus possesses significantly more dentary teeth (minimum 18) than Planocephalosaurus (maximum 14), the anterior teeth lack the striations typical of Planocephalosaurus, and posterior teeth lack more than the rudimentary flange or divot described in the posterior teeth (Pl. 1, fig. 5; Text-fig. 3C): there is also more uniformity in posterior teeth in Whitakersaurus than illustrated for Planocephalosaurus (Fraser 1982). Heckert (2004) erected P. lucasi for admittedly fragmentary specimens of Planocephalosaurus from the Tecovas Formation in West Texas. Teeth 7–9 and 11–13 in the left lower jaw of Whitakersaurus (Pl. 2, fig. 4) are similar in shape to those of P. lucasi, but not as distinctly ridged on their lingual surface, and teeth 11–13 are acrodont as opposed to the pleurodont condition observed in P. lucasi (Heckert, 2004). Thus, the accumulated differences we document here are at least as great as the differences between Planocephalosaurus and other existing sphenodontian genera (e.g. Fraser 1982, 1986).
Whitakersaurus is significantly larger than Rebbanasaurus jainiEvans et al. (2001). Other distinguishing characteristics include the fact that teeth of Whitakersaurus are more strongly striated and lack the anterior faceting described in Rebbanasaurus by Evans et al. The dentary of Whitakersaurus also lacks both the pronounced symphyseal notch and the prominent subdental shelf of Rebbanasaurus.
Whitakersaurus is larger than Godavarisaurus lateefiEvans et al. (2001). It has teeth that are more conical (less backswept or recurved and lack the asymmetrical texture: lingual ridges but labially polished surfaces) of Godavarisaurus. The anterior dentary teeth of Whitakersaurus are smaller than those of Godavarisaurus, and the dentary is proportionately thicker as well.
The most obvious difference between the holotype of Whitakersaurus bermani and known specimens of Clevosaurus is the lack of the distinct flange in the New Mexican specimen that typifies Clevosaurus teeth (Fraser 1988). The holotype skull of Clevosaurus brasiliensisBonaparte and Sues, 2006, while similar in size to P-43125, completely lacks an anterior dentition and, therefore, represents an adult. Thus the similar-sized P-43125, which does possess an anterior dentition, clearly represents a taxon distinct from C. brasiliensis. The left dentary appears too low and elongate to belong to Clevosaurus (cf. Bonaparte and Sues 2006). The anterior teeth of P-43125 are larger and better developed than the hatchling teeth of C. cornvallis (e.g. Säilä 2005, text-fig. 6).
Details of the dentition readily distinguish Whitakersaurus from both Sigmala sigmala Fraser and Pelecymala robustus Fraser. Whitakersaurus lacks the well-developed wear facets and mesiodistal flanges that typify Sigmala teeth, and the increase in additional tooth size seen in Whitakersaurus also differs from that of Sigmala (Fraser 1986). No dentaries of Pelecymala were reported by Fraser (1986), but the maxillary dentition is particularly distinct, with teeth ‘displaying a slight tendency towards transverse broadening’ (Fraser 1986, p. 172). This feature is difficult to ascertain in the New Mexican specimen, but the upper teeth preserved in the putative maxilla fragment on Block A lack any evidence of transverse broadening, and this is not seen at all in the lower dentition of Whitakersaurus, or indeed in most sphenodontians.
Whitakersaurus is broadly similar in size to Polysphenodon mulleri Jaekel from the Middle Keuper (Carnian or lower Norian) near Hannover (Hoffmannthal near Fallersleben), for which there is no dentary (Fraser and Benton 1989). However, the premaxilla in Polysphenodon has as few as two teeth, and the maxilla has 8–10 teeth, so it appears that the dentary tooth count of Whitakersaurus (c. 19) is significantly higher than that expected of Polysphenodon (c. 14).
Whitakersaurus is similarly distinct from Brachyrhinodon taylori Huene from the Lossiemouth Sandstone in Scotland for similar reasons. The dentary of Whitakersaurus contains many more teeth, and more posterior teeth, than observed in the upper dentition of Brachyrhinodon (its lower dentition is hidden by the occlusion of the holotype skull; Fraser and Benton 1989).
The only other broadly contemporaneous sphenodontians with which to compare the Ghost Ranch specimen are fragmentary specimens from the Lower Jurassic Kayenta Formation first reported by Meszoely et al. (1987) and briefly described by Curtis and Padian (1999). The Kayenta sphenodontians most obviously differ from Whitakersaurus in that their posteriormost teeth decrease in size from the anterior dentition (Curtis and Padian 1999), whereas these teeth are the largest in Whitakersaurus. The Kayenta specimens also have well-developed alternation in tooth size throughout, where this feature is only evident in the posteriormost teeth of Whitakersaurus.
Whitakersaurus is superficially similar to a few squamate taxa. It is differentiated from acrodont iguanians (e.g. Bharatagama, Priscagama, Tinosaurus) by the presence of a continuous tooth row (absence of a diastemma) and by a higher dentary tooth count than seen in these taxa (e.g. Evans et al. 2002). It is also distinct from Tikiguania and Tinosaurus in the absence of polycuspate teeth and the fact that the dentary is more markedly heterodont than in either of those taxa (Datta and Ray 2006). Whitakersaurus possesses many more teeth that are much simpler in construction than those of the acrodont squamate Uromastyx (Robinson 1976).
Whitakersaurus is significant for many reasons. It is one of the few tetrapods known from the Rock Point Formation, the most complete sphenodontian thus far recovered from the Chinle Group, and appears to represent another taxon with a mosaic of features that appears relatively early in sphenodontian evolution, which highlights the apparently rapid global diversification and radiation of sphenodontians. Although there are some correlations based on sphenodontian fossils, their long stratigraphic ranges preclude more precise biostratigraphic correlations in spite of this diversification event.
Chinle Group sphenodontians
All previous records of sphenodontians from the Chinle Group consist of small, dentulous fragments recovered during screenwashing and assigned to the Sphenodontia on the basis of acrodont tooth implantation. The first such records were fragmentary specimens described by Murry (1982, 1986) from the Tecovas Formation of West Texas and the Redonda Formation of east-central New Mexico. Murry (1987) also described several dentulous fragments of sphenodontians from the Placerias quarry (Bluewater Creek Formation, east-central Arizona), and Kirby (1989, 1991) described even more fragmentary material from the Owl Rock Formation in north-central Arizona. Fraser (1993) identified a probable occurrence of Clevosaurus from the Placerias quarry that was subsequently illustrated by Kaye and Padian (1994). Harris et al. (1999) briefly re-examined this material, assigning it to four tooth morphotypes (A–D), some of which were broadly similar, but not assignable, to Diphydontosaurus and Planocephalosaurus (see preceding section). Heckert (2004) documented numerous occurrences of sphenodontians from lower Chinle Group strata, principally the Tecovas Formation. These records included taxa he named Paleollanosaurus fraseri and Planocephalosaurus lucasi as well as more problematic fossils assigned to aff. Diphydontosaurus, aff. Clevosaurus, and other, indeterminate sphenodontians. All of these fossils are so small and fragmentary that many are not even identified to element.
As shown in the diagnosis and comparisons documented previously, Whitakersaurus is distinct from all named Chinle taxa. It is also a much more complete specimen than any previously known Chinle lepidosauromorph (the putative lepidosauromorph Dolabrosaurus aquitilisBerman and Reisz, 1992, is now recognized as a drepanosaurid and, therefore, either an archosauriform or else a basal neodiapsid; Renesto 2003; Senter 2004). Indeed, Whitakersaurus is the only sphenodontian from the Chinle Group recovered from ‘traditional’ excavation and preparation, not screenwashing. The most complete previously described Chinle sphenodontian is the holotype of Paleollanosaurus fraseri, a left dentary fragment with ten preserved tooth positions in the c. 3 mm of preserved bone. All other Chinle sphenodontians documented to date preserve six or fewer (typically 1–3) teeth on similarly sized or smaller fragments.
Historically the phylogenetic position of rhynchocephalians was poorly understood. The advent of modern biological methods and the application of cladistic methods to fossils in the 1980s led to the now-accepted crown group Lepidosauria consisting of Sphenodontia (e.g. Evans 1980, 1984, 1988, 2003; Benton 1985; Gauthier et al. 1988; Fraser and Benton 1989) and Squamata (e.g. Gauthier et al. 1988; Benton 1991; Carroll and Currie 1991); the latter is not germane to discussion here (but see Evans 2003 for a recent review). Various taxa outside the crown group (e.g. eolacertilians, kuhneosaurs) are typically included within the Lepidosauromorpha by these authors. Numerous soft-tissue synapomorphies support the hypothesis of a monophyletic Lepidosauria (Gauthier 1994), as do molecular data (Hedges and Poling 1999). Subsequent work has not challenged the monophyly of the crown group, and since the late 1980s, most phylogenetic work on sphenodontians has focused on ingroup relationships (e.g. Fraser and Benton 1989; Sues et al. 1994; Wu 1994; Reynoso 1996; Wilkinson and Benton 1996; Evans et al. 2001). A fully fledged phylogenetic analysis is beyond the scope of this paper, especially as the new specimen lacks most of the skull and all postcrania, but it is still possible to evaluate qualitatively its relationships, drawing mostly on the analyses of Apesteguía and Novas (2003) with additional insight from Evans (2003) and Datta and Ray (2006) (Text-fig. 4).
Whitakersaurus possesses several distinctly sphenodontian features: the teeth exhibit acrodont implantation (at least posteriorly) and are conical and striated, with the total tooth count reduced to c. 20 positions in each jaw quadrant. Other characteristics, however, are more primitive than in many coeval sphenodontians. These include the relatively simple, conical tooth crowns that lack a distinct flange on the posterior teeth, the pleurodont anterior dentary dentition, the larger number of tooth positions (>14), the weakly developed striations on the teeth, and the apparent lack of development of the more derived sphenodontian tooth replacement pattern. Thus, while Whitakersaurus resembles more primitive taxa such as Diphydontosaurus, it also resembles more derived sphenodontians such as Planocephalosaurus; hence, our tentative placement of Whitakersaurus in Text-figure 4, which also shows that numerous relatively derived taxa, including Clevosaurus, Brachyrhinodon, and Polysphenodon, have first appearances before comparatively more primitive rhynchocephalian taxa such as Gephyrosaurus and Diphydontosaurus. Indeed, the stratigraphically disjunct appearance of hypothesized primitive taxa later than derived taxa is probably a reflection of just how incomplete the fossil record of lepidosaurian evolution is for the Triassic–Jurassic interval. We anticipate that additional collecting will help to resolve these discrepancies, but in the meantime it appears that sphenodontians are an outstanding example of mosaic evolution. Our reading of the Triassic–Jurassic lepidosaurian fossil record suggests that numerous taxa independently acquired (and may have subsequently lost) varied traits associated with a more derived position. Specifically, in addition to the apparent rhynchocephalian ghost lineage leading to Gephyrosaurus, key features such as acrodonty, reduction of tooth count and breadth of the mandibular symphysis, developed in both Squamata (e.g. Tikiguania) and throughout many Triassic–Jurassic sphenodontians (see character state distribution in Apesteguía and Novas 2003). Thus, Tikiguania and sphenodontians more derived than Planocephalosaurus are fully acrodont, with varying degrees of acrodonty expressed by Diphydontosaurus, Planocephalosaurus, and Whitakersaurus. Similarly, tooth counts are variable, with early occurrences of taxa with greatly reduced and specialized dentitions (e.g. Clevosaurus) and later-appearing taxa retaining higher tooth counts and more generalized dentitions (e.g. Gephyrosaurus, Diphydontosaurus). Similarly, tooth morphology varies considerably between these taxa, with much of the tooth morphospace ever occupied by sphenodontians (e.g. Jones 2006b) utilized by Early Jurassic time. One consequence of this pattern of evolution is that fragmentary specimens exhibiting acrodont dentition are no longer safely assigned to Sphenodontia (e.g. Murry 1987; Kaye and Padian 1994; Heckert 2004) unless they bear additional features of Sphenodontia (e.g. tooth crown shape and ornamentation) that distinguish them from the acrodont squamate Tikiguania (Datta and Ray 2006).
Text-figures 4 and 5 establish several facts about our present knowledge of early lepidosaur evolution: (1) by Late Triassic time lepidosaurs had a Pangaean distribution; (2) there are more localities yielding Late Triassic remains than there are records from strata of Early Jurassic age, although stratigraphic uncertainties associated with fissure-fill deposits in Europe somewhat complicate this assessment; (3) by the Late Triassic sphenodontians had diversified extensively, and are represented by taxa ranging from basal (e.g. Diphydontosaurus) to relatively derived (Clevosaurus spp.); (4) Late Triassic sphenodontians (nine genera, ten or more species) are more diverse and globally distributed than are those of Early Jurassic age (seven genera, eight or more species as well as the rhynchocephalian Gephyrosaurus); (5) the fossil record is too fragmentary to correlate the reduced diversity of Early Jurassic sphenodontians with any sort of putative Triassic–Jurassic extinction event; and finally (6) the Middle Jurassic record of sphenodontians is particularly poor, consisting entirely of unnamed, fragmentary fossils from screenwash localities (e.g. Evans 1992; Evans and Milner 1994). This last point is especially relevant now that the La Boca Formation of Mexico, and thus its sphenodontian records (Reynoso 1996, 2003, 2005; Reynoso and Clark 1998) have been shown to be of Early Jurassic age (Barboza-Gudino et al. 1998, 1999).
We suspect that the majority of these observations are artefacts of the Triassic fossil record, which is especially incomplete with regard to small terrestrial tetrapods. As previous authors have demonstrated, a particularly frustrating aspect of the rhynchocephalian record is that the best (most abundant and completely preserved) records tend to come from fissure fills or other deposits for which stratigraphic relationships are difficult to ascertain (e.g. Evans and Kermack 1994; Fraser 1994). Conversely, well-constrained, reliably dated stratigraphic units yield few, if any, reasonably complete sphenodontian specimens.
Combined with the relatively long stratigraphic ranges of lepidosaurs generally and sphenodontian genera in particular, this hinders attempts to use them to correlate strata. As an example, Clevosaurus would appear to be an excellent candidate as an index fossil with which to correlate strata. Teeth and jaw fragments of Clevosaurus are readily identifiable, relatively common, and widely distributed, with occurrences in multiple basins across Pangaea. In fact, the genus is known from North and South America, the UK, South Africa, and, probably, China (Fraser 1988, 1994; Sues and Baird 1993; Wu 1994; Sues and Reisz 1995; Bonaparte and Sues 2006; Jones 2006a). However, records of Clevosaurus range in age from fairly early in the Late Triassic in the south-western USA (Fraser 1993) to the Early Jurassic of China (Wu 1994; Jones 2006a), with the best records from fissure fills of somewhat more ambiguous age in south-west England (e.g. Fraser 1988). Worse, almost all Clevosaurus occurrences outside the UK represent species distinct from C. hudsoni and C. minor, which occur in many fissure fills (Fraser 1988). Consequently, Clevosaurus cannot be used as an index fossil to correlate strata reliably to a level of resolution better than Late Triassic–Early Jurassic, which may represent as much as 30–35 myr of sphenodontian evolution. It appears likely that subsequent collection will probably result in similar range extensions for the few early sphenodont genera that have short ranges (many of which, like Whitakersaurus, are known only from a single locality at present). Consequently, as demonstrated by Heckert and Lucas (2006), other small- or microvertebrate fossils, principally archosauriforms, hold the greatest potential for subdividing Triassic time and correlating strata.
Whitakersaurus represents an important new record of sphenodontians from the Upper Triassic Coelophysis (Whitaker) quarry in the western USA. It is a relatively rare non-dinosaurian fossil in the Rock Point Formation, the most complete sphenodontian yet recovered from the Chinle Group, and of Apachean (late Norian–Rhaetian) age. Although clearly a sphenodontian, it possesses some characteristics typical of relatively primitive taxa and others more indicative of a more advanced position. Therefore, we hypothesize that, with more complete remains, it will probably be located near Diphydontosaurus and Planocephalosaurus in the current phylogenetic hypothesis of sphenodontian relationships (Text-fig. 4). The global record of rhynchocephalians, while not amenable to precise biostratigraphy, demonstrates mosaic evolution and provides strong evidence for an extensive diversification event prior to the Late Triassic.
Acknowledgements. The micrographs published here were obtained at both the Institute of Meteoritics in the Department of Earth and Planetary Sciences, University of New Mexico, and the College of Arts and Sciences Microscopy Facility at Appalachian State University. Discussions with N. C. Fraser and L. Säilä influenced the ideas we put forth here. Ben Edwards provided comparative SEM images of many taxa from fissure-fills in the UK. Reviews by Susan Evans and Johannes Müller improved the manuscript.