(upper triassic) of south africa - CiteSeerX

J. Paleont., 80(2), 2006, pp. 333–342 Copyright q 2006, The Paleontological Society 0022-3360/06/0080-333$03.00

ELLIOTHERIUM KERSTENI, A NEW TRITHELEDONTID FROM THE LOWER ELLIOT FORMATION (UPPER TRIASSIC) OF SOUTH AFRICA C. A. SIDOR1

P. J. HANCOX2

AND

Burke Museum and Department of Biology, University of Washington, Seattle 98195, ,[email protected]. and 2Bernard Price Institute for Palaeontological Research, University of the Witwatersrand, Johannesburg, WITS 2050, South Africa, ,[email protected].

1

ABSTRACT—We describe a new tritheledontid, Elliotherium kersteni n. gen. and sp., on the basis of a partial skull collected from the lower Elliot Formation (Upper Triassic; Euskelosaurus Range Zone) on the farm Beatrix, Free State Province, South Africa. Although similar to Pachygenelus, the genus differs from other South African tritheledontids in its higher maxillary postcanine tooth count, lack of labial cingula on the upper postcanine teeth, shorter secondary palate, retention of a median vomerine ridge behind the secondary palate, and smaller interpterygoid vacuity. With regard to these features, Elliotherium is most similar to the South American tritheledontid Chaliminia. Elliotherium is autapomorphic in its possession of a long parasphenoid rostrum, ventrally depressed secondary palate, medially convex margins of the interpterygoid vacuity, and a thin lamina of palatine covering the interpterygoid vacuity. Although Elliotherium is the stratigraphically lowest positioned tritheledontid from southern Africa, it is still probably later occurring than the South American taxa Chaliminia or Riograndia. Based on a revision of a previous analysis of advanced cynodont relationships, we propose a sister-group relationship between Elliotherium and Chaliminia, although support for this pairing is weak.

INTRODUCTION

years the perennial discussion of mammal origins among paleontologists has seen a polarization of phylogenetic thought. One camp maintains the traditional hypothesis of tritheledontids (sometimes also called ictidosaurs) being the cynodonts most closely related to early mammals (Sues, 1985; Hopson and Barghusen, 1986; Shubin et al., 1991; Crompton and Luo, 1993; Luo and Crompton, 1994; Hopson and Kitching, 2001; Rubidge and Sidor, 2001). The second group supports the more recently advanced cladistic pairing of early mammals with tritylodontids. The latter sister grouping was first proposed by Kemp (1983) and then later gained support from the computer-based analyses of Rowe (1988, 1993) and Gauthier et al. (1988) (see also Wible, 1991). Both hypotheses are not without significant drawbacks in terms of the number of evolutionary reversals or homoplasies they imply (Luo, 1994). For example, tritheledontids appear to possess a contact between the dentary and squamosal, although they are remarkably primitive in their retention of an interpterygoid vacuity and in their lack of an anterior lamina of the prootic. Alternatively, tritylodontids possess an anterior lamina, ossified orbitosphenoid, and several mammal-like postcranial characteristics, but have such a strikingly derived dentition that drawing homologies with the teeth of early mammals seems impossible. Despite the degree of controversy surrounding their position relative to mammals, tritheledontids are not a particularly wellknown group. The six monospecific genera currently recognized are based on only a handful of specimens (Table 1). In this contribution, we describe a new tritheledontid genus that extends this family’s range in Africa from Lower Jurassic into Upper Triassic strata. We also review previous phylogenetic work on tritheledontids and incorporate our new taxon into a modified version of the cladistic analysis of Martinelli et al. (2005). Taxonomic background.Broom (1912) named Tritheledon riconoi as the first tritheledontid on the basis of a partial skull with a peculiar transversely expanded postcanine dentition. Pachygenelus monus, based on a partial lower jaw, was named shortly thereafter (Watson, 1913). Although Broom created the family Tritheledontidae to receive Tritheledon Broom, 1912, he later (1929) named the therapsid suborder Ictidosauria, to encompass Pachygenelus Watson, 1913, Tritheledon, and Karoomys browni Broom, 1903 (which was later synonymized under Cynognathus Seeley, 1895), in addition to two unnamed specimens. From these specimens, which were referred to as ‘‘Ictidosaurians A and B’’

I

N RECENT

by Broom (1932), Crompton (1958) used the latter to name Diarthrognathus broomi. This famous specimen was argued to possess both a quadrate-articular and dentary-squamosal jaw articulation, and thus be transitional between the conditions present in modern mammals and reptiles (Crompton, 1963; Gow, 1981). Lycorhinus angustidens Haughton, 1924 was thought to represent an additional South African ictidosaur (Haughton, 1924; Broom, 1932; Parrington, 1946), until it was demonstrated to be based on the lower jaw of an ornithischian dinosaur (Thulborn, 1970; Hopson, 1975). Tritheledontids were known only from southern Africa until 1980, when Bonaparte (1980) described Chaliminia musteloides from the Upper Triassic Los Colorados Formation of Argentina as the first South American member of the family. Later, Shubin et al. (1991) described material indistinguishable from that of Pachygenelus from the McCoy Brook Formation of Canada. More recently, Bonaparte et al. (2001) named Riograndia guaibensis, from the Upper Triassic Caturrita Formation of Brazil, as the first member of the family Riograndidae. Riograndidae was informally allied with Tritheledontidae by virtue of several similarities with Chaliminia Bonaparte, 1980. Finally, Martinelli et al. (2005) have described a second tritheledontid from Brazil and provided the first computer-assisted cladistic analysis of the family. Tritheledontid taxonomy has been subject to several revisions. Early considerations include those of von Huene (1933), Parrington (1946), Young (1947), and Watson and Romer (1956). In their widely-cited classification of cynodonts, Hopson and Kitching (1972) advocated the use of Tritheledontidae over Ictidosauria. Gow (1980) included Pachygenelus, Diarthrognathus Crompton, 1958, and Tritheledon Broom, 1912 all within Tritheledontidae, whereas Bonaparte (1980) recognized two families, Tritheledontidae and Pachygenelidae, within the superfamily Tritheledontoidea. Although described as a ‘‘new primitive ictidosaur,’’ Bonaparte et al.’s (2001) Riograndidae was placed only at an indeterminate level within the Cynodontia (i.e., within neither Tritheledontidae nor Ictidosauria). Most recently, Martinelli et al. (2005) have phylogenetically defined Tritheledontidae as the most recent common ancestor of Riograndia Bonaparte, Ferigolo, and Ribeiro, 2001 and Pachygenelus and all of its descendants, thereby settling the ambiguous position of Riograndia. At the alpha taxonomic level, Hopson and Kitching (1972) considered Diarthrognathus to be a subjective junior synonym of Pachygenelus, although some more recent workers have maintained their separation (Gow, 1980; Shubin et al., 1991; Gow,

333

334

JOURNAL OF PALEONTOLOGY, V. 80, NO. 2, 2006

TABLE 1—Stratigraphic and geographic distribution of tritheledontid taxa. ‘‘Hett.’’ refers to the Hettangian Age of the Jurassic. Taxon

Formation

Chaliminia musteloides Diarthrognathus broomi New Brazilian Genus Pachygenelus monus Pachygenelus cf. monus Elliotherium kersteni n. gen. and sp. Riograndia guaibensis Tritheledon riconoi

Los Colorados Fm. upper Elliot and Clarens Fms. mid-upper Caturrita Fm. upper Elliot and Clarens Fms. McCoy Brook Fm. lower Elliot Fm. mid-upper Caturrita Fm. upper Elliot Fm.

1994). Hopson (personal commun., 2003) now considers them generically distinct. A jaw assigned to cf. Pachygenelus from the Upper Triassic Dockum Group of Texas (Chatterjee, 1983) has been considered indeterminate by more recent workers (Shubin et al., 1991) and is probably based on fish material (Hopson, personal commun., 2004). ABBREVIATIONS

The following institutional abbreviations are used: BP, Bernard Price Institute for Palaeontological Research, Johannesburg; NMQR, National Museum, Bloemfontein; SAM, South African Museum, Cape Town. SYSTEMATIC PALEONTOLOGY

Infraorder CYNODONTIA Owen, 1861 Suborder ICTIDOSAURIA Broom, 1929 Family TRITHELEDONTIDAE Broom, 1912 Definition.The clade including the most recent common ancestor of Riograndia and Pachygenelus, and all its descendents (from Martinelli et al., 2005). Diagnosis.Derived cynodonts with three upper incisors; first lower incisor enlarged and others small; narrow interpterygoid vacuity present; palatine longer than maxilla in bony secondary palate; bony secondary palate extends posterior to level of anterior orbital margin (also present in basal mammals). ELLIOTHERIUM new genus Figures 2–5 Type species.Elliotherium kersteni, n. sp. Diagnosis.As for the species, by monotypy. Etymology.Therio (Greek), beast. Named for the geological formation in which it was found and a suffix commonly associated with early mammals and their relatives. ELLIOTHERIUM KERSTENI new species Diagnosis.Tritheledontid autapomorphic in its possession of a long parasphenoid rostrum, medially convex margins of the interpterygoid vacuity, ventrally depressed secondary palate, and a thin lamina of palatine covering the interpterygoid vacuity. Differs from Pachygenelus monus in its greater number of upper postcanine teeth (at least 13), larger overall skull size, lack of complete labial cingula on the upper postcanine teeth, bony secondary palate extending only to level of last upper postcanine tooth; presence of median vomerine ridge behind bony secondary palate, and smaller interpterygoid vacuity. Etymology.Named for Mrs. Olga Kersten, farm owner and discoverer of the fossil. Type.BP/1/6106, partial skull lacking tip of snout, occiput, and mandible. Occurrence.BP/1/6106 was discovered in red overbank siltstones of the Upper Triassic lower Elliot Formation (Fig. 1) on the farm Beatrix (coordinates 28833.2339S, 27857.8169E), Rosendal District, Free State Province, South Africa.

Country Argentina South Africa Brazil South Africa Canada South Africa Brazil South Africa

Epoch Late Triassic Early Jurassic Late Triassic Early Jurassic Early Jurassic Late Triassic Late Triassic Early Jurassic

Age Norian Hett.-Sinemurian early Norian Hett.-Sinemurian Hettangian Rhaetic? early Norian Hett.-Sinemurian

GEOLOGIC SETTING

BP/1/6106 was recovered from rocks of the lower part of the Elliot Formation in the northern part of the main Karoo Basin of South Africa (Fig. 1). The formation is part of the Upper Carboniferous to Middle Jurassic Karoo Supergroup that outcrops in the main Karoo Basin of South Africa and Lesotho, as well as in several other separated outcrop areas in central and southern Africa (Johnson et al., 1996; Bordy et al., 2004). Previous workers have recognized three lithostratigraphic subdivisions to the Elliot Formation (Kitching and Raath, 1984), but recent research has shown that only two subdivisions are recognizable on a basinal scale (Bordy et al., 2004). The latter authors have shown that the Lower Elliot Formation (LEF) correlates to the Lower informal subdivision of Kitching and Raath (1984), whereas the Upper Elliot Formation (UEF) encompasses both the Middle and Upper subdivisions. As such, the LEF also corresponds to the Euskelosaurus Range Zone within the preliminary biostratigraphy of Kitching and Raath (1984), and the UEF to most of the Massospondylus Range Zone (including the Tritylodon Acme Zone). We follow Bordy et al. (2004) in their usage of the terms LEF and UEF for the Elliot Formation in the north of the main Karoo Basin. The sandstones of the LEF resemble multistory channel fills and have been interpreted as deposits of perennial, moderately meandering fluvial systems deposited under mesic conditions (Bordy et al., 2004). By contrast, the UEF is characterized by mostly tabular, multistory sheet sandstones that resulted from ephemeral fluvial processes in a semiarid to arid setting (Bordy et al., 2004). In the north of the basin, the Elliot Formation is much thinner (between 80 and 100 m) than in the more proximal reaches in the south (around 450 m) (Visser and Botha, 1980; Bordy et al., 2004). However, the twofold subdivision is still applicable in the north, with the base of the UEF defined by the first appearance of a laterally persistent sandstone with a pedogenic glaebule lag conglomerate overlying an erosive base. DESCRIPTION

General preservation.The specimen consists of a partial skull that lacks the tip of the snout, most of both zygomatic arches, much of the occiput, and the entire mandible. The following skull elements are missing: premaxillae, septomaxillae, squamosals, quadrates, quadratojugals, postparietal, tabulars, and most of the braincase (prootics, opisthotics, exoccipitals, supraoccipital). The bone is typically ivory-colored, although in certain areas a black, presumably iron- and manganese-based, staining occurs. As a whole, skull is riddled with small fractures and cracks that make the recognition of sutures very difficult. The dorsal surface of the skull roof appears to be minimally crushed, although the junction of the pterygoids and palatines has been plastically deformed (Fig. 2.2). Maxilla.The right maxilla, despite missing the base of its zygomatic process, is the more complete of the two sides (Fig. 3.1). It preserves the root of the enlarged canine along its jagged

SIDOR AND HANCOX—NEW TRITHELEDONTID

335

FIGURE 1—Geographic position of study area (1 and 2) and representative geological section (3) of the type locality on the farm Beatrix (Rosendal District, Free State Province, South Africa). Light line in 3 indicates margin of the Karoo Basin. The informally recognized ‘Stormberg Group’ (in 2) includes the Molteno, Elliot, and Clarens formations. The base of the Clarens Formation is shown above the upper Elliot Formation at the 119 m mark (in 3). Facies codes for 3 are as follows: Fr, ripple cross-laminated fines; Fm, massive fines; Sei, intraformational lag pebble conglomerate sandstone; Sh, horizontally stratified sandstone; Sm, massive sandstone; Sp, planar cross-stratified sandstone; Sr, ripple cross-stratified sandstone; St, trough cross-stratified sandstone; Stl, large-scale trough cross-stratified sandstone.

anterior margin, medial to which is the matrix-filled nasal cavity. The lateral, facial portion of the maxilla curves upwards and outwards from the relatively horizontal and medially positioned toothrow. An infraorbital foramen cannot be recognized on either side. Although poorly demarcated, the maxilla and nasal contact where the lateral and dorsal surfaces of the snout curve to meet one another. The maxilla-lacrimal contact is poorly preserved on both sides, but appears to have been positioned just anterior to the orbit (Fig. 4). On the left side, the maxilla contacts the zygomatic portion of the jugal below the orbit. In palatal view (Figs. 2.2, 4.2), the maxilla contributes to the rostral half of the preserved bony secondary palate. Anteriorly, the hard palate is broken to reveal the infilled passageway for inspired air, which the premaxilla would normally cover ventrally. Posteriorly, the maxilla contacts the palatine along a posterolateral

suture that becomes obscured in the vicinity of the posterior postcanine teeth. An autapomorphy of Elliotherium n. gen. is the extent to which the secondary palate descends posteriorly to more than twice the height of the postcanine teeth (Fig. 5). Although some degree of depression is found in other tritheledontid palates, we believe that this feature is real in Elliotherium, although it is probably accentuated by the small size of the postcanine teeth. There are the remains of at least 13 postcanine teeth preserved in the right maxilla (Fig. 5) and 10 in the left maxilla. In occlusal view, the toothrows are oriented parasagittally and do not diverge posteriorly to parallel the margin of the maxilla, as in basal cynodonts and, to a lesser degree, Riograndia (Bonaparte et al., 2001; Sidor and Smith, 2004). Especially on the left side, the long axis of each postcanine tooth is oriented somewhat obliquely to the sagittal plane, although not to the degree seen in Tritheledon

336


FIGURE 2—Stereophotographs of Elliotherium kersteni n. gen. and sp., (BP/1/6106) in dorsal (1) and ventral (2) views. Scale bar equals 20 mm.

(Gow, 1980). Compared to Pachygenelus, the upper postcanine teeth in Elliotherium are remarkably slender in profile and lack the labial cingula that characterize the former (Gow, 1980). The majority of the teeth appear tricuspate in lateral view, several with very prominent central cusps and poorly developed mesial and distal accessory cuspules. A similar condition was illustrated by Gow (2001) for a Pachygenelus sp. specimen from the upper portion of the Elliot Formation. Prominent wear facets truncate the medial margin of most well-exposed teeth. We interpret the postcanines as showing alternating replacement on the basis on their size distribution. Finally, we are unable to determine with the present material the morphology of postcanine roots. Nasal.The nasal is a large bone of the anterior part of the skull roof (Figs. 2.1, 4). As in other cynodonts, its posterior end is transversely expanded where it contacts the lacrimal posterolaterally and the parietal posteriorly. In addition, in this region the surface of the nasal has a lightly striated surface. In lateral view, the nasal-maxilla suture is difficult to determine. Anteriorly, the nasal is broken before the region of its contact with the premaxilla and external naris. Lacrimal.During the transition from Early to Late Triassic cynodonts, the lacrimal became a bone of larger dimensions and greater exposure on the face. In Elliotherium, the left lacrimal is

exposed on the lateral surface of the face as a triangular bone originating from the anterior margin of the orbit (Figs. 3.2, 4). In addition, the lacrimal forms the anterior surface of the orbital cavity anterior and lateral to the descending process of the frontal. Probably because of the numerous small cracks in this region, a lacrimal foramen is not visible. Jugal.Only a small portion of the left jugal is preserved in BP/1/6106. Posteriorly, at what would be near the midpoint of the zygomatic arch, the jugal is preserved as an impression in the matrix (Figs. 3.2, 4). Based on the impression and its bony continuation, the zygomatic arch becomes shallow anteriorly towards its junction with the maxilla. Thus in contrast to the condition in Pachygenelus and Diarthrognathus, where the zygomatic arch is uniformly thin throughout its length, Elliotherium retains the condition common among more primitive cynodonts, where the zygomatic arch deepens posteriorly. This deepening, although not as pronounced as in basal cynodonts or tritylodontids, is an important feature because a narrow zygomatic arch has been considered a derived character uniting tritheledontids and early mammals (Hopson and Barghusen, 1986; Luo, 1994). Near its contact with the maxilla, the anterior portion of the jugal bears a ridge on it ventrolateral surface. Frontal.The frontal occupies a relatively small area on the


337

FIGURE 3—Stereophotographs of Elliotherium kersteni n. gen. and sp., (BP/1/6106) in right lateral (1) and left lateral (2) views. Scale bar equals 20 mm.

dorsal skull roof, where it is squeezed between nasals anteriorly and parietals posteriorly (Figs. 2.1, 4). The degree to which the frontal is narrowly exposed between the orbit and posterolateral portion of the nasal is not seen in other tritheledontids. As in other derived cynodonts, the frontal contributes to the medial aspect of the orbit with a large descending lamina that is positioned medial to the lacrimal and contacts the palatine ventrally. In more basal cynodonts, this lamina is formed by the prefrontal (Sidor and Smith, 2004). The frontal and epipterygoid make contact along the dorsal margin of the interorbital fenestra, as in all other cynodonts (Hopson and Barghusen, 1986). Parietal.On the right side, the frontoparietal suture can be seen to be oriented posteromedially on the skull roof, although a corresponding suture cannot be made out on the left side. Poor preservation precludes determination whether the parietals were fused across the midline or joined by a suture. In addition, the dorsal surface of the parietal is broken to such a degree that it is difficult to determine if a sagittal crest was formed in the temporal region. In lateral view, the parietal is separated from the epipterygoid by a thin extension of the frontal. Vomer.When the skull is viewed from the front, a sliver of vomer can be seen in the matrix between the left and right nasal passageways on the left side. As in other cynodonts, the vomer was presumably clasped ventrally by the maxillae. More posteriorly, the vomer emerges from above the caudal end of the hard palate as a midline ridge, which diminishes in height towards the parasphenoid (Fig. 2.2). The ridge has been deformed postmortem to curve slightly to the right and then back towards the midline as it passes posteriorly. A small portion of the primary palate is

formed by the vomer, just medial to the paired palatine ridges. Pachygenelus is shown to lack a midline vomerine ridge (Bonaparte et al., 2003, fig. 19) whereas it is present in Chaliminia (Bonaparte, 1980, fig. 4). Palatine.The palate forms approximately the posterior half of the preserved bony secondary palate (Figs. 2.2, 4.2). Despite containing numerous small cracks, the secondary palate has a flat ventral surface, which then turns sharply upwards at its lateral margins, where it approaches the toothrow. Deformation may have accentuated the degree to which the posterior portion of the hard palate is depressed ventrally, although Pachygenelus also shows a deep groove between the hard palate and toothrow. The contact between the palatine and maxilla is presumably obscured by matrix just medial to the toothrow. The position of both the major and minor palatine foramina is uncertain. The posterior margin of the secondary palate is undamaged on the right side and retains its slightly arched shape. On the roof of the primary palate, the palatines contribute to well-developed paired pterygopalatine ridges. In Pachygenelus, these ridges end posteriorly at the anterolateral corners of the interpterygoid vacuity. In Elliotherium, these structures terminate near the base of the pterygoid’s transverse flange, but the margins of the vacuity remain obscured by matrix. Pterygoid.The pterygoid is best shown on the left side, where it preserves the transverse flange and a vestigial quadrate ramus. Although the contact between the palatine and pterygoid is obscured by matrix, more posteriorly the medial margin of the pterygoid can be seen to curve towards the cultriform process of the parasphenoid. This margin is slightly raised and extremely thin

338


FIGURE 4—Interpretive drawing of Elliotherium kersteni n. gen. and sp., (BP/1/6106) in dorsal (1), ventral (2), left lateral (3), and right lateral (4) views. Anatomical abbreviations: 1–13, upper postcanine tooth positions; c, canine; epi, epipterygoid; f, frontal; imp, impression of zygomatic arch; j, jugal; l, lacrimal; m, maxilla; n, nasal; orb, orbitosphenoid; p, parietal; pal, palatine; pbs, parabasisphenoid; ps r, parasphenoid rostrum; pt, pterygoid; v, vomer; vac, interpterygoid vacuity.

and approaches its mate posteriorly, thereby greatly restricting the potential size of the interpterygoid vacuity. On the right side, a portion of the pterygoid has been lost, revealing the underlying epipterygoid. Tritheledontids are characterized by the phylogenetic reacquisition of an interpterygoid vacuity (Fig. 4). This opening is present in the Late Permian cynodonts Procynosuchus Broom, 1937 and Dvinia Amalitzky, 1922, but is closed off starting in the Early Triassic galesaurids and Thrinaxodon Seeley, 1894 (Tatarinov, 1968; Fourie, 1974; Kemp, 1979). Sidor and Smith (2004) note that specimens of juvenile Thrinaxodon possess a parasphenoid and interpterygoid vacuity similar to those of their Late Permian relatives.

The interpterygoid vacuity has been figured as a relatively broad opening in Pachygenelus (Bonaparte et al., 2003) as well as in Crompton’s (1958) reconstruction of Diarthrognathus. In Chaliminia (Bonaparte, 1980), the basicranial rami of the pterygoids are less foreshortened, so the interpterygoid vacuity is longer and narrower. Elliotherium appears to possess the smallest interpterygoid vacuity among tritheledontids, although the condition in Riograndia and the new Brazilian taxon (Martinelli et al., 2005) is unknown. Parabasisphenoid.The parasphenoid and basisphenoid are fused in cynodonts and so the term parabasisphenoid will be used to describe both. This element is very similar in Elliotherium and Pachygenelus. On either side of the cultriform process, a


339

approximates the condition seen in early mammals (Hopson, 1964; Sues, 1986; Luo, 1994; Luo et al., 2002). DISCUSSION

FIGURE 5—Stereophotographs of the right postcanine dentition of Elliotherium kersteni n. gen. and sp., (BP/1/6106) in lateral view. Note the relatively small postcanine teeth and the ventrally depressed secondary palate. Scale bar equals 2 mm.

basipterygoid process sutures to the posterior margin of the corresponding pterygoid. This suture is roughly transverse. Because of the close opposition of the pterygoids, the cultriform process of the parabasisphenoid is extremely narrow. Furthermore, although it is somewhat damaged along its length, the process is continuous with the median vomerine ridge, as in Chaliminia (Bonaparte, 1980). Behind the level of the basipterygoid processes, the ventral surface of the parabasisphenoid is poorly preserved. Further posteriorly, a large oval hole is present on the midline, which might indicate the area of contact for the missing basioccipital. Epipterygoid.The right epipterygoid is well exposed, due to the loss of the zygomatic arch on this side (Figs. 3.1, 4.4). The anterior margin of this element is incomplete, although its impression is preserved in places. Its leading edge is tipped anterodorsally to a greater extent than depicted by Hopson and Rougier (1993) for Pachygenelus. The posterior margin of the epipterygoid is more poorly preserved, lacking any remnant of its connection with the prootic and the margin of the foramen for the maxillary and mandibular components of the trigeminal nerve. Dorsally, the epipterygoid contacts the parietal and frontal. Ventrally, the epipterygoid rests on the pterygoid. The quadrate ramus of the epipterygoid is preserved on the left side and visible in ventral view (Fig. 2). Orbitosphenoid.A small, delicate bone is present in the right orbital vacuity between the epipterygoid and descending lamina of the frontal (Figs. 3.1, 4.4). A deep notch on its leading surface demarcates dorsal and ventral portions of this element. Based on its location and the fact that it lies at a slightly deeper level than the epipterygoid, we identify this bone as an ossified orbitosphenoid. We suggest that the dorsal portion represents the main body of the orbitosphenoid whereas the ventral one represents this element’s ventral keel. Crompton (1958) described an orbitosphenoid in Diarthrognathus. However, given the position of the element (ventral and posterior to the ascending process of the epipterygoid), we believe that this identification is in error and that the element likely represents part of either the epipterygoid or prootic. Nonetheless, the recognition of a definitive orbitosphenoid in Elliotherium means that Tritheledontidae is polymorphic for this feature. It should be noted, however, that the tritylodontid orbitosphenoid is much larger and more closely

Geological implications.The holotype of Elliotherium kersteni n. gen. and sp. (BP/1/6106) was discovered in red siltstones below a zone of tree root haloes, which in turn are overlain by a strongly erosively based sandstone with a thick accumulation of bone-bearing pedogenic carbonate glaebule conglomerate (facies Sei), a lithofacies believed to be unique to the UEF within the Elliot Formation (Bordy et al., 2004). Also recovered from the same level are a number of large prosauropod vertebrae and long bones, which have yet to be catalogued. Based on its stratigraphic position, BP/1/6106 may be assigned to the LEF and forms part of the Euskelosaurus Range Zone fauna. Significantly, it is the first record of a tritheledontid from this fauna. All previously collected tritheledontid material in South Africa and Lesotho has come from the UEF and forms part of the Massospondylus Range Zone fauna. A number of specimens (e.g., BP/ 1/4741, 4882, 5109, 5162, 5330) have been recovered from a reworked paleosol informally termed the Tritylodon Acme Zone (Smith and Kitching, 1997). There also seems to be an abundance of specimens from the uppermost levels of the UEF, in the finegrained sandstone unit below the base of the overlying Clarens Formation. For example, Pachygenelus specimens BP/1/4982, 5110, 5292, and 5691 are all recorded as being from within 3 m of the Elliot–Clarens contact. No specimens of Pachygenelus have been collected from these upper levels on the farm Beatrix. However, a small lag containing mixed dinosaurian remains and a well-preserved prosauropod skull have been recovered. The ages of the LEF and UEF are still the subject of some debate and better taxonomic resolution is required before the matter can be resolved. The fauna of the LEF contains at least five sauropodomorph species, including the earliest true sauropod (Yates, 2003; Yates and Kitching, 2003). Based on correlation of this fauna with better-dated sequences elsewhere, most authors believe the LEF (and its included Euskelosaurus Range Zone fauna) to be Late Triassic (Norian) in age (e.g., Hopson, 1984; Gow and Hancox, 1993; Lucas and Hancox, 2001). The UEF and its included Massospondylus Range Zone fauna is presently accepted as being Early Jurassic (Hettangian–Sinemurian) in age (Olsen and Galton, 1984; Smith and Kitching, 1997) as it certainly occurs below the Toarcian–Aalenian volcanics that terminate the Karoo sequence. The recent discovery of a specimen (BP/1/6105) assignable to the sauropod dinosaur Vulcanodon Raath, 1972 from the upper UEF on the farm Spionkop (also in the Rosendal District), may, however, suggest a slightly younger age for the upper part of the UEF (Yates et al., 2004). Previously, Vulcanodon was known only from a single specimen found in rocks sandwiched between two basaltic flows in Lake Kariba (Zimbabwe). These basalts may be correlated to the welldated continental flood basalts exposed in South Africa, which have been dated at 183 6 1 Ma (Duncan et al., 1997). The occurrence of Vulcanodon in South Africa therefore suggests that relatively little time elapsed between the deposition of the UEF and the extrusion of the continental flood basalts in which the type species of Vulcanodon was found (Yates et al., 2004). Phylogenetic comparisons.Despite the lack of a complete description of its cranial anatomy, Pachygenelus monus is still the best-known tritheledontid and consequently the taxon most commonly employed in broad-scale analyses of mammalian origins (e.g., Shubin et al., 1991; Luo, 1994; Hopson and Kitching, 2001). However, as noted by several workers, some features of Pachygenelus clearly demonstrate that it is specialized in aspects of its structure. These features include the: 1) presence of a large, anteriorly rounded interpterygoid vacuity; 2) absence of the first

340


upper incisor leading to an edentulous anterior portion of the premaxilla; 3) presence of enlarged and procumbent first lower incisors, passing medial to the upper incisors; 4) presence of welldefined labial cingula on most of the upper postcanine teeth; and 5) lower postcanines with a large mesial cusp and several progressively smaller distal cusps. Until the present work, the other known South African tritheledontids were Tritheledon riconoi and Diarthrognathus broomi. The former is represented only by the holotype, which preserves two incomplete maxillae bearing peculiar, transversely expanded teeth (Broom, 1912; Gow, 1980). Although diagnostic, the material currently known for Tritheledon sheds little light on the distribution of the aforementioned features. Similarly, Diarthrognathus is represented by only three specimens (BP/1/4882; NMQR C249, C253). The presence of a large interpterygoid vacuity in this taxon was reconstructed by Crompton (1958, fig. 6), but little is known of the anterior region of the skull. The dentition of Diarthrognathus is similar to that of Pachygenelus, but distinctive by virtue of the lower postcanines being more cylindrical in cross section, being oriented more transversely when fully erupted, and having a more pronounced lingual cingulum (Gow, 1994). Bonaparte (1980) suggested that Chaliminia musteloides and Pachygenelus monus were close relatives and belonged to a lineage distinct from that of Tritheledon. In support of this proposition, he noted shared derived similarities including the edentulous region of the premaxilla which allowed the lower incisors to pass medial to the uppers (Bonaparte, 1980, fig. 6) and the lack of transversely expanded upper postcanine teeth. More recently, Bonaparte et al. (2003) have concluded that the Late Triassic forms Brasilodon Bonaparte et al., 2003 and Brasilitherium Bonaparte et al., 2003 represent the closest known relatives of early mammals, with both tritheledontids and tritylodontids falling more basally. In the only cladistic analysis of Tritheledontidae, Martinelli et al. (2005) found Chaliminia and Riograndia to be sequential outgroups to an unresolved terminal trichotomy (Diarthrognathus, Pachygenelus, and their new Brazilian form). Studying the results of this analysis shows that the distribution of dental features is responsible for the basal position of Riograndia and Chaliminia. For example, they both lack a bulbous central cusp on the upper postcanine teeth (character 12). The position of Chaliminia was further supported by aspects of its lower postcanine morphology (character 14) and the alignment and position of its upper toothrow (characters 16 and 17). It should be noted that the analysis by Martinelli et al. did not include Tritheledon, presumably because of its fragmentary nature. We added Elliotherium n. gen. to the cladistic data matrix of Martinelli et al. (2005) (see Appendix). In addition, we have added three characters and modified another to help resolve tritheledontid relationships. The resulting majority-rule consensus tree of our analysis is presented in Figure 6. A close relationship between Elliotherium and Chaliminia is supported by the apomorphic distribution of characters 20, 65, and 66. Conclusions.The discovery of Elliotherium kersteni improves our understanding of an important group of Mesozoic cynodonts, the Tritheledontidae. The cladogram presented in Figure 6 suggests that tritheledontids may have first diversified in Gondwana during the Norian. If the sediments of the lower Elliot Formation are indeed Late Triassic in age, then tritheledontids must have been established in both African as well as South American sections of Gondwana only shortly after their first appearance in the fossil record. By the Early Jurassic, with the discovery of Pachygenelus cf. monus in Canada, tritheledonts were a geographically widespread clade.

FIGURE 6—Majority rule consensus tree of cynodont relationships. Tree statistics: 27 primary trees; 138 steps; CI 5 0.624 (excluding uninformative characters); RI 5 0.731. Percentages of trees with the topology depicted are given above corresponding branches. Details of the analysis are presented in the Appendix.

As with Elliotherium, future tritheledontid fossils will undoubtedly advance our anatomical knowledge of this poorly understood group. This knowledge will, in turn, lead to a better understanding of the group’s phylogenetic history, including a possible sistergroup relationship with basal mammals. ACKNOWLEDGMENTS

We thank M. Raath, B. Rubidge, and the South African Heritage Resource Agency for arranging and permitting loan of the specimen. Mrs. O. Kersten discovered the specimen and brought it to the attention of PJH. We thank V. Heisey and the late G. Modese for their excellent preparation of the specimen. We acknowledge the assistance of J. Hopson in early aspects of this project and for his critique of an early version of the manuscript. The careful reviews provided by Z.-X. Luo and H.-D. Sues substantially improved the final manuscript. CAS was able to visit the Bernard Price Institute in 2002 with support from the New York College of Osteopathic Medicine. REFERENCES

AMALITZKY, V. 1922. Diagnoses of the new forms of vertebrates and plants from the Upper Permian of North Dvina. Bulletin of the Academy of Science, St. Petersburg, 16:329–340. BONAPARTE, J. F. 1980. El primero Ictidosaurio (Reptilia–Therapsida) de America del Su, Chaliminia musterloides, del Triasico Superior de La Rioja, Republica Argentina. Actas II Congreso Argentino de Paleontologıá y Bioestratigrafıá y I Congreso Latinamericano de Paleontologıá, p. 123–133. BONAPARTE, J. F., J. FERIGOLO, AND A. M. RIBEIRO. 2001. A primitive Late Triassic ‘ictidosaur’ from Rio Grande do Sul, Brazil. Palaeontology, 44:623–635.

SIDOR AND HANCOX—NEW TRITHELEDONTID BONAPARTE, J. F., A. G. MARTINELLI, C. L. SCHULTZ, AND R. RUBERT. 2003. The sister group of mammals: small cynodonts from the Late Triassic of southern Brazil. Revista Brasileira de Paleontologia, 5:5– 27. BORDY, E. M., P. J. HANCOX, AND B. S. RUBIDGE. 2004. Fluvial style variations in the Late Triassic–Early Jurassic Elliot Formation, main Karoo Basin, South Africa. Journal of African Earth Sciences, 38:383– 400. BROOM, R. 1903. On the lower jaw of a small mammal from the Karroo beds of Aliwal North, South Africa. Geological Magazine, 10:345. BROOM, R. 1912. On a new type of cynodont from the Stormberg. Annals of the South African Museum, 7:334–336. BROOM, R. 1929. On some recent new light on the origin of mammals. Proceedings of the Linnean Society, New South Wales, 54:688–694. BROOM, R. 1932. The Mammal-Like Reptiles of South Africa and the Origin of Mammals. Witherby, London, 376 p. BROOM, R. 1937. A further contribution to our knowledge of the fossil reptiles of the Karoo. Proceedings of Zoological Society of London B, 107:299–318. CHATTERJEE, S. 1983. An ictidosaur fossil from North America. Science, 220:1151–1153. CROMPTON, A. W. 1958. The cranial morphology of a new genus and species of ictidosaurian. Proceedings of the Zoological Society of London, 130:183–215. CROMPTON, A. W. 1963. On the lower jaw of Diarthrognathus and the origin of the mammalian lower jaw. Proceedings of the Zoological Society of London, 140:697–750. CROMPTON, A. W., AND Z. LUO. 1993. Relationship of the Liassic mammals Sinoconodon, Morganucodon oehleri, and Dinnetherium, p. 30– 44. In F. S. Szalay, M. J. Novacek, and M. C. McKenna (eds.), Mammal Phylogeny: Mesozoic Differentiation, Multituberculates, Monotremes, Early Therians, and Marsupials. Springer-Verlag, New York. DUNCAN, R. A., P. R. HOOPER, J. REHACEK, J. S. MARSH, AND A. R. DUNCAN. 1997. The timing and duration of the Karoo igneous event, southern Gondwana. Journal of Geophysical Research, 102:18127– 18138. FOURIE, S. 1974. The cranial morphology of Thrinaxodon liorhinus Seeley. Annals of the South African Museum, 65:337–400. GAUTHIER, J., A. KLUGE, AND T. ROWE. 1988. Amniote phylogeny and the importance of fossils. Cladistics, 4:105–209. GOW, C. E. 1980. The dentition of the Tritheledontidae (Therapsida: Cynodontia). Proceedings of the Royal Society of London B, 208:461– 481. GOW, C. E. 1981. Pachygenelus, Diarthrognathus and the double jaw articulation. Palaeontologia Africana, 24:15. GOW, C. E. 1994. New find of Diarthrognathus (Therapsida: Cynodontia) after seventy years. Palaeontologia Africana, 31:51–54. GOW, C. E. 2001. A partial skeleton of the tritheledontid Pachygenelus (Therapsida: Cynodontia). Palaeontologia Africana, 37:93–97. GOW, C. E., AND P. J. HANCOX. 1993. First complete skull of the Late Triassic Scalenodontoides (Reptilia Cynodontia) from southern Africa, p. 161–168. In S. G. Lucas and M. Morales (eds.), The Nonmarine Triassic. New Mexico Museum of Natural History & Science Bulletin, No. 3. HAUGHTON, S. H. 1924. The fauna and stratigraphy of the Stormberg Series. Annals of the South African Museum, 12:323–497. HOPSON, J. A. 1964. The braincase of the advanced mammal-like reptile Bienotherium. Postilla, 87:1–30. HOPSON, J. A. 1975. On the generic separation of the ornithischian dinosaurs Lycorhinus and Heterodontosaurus from the Stormberg Series (Upper Triassic) of South Africa. South African Journal of Science, 71: 302–305. HOPSON, J. A. 1984. Late Triassic traversodont cynodonts from Nova Scotia and southern Africa. Palaeontologia Africana, 25:181–201. HOPSON, J. A., AND H. BARGHUSEN. 1986. An analysis of therapsid relationships, p. 83–106. In N. Hotton, P. D. MacLean, J. J. Roth, and E. C. Roth (eds.), The Ecology and Biology of the Mammal-Like Reptiles. Smithsonian Institution Press, Washington, DC. HOPSON, J. A., AND J. W. KITCHING. 1972. A revised classification of cynodonts (Reptilia; Therapsida). Palaeontologia Africana, 14:71–85. HOPSON, J. A., AND J. W. KITCHING. 2001. A probainognathian cynodont from South Africa and the phylogeny of nonmammalian cynodonts. Bulletin of the Museum of Comparative Zoology, 156:3–35.

341

HOPSON, J. A., AND G. W. ROUGIER. 1993. Braincase structure in the oldest known skull of a therian mammal: Implications for mammalian systematics and cranial evolution. American Journal of Science, 293A: 268–299. JOHNSON, M. R., C. J. VAN VUUREN, W. F. HEGENBERGER, R. KEY, AND U. SHOKO. 1996. Stratigraphy of the Karoo Supergroup in southern Africa: An overview. Journal of African Earth Sciences, 23:3–15. KEMP, T. S. 1979. The primitive cynodont Procynosuchus: functional anatomy of the skull and relationships. Philosophical Transactions of the Royal Society of London B, 285:73–122. KEMP, T. S. 1983. The relationships of mammals. Zoological Journal of the Linnean Society, 77:353–384. KITCHING, J. W., AND M. A. RAATH. 1984. Fossils from the Elliot and Clarens formations (Karoo Sequence) of the northeastern Cape, Orange Free State and Lesotho, and a suggested biozonation based on tetrapods. Palaeontologia Africana, 25:111–125. LUCAS, S. G., AND P. J. HANCOX. 2001. Tetrapod-based correlation of the nonmarine Upper Triassic of southern Africa. Albertiana, 25:5–9. LUO, Z. 1994. Sister-group relationships of mammals and transformations of diagnostic mammalian characters, p. 98–128. In N. C. Fraser and H.-D. Sues (eds.), In the Shadow of the Dinosaurs: Early Mesozoic Tetrapods. Cambridge University Press, New York. LUO, Z., AND A. W. CROMPTON. 1994. Transformation of the quadrate (incus) through the transition from non-mammalian cynodonts to mammals. Journal of Vertebrate Paleontology, 14:341–374. LUO, Z., Z. KIELAN-JAWOROSKA, AND R. L. CIFELLI. 2002. In quest for a phylogeny of Mesozoic mammals. Acta Palaeontologica Polonica, 47:1–78. MARTINELLI, A. G., J. F. BONAPARTE, C. L. SCHULTZ, AND R. RUBERT. In press. A new tritheledontid (Therapsida, Eucynodontia) from the Late Triassic of Rio Grande do Sul (Brazil), and its phylogenetic relationships among carnivorous non-mammalian eucynodonts. Ameghiniana, 42:191–208. OLSEN, P. E., AND P. M. GALTON. 1984. A review of the reptile and amphibian assemblages from the Stormberg of southern Africa, with special reference on the footprints and the age of the Stormberg. Palaeontologia Africana, 25:87–110. OWEN, R. 1861. Palaeontology, or, A Systematic Summary of Extinct Animals and Their Geologic Relations. Adam and Charles Buck, Edinburgh, 463 p. PARRINGTON, F. R. 1946. On a collection of Rhaetic mammalian teeth. Proceedings of the Zoological Society of London, 116:707–728. RAATH, M. 1972. Fossil vertebrate studies in Rhodesia: a new dinosaur (Reptilia, Saurischia) from near the Trias–Jurassic boundary. Arnoldia, 5:1–37. ROWE, T. 1988. Definition, diagnosis, and the origin of Mammalia. Journal of Vertebrate Paleontology, 8:241–264. ROWE, T. 1993. Phylogenetic systematics and the early history of mammals, p. 129–145. In F. S. Szalay, M. J. Novacek, and M. C. McKenna (eds.), Mammal Phylogeny: Mesozoic Differentiation, Multituberculates, Monotremes, Early Therians, and Marsupials. Springer-Verlag, New York. RUBIDGE, B. S., AND C. A. SIDOR. 2001. Evolutionary patterns among Permo–Triassic therapsids. Annual Review of Ecology and Systematics, 32:449–480. SEELEY, H. G. 1894. Researches on the structure, organisation, and classification of the fossil Reptilia. Pt. IX. Section 1. On the Therosuchia. Philosophical Transactions of the Royal Society of London B, 185: 987–1018. SEELEY, H. G. 1895. Researches on the structure, organization, and classification of the fossil Reptilia. Pt. 9, Sect. 5. On the skeleton in new Cynodontia from the Karroo rocks. Philosophical Transactions of the Royal Society B, 198:59–148. SHUBIN, N. H., A. W. CROMPTON, H.-D. SUES, AND P. E. OLSEN. 1991. New fossil evidence on the sister-group of mammals and early Mesozoic faunal distributions. Science, 251:1063–1065. SIDOR, C. A., AND R. M. H. SMITH. 2004. A new galesaurid (Therapsida: Cynodontia) from the Lower Triassic of South Africa. Palaeontology, 46:535–556. SMITH, R. M. H., AND J. W. KITCHING. 1997. Sedimentology and vertebrate taphonomy of the Tritylodon Acme Zone: a reworked palaeosol in the Lower Jurassic Elliot Formation, Karoo Supergroup, South Africa. Palaeogeography, Palaeoclimatology, Palaeoecology, 131:29–50.

342


SUES, H.-D. 1985. The relationships of the Tritylodontidae (Synapsida). Zoological Journal of the Linnean Society, 85:205–217. SUES, H.-D. 1986. The skull and dentition of two tritylodontid synapsids from the Lower Jurassic of western North America. Bulletin of the Museum of Comparative Zoology, 151:217–268. SWOFFORD, D. L. 1993. PAUP: Phylogenetic Analysis Using Parsimony. Illinois Natural History Survey, Champaign. TATARINOV, L. P. 1968. Morphology and systematics of the Northern Dvinia cynodonts (Reptilia, Therapsida; Upper Permian). Postilla, 126: 1–15. THULBORN, R. A. 1970. The systematic position of the Triassic ornithischian dinosaur Lycorhinus angustidens. Zoological Journal of the Linnean Society, 49:235–245. VISSER, J. N. J., AND B. J. V. BOTHA. 1980. Meander channel, point bar, crevasse splay and aeolian deposits from the Elliot Formation in Barkly Pass, north-eastern Cape. Transactions of the Geological Society of South Africa, 83:55–62. VON HUENE, E. 1933. Zur kenntnis des Wu ¨ rtembergishen Ra¨tbonebeds, etc. Jahreshefte des Vereins fu¨r vaterla¨ndische Naturkunde in Wu¨rttemberg, 89:65–128. WATSON, D. M. S. 1913. On a new cynodont from the Stormberg. Geological Magazine, 10:145–148. WATSON, D. M. S., AND A. S. ROMER. 1956. A classification of therapsid reptiles. Bulletin of the Museum of Comparative Zoology, 114:37–89. WIBLE, J. R. 1991. Origin of Mammalia: the craniodental evidence reexamined. Journal of Vertebrate Paleontology, 11:1–28. YATES, A. M. 2003. A definite prosauropod dinosaur from the lower Elliot Formation (Norian: Upper Triassic) of South Africa. Palaeontologia Africana, 39:63–68. YATES, A. M., AND J. W. KITCHING. 2003. The earliest known sauropod dinosaur and the first steps towards sauropod locomotion. Proceedings of the Royal Society of London B, 270:1753–1758. YATES, A. M., P. J. HANCOX, AND B. S. RUBIDGE. 2004. The first record of a sauropod dinosaur from the upper Elliot Formation (Early Jurassic) of South Africa. South African Journal of Science, 100:504–506. YOUNG, C.-C. 1947. Mammal-like reptiles from Lufeng, Yunnan, China. Proceedings of the Zoological Society of London, 117:537–597.

ACCEPTED 25 JANUARY 2005 APPENDIX

Martinelli et al. (2005) published the first cladistic data matrix of tritheledontid relationships, which included 63 characters (50 cranial, 13 postcranial) in 16 cynodont taxa. We made no attempt to remove parsimony uninformative characters in Martinelli et al.’s (2005) data. To their analysis, we added three characters, modified one character, and added one taxon, Elliotherium kersteni n. gen. and sp. We analyzed the revised matrix with PAUP (Swofford, 1993), with multistate taxa representing polymorphism and all characters given equal weight and considered unordered. The first new character (#64) deals with the morphology of the premaxilla and is defined as: premaxilla without (0) or with (1) edentulous gap between anteriormost incisors. From the top of the original data matrix in Martinelli et al. (2005), which includes 16 terminal taxa, our coding is: 0000000??01?1?00. The second new character (#65) deals with the vomer and is defined as: vomer with (0) or without (1) vertical septum extending posterior to level of secondary palate. From the top of the original data matrix in Martinelli et al. (2005), our coding is: 0011111??10?1?11. The third new character (#66) deals with the number of upper postcanine teeth and is defined as: number of upper postcanine teeth in adult: no greater than 9 (0), or greater than 11 (1). From the top of the original data matrix Martinelli et al. (2005), our coding is: 0000000??0100000. We modified the character states for character 29 by making a distinction between narrow and enlarged interpterygoid vacuities. The modified character reads: interpterygoid vacuity in adult between pterygoid flanges: absent (0), present and narrow (1), and present and enlarged into semicircular fenestra (2). From the top of the original data matrix Martinelli et al. (2005), the revised coding is: 0001000?011?2200. Including the character additions and modifications described above, our coding for Elliotherium kersteni follows. Note that the final three values (?01) represent the three new characters added here: ?????????0 ?10??12111????11?1112??0?0?1????1??0???1??????????????01.