The vertebrate fauna of the Upper Permian of Niger. V. The primitive temnospondyl Saharastega moradiensis

Journal of Vertebrate Paleontology 26(3):559–572, September 2006 © 2006 by the Society of Vertebrate Paleontology

THE VERTEBRATE FAUNA OF THE UPPER PERMIAN OF NIGER. V. THE PRIMITIVE TEMNOSPONDYL SAHARASTEGA MORADIENSIS ROSS DAMIANI1,*, CHRISTIAN A. SIDOR2, J. SÉBASTIEN STEYER3, ROGER M. H. SMITH4, HANS C. E. LARSSON5, ABDOULAYE MAGA6, and OUMAROU IDE6 1 Bernard Price Institute for Palaeontological Research, School of Geosciences, University of the Witwatersrand, Private Bag 3, Wits 2050, Johannesburg, South Africa, and Staatliches Museum für Naturkunde Stuttgart, Rosenstein 1, D-70191 Stuttgart, Germany, [email protected]; 2 Burke Museum and Department of Biology, University of Washington, Seattle, Washington 98195, U.S.A.; 3 Bâtiment de Paléontologie, UMR 5143 CNRS, Département Histoire de la Terre, Muséum national d’Histoire naturelle, 8, rue Buffon, F-75008, Paris, France; 4 Department of Karoo Palaeontology, IZIKO: South African Museum, PO Box 61, Cape Town 8000, South Africa; 5 Redpath Museum, McGill University, Montreal, Quebec H3A 2K6, Canada; 6 Institut de Recherches en Sciences Humaines, Université Abdou Moumouni de Niamey, Republic of Niger

ABSTRACT—The skull of the temnospondyl amphibian Saharastega moradiensis, from the Upper Permian Moradi Formation (Izégouandane Group, Izégouandane Basin) of northwestern Niger, is described in detail. Saharastega moradiensis is the most primitive known temnospondyl from Gondwana and possesses a combination of plesiomorphic and apomorphic character states, which suggest affinities with the Edopoidea, a clade of basal temnospondyls from the Upper Carboniferous and Lower Permian of Euramerica. These include the exclusion of the lacrimal from the orbital margin, the exclusion of the vomers and palatines from the interpterygoid vacuities, and the presence of an intertemporal ossification. Autapomorphies of the new taxon include the presence of narrow and elongated, transversely oriented nostrils; an extensive tongue-and-groove contact between the premaxillae and maxillae; tabulars that possess exceptionally large, laterally and ventrally directed ‘horns’; and an extraordinary ‘occipital plate’ that may be formed, at least in part, by a supraoccipital ossification. A phylogenetic analysis of select Paleozoic temnospondyls indicates that S. moradiensis is the sister taxon to the edopoids, represented in this analysis by Chenoprosopus and Edops. This suggests that S. moradiensis represents a late-surviving member of a clade that is the sister group of the Edopoidea. Members of this clade may have been restricted to equatorial northwest Africa during the Late Carboniferous and Early Permian, an area that was not affected by the extensive glaciation that covered much of southern Pangea.

INTRODUCTION Our knowledge of temnospondyl amphibian diversity during the latest Paleozoic is largely based on the faunas of the nonmarine Permian of European Russia and, to a lesser extent, South Africa (Anderson and Cruickshank, 1978; Milner, 1993). The Russian fauna (Ivakhnenko et al., 1997; Golubev, 2000), of Middle to Late Permian age (Lucas, 1998), is dominated by members of the Archegosauroidea (sensu Yates and Warren, 2000), with lesser representation by the Dvinosauria (sensu Yates and Warren, 2000), Eryopoidea and Dissorophoidea. In contrast, the largely contemporaneous South African record (Rubidge, 1995; Lucas, 1998) is at present limited to the Rhinesuchidae (Damiani and Rubidge, 2003; Damiani, 2004). The Archegosauroidea, the sister-taxon of the Stereospondyli (Milner, 1990; Gubin, 1997; Yates and Warren, 2000), ranges in age from the Early to Late Permian (Schoch and Milner, 2000). The Rhinesuchidae is the most basal group within the Stereospondyli, and, with the exception of a single relict taxon in the Lower Triassic of South Africa (Shishkin and Rubidge, 2000), ranges in age from Middle to Late Permian (Lucas, 1998). Elsewhere in Pangea, Late Permian temnospondyls, mainly archegosauroids or rhinesuchids, are known from Australia (Marsicano and Warren, 1998), Brazil (Malabarba et al., 2003), Tanzania (Panchen, 1959), Malawi (Watson, 1962), Madagascar (Piveteau, 1926), China (Li and Cheng, 1999), and India (Werneburg and Schneider, 1996), although these records are scanty. Members of the clades Dissorophoidea, Eryopoidea and Dvinosauria have

* Present address: Staatliches Museum fu¨r Naturkunde Stuttgart, Rosenstein 1, D-70191 Stuttgart, Germany.

not previously been recorded in the Upper Permian of Gondwana. During recent fieldwork near Arlit in the northwest of the Republic of Niger (Fig. 1), a moderately diverse tetrapod fauna was recovered from outcrops of the non-marine, Upper Permian Moradi Formation (Sidor et al., 2003a). The fauna recovered includes temnospondyl, pareiasaurian, and captorhinid remains (Sidor et al., 2003a), the pareiasaur being recently described as Bunostegos akokanensis (Sidor et al., 2003b). Previous fieldwork in the Moradi Formation by French paleontologists in the late 1960s yielded the remains of a large, multiple-tooth rowed captorhinid, Moradisaurus grandis (Taquet, 1969; de Ricqlès and Taquet, 1982). In addition, dicynodont therapsid and “labyrinthodont” amphibian material has been mentioned (Taquet, 1967; de Ricqlès and Taquet, 1982), and a pareiasaur skull was figured, but not described, by Taquet (1976). The Moradi temnospondyl fauna includes at least two new, large-bodied taxa that were briefly described and named by Sidor et al. (2005). The first, Nigerpeton ricqlesi, is a long-snouted taxon that is represented by cranial and postcranial remains, and is a member of the Cochleosauridae (Steyer et al., 2006). The second, Saharastega moradiensis, is a bizarre temnospondyl that displays a peculiar combination of plesiomorphic, apomorphic, and autapomorphic character states; it is considered to be closely related to, but not strictly a member of, the Edopoidea, because it shares the greatest number of character states with members of that group. The Edopoidea originally (Romer, 1947) included a variety of early temnospondyls now recognized as unrelated, and recent revisions have established that the group is a clade of basal temnospondyls (Milner, 1980; Godfrey et al., 1987; Milner and Sequeira, 1993, 1994; Holmes, 2000; Ruta et al., 2003; Sequeira, 2004; but see Boy, 1990 for an alternative view). Edopoids are known exclusively from Upper Carboniferous and Lower Per-

559

560

JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 26, NO. 3, 2006

mian lowland coalswamp deposits of Euramerica (Milner and Sequeira, 1998), and were superficially crocodile-like in appearance and presumably lifestyle. The group is currently divided into the Cochleosauridae, comprising Cochleosaurus bohemicus (Fritsch) 1885 from the Upper Carboniferous of the Czech Republic, Cochleosaurus florensis Rieppel, 1980 from the Upper Carboniferous of Novia Scotia, Chenoprosopus milleri Mehl, 1913 from the Upper Carboniferous-Lower Permian of New Mexico, Chenoprosopus lewisi Hook, 1993 from the Upper Carboniferous-Lower Permian of Texas, Procochleosaurus jarrowensis Sequeira, 1996 from the Upper Carboniferous of Ireland, and Adamanterpeton ohioensis Milner and Sequeira, 1998 from the Upper Carboniferous of Ohio, and the Edopidae, comprising the monotypic Edops craigi Romer and Witter, 1942 (first named by Romer, 1936 as a nomen nudum) from the Upper Carboniferous-Lower Permian of Texas. Milner and Sequeira (1998) and Sequeira (2004) provide the most recent, and detailed, accounts of the composition and phylogeny of the Edopoidea. Saharastega moradiensis and Nigerpeton ricqlesi are the first basal temnospondyl (i.e., all temnospondyls basal to the clade comprising all descendants of the common ancestor of Eryops and Parotosuchus, sensu Yates and Warren, 2000) recorded outside of Euramerica. S. moradiensis is described in detail in this paper, with emphasis placed on comparison with edopoids. Its relationships are evaluated by means of a phylogenetic analysis modified from that of Holmes et al. (1998). Institutional Abbreviations—AMNH, American Museum of Natural History, New York; MCZ, Museum of Comparative Zoology, Harvard University, Cambridge; MNN, Musée National du Niger, Niamey.

Anatomical Abbreviations—al, alar process of the jugal; apv, anterior palatal vacuity; arp, ascending ramus of the pterygoid; bo, basioccipital; ch, choana; cp, cultriform process of the parasphenoid; ec, ectopterygoid; eo, exoccipital; f, frontal; fm, foramen magnum; fsm, fossa subrostralis media; ipv, interpterygoid vacuity; it, intertemporal; j, jugal; la, lacrimal; mx, maxilla; n, nasal; p, parietal; pa, palatine; pf, prefrontal; pmx, premaxilla; po, postorbital; pof, postfrontal; pop, paroccipital process; pp, postparietal; psp, parasphenoid; pt, pterygoid; q, quadrate; qj, quadratojugal; so, supraoccipital; sph, sphenethmoid; sq, squamosal; st, supratemporal; stv, subtemporal vacuity; t, tabular; tp, tubera parasphenoidales; v, vomer. SYSTEMATIC PALEONTOLOGY AMPHIBIA Linnaeus, 1758 TEMNOSPONDYLI Zittel, 1887-1890 incertae sedis SAHARASTEGA MORADIENSIS Sidor et al., 2005 (Figs. 2–4) Holotype—MNN MOR73, a near-complete, adult skull lacking lower jaws. Locality—A locality (18° 47.515’ N, 7° 11.160’ E) near Ibadanane, approximately 20 km due west of Arlit, Agadès Department, northwestern Niger (Fig. 1). Horizon—The specimen was found within a thick succession of almost flat-lying, friable, dark reddish brown mudrocks interspersed with thin beds of indurated, matrix-supported intraformational conglomerate. The locality lies within the upper strata

FIGURE 1. Map of Niger (left) showing the geology of the study area and the locality of Saharastega moradiensis within the Permian rocks of the Moradi Formation, and the late Paleozoic and early Mesozoic stratigraphy of Niger (right), showing the position of the Moradi Formation.

DAMIANI ET AL.—PERMIAN TEMNOSPONDYL FROM NIGER of the Moradi Formation, approximately 25 m below an unconformity at the base of the Téloua 1 Formation (Fig. 1). The Moradi Formation is the uppermost unit of the Permian-aged, non-marine Izégouandane Group of the Izégouandane Basin (de Ricqlès and Taquet, 1982). Deposition of the upper part of the Moradi succession evidently occurred by episodic overbank flooding of meandering streams that flowed into a closed, semiarid continental basin. The age of the Moradi Formation is poorly constrained, but paleontological evidence indicates a Late Permian age (Taquet, 1972; de Ricqlès and Taquet, 1982). This is based mainly on the presence of the Triassic ichnotaxon Chirotherium in the overlying Téloua 1 Formation (Ginsburg et al., 1966), and on the presence of pareiasaurs and captorhinids in the Moradi Formation, all other known Gondwanan captorhinids also being of supposedly Late Permian age (Modesto and Smith, 2001). Diagnosis—Medium to large-sized temnospondyl characterized by the following autapomorphic characters: Nostrils narrow and elongated, with their long axes transversely oriented; premaxillae-maxillae articulation via extensive tongue-and-groove contact; tabulars with exceptionally large, flap-like ‘horns’ that are directed both laterally and ventrally; dorsal region of occiput comprises a complex ‘occipital plate’ that may be formed, in part, by a supraoccipital ossification. Apomorphic characters shared with some other temnospondyls: Orbits located in posterior half of skull, and close to skull margins; pineal foramen absent; squamosal embayment absent; lacrimal excluded from orbital margin by prefrontal-jugal contact; anterior palatal vacuities present as paired foramina; interpterygoid vacuities small, located in posterior half of skull; fossa subrostralis media present; transvomerine tooth row present; basicranial articulation sutural; long, well developed subotic process of the exoccipital; short suspensorium; quadrate condyles at same transverse level as occipital condyles. DESCRIPTION Preservation MNN MOR73 was preserved in a dark reddish-brown mudstone matrix that was removed largely through mechanical preparation. The skull is heavily fractured throughout, the larger cracks being occupied by the imbedding matrix or later filled with glue for consolidation. Countless fracture ‘veins’ are also present throughout the skull, complicating the identification of sutures. Dorsally, the surface ornament is quite weathered so that few areas of pristine ornament remain. Ventrally, the palatal bones remain partially covered by matrix which could not be fully removed without damage to the denticles. A moderate degree of post-mortem distortion is evident. The mid-point of the snout has been compressed in the dorso-ventral plane, resulting in a bowed appearance in lateral view. The left posterior region of the skull is compressed in the medio-lateral plane, resulting in slight asymmetry of the left and right skull halves. In addition, the left tabular/supratemporal region has been crushed against the skull. Areas of bone not preserved are limited to the regions surrounding the nostrils anteriorly, and the left cheek region. Broken bone surfaces are indicated by cross-hatching in the drawings. Unidentified bone fragments are attached to the right postfrontal, the right pterygoid, the left choanal margin, and in the right subtemporal vacuity. In addition, what appears to be a partial neural arch is attached to the left maxilla anteriorly, and two partial ribs are preserved in the basicranial region. Skull—General Morphology As preserved, the holotype skull of Saharastega moradiensis (Figs. 2–4) measures 368 mm from the tip of the snout to the back of the skull table in the midline, and an estimated 300 mm across the quadratojugals where the width of the skull is greatest.

561

This is moderately larger than those cochleosaurids that are known from adult specimens, namely Chenoprosopus milleri (Langston, 1953) and Cochleosaurus bohemicus (Sequeira, 2004). However, S. moradiensis is considerably smaller than the largest known skull of Edops craigi (Romer and Witter, 1942), which measures 630 mm from the tip of the snout to the back of the quadrates. Although exaggerated to some degree by compression, the skull of S. moradiensis was clearly rather flat, in profile tapering continuously from the occiput toward the snout (Figs. 2D, 4D). In overall appearance, the skull is crocodile-like in having an elongated, broadly triangular outline, posteriorly positioned orbits, and a long snout. The snout margins are straight to slightly convex and taper moderately towards a blunt, rounded tip. The posterior skull margin is gently concave. The overall outline is broadly similar to that of edopoids and rhinesuchids. The skull is likely that of a mature adult, judging from its overall large size, the high degree of ossification, the nature of the dermal ornamentation, and the nature of sutural attachment between bones (see Steyer, 2000 for a review of age criteria in temnospondyls). Sutures are extremely difficult to trace and individual bones appear to be solidly fused to one another, as might be expected in an older individual. The dermal bone of the skull roof varies in thickness from a few millimeters up to about 10 mm in the region surrounding the orbits, averaging about 5 mm. Skull Roof Ornamentation on the skull roof (Figs. 2A, 3A) of Saharastega moradiensis consists of the typical temnospondyl pattern of pits and ridges. Dermal sculpturing is relatively fine, with uniformly distributed pits ranging up to a few millimeters in diameter and up to 1 mm in depth. There are no areas of elongated grooves or ‘zones of intensive growth’ (Bystrow, 1935), a fairly common feature of the skulls of large temnospondyls, indicating that the skull is that of a mature individual. Centres of ossification of individual bones, normally indicated by a characteristic radiating pattern of ornamentation, are not apparent. The ornamentation differs markedly from that of cochleosaurids in that the latter possess a distinct region of subdued ornamentation in the midline that usually runs the length of the skull and is bounded on either side by dermal ridges. This character has been considered a synapomorphy of cochleosaurids (e.g., Milner and Sequeira, 1994, 1998; Godfrey and Holmes, 1995; Sequeira, 2004), although it was rejected outright by Hook (1993), who considered the character to have a wider distribution amongst early tetrapods. In the edopid Edops, a single, bifurcated ridge, accompanied by an area of subdued sculpturing, is present anterior to each orbit (Romer and Witter, 1942), a condition that may or may not be related to that in cochleosaurids. As in edopoids, there are no lateral line sulci on the skull roof. The absence of sensory sulci is usually regarded as indicative of terrestrial habits (e.g., Boy and Sues, 2000), although Schoch (2001) has pointed out that the lateral line system may shift to a sub-dermal position during ontogeny in some temnospondyls. Terrestrially adapted temnospondyls usually also possess skulls that are relatively deep and show well developed squamosal embayments (sensu Godfrey et al., 1987), such as Eryops, Balanerpeton, and dissorophoids. The large, flat, notchless skull of Saharastega moradiensis instead suggests a largely aquatic lifestyle. Although the nostrils are incompletely preserved, their shape can be deduced. They are relatively large, sub-rectangular in outline, located close to the margins of the skull, and face dorsally. Most interestingly, the long axis of each nostril is oriented nearly perpendicular to the midline, a condition which is unique to Saharastega moradiensis. In general, the nostrils of most temnospondyls are either circular or oval in outline, with their long

562

JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 26, NO. 3, 2006

FIGURE 2. Photographs of the skull of Saharastega moradiensis (MNN MOR73, holotype) in A, dorsal; B, ventral; C, occipital; and D, right lateral views.

axes parallel or sub-parallel to the midline. The narial morphology of S. moradiensis contrasts strongly with that of edopoids, in which the nostrils are relatively small, circular in outline, and located well posterior to the tip of the snout (Milner and Sequeira, 1998). The orbits of Saharastega moradiensis are small, sub-circular in outline, and are oriented with their long axes diverging slightly from the midline. They are located close to the skull margins a little posterior to the mid-length of the skull, leaving a broad and expansive skull table between and posterior to them. The orbits face partly laterally, as in some long-snouted Permian archegosauroids and Triassic trematosauroids (Gubin, 1991; Steyer, 2002). This differs from the more medially located orbits of edopoids and eryopoids. There is no pineal foramen in Saharastega moradiensis; the area where the foramen might be expected instead shows a patch of spongy bone. This suggests that the foramen had been obliterated by bone during growth, as is thought to occur in medium-

to large-sized skulls of cochleosaurids (Milner and Sequeira, 1998). Smaller cochleosaurid specimens, such as the type skulls of Chenoprosopus lewisi (Hook, 1993) and Adamanterpeton ohioensis (Milner and Sequeira, 1998), retain a pineal opening. The large-growing Edops also possesses a pineal foramen (Romer and Witter, 1942). The absence of a pineal foramen has been considered a derived character state of some or all adult cochleosaurids (Hook, 1993; Sequeira and Milner, 1993; Milner and Sequeira, 1994; Sequeira, 2004). The suspensorium of Saharastega moradiensis is derived in being strongly abbreviated posteriorly, so that its margin is positioned well anterior to the posterior margin of the skull table. As a result, the quadrate condyles are located anteriorly, level with the occipital condyles, and the squamosal embayment is virtually eliminated, the latter an apomorphic feature shared convergently with some dvinosaurians and brachyopoids (Milner, 1990; Yates and Warren, 2000). This suspensorium morphology differs dramatically from that of edopoids and eryopoids in

DAMIANI ET AL.—PERMIAN TEMNOSPONDYL FROM NIGER

563

FIGURE 3. Interpretive drawings of the skull of Saharastega moradiensis (MNN MOR73, holotype) in A, dorsal; B, ventral; and C, occipital views. Only clearly visible teeth and tooth loci are shown in B. Areas of light stippling in A represent damaged bone surface; cross-hatching represents broken bone; grey shading represents matrix.

which the suspensorium is rather long and there is a well developed squamosal embayment. The complement and arrangement of dermal roofing bones (Figs. 3A, 4A) is that of a typically primitive temnospondyl (e.g., Milner and Sequeira, 1994). The premaxillae, which form the tip of the snout, cannot be fully delimited because of poor/ incomplete preservation. The premaxillae enclose a small, midline interpremaxillary foramen (⳱internarial fenestra of Holmes et al., 1998), a structure which occurs sporadically in Paleozoic and Mesozoic temnospondyls. Given the size and position of the nostrils, it is unlikely that the premaxillae had the derived morphology of edopoids. In the latter, the premaxillae are large,

marginally but not medially elongated, and have a long, diagonally oriented suture with the nasal (Godfrey and Holmes, 1995; Milner and Sequeira, 1998). Similarly, the presence or absence of a septomaxilla cannot be determined because of poor preservation. The nasals are distinctly elongated bones, the elements together being noticeably longer than wide. Similarly elongated nasals are present in adult skulls of Chenoprosopus milleri (Langston, 1953). In contrast, in other cochleosaurids and in Edops, the paired nasals are only marginally longer than broad, or marginally shorter than broad. This feature is related to the relatively longer snouts in Saharastega moradiensis and C. milleri.

564

JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 26, NO. 3, 2006

FIGURE 4. Reconstruction of the skull of Saharastega moradiensis (MNN MOR73, holotype) in A, dorsal; B, ventral; C, occipital; and D, lateral views. Denticles are omitted from the left side of the palate so that the sutural pattern can be observed. Although there is little or no evidence for the presence of palatal fangs in the holotype (see text for discussion), they are tentatively included in the reconstruction. Marginal tooth count is hypothetical.

As in edopoids, the lacrimals are excluded from entering the anterior margin of the orbits by a contact between the prefrontal and the jugal. This condition has been considered an edopoid synapomorphy (Milner and Sequeira, 1994; Sequeira, 2004), although it was also noted to occur in eryopoids (Milner and Sequeira, 1998). In other primitive temnospondyls, such as Capetus (Sequeira and Milner, 1993), Balanerpeton (Milner and Sequeira, 1994), and most dvinosaurians (Holmes, 2000), the lacrimal enters the orbital margin. The anterior boundaries of the lacrimals in Saharastega moradiensis could not be determined. With the exception of both species of Chenoprosopus (Langston, 1953; Hook, 1993), in all other edopoids the lacrimal contacts the septomaxilla posterior to the narial margin, precluding a nasalmaxilla contact. In Chenoprosopus, the lacrimal does not reach the septomaxilla, so that a nasal-maxilla contact is present. In contrast, in other Paleozoic temnospondyls, such as the eryopoids Eryops (Sawin, 1941) and Parioxys (Moustafa, 1955; Carroll, 1964a), the septomaxilla is an exclusively intranarial ossification and the lacrimal contacts the narial margin.

The frontals and parietals of Saharastega moradiensis are very broad elements, slightly wider than long when taken together. This is the reverse of the situation present in edopoids. The width of these elements in S. moradiensis contributes to the great breadth of the skull roof. The postparietals do not possess a lappet on their posterior margin, as is present in Cochleosaurus bohemicus (Steen, 1938; Sequeira, 2004) and C. florensis (Rieppel, 1980; Godfrey and Holmes, 1995). The circumorbital series of bones comprises the prefrontals anteromedially, the postfrontals medially, the postorbitals posteriorly, and the jugals laterally, all of which appear to have been rather large elements. The anterolateral part of the postorbital forms a distinct flange that would likely have inserted into an embayment of the jugal, as in Cochleosaurus (Sequeira, 2004), Procochleosaurus (Sequeira, 1996) and Edops (Romer and Witter, 1942). An identical ‘hooking’ of the postorbital occurs in Triassic mastodonsauroids (Damiani, 2001). It cannot be determined, however, whether the posterior part of the postorbital formed the finger-like process synapomorphic for edopoids (Se-

DAMIANI ET AL.—PERMIAN TEMNOSPONDYL FROM NIGER queira, 2004). The jugal sends a relatively long, anterior tongue which broadly contacts the posterolateral part of the lacrimal, thus forming a significant part of the antorbital region of the snout, as in edopoids (Milner and Sequeira, 1998). The phylogenetic significance of this character (Milner and Sequeira, 1994), however, is uncertain, because the jugal extends well forward of the orbits in many, apparently unrelated temnospondyls, such as Capetus (Sequeira and Milner, 1993) and Parioxys (Moustafa, 1955). On the skull table, a well defined intertemporal ossification is present, indicating that Saharastega moradiensis is a primitive temnospondyl (Milner and Sequeira, 2003). An intertemporal is known in a number of unrelated Paleozoic temnospondyls, such as Capetus (Sequeira and Milner, 1993), Dendrerpeton (Holmes et al., 1998), the dvinosaurian Neldasaurus (Chase, 1965), and in all known edopoids. Further occurrences of an intertemporal have been documented by Gubin et al. (2000). The large supratemporals are located posterolateral to the intertemporals; each has a ‘free’ posterior margin anterolateral to the tips of the tabular ‘horns’. A supratemporal with a free posterior margin (i.e., with a contribution to the rim of the squamosal embayment or to the margin of the skull) characterizes many primitive temnospondyls, including Balanerpeton (Milner and Sequeira, 1994), Dendrerpeton (Holmes et al., 1998), Capetus (Sequeira and Milner, 1993), dissorophids (Carroll, 1964b), and edopoids. The derived condition, in which the supratemporal is excluded from the squamosal embayment by a squamosal/tabular contact, occurs in most dvinosaurians (Holmes, 2000) and most Mesozoic stereospondyls (Schoch and Milner, 2000; Warren, 2000). The tabular morphology of Saharastega moradiensis is strikingly different from that of all other temnospondyls. In most Paleozoic temnospondyls, the tabular is a small, triangular element lacking a well developed ‘horn’. In S. moradiensis, the ‘horn’ is a very long, broad and thick flange of bone with a blunt end; the tips of both ‘horns’ may or may not be complete, although both have an identical morphology. Most remarkable is the orientation of the ‘horn’ (best observed in occipital (Figs. 2C, 3C) and lateral (Figs. 2D, 4D) views), which projects in a ventrolateral direction, partially underlapping the posterior margin of the supratemporal. This extraordinary tabular morphology is not a taphonomic artefact, because the right cheek region of the skull is relatively undistorted. Ornamentation is restricted to the dorsal surface of the tabular, whereas in both species of Cochleosaurus ornament extends onto the ventral surface of the ‘horn’ (Godfrey and Holmes, 1995; Sequeira, 2004). Palate The palate of Saharastega moradiensis (Figs. 2B, 3B, 4B) is, for the most part, that of a typically primitive temnospondyl. Anteriorly, there are paired, anterior palatal vacuities, each circular in outline and partially contiguous with the nostril dorsally. The presence of an anterior palatal vacuity is a derived character for temnospondyls (Milner, 1990; Yates and Warren, 2000), and its presence in S. moradiensis is unusual for what is otherwise a very primitive palatal morphology. Among Paleozoic temnospondyls, paired vacuities are restricted to the short-faced dvinosaurians (Holmes, 2000) and the more derived archegosauroids (Gubin, 1991). A few dissorophoids possess a small, midline vacuity (e.g., Kamacops: Gubin, 1980; Phonerpeton: Dilkes, 1990), but, more commonly, an anterior palatal vacuity is lacking altogether. Edopoids lack an anterior palatal vacuity, although paired, anterior palatal depressions are apparently present in Chenoprosopus lewisi (Hook, 1993). The choanae are very large, long and rectangular in outline, with a rounded anterior margin and a squared-off posterior margin. Anteriorly, there is a marked constriction of the choanae from a lateral lappet of the vomers in the region of the transvo-

565

merine tooth row. Large and elongated choanae, like those of Saharastega moradiensis, clearly evolved independently in many Paleozoic temnospondyls, such as cochleosaurids, the eryopoid Parioxys (Moustafa, 1955), the archegosaurid Archegosaurus (Gubin, 1997), and the dissorophoid Kamacops (Gubin, 1980). However, the choanae of cochleosaurids are considered distinct in being wider anteriorly than posteriorly (although only marginally so in Chenoprosopus milleri), a synapomorphy of that group (Hook, 1993; Milner and Sequeira, 1998; Sequeira, 2004). In Edops (Romer and Witter, 1942), the choanae are relatively small and sub-circular in outline. Saharastega moradiensis has very small, rounded interpterygoid vacuities that are positioned mainly in the posterior half of the skull, leaving an expansive palatal area anterior to them. This arrangement is considered a synapomorphy of cochleosaurids (Hook, 1993; Milner and Sequeira, 1994, 1998). In addition, Milner and Sequeira (1994) distinguished between two character states in primitive temnospondyls: a primitive state in which the vacuities taper to a point anteriorly (considered by Milner and Sequeira to be restricted to edopoids), and a derived state in which the vacuities are rounded anteriorly (e.g., Balanerpeton: Milner and Sequeira, 1994; Eryops: Sawin, 1941). The condition in S. moradiensis is unclear because the vacuities are not well preserved anteriorly, but we suggest that among temnospondyls in general the condition may be strongly influenced by preservation. The subtemporal vacuities are remarkably small, being limited by two factors: the shortness of the suspensorium as a whole, and by the encroaching pterygoids which form their medial border. The vacuities are considerably narrower anteriorly than posteriorly, although the anterior and posterior sections are of approximately equal length. This is unusual, if not unique, among temnospondyls. Although small subtemporal vacuities characterize various Paleozoic temnospondyls, most notably the eryopoid Zatrachys (Schoch, 1997), none are as relatively small as those of Saharastega moradiensis. The premaxillae suture with the maxillae marginally (described below) and the vomers posteriorly. As in most, but not all (e.g., Isodectes: Sequeira, 1998), temnospondyls with paired anterior palatal vacuities, the premaxillae form their anterior and lateral margin. Primitively, the premaxillae also form the anterior margin of the choanae, mainly through the presence of a sunken, medial shelf that is clearly demarcated from the marginal dentition. Edopoids lack this medial premaxillary shelf, so that the premaxilla is excluded from (Cochleosaurus florensis), or makes only point contact with (all other edopoids), the choanal margin. A weakly developed fossa subrostralis media (⳱premaxillary tubercle or median subrostral tubercle of other authors) is present in the midline of the premaxillae behind the dentition. This structure, of sporadic occurrence in temnospondyls, is known in archegosauroids (Gubin, 1991), rhinesuchids (RD pers. obs.), and several Mesozoic stereospondyls (e.g., Thoosuchus yakovlevi: Damiani and Yates, 2003). As far as can be determined, it has not previouly been reported in a basal temnospondyl. The tongue-and-groove contact between the premaxilla and the maxilla is, as far as can be determined, unique among temnospondyls. The anteriormost portion of the maxilla forms a longitudinal trough, some 25 mm in length, for accommodation of the premaxilla. On the right side of the skull, the premaxilla has been dislodged, clearly exposing the maxillary trough within which the premaxilla was seated (Fig. 3B); on the left side, the premaxilla remains in articulation, but the maxilla clearly clasps the premaxilla in a similar manner. Posteriorly, the maxilla appears to have reached the level of the subtemporal vacuity, but it is unclear whether the maxilla contacted the quadratojugal, or whether the jugal intervened between those bones. The latter condition occurs in cochleosaurids (Hook, 1993; Godfrey and

566

JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 26, NO. 3, 2006

Holmes, 1995; Milner and Sequeira, 1998), but not in Edops, where a point contact is maintained (Romer and Witter, 1942). The massive vomers form a large, elongated vomerine plate anterior to the interpterygoid vacuities, as in cochleosaurids (Milner and Sequeira, 1998). However, the vomers are not markedly elongated anterior to the choanae, the latter being located in a relatively anterior position, unlike in cochleosaurids. With respect to vomerine proportions and choanal position, Saharastega moradiensis most closely resembles Parioxys (Moustafa, 1955). The vomers of S. moradiensis form the posteromedial margins of the anterior palatal vacuities, and the medial margin of the choanae. It seems likely that the vomers were excluded from entering the margins of the interpterygoid vacuities, judging from the arrangement of nearby sutures with adjoining bones and by comparison with other primitive temnospondyls such as edopoids and Parioxys. Judging from the position of the left vomer/palatine suture, the palatine bone was probably considerably broader anteriorly than posteriorly. The sutures with the pterygoid medially and ectopterygoid posteriorly could not be determined. Part of the suture between the ectopterygoid and the pterygoid is present on the left side of the palate. Almost certainly, the ectopterygoid was excluded from the margin of the subtemporal vacuity by a large, alar process of the jugal (⳱insula jugalis of other authors). The alar process is bordered laterally by the maxilla probably along its full length, whereas in some cochleosaurids (e.g., Adamanterpeton) a broad expansion of the alar process limits the posterior extent of the maxilla (Godfrey and Holmes, 1995; Milner and Sequeira, 1998). The distribution of this character within temnospondyls is uncertain, as this area of the skull is often poorly preserved. The large, triradiate pterygoid, the most complex bone of the palate, consists of a median basicranial process, a long palatal ramus, and a short quadrate ramus. As in all primitive temnospondyls, the basicranial process is an anteroposteriorly narrow structure. It articulates with the parasphenoid by way of a distinctly bulbous sutural junction, rather than the ‘mobile’ articulation common to many primitive temnospondyls, including all edopoids (except adult specimens of Chenoprosopus milleri; Langston, 1953). Although its sutural boundaries are unclear, the palatal ramus of the pterygoid quite likely formed the entire lateral margin of the interpterygoid vacuity, but appears to have been prevented from contacting the contralateral pterygoid in the midline by the anterior end of the cultriform process. This arrangement occurs in various primitive temnospondyls with small interpterygoid vacuities, such as cochleosaurids, Parioxys (Moustafa, 1955) and Capetus (Sequeira and Milner, 1993). In Edops, the pterygoids broadly meet in the midline anterior to the interpterygoid vacuities, a primitive tetrapod condition (Milner and Sequeira, 1994). A striking aspect of the palatal ramus of the pterygoid is the very large posteroventral flange bordering the subtemporal vacuity. The quadrate ramus of the pterygoid is short and robust, partly as a result of the expansion of the posteroventral flange and partly a result of the abbreviated suspensorium. As a result, the quadrates, neither of which is preserved, were located at the same transverse level as the occipital condyles, a derived condition (Damiani, 2001) often occurring in Mesozoic stereospondyls. The parasphenoid consists of a main body, lateral basipterygoid processes, and a cultriform process. Unlike most primitive temnospondyls in which the basipterygoid process of the parasphenoid is rather short and indistinct, the basipterygoid process in Saharastega moradiensis forms a prominent protuberance. The cultriform process is partially overlapped on the left side by the sphenethmoid (described below); the process is narrow and deep posteriorly but laterally expanded anteriorly, reaching its maximum width anterior to the interpterygoid vacuities. Its ventral surface is convex for most of its length.

The main parasphenoid body is distinctly longer than wide, and has a moderate constriction at the level of the posterior margin of the basicranial articulation. At the same level, on either side of the parasphenoid, are paired, sharp-rimmed crests, the tubera parasphenoidales (Clack and Holmes, 1988; ⳱tubera basisphenoidales, crista musculari, muscular/parasphenoid ‘pockets’, or transverse ridges of other authors), which border deep excavations (‘pockets’). Parasphenoid tubera of diverse morphology are widespread amongst early tetrapods including the anthracosaur Archeria (Clack and Holmes, 1988), the colosteid Greererpeton (Smithson, 1982), the baphetid Megalocephalus (Beaumont, 1977), and Paleozoic and Mesozoic temnospondyls (Holmes, 2000; Warren, 2000). Parasphenoid tubera are present in all edopoids in which the area is well preserved, but in Adamanterpeton the parasphenoid possesses three, shallow depressions (Milner and Sequeira, 1998) which may or may not be homologous with the parasphenoid tubera of other edopoids. A large foramen, likely for the internal carotid artery, pierces the sub-vertical side wall of the parasphenoid immediately lateral to the tubera parasphenoidales, as in eryopoids and stereospondyls (Bystrow and Efremov, 1940; Shishkin, 1968; Schoch, 1999a). In other primitive temnospondyls, including Edops (Yates and Warren, 2000) and Trimerorhachis (Schoch, 1999a), the artery enters the parasphenoid through its ventral surface, which is the primitive tetrapod condition (Shishkin, 1968). The posteriormost part of parasphenoid bears a shallow, midline ridge which may be a continuation of the medial part of the parasphenoid tubera. A convex, sharp-edged rim marks the boundary between the parasphenoid and the basioccipital. Posterolaterally, the parasphenoid sutures with the exoccipitals. The exoccipitals may be divided into a condylar portion (described below), and a long, subotic process which clasps the lateral margin of the parasphenoid body and reaches the level of the parasphenoid tubera. A similarly long, well developed subotic process is known only in some derived temnospondyls (e.g., brachyopoids), although Eryops (Sawin, 1941) also shows moderate development of this process. A fully ossified, rhomboidal sphenethmoid is preserved, the most prominent feature of which are the large lateral ‘wings’. The sphenethmoid is comparable in shape and relative size to that of Edops (Romer and Witter, 1942) and Eryops (Sawin, 1941), but it lacks the well defined morphology of those taxa. Although subject to variable degrees of ossification (Romer, 1947), a large, rhomboidal or rectangular sphenethmoid is present in all edopoids in which the area is preserved, whereas the sphenethmoid in most other temnospondyls seems to be relatively smaller. Dentition The premaxillary and maxillary dentition of Saharastega moradiensis is poorly preserved (Fig. 3B). Most tooth loci are empty alveoli, and none of the few teeth that are preserved possess complete tooth crowns. Tooth bases are elliptical in crosssection, show typical labyrinthine infolding, and possess notably large pulp cavities. Large pulp cavities were also reported in Eryops (Warren and Davey, 1992), but the significance of this is uncertain. As in all temnospondyls, teeth are implanted by bone of attachment on the labial and lingual walls of the jaw, and tooth alveoli remain distinct. There appears to have been a moderately high tooth count judging from the relatively small size and close spacing of the tooth loci, and there is a slight increase in tooth size from posterior to anterior. Together with the elliptical crosssection of the teeth, this arrangement is more typical of stereospondyls than of primitive temnospondyls (Damiani, 2001; Schoch and Milner, 2000). This contrasts with the condition in Eryops (Sawin, 1941) and Edops (Romer and Witter, 1942) in which there are variably-sized marginal teeth. Pseudocanine

DAMIANI ET AL.—PERMIAN TEMNOSPONDYL FROM NIGER peaking in more anterior marginal teeth, as observed in Edops (Romer and Witter, 1942), Adamanterpeton (Milner and Sequeira, 1998), and Cochleosaurus (Godfrey and Holmes, 1995; Sequeira, 2004), is also unlikely to have been present judging from the uniform diameter of the alveoli. As in many primitive temnospondyls, a dense shagreen of small denticles uniformly covers the vomers, palatines, ectopterygoids, and pterygoids. The denticles are tiny domes with pulp cavities and vary in size from under 1 mm to 2 mm in diameter, although a few may reach 4 mm in diamater. On the pterygoids, the denticles thin out on the posterior parts of the quadrate rami and on the basicranial process, while narrow, nondenticulated strips are present adjacent to the interpterygoid vacuities both anteriorly and posteriorly. The latter areas may represent attachment sites for skin covering the vacuities in life. The surface of the parasphenoid is poorly preserved but the anteriormost part of the cultriform process clearly bore denticles. In addition, two, small, isolated denticle patches are present on the body of the parasphenoid near the base of the cultriform process (Fig. 3B). These patches may be preserved as they were in life; alternatively, they may be isolated, denticle-bearing platelets that were fortuitously preserved, or they may have originally formed part of a single, larger patch that was subsequently fragmented post-mortem. In most primitive temnospondyls, the parasphenoid bears a large field of denticles, suggesting that the latter alternative is most likely. Saharastega moradiensis is unique among primitive temnospondyls in possessing a transvomerine tooth row, which is a derived character of more advanced temnospondyls (Milner, 1990; Yates and Warren, 2000). This consists of a single, arcuate row of teeth that is intermingled with the denticles on the anterior part of the vomerine plate. Curiously, there is no trace of the usual palatal fang/replacement pit pairs on either the vomers or palatines, and only tenuous evidence for the presence of ectopterygoid fangs, although unfavourable preservation (e.g., post-mortem damage or concealment by matrix and/or ‘loose’ denticles) may account for their absence. In temnospondyls, vomerine and palatine tusks are always present, whereas ectopterygoid tusks are highly reduced or even absent in some Triassic groups, such as in mastodonsauroids (Damiani, 2001). In anthracosaurs, vomerine tusks are absent, whereas palatine and ectopterygoid tusks are always present (Smithson, 2000). It is therefore not inconceivable that S. moradiensis lacked at least some of the usual fang pairs found in temnospondyls, but the evidence is equivocal. Occiput Limited data is available on the braincase morphology of the most primitive temnospondyls, and hence little is known of structural variability. Partly because of that, and partly because of its highly unusual morphology, the occiput of Saharastega moradiensis (Figs. 2C, 3C, 4C) proved difficult to interpret, and the following description should therefore be regarded as tentative. As in other large-skulled, Paleozoic temnospondyls such as Eryops (Sawin, 1941) and Edops (Romer and Witter, 1942), the braincase of Saharastega moradiensis is a solidly constructed, well ossified unit. Although the skull has undergone some degree of compaction, the occiput was clearly rather shallow. This contrasts with the relatively deep occiput present in, for example, Eryops and some dissorophoids. The occiput is deeply recessed from the posterior margin of the skull table, which forms a prominent, ‘free’ ledge. Directly below the ledge in the central part of the occiput is an extraordinarily large plate of bone of peculiar morphology, hereafter (informally) termed the occipital plate. The occipital plate has a low, midline ridge dorsally, which merges ventrally with a robust, convex ridge that spans the width of the plate and evidently forms the dorsal margin of the fora-

567

men magnum. Although there are no apparent sutures subdividing the occipital plate, the inverted ‘T’-shaped ridge system may represent a separate ossification. Laterally, the occipital plate on each side forms a well defined suture with a bone most reasonably identified as an exoccipital. The nature of the contact between the occipital plate and the skull table cannot be determined. The identification of the occipital plate in Saharastega moradiensis is perplexing, because we are unaware of a similar structure in any other temnospondyl. In almost all temnospondyls in which the braincase structure is known, the corresponding area is occupied by a large, midline fenestra—thought to have been filled in life by a cartilaginous supraoccipital (e.g., Romer, 1947)—that is flanked by short, occipital processes of the postparietals. In this case, there is no reason to suggest that the plate is formed by separate occipital flanges of the postparietals because of the apparent lack of a midline suture. The plate may be a separate, large supraoccipital ossification, as is present in some early crown-group amniotes (Carroll, 1980; Berman, 2000), but this hypothesis seems unlikely. A more plausible hypothesis is that the plate is formed from separate ossifications, with the median, ridge system being supraoccipital and the remaining bone, on either side of the putative supraoccipital, being postparietal. That identification is at least partially consistent with the condition in the baphetids Megalocephalus (Beaumont, 1977) and, possibly, Kyrinion (Clack, 2003), in which the supraoccipital and opisthotics contribute to a broad, undivided occipital plate across the midline of the occiput dorsally (Carroll, 1980; Berman, 2000). The presence of an ossified supraoccipital, either as a separate ossification or in association with the opisthotics, is thought to characterize baphetids, anthracosaurs, diadectomorphs, some lepospondyls, and crown-group amniotes, but not temnospondyls (Berman, 2000). However, a separate, ossified supraoccipital has been described in the dissorophoids Dissorophus angustus (Carroll, 1964b) and Kamacops acervalis (Gubin, 1980; Schoch, 1999b), and the dvinosaurians Isodectes obtusus (Watson, 1956; Sequeira, 1998) and Dvinosaurus primus (Shishkin, 1973). In particular, the putative supraoccipital of S. moradiensis is similar to that of I. obtusus in resembling an inverted ‘T’. Whatever the correct identification, the braincase morphology of S. moradiensis is unlike that of any other known temnospondyl. As with the rest of the braincase, the exoccipitals are well ossified. Their sutures with neighboring bones cannot be determined but, in addition to a long subotic process (described earlier), they appear to possess robust dorsal and paroccipital processes. Both tabulars display clear evidence of a distinct paroccipital process which is undoubtedly in firm sutural contact (not visible) with the corresponding process on the exoccipitals. This is in contrast with some other temnospondyls, including Dendrerpeton (Steen, 1934), Edops (Romer and Witter, 1942), and, possibly, Chenoprosopus milleri (Langston, 1953), in which the paroccipital process remains partly unossified so that a gap persists between the tabular and exoccipital. Posttemporal fenestrae appear to have been present but their size and shape cannot be determined because of compaction. The unitary occipital condyle comprises paired exoccipitals and a median basioccipital. Although affected by compaction, it is clear that the basioccipital formed most of the cotyle, as in many primitive temnospondyls. In more derived temnospondyls the exoccipitals become the dominant condylar elements, with a concomitant reduction in the basioccipital (Watson, 1919; Shishkin, 1973). In all temnospondyls in which the area is well preserved, the quadrate ramus of the pterygoid sends a dorsally directed ascending ramus that usually contacts the cheek region of the skull roof via a descending flange from the squamosal. Saharastega moradiensis has a well developed ascending ramus but, in contrast to the condition in other temnospondyls, the contact with the cheek has little occipital exposure because of the short suspensorium and the total lack of a squamosal embayment. In-

568

JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 26, NO. 3, 2006

stead, most of the contact appears to be achieved with the ventral surface of the skull table, rather than the cheek, in broad sutural contact anterior to the tabulars (Figs. 3B, 4B). The lateral region of the occiput (i.e., the rear wall of the suspensorium) on the preserved right side lacks sutural detail, and is structureless except for the presence of a shallow fossa directly below margin of the supratemporal, on what is presumably the squamosal. DISCUSSION Relationships of Saharastega moradiensis The extended time range (Early Carboniferous-Early Cretaceous) and tremendous adaptive diversity of temnospondyls renders any analysis of their interrelationships a challenging task. With respect to Paleozoic temnospondyls, a number of phylogenetic analyses that cover a spectrum of taxa have been published (Godfrey et al., 1987; Boy, 1990, 1996; Milner, 1990; Trueb and Cloutier, 1991; Milner and Sequeira, 1994; Holmes et al., 1998; Holmes, 2000; Steyer, 2000; Yates and Warren, 2000; Sequeira, 2004). In those analyses in which the Edopoidea is included, there is broad consensus as regards to their position as the most basal clade of temnospondyls. Five of those analyses are based on parsimony analysis of a data matrix: the analysis of Trueb and Cloutier (1991) focussed on the (putative) relationship between dissorophoids and lissamphibians, while that of Steyer (2000) and of Yates and Warren (2000) covered a broader range of taxa but excluded the Edopoidea. Only the analyses of Holmes et al. (1998) and Sequeira (2004) covered a moderate range of taxa which included the Edopoidea. The phylogenetic analysis performed here is intended to indicate the closest relatives of Saharastega moradiensis among Paleozoic temnospondyls, rather than an analysis of the interrelationships of Paleozoic temnospondyls in general. To this end, the taxon was incorporated into the analysis of Holmes et al. (1998), rather than that of Sequeira (2004), because the former included a slightly broader spectrum of basal temnospondyls. A more comprehensive analysis that includes S. moradiensis will be included with the forthcoming description of the cochleosaurid Nigerpeton ricqlesi (Steyer et al., 2006). The Holmes et al. (1998) data matrix was coded for S. moradiensis (Appendix 1) and entered into MacClade 3 (Maddison and Maddison, 1992). The matrix was analysed by PAUP 3.1.1 (Swofford, 1993), following the same parameters employed by Holmes et al. (1998). It should be noted that an initial run of the data excluding S. moradiensis produced the same number (i.e., three) of most parsimonious trees (MPTs) and identical tree topologies as Holmes et al. (1998), but slightly different tree statistics (Holmes et al., 1998: 34 steps, CI⳱0.53; our analysis: 35 steps, CI⳱0.514). Rerunning the data with S. moradiensis included produced 12 equally parsimonious trees (CI⳱0.486, RI⳱0.612) of 37 steps each. These are briefly described below. In all 12 MPTs, Capetus, Dendrerpeton, Balanerpeton, Trimerorhachis, Isodectes (⳱Saurerpeton of Holmes et al., 1998), Zatrachys, Eoscopus, Sclerocephalus, Eryops and Paryoxis form a clade to the exclusion of the edopoids Edops and Chenoprosopus. However, as reflected in a strict consensus tree (Fig. 5A), there is some instability in their internal relationships. Significantly, Saharastega moradiensis always forms a clade with Edops and Chenoprosopus, either as the sister taxon to the clade [Edops + Chenoprosopus], or as the sister taxon to Chenoprosopus with Edops as their respective sister taxon. The strict consensus (Fig. 5A) shows a polytomy comprising these three taxa, and the node is supported by one unambiguous synapomorphy, the exclusion of the lacrimal from the orbital margin by a prefrontal-jugal contact. Our preferred phylogenetic hypothesis (Fig. 5B) was selected from among the MPTs on the basis that it is both fully resolved, and maintains the integrity of the clade

FIGURE 5. Phylogenetic position of Saharastega moradiensis among Paleozoic temnospondyls, based on a PAUP analysis of the data matrix of Holmes et al. (1998). A, Strict consensus of 12 MPTs; in all the MPTs, S. moradiensis forms a clade with the edopoids Chenoprosopus and Edops. B, Preferred phylogeny representing one of the 12 MPTs, in which S. moradiensis is the sister taxon of the edopoids Chenoprosopus and Edops.

Edopoidea (i.e., [Edops + Chenoprosopus] in this analysis) as widely conceived. The monophyly of the Edopoidea is supported by one unambiguous synapomorphy, the presence of an anteroposteriorly elongated premaxilla. Other characters of S. moradiensis and edopoids not included in the analysis are discussed below. In a series of papers by A. R. Milner and S. E. K. Sequeira (Milner and Sequeira, 1993, 1994, 1998; Sequeira and Milner, 1993; Sequeira, 1996, 2004), the Edopoidea (Romer, 1945 as Edopsoidea, emended Langston, 1953) was defined as a clade consisting of the Edopidae and Cochleosauridae, and argued to be the most primitive group of temnospondyls (cf. Romer, 1947). The phylogenetic analysis indicates that Saharastega moradiensis is more closely related to edopoids than to other Paleozoic temnospondyls. In addition to the synapomorphy of exclusion of the lacrimal from the orbital margin by a prefrontal-jugal contact, S. moradiensis shares with the Edopoidea the broad contact between the jugal and lacrimal anterior to the orbit (considered an edopoid synapomorphy by Milner and Sequeira, 1994), as well as a number of primitive character states otherwise unusual within the Temnospondyli. These include the retention of an intertemporal and, in all likelihood, the arrangement of the palatal rami of the pterygoids whereby the vomers and palatines are excluded from the margin of the interpterygoid vacuities. However, S. moradiensis lacks what is arguably the most critical synapomorphy of edopoids, namely the presence of an anteroposteriorly

DAMIANI ET AL.—PERMIAN TEMNOSPONDYL FROM NIGER elongate premaxilla that is expanded marginally but not medially and which is associated with small, posteriorly located nostrils (Hook, 1993; Milner and Sequeira, 1994, 1998). Intriguingly, S. moradiensis shares a number of synapomorphies with the Cochleosauridae. These include the presence of elongate vomers that are associated with long choanae, small, rounded interpterygoid vacuities that are located in the posterior half of the skull, and the absence of a pineal in adults (Hook, 1993; Godfrey and Holmes, 1995; Milner and Sequeira, 1994, 1998). Note, however, that unlike in cochleosaurids the vomers are not elongated anterior to the choanae, and the choanae are not wider anteriorly than posteriorly. In conclusion, Saharastega moradiensis is clearly a rather primitive temnospondyl as it possesses the following plesiomorphic character states: on the skull roof, the presence of an intertemporal and the lack of (exposed) sensory sulci; on the palate, the extensive denticle cover, the entry of the premaxilla into the choanal margin, the narrow pterygoid-parasphenoid basicranial articulation, the anterior extension of the palatal rami of the pterygoids to exclude the vomers and palatines from the interpterygoid vacuities, the dominant basioccipital, and, possibly, the large, rhomboidal sphenethmoid. Among primitive temnospondyls, S. moradiensis appears to be most closely related to edopoids, and can provisionally be regarded as a late-surviving member of a clade that includes as a sub-clade the Edopoidea. This late occurrence accounts for the presence of a number of apomorphies in an otherwise primitive temnospondyl. Biogeographic Considerations Saharastega moradiensis and its temnospondyl contemporaries in the Moradi Formation of Niger represent the first record of Late Permian temnospondyls from northern Africa. This expands the known range of Gondwanan Late Permian temnospondyls from southern and eastern Africa to the equatorial region of northern Africa. However, the Moradi temnospondyl fauna is distinctly Euramerican in aspect, and most closely resembles Euramerican faunas of Pennsylvanian to Early Permian age (Milner, 1993; Sidor et al., 2005). This is in marked contrast to the stereospondyl dominated southern and eastern African faunas (Anderson and Cruickshank, 1978; Milner, 1993; Damiani and Rubidge, 2003). It is also significant that a fauna of similarly Euramerican aspect is present in the supposed Upper Permian levels of the Argana Formation of Morocco (cf. Milner, 1993; Jalil, 1999). Together, the Moradi and Argana faunas hint at an unexpected degree of biogeographic endemism for northwest Africa during the Late Permian (Sidor et al., 2005). Most continental tetrapods of Pennsylvanian-Early Permian age appear to have been restricted to an equatorial belt of Pangea which Milner (1993) termed the Edaphosaur-Nectridian Province. This faunal province, which is almost exclusively Euramerican, includes the Edopoidea. Milner (1993) predicted that Euramerican faunal elements might be expected to occur in equatorial regions of Gondwana during the Early Permian, because Laurasia and Gondwana had been in contact by that time. The Moradi tetrapod fauna (Ricqles and Taquet, 1982; Sidor et al., 2003a, b, 2005), along with the contemporaneous fauna from the lower levels of the Argana Formation of Morocco (Jalil, 1996, 1999), partially fulfils that prediction. However, both of those faunas are apparently of Late Permian age, and therefore postdate the edaphosaur-nectridian faunal province. The simplest explanation for the presence of Euramerican faunal elements (i.e., edopoid-relatives and cochleosaurids) in the Upper Permian of northern Africa is that they represent a range extension of the Edaphosaur-Nectridian Province (Milner, 1993). These extensions appear to have developed vicariantly. Hence, primitive temnospondyls may have been present in the northern, equatorial region of Africa during the Pennsylvanian-

569

Early Permian. Paleogeographic reconstructions of Late Carboniferous Pangea support this scenario in that the ice sheet which covered much of the rest of Gondwana at that time did not extend to northwestern Africa (Scotese, 2001). Temnospondyls present in this region included members of a clade which is the sister-group of the Edopoidea. Late-surviving members of that clade, such as Saharastega moradiensis, do not appear to have expanded their geographic range into southern Gondwana following climatic amelioration in the Permian. Milner (1993) argued for the presence of two amphibian faunal provinces during the Late Permian, namely a northern Dvinosaurid-Chroniosuchid Province, and a southern (i.e., Gondwanan) Rhinesuchid Province (cf. Schoch, 2000). The Moradi temnospondyl fauna evidently does not pertain to the Rhinesuchid Province, and Sidor et al. (2003a, b, 2005) have hypothesized that north Africa hosted an endemic tetrapod fauna that formed part of a distinct biogeographic province during the Late Permian. Evidently, this fauna combined elements from the more primitive Pennsylvanian-Upper Permian equatorial faunas of Euramerica (i.e., the temnospondyls), with those from groups having a much broader distribution during the Permian (i.e., captorhinids and pareiasaurs). ACKNOWLEDGMENTS We thank F. R. O’Keefe, A. Dindine, B. Gado, and D. Sindy for their assistance in the field, A. Crean of the South African Museum for preparation of the specimen, and A. M. Yates for transporting the specimen from Cape Town to Johannesburg on our behalf. R. B. Holmes, A. R. Milner, A. M. Yates and F. R. O’Keefe provided valuable comments and discussions on the new amphibian, and A. R. Milner provided a copy of Sequeira’s manuscript on Cochleosaurus bohemicus prior to publication. A. M. Jeannot assisted with Figure 3A; all other drawings are by the senior author. For permission to study specimens in their care the senior author would like to thank F. A. Jenkins Jr. and C. Schaff (Museum of Comparative Zoology, Cambridge), and E. S. Gaffney and C. Mehling (AMNH, New York). The senior author’s research was supported by funding from B. S. Rubidge and the University of the Witwatersrand, the Palaeontological Scientific Trust (PAST), and a Collections Study Grant from the AMNH. Comments on the manuscript by R. B. Holmes and an anonymous referee improved the quality of this work. We are indebted to the United States Embassy and Cultural Center (Niamey) for diplomatic assistance and hospitality, and H. M. Salissou for permission to conduct research in Niger. Our 2003 fieldwork was supported by grant number 7258-02 from the National Geographic Society. LITERATURE CITED Anderson, J. M., and A. R. I. Cruickshank. 1978. The biostratigraphy of the Permian and the Triassic. Part 5. A review of the classification and distribution of Permo-Triassic tetrapods. Palaeontologia africana 21:15–44. Beaumont, E. H. 1977. Cranial morphology of the Loxommatidae (Amphibia: Labyrinthodontia). Philosophical Transactions of the Royal Society of London B 280:29–101. Berman, D. S. 2000. Origin and early evolution of the amniote occiput. Journal of Paleontology 74:938–956. Boy, J. A. 1990. Über einige Vertreter der Eryopoidea (Amphibia: Temnospondyli) aus dem europäischen Rotliegend (?höchstes KarbonPerm). Paläontologische Zeitschrift 64:287–312. Boy, J. A. 1996. Ein neuer Eryopoide (Amphibia: Temnospondyli) aus

570

JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 26, NO. 3, 2006

dem saarpfälzischen Rotliegend (Unter-Perm; Südwest-Deutschland). Mainzer geowissenschaftliche Mitteilungen 25:7–26. Boy, J. A., and H.-D. Sues. 2000. Branchiosaurs: larvae, metamorphosis and heterochrony in temnospondyls and seymouriamorphs; pp. 1150–1197 in H. Heatwole and R. L. Carroll (eds.), Amphibian Biology, Volume 4. Palaeontology, the evolutionary history of amphibians. Surrey Beatty & Sons, Chipping Norton. Bystrow, A. P. 1935. Morphologische Untersuchungen der Deckknochen des Schädels der Wirbeltiere. Acta Zoologica 16:65–141. Bystrow, A. P., and J. A. Efremov. 1940. Benthosuchus sushkini Efr.—A labyrinthodont from the Eotriassic of Sharzhenga River. Trudy Paleontologicheskogo Instituta 10:1–152. [Russian] Carroll, R. L. 1964a. The relationships of the rhachitomous amphibian Parioxys. American Museum Novitates 2167:1–11. Carroll, R. L. 1964b. Early evolution of the dissorophid amphibians. Bulletin of the Museum of Comparative Zoology at Harvard University 131:161–250. Carroll, R. L. 1980. The hyomandibular as a supporting element in the skull of primitive tetrapods; in A. L. Panchen (ed.), The Terrestrial Environment and the Origin of Land Vertebrates. Systematics Association Special Volume 15:293–317. Chase, J. N. 1965. Neldasaurus wrightae, a new rhachitomous labyrinthodont from the Texas Lower Permian. Bulletin of the Museum of Comparative Zoology at Harvard University 133:153–225. Clack, J. A. 2003. A new baphetid (stem tetrapod) from the Upper Carboniferous of Tyne and Wear, U.K., and the evolution of the tetrapod occiput. Canadian Journal of Earth Sciences 40: 483–498. Clack, J. A., and R. Holmes. 1988. The braincase of the anthracosaur Archeria crassidisca with comments on the interrelationships of primitive tetrapods. Palaeontology 31:85–107. Damiani, R. J. 2001. A systematic revision and phylogenetic analysis of Triassic mastodonsauroids (Temnospondyli: Stereospondyli). Zoological Journal of the Linnean Society 133:379–482. Damiani, R. J. 2004. Temnospondyls from the Beaufort Group (Karoo Basin) of South Africa and their biostratigraphy. Gondwana Research 7:165–173. Damiani, R. J., and B. S. Rubidge. 2003. A review of the South African temnospondyl amphibian record. Palaeontologia africana 39:21– 36. Damiani, R. J., and A. M. Yates. 2003. The Triassic amphibian Thoosuchus yakovlevi and the relationships of the Trematosauroidea (Temnospondyli: Stereospondyli). Records of the Australian Museum 55: 331–342. Dilkes, D. W. 1990. A new trematopsid amphibian (Temnospondyli: Dissorophoidea) from the Lower Permian of Texas. Journal of Vertebrate Paleontology 10:222–243. Fritsch, A. 1885. Fauna der Gaskohle und der Kalksteine der Permformation Böhmens. II. Rivnac, Prague, 64 pp. Ginsburg, L., A. F. de Lapparent, B. Loiret, and P. Taquet. 1966. Empreintes de pas de vertébrés tétrapodes dans la série continentale à l’ouest d’Agadès (République du Niger). Comptes Rendus de l’Académie des Sciences de Paris, Série D 263:28–31. Godfrey, S. J., A. R. Fiorillo, and R. L. Carroll. 1987. A newly discovered skull of the temnospondyl amphibian Dendrerpeton acadianum Owen. Canadian Journal of Earth Sciences 24:796–805. Godfrey, S. J., and R. Holmes. 1995. The Pensylvanian temnospondyl Cochleosaurus florensis Rieppel, from the lycopsid stump fauna at Florence, Nova Scotia. Breviora 500:1–25. Golubev, V. K. 2000. The faunal assemblages of Permian terrestrial vertebrates from eastern Europe. Paleontological Journal (34, Supplement 2):211–224. Gubin, Y. M. 1980. New Permian dissorophids of the Ural forelands. Paleontological Journal 1980(3):82–90. Gubin, Y. M. 1991. Permian archegosauroid amphibians of the USSR. Trudy Paleontologicheskogo Instituta 249:1–141. [Russian] Gubin, Y. M. 1997. Skull morphology of Archegosaurus decheni Goldfuss (Amphibia, Temnospondyli) from the Early Permian of Germany. Alcheringa 21:103–121. Gubin, Y. M., I. V. Novikov, and M. Morales. 2000. A review of anomalies in the structure of the skull roof of temnospondylous labyrinthodonts. Paleontological Journal (34, Supplement 2):154–164. Holmes, R. 2000. Palaeozoic temnospondyls; pp. 1081–1120 in H. Heatwole and R. L. Carroll (eds.), Amphibian Biology, Volume 4. Pal-

aeontology, the evolutionary history of amphibians. Surrey Beatty & Sons, Chipping Norton. Holmes, R. B., R. L. Carroll, and R. R. Reisz. 1998. The first articulated skeleton of Dendrerpeton acadianum (Temnospondyli, Dendrerpetontidae) from the Lower Pennsylvanian locality of Joggins, Novia Scotia, and a review of its relationships. Journal of Vertebrate Paleontology 18:64–79. Hook, R. W. 1993. Chenoprosopus lewisi, a new cochleosaurid amphibian (Amphibia: Temnospondyli) from the Permo-Carboniferous of north-central Texas. Annals of Carnegie Museum 62:273–291. Howie, A. A. 1970. A new capitosaurid labyrinthodont from East Africa. Palaeontology 13:210–253. Ivakhnenko, M. F., V. K. Golubev, Y. M. Gubin, N. N. Kalandadze, I. V. Novikov, A. G. Sennikov, and A. S. Rautian. 1997. Permian and Triassic tetrapods of Eastern Europe. GEOS, Moscow, 216 pp. [Russian] Jalil, N.-E. 1996. Les Vertébrés permiens et triasiques de la Formation d’Argana (Haut Atlas occidental): liste faunique préliminaire et implications stratigraphiques; pp. 227–250 in F. Medina (ed.), Le Permien et le Trias du Maroc: état des connaissances. Editions Pumag, Marrakech. Jalil, N.-E. 1999. Continental Permian and Triassic vertebrate localities from Algeria and Morocco and their stratigraphical correlations. Journal of African Earth Sciences 29:219–226. Langston, W., Jr. 1953. Permian amphibians from New Mexico. University of California Publications in Geological Sciences 29:349– 416. Li, J.-L., and Z.-W. Cheng. 1999. New anthracosaur and temnospondyl amphibians from Gansu, China—the fifth report on Late Permian Dashankou lower tetrapod fauna. Vertebrata PalAsiatica 37: 234–237. [Chinese 234–241; English 242–247] Linnaeus, C. 1758. Systema naturae per regna tria naturae, secundum classes, ordines, genera, species cum characteribus, differentiis, synonymis, locis. Editio decima, reformata, Tom. I. Laurentii Salvii, Stockholm, 824 pp. Lucas, S. G. 1998. Toward a tetrapod biochronology of the Permian. New Mexico Museum of Natural History and Science Bulletin 12:71– 91. Maddison, W. P., and D. R. Maddison. 1992. MacClade Version 3. Sinauer Associates, Sunderland, Massachusetts. Malabarba, M. C., F. Abdala, F. E. Weiss, and P. A. Perez. 2003. New data on the Late Permian vertebrate fauna of Posto Queimado, Rio do Rasto Formation, southern Brazil. Revista Brasileira de Paleontologia 6:49–54. Marsicano, C. A., and A. Warren. 1998. The first Palaeozoic rhytidosteid: Trucheosaurus major (Woodward, 1909) from the Late Permian of Australia, and a reassessment of the Rhytidosteidae (Amphibia, Temnospondyli). Bulletin of the Natural History Museum, London (Geology) 54:147–154. Mehl, M. G. 1913. A description of Chenoprosopus milleri. Publications of the Carnegie Institute, Washington 186:11–16. Milner, A. R. 1980. The temnospondyl amphibian Dendrerpeton from the Upper Carboniferous of Ireland. Palaeontology 23:125–141. Milner, A. R. 1990. The radiations of temnospondyl amphibians; in P. D. Taylor and G. P. Larwood (eds.), Major Evolutionary Radiations. Systematics Association Special Volume 42:321–349. Milner, A. R. 1993. Biogeography of Palaeozoic tetrapods; pp. 324–353 in J. A. Long (ed.), Palaeozoic vertebrate biostratigraphy and biogeography. Belhaven Press, London. Milner, A. R., and S. E. K. Sequeira. 1993. Capetus and the problems of primitive temnospondyl relationships; in U. Heidtke (ed.), New Research on Permo-Carboniferous Faunas. Pollichia-Buch 29: 193–199. Milner, A. R., and S. E. K. Sequeira. 1994. The temnospondyl amphibians from the Viséan of East Kirkton, West Lothian, Scotland. Transactions of the Royal Society of Edinburgh: Earth Sciences 84:331–361. Milner, A. R., and S. E. K. Sequeira. 1998. A cochleosaurid temnospondyl amphibian from the Middle Pennsylvanian of Linton, Ohio, U.S.A. Zoological Journal of the Linnean Society 122:261–290.

DAMIANI ET AL.—PERMIAN TEMNOSPONDYL FROM NIGER Milner, A. R., and S. E. K. Sequeira. 2003. On a small Cochleosaurus described as a large Limnogyrinus (Amphibia, Temnospondyli) from the Upper Carboniferous of the Czech Republic. Acta Palaeontologica Polonica 48:143–147. Modesto, S. P., and R. M. H. Smith. 2001. A new Late Permian captorhinid reptile: a first record from the South African Karoo. Journal of Vertebrate Paleontology 21:405–409. Moustafa, Y. S. 1955. The skeletal structure of Parioxys ferricolus, Cope. Bulletin de 1’Institut d’Egypte 36:41–76. Panchen, A. L. 1959. A new armoured amphibian from the Upper Permian of east Africa. Philosophical Transactions of the Royal Society of London B 242:207–281. Piveteau, J. 1926. Amphibiens et reptiliens permiens. Annales de Paléontologie 15:55–179. Ricqles, A. de., and P. Taquet. 1982. La faune de vertébrés du Permien supérieur du Niger. I. Le Captorhinomorphe Moradisaurus grandis (Reptilia, Cotylosauria)—Le crâne. Annales de Paléontologie (Vert.-Invert.) 68:33–106. Rieppel, O. 1980. The edopoid amphibian Cochleosaurus from the Middle Pennsylvanian of Novia Scotia. Palaeontology 23: 143–149. Romer, A. S. 1936. Studies on American Permo-Carboniferous tetrapods. Problems of Paleontology 1:85–93. Romer, A. S. 1945. Vertebrate Paleontology, 2nd Edition. The University of Chicago Press, Chicago, Illinois, 687 pp. Romer, A. S. 1947. Review of the Labyrinthodontia. Bulletin of the Museum of Comparative Zoology at Harvard College 99:1–368. Romer, A. S., and R. V. Witter. 1942. Edops, a primitive rhachitomous amphibian from the Texas red beds. Journal of Geology 50: 925–960. Rubidge, B. S. 1995. Biostratigraphy of the Beaufort Group (Karoo Supergroup). South African Committee for Stratigraphy Biostratigraphic Series No. 1. Council for Geoscience, Pretoria, 46 pp. Ruta, M., M. I. Coates, and D. L. J. Quicke. 2003. Early tetrapod relationships revisited. Biological Reviews 78:251–345. Sawin, H. J. 1941. The cranial anatomy of Eryops megacephalus. Bulletin of the Museum of Comparative Zoology at Harvard College 88: 407–463. Schoch, R. R. 1997. Cranial anatomy of the Permian temnospondyl amphibian Zatrachys serratus Cope 1878, and the phylogenetic position of the Zatrachydidae. Neues Jahrbuch für Geologie und Paläontologie, Abhandlungen 206:223–248. Schoch, R. R. 1999a. Studies on braincases of early tetrapods: Structure, morphological diversity, and phylogeny—1. Trimerorhachis and other primitive temnospondyls. Neues Jahrbuch für Geologie und Paläontologie, Abhandlungen 213:233–259. Schoch, R. R. 1999b. Studies on braincases of early tetrapods: Structure, morphological diversity, and phylogeny—2. Dissorophoids, eryopids, and stereospondyls. Neues Jahrbuch für Geologie und Paläontologie, Abhandlungen 213:289–312. Schoch, R. R. 2000. Biogeography of stereospondyl amphibians. Neues Jahrbuch für Geologie und Paläontologie, Abhandlungen 215: 201–231. Schoch, R. R. 2001. Can metamorphosis be recognised in Palaeozoic amphibians? Neues Jahrbuch für Geologie und Paläontologie, Abhandlungen 220:335–367. Schoch, R. R., and A. R. Milner. 2000. Stereospondyli. Handbuch der Paläoherpetologie, Teil 3B. Verlag Dr. Friedrich Pfeil, München, 203 pp. Scotese, C. R. 2001. Atlas of Earth History, Volume 1, Paleogeography. Paleomap Project, Arlington, Texas, 52 pp. Sequeira, S. E. K. 1996. A cochleosaurid amphibian from the Upper Carboniferous of Ireland. Special Papers in Palaeontology 52: 65–80. Sequeira, S. E. K. 1998. The cranial morphology and taxonomy of the saurerpetontid Isodectes obtusus comb. nov. (Amphibia: Temnospondyli). Zoological Journal of the Linnean Society 122: 237–259. Sequeira, S. E. K. 2004. The skull of Cochleosaurus bohemicus Fricˇ, a temnospondyl from the Czech Republic (Upper Carboniferous) and cochleosaurid interrelationships. Transactions of the Royal Society of Edinburgh: Earth Sciences 94:21–43. Sequeira, S. E. K., and A. R. Milner. 1993. The temnospondyl amphibian

571

Capetus from the Upper Carboniferous of the Czech Republic. Palaeontology 36:657–680. Shishkin, M. A. 1968. On the cranial arterial system of the labyrinthodonts. Acta Zoologica 49:1–22. Shishkin, M. A. 1973. The morphology of the early Amphibia and some problems of the lower tetrapod evolution. Trudy Paleontologicheskogo Instituta 137:1–260. [Russian] Shishkin, M. A., and B. S. Rubidge. 2000. A relict rhinesuchid (Amphibia: Temnospondyli) from the Lower Triassic of South Africa. Palaeontology 43:653–670. Sidor, C. A., H. C. E. Larsson, J.-S. Steyer, F. R. O’Keefe, and R. M. H. Smith. 2003a. Late Permian vertebrates from the Sahara. Journal of Vertebrate Paleontology (23, Supplement):97A. Sidor, C. A., D. C. Blackburn, and B. Gado. 2003b. The vertebrate fauna of the Upper Permian of Niger—II, Preliminary description of a new pareiasaur. Palaeontologia Africana 39:45–52. Sidor, C. A., F. R. O’Keefe, R. Damiani, J.-S. Steyer, R. M. H. Smith, H. C. E. Larsson, P. C. Sereno, O. Ide, and A. Maga. 2005. Permian tetrapods from the Sahara show climate-controlled endemism in Pangaea. Nature 434:886–889. Smithson, T. R. 1982. The cranial morphology of Greererpeton burkemorani Romer (Amphibia: Temnospondyli). Zoological Journal of the Linnean Society 76:29–90. Smithson, T. R. 2000. Anthracosaurs; pp. 1121–1149 in H. Heatwole and R. L. Carroll (eds.), Amphibian Biology, Volume 4. Palaeontology, the evolutionary history of amphibians. Surrey Beatty & Sons, Chipping Norton. Steen, M. C. 1934. The amphibian fauna from South Joggins, Novia Scotia. Proceedings of the Zoological Society of London 1934: 465–504. Steen, M. C. 1938. On the fossil Amphibia from the Gas Coal of Nýrˇany and other deposits in Czechoslovakia. Proceedings of the Zoological Society of London, Series B 108:205–283. Steyer, J.-S. 2000. Ontogeny and phylogeny in temnospondyls: a new method of analysis. Zoological Journal of the Linnean Society 130: 449–467. Steyer, J.-S. 2002. The first articulated trematosaur “amphibian” from the Lower Triassic of Madagascar: implications for the phylogeny of the group. Palaeontology 45:771–793. Steyer, J.-S., R. Damiani, C. A. Sidor, F. R. O’Keefe, H. C. E. Larsson, A. Maga, and O. Ide. 2006. The vertebrate fauna of the Upper Permian of Niger. IV. Nigerpeton ricqlesi (Temnospondyl: Cochleosauridae), and the edopoid colonization of Gondwana. Journal of Vertebrate Paleontology 26:18–28. Swofford, D. L. 1993. PAUP: Phylogenetic Analysis Using Parsimony, Version 3.1.1. Illinois Natural History Survey, Champaign. Taquet, P. 1967. Découvertes paléontologiques récentes dans le Nord du Niger. Colloques internationaux du Centre National de la Recherche Scientifique 163:415–418. Taquet, P. 1969. Première découverte en Afrique d’un Reptile Captorhinomorphe (Cotylosaurien). Comptes Rendus de l’Académie des Sciences de Paris, Série D 268:779–781. Taquet, P. 1972. Un exemple de datation et de corrélation stratigraphique basé sur les Captorhinomorphes (Reptiles cotylosauriens). Mémoires du Bureau de Recherches Géologiques et Minières 77:407–409. Taquet, P. 1976. Géologie et paléontologie du gisement de Gadoufaoua (Aptien du Niger). Cahiers de Paléontologie, Paris, 191 pp. Trueb, L., and R. Cloutier. 1991. A phylogenetic investigation of the inter- and intrarelationships of the Lissamphibia (Amphibia: Temnospondyli); pp. 223–313 in H. P. Schultze and L. Trueb (eds.), Origins of the higher groups of tetrapods—controversy and consensus. Cornell University Press, Ithaca. Warren, A. A. 2000. Secondarily aquatic temnospondyls of the Upper Permian and Mesozoic; pp. 1121–1149 in H. Heatwole and R. L. Carroll (eds.), Amphibian Biology, Volume 4. Palaeontology, the evolutionary history of amphibians. Surrey Beatty & Sons, Chipping Norton. Warren, A. A., and L. Davey. 1992. Folded teeth in temnospondyls—a preliminary study. Alcheringa 16:107–132.

572

JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 26, NO. 3, 2006

Warren, A., and C. Marsicano. 2000. A phylogeny of the Brachyopoidea (Temnospondyli, Stereospondyli). Journal of Vertebrate Paleontology 20:462–483. Watson, D. M. S. 1919. The structure, evolution and origin of the Amphibia. The “Orders” Rhachitomi and Stereospondyli. Philosophical Transactions of the Royal Society of London B 209:1–73. Watson, D. M. S. 1956. The brachyopid labyrinthodonts. Bulletin of the British Museum (Natural History), Geology 2:317–391. Watson, D. M. S. 1962. The evolution of the labyrinthodonts. Philosophical Transactions of the Royal Society of London B 245: 219–265. Werneburg, R., and J. Schneider. 1996. The Permian temnospondyl amphibians of India. Special Papers in Palaeontology 52:105–128. Yates, A. M., and A. A. Warren. 2000. The phylogeny of the ‘higher’ temnospondyls (Vertebrata: Choanata) and its implications for the monophyly and origins of the Stereospondyli. Zoological Journal of the Linnean Society 128:77–121. Zittel, K. A. von. 1887–1890. Handbuch der Paläontologie. Abteilung 1. Paläozoologie Band III. Vertebrata (Pisces, Amphibia, Reptilia, Aves). Oldenbourg, Munich and Leipzig, 900 pp. Submitted 2 February 2005; accepted 21 March 2006.

APPENDIX 1. Data matrix of Holmes et al. (1998), incorporating Saharastega moradiensis. The colosteid Greererpeton and the anthracosaur Proterogyrinus are the outgroups. Saurerpeton, listed in Holmes et al. (1998), is a junior synonym of Isodectes (Sequeira, 1998). Characters Taxon Chenoprosopus Edops Dendrerpeton Balanerpeton Trimerorhachis Capetus Eryops Greererpeton Proterogyrinus Isodectes Paryoxis Zatrachys Eoscopus Sclerocephalus Saharastega

12345

1 67890

11111 12345

111 678

11101 11001 11000 11000 11010 11000 11000 00010 00100 11110 11000 11001 11000 11000 11110

1?000 11000 00111 00111 01101 0?110 11111 0000– 000–0 01011 11010 0?101 01111 11101 1???0

00000 00000 11000 00000 –0100 0000? 00001 –0––0 00000 –011? 00001 00001 11111 00001 –?000

0?0 0?0 000 000 000 0?0 110 000 000 000 111 1?1 101 100 0?0