651
The enigmatic diapsid Acerosodontosaurus piveteaui (Reptilia: Neodiapsida) from the Upper Permian of Madagascar and the paraphyly of ‘‘younginiform’’ reptiles Constanze Bickelmann, Johannes Mu¨ller, and Robert R. Reisz
Abstract: A restudy of the Upper Permian diapsid Acerosodontosaurus piveteaui from Madagascar indicates that the bone formerly identified as the quadratojugal is a fragment of a rib. This in turn implies that, in contrast to previous studies, the lower temporal arcade must be considered incomplete and derived relative to the ancestral condition. Since the phylogenetic position of Acerosodontosaurus is poorly understood, the taxon was entered into a modified phylogenetic data matrix of diapsid reptiles, and the purported monophyly of ‘‘Younginiformes’’ was tested for the first time by including all potential members of the clade as separate taxa, as well as other taxa from the same deposits. The results of the phylogenetic analysis do not support the monophyly of ‘‘younginiform’’ reptiles. Instead, most taxa cluster unresolved at the base of Neodiapsida, a finding that has important implications for the understanding of early diapsid evolution because it suggests that early neodiapsids represent several distinct evolutionary lineages. Acerosodontosaurus and Hovasaurus do form a clade, a finding consistent with the stratigraphic age and biogeography of these taxa. Re´sume´ : Une nouvelle e´tude du diapside du Permien supe´rieur Acerosodontosaurus piveteaui de Madagascar indique que l’os pre´ce´demment identifie´ comme e´tant le quadrajugal est en fait un fragment de coˆte, ce qui signifie, a` l’encontre des interpre´tations d’e´tudes pre´ce´dentes, que l’arcade temporale infe´rieure doit eˆtre conside´re´e incomple`te et de´rive´e par rapport a` l’e´tat ancestral. E´tant donne´ que la position phyloge´ne´tique d’Acerosodontosaurus est mal comprise, le taxon a e´te´ verse´ dans une matrice de donne´es phyloge´ne´tiques modifie´e des reptiles diapsides et la pre´tendue monophylie des « Younginiformes » a pour la premie`re fois fait l’objet d’un test consistant a` inclure tous les membres potentiels du clade en tant que taxons distincts, ainsi que d’autres taxons provenant des meˆmes de´poˆts. Les re´sultats de l’analyse phyloge´ne´tique n’appuient pas l’hypothe`se de la monophylie des reptiles « younginiformes ». La plupart des taxons forment plutoˆt des groupes non re´solus a` la base des Ne´odiapside´s, une conclusion qui a d’importantes re´percussions sur la compre´hension de l’e´volution pre´coce des diapsides, puisqu’elle porte a` croire que les ne´odiapsides pre´coces repre´sentent plusieurs ligne´es e´volutionnaires distinctes. Acerosodontosaurus et Hovasaurus constituent effectivement un clade, ce qui concorde avec l’aˆge stratigraphique et la bioge´ographie de ces taxons. [Traduit par la Re´daction]
Introduction The so-called ‘‘younginiform’’ reptiles (Benton 1985; Romer 1945) are known exclusively from the Upper Permian and Late Triassic of Madagascar and East Africa (Broom 1914; Carroll 1981; Currie 1980, 1981, 1982; Harris and Carroll 1977; Haughton 1924; Piveteau 1926). Originally thought to be the most basal diapsids, this view was modiReceived 8 April 2009. Accepted 31 July 2009. Published on the NRC Research Press Web site at cjes.nrc.ca on 22 September 2009. Paper handled by Associate Editor H.-D. Sues. C. Bickelmann1 and J. Mu¨ller. Museum fu¨r Naturkunde – Leibniz-Institut fu¨r Evolutions- und Biodiversita¨tsforschung an der Humboldt-Universita¨t zu Berlin, Invalidenstrasse 43, D10115 Berlin, Germany. R.R. Reisz. Department of Biology, University of Toronto at Mississauga, 3359 Mississauga Road, Mississauga, ON L5L 1C6, Canada. 1Corresponding
author (e-mail:
[email protected]).
Can. J. Earth Sci. 46: 651–661 (2009)
fied after careful investigations of Petrolacosaurus and other araeoscelid reptiles (Reisz 1977, 1981; Reisz et al. 1984). Although ‘‘younginiform’’ reptiles were occasionally considered as lepidosauromorphs (Benton 1985; Evans 1988), Gaffney (1980), and later Laurin (1991) eventually placed them in their currently accepted position at the base of Neodiapsida. However, the in-group relationships of ‘‘Younginiformes,’’ as well as their monophyletic status, are neither understood nor have they been tested in a modern phylogenetic framework. Currie suggested their monophyly in 1982, including Youngina capensis (Broom 1914), Hovasaurus boulei (Currie 1981), Tangasaurus mennelli (Currie 1982), Kenyasaurus mariakanensis (Harris and Carroll 1977), as well as Thadeosaurus colcanapi (Carroll 1981), but this interpretation was solely based on Hennigian argumentation, which was the only technique available at that time. The same methodology was also used in studies by Benton (1985) and Evans (1988) on diapsid ‘‘younginiform’’ phylogeny, in which both authors also included another taxon from the same deposits, Acerosodontosaurus piveteaui (Currie 1980). While Currie (1982) had surprisingly ignored this taxon in his phylogenetic study, both Benton (1985) and
doi:10.1139/E09-038
Published by NRC Research Press
652
Evans (1988) placed it as sister taxon to all other ‘‘Younginiformes.’’ Acerosodontosaurus piveteaui is known from only a single specimen housed at the Muse´um national d’histoire naturelle, Paris (MNHN; collection no. 1908-32-57), which preserves an almost complete and articulated right skull side, as well as vertebrae, ribs, gastralia, parts of the limbs, and parts of pectoral and pelvic girdle in a nodule of fine sandstone. The specimen comes from the Lower Sakamena Formation (Upper Permian) of the Sakamena River Valley in southern Madagascar. The formation is dominated by muddy conglomerates and sandstones with sandy interbeds and thin limestone beds, indicating nearshore and offshore facies of linear rift-valley lakes (Wescott and Diggens 1998; Smith 2000). The deposits yield a diverse vertebrate fauna including many reptilian taxa, such as the basal diapsids Claudiosaurus (Carroll 1981) and Coelurosauravus (Piveteau 1926), the procolophonoid Barasaurus (Piveteau 1955), as well as the ‘‘younginiforms’’ Hovasaurus and Thadeosaurus (Smith 2000; Ketchum and Barrett 2004). In the present study, the anatomy and phylogenetic affinities of Acerosodontosaurus are reassessed in the light of recent increases in our knowledge of early diapsid anatomy and phylogeny. In addition, we test for the first time the traditional view that ‘‘Younginiformes’’ form a monophyletic group within diapsid reptiles.
Material and methods Specimen MNHN 1908-32-57 consists of a natural mold, which preserves both sides of the skeleton (Fig. 1). It comprises an almost complete and articulated right side of the skull roof (Fig. 2), parts of the thoracic vertebral column with ribs and gastralia (Fig. 3B–3E), as well as parts of the front and hind limbs (Fig. 3A), and parts of pectoral and pelvic girdle (Fig. 1). We assume that the specimen was at least subadult, as indicated by the complete ossification of the manus and the absence of neurocentral sutures in the vertebrae, of which only one posterior dorsal shows some doubtful traces of a suture (see later in the text). Two new latex casts were made for this study, since the original latex casts have deteriorated over time. The new mold that was produced by the staff of the MNHN is of better quality and higher fidelity than the original latex cast as it is made of a particularly delicate silicone that not only reproduces very well the anatomy of the skeleton, but also is less intrusive than the original latex mold. In addition, the matrix is very resistant, and there is no noticeable deterioration in its fidelity because of the molding process. Pictures were taken of the casts, outlines produced in Adobe Illustrator, and stipple drawings were made by hand. For the phylogenetic analyses, eight additional taxa were added to an existing matrix from Mu¨ller (2004), resulting in a total of 38 taxa. The additional taxa are Acerosodontosaurus, Youngina, Thadeosaurus, Kenyasaurus, Tangasaurus, Hovasaurus, Lanthanolania ivakhnenkoi (Modesto and Reisz 2002), and Galesphyrus capensis (Carroll 1976). The matrix from Mu¨ller (2004), which included 184 informative characters, was expanded by four new characters (see Appendices A–C for character definitions and additional scorings). Characters for Permian and ‘‘younginiform’’ taxa other
Can. J. Earth Sci. Vol. 46, 2009
than Acerosodontosaurus were scored based on information from the literature (Carroll 1976, 1981; Currie 1980, 1981, 1982; Gow 1975; Harris and Carroll 1977; Ketchum and Barrett 2004; Modesto and Reisz 2002; Reisz 1977, 1981), as well as personal observations. Regarding the additional scorings for the remaining taxa, see Mu¨ller (2004) for a complete list of references. Anatomical abbreviations a, angular; art, articular; atl-c, atlantal centrum; atl-na, atlantal neural arch; ax, axial intercentrum; cl, cleithrum; co, coronoid; d, dentary; dc, distal carpal (1, 2, 3, 4, 5 indicate first, second, third, forth, fifth); ec, ectopterygoid; f, frontal; fe, femur; fi, fibula; g, gastralia; hu, humerus; i, intermedium; il, ilium; is, ischium; j, jugal; l, lacrimal; lc, lateral centrale; m, maxilla; mc, medial centrale; mtc, metacarpal (1, 2, 3, 4, 5 indicate first, second, third, forth, fifth); p, pisiform; pf, postfrontal; ph, phalange; po, postorbital; pr, prearticular; prf, prefrontal; pu, pubis; q, quadrate; r, radius; ra, radiale; rf, rib fragment; sa, surangular; sc, scleral plate; sp, splenial; sq, squamosal; sv, sacral vertebra; t, tooth; ti, tibia; u, ulna; ul, ulnare.
Systematic paleontology Reptilia Laurenti, 1768 Neodiapsida Gauthier, 1984 Younginiformes Romer, 1945 Acerosodontosaurus piveteaui Currie, 1980 REVISED DIAGNOSIS: Aquatic neodiapsid reptile characterized by the absence of a quadratojugal and the presence of a short posterior process of the jugal, resulting in an open lower temporal arcade; at least 36 teeth in the maxilla and 34 teeth in the lower jaw. Differs from other ‘‘Younginiformes’’ on the basis of the following combination of features: neural spines of dorsal vertebrae intermediate in height between those of Thadeosaurus and Hovasaurus; transverse processes shorter than in Kenyasaurus; cleithrum present, in contrast to Galesphyrus, Thadeosaurus, and Youngina; postaxial edge of the shaft of the humerus straighter than in Thadeosaurus and preaxial edge of the shaft not as concave; radius of twisted appearance; ulna lacking an ossified olecranon; metacarpals longer than in Hovasaurus but not as long as in Thadeosaurus; intermedium more polygonal than in Galesphyrus; pubis with long tubercle, longer than in Thadeosaurus; femur not as slender as in Kenyasaurus. HOLOTYPE: MNHN 1908-32-57. LOCALITY AND HORIZON: Sakamena River Valley, Madagascar. Lower Sakamena Formation. Upper Permian (Currie 1980).
Results Anatomical description Skull The right side of the skull is preserved in internal view (Fig. 2). The orbit is large, and two rectangular scleral plates are lying close to the lower orbital rim. The maxilla is long and anteriorly tall when viewed on its Published by NRC Research Press
Bickelmann et al.
653
Fig. 1. Acerosodontosaurus piveteaui (Currie 1980). Specimen MNHN 1908-32-57 in (A) ventral and (B) dorsal views. Scale bar = 10 mm. Diagonal hatching indicates broken bone. See ‘‘Anatomical abbreviations’’ subsection of text.
Published by NRC Research Press
654
Can. J. Earth Sci. Vol. 46, 2009
Fig. 2. Right skull side of Acerosodontosaurus piveteaui (MNHN 1908-32-57; Currie 1980) in internal view. Scale bar = 10 mm. See ‘‘Anatomical abbreviations’’ subsection of text.
medial aspect, and it is unclear if the anteriormost portion of the bone contacting the premaxilla is not preserved or whether it was absent. The posterior half of the bone decreases significantly in height and forms a narrow tip reaching into the posterior half of the orbit, thereby contacting the jugal. The lacrimal forms part of the anteroventral orbital rim. It is triangular, with a rather short posterior process. Anterior to the posterior process, it has a buttress that bears the opening for the lacrimal duct. Based on the short anterior process of the lacrimal and the anteriorly tall maxilla, we infer that the lacrimal was excluded from the narial opening by a contact of maxilla and prefrontal, as is the case in Youngina (Broom 1914). The left lacrimal is also completely preserved lying below the left radius. The prefrontal forms the anterodorsal portion of the orbital rim, being roughly quadrangular in shape. The anteroventral edge of the bone appears curved, but we infer that the tip is broken off. In its central portion, the prefrontal is heavily depressed. The elongate frontal is incompletely preserved, with the posterior part and the anterior margins missing. The central part of the frontal, forming the dorsal margin of the orbit, is as long as the preserved anterior portion, but more slender and much thicker. Close to the contact with the prefrontal, the anterior part of the bone bears a rounded depression. Posteriorly, the frontal contacts the anterodorsal cusp of the small postfrontal. The postfrontal forms part of the orbital rim and is a small, triangular, and curved element, which is very similar in shape to the postfrontal of Hovasaurus from the same locality (Currie 1981). Both elements are preserved, but the
left postfrontal is dissociated from the skull and is lying next to the ulna. The central part of the element is fairly thick. Because of the poor preservation of the posterior skull roof, we cannot determine to what extent the postfrontal contributed to the upper temporal fenestra. The T-shaped postorbital has a thickened anterior portion with the dorsal and ventral ends being slightly elevated. By contrast, the elongate posterior process, which contacts the squamosal, is flattened and long. The element appears overall to be somewhat stouter than the postorbital of Hovasaurus (Currie 1981). The jugal is very long and slender. In contrast to Hovasaurus, the anterior process of the bone forms most of the ventral orbital rim, and the distally tapering dorsal process contacts the postorbital. The posterior process of the jugal is very short and ends in a sharp tip. There is no indication of a contact with a quadratojugal, the presence of which was suggested by Currie (1980). In fact, Currie (1980) identified a thin element located in between the jugal and the squamosal as quadratojugal, but this element is here reinterpreted as a fragment of a rib head, based on its triangular outline and its dorsal edge forming an articulation surface, which is very close to the condition observed in the cervical ribs of Hovasaurus (Currie 1981, figs. 9, 11). In conjunction with the morphology of the posterior process of the jugal, we assume that the quadratojugal was absent in Acerosodontosaurus, indicating that the lower temporal arcade was open. The tall and slender squamosal is not completely preserved. There is no anterior process of the squamosal functioning as a rim between the two temporal openings as in Petrolacosaurus (Reisz 1977). The posterior edge is highly elevated, forming a thin ridge that articulated with the convex part of the quadrate. Published by NRC Research Press
Bickelmann et al.
655
Fig. 3. Acerosodontosaurus piveteaui (MNHN 1908-32-57; Currie 1980). (A) Right forelimb in ventral view. (B) Rib in lateral view. Vertebrae in (C) posterior, (D) lateral, and (E) anterior views. Each scale bar = 10 mm. See ‘‘Anatomical abbreviations’’ subsection of text.
The quadrate is a massive bone. The anterior part is convex, having a large contact with the squamosal, and the posterior edge is straight. At the mid level of the posterior edge there is a small bump of unknown function. The only preserved bone of the palate is the ectopterygoid, which is a small and curved tubular element. Its anterior tip is broken off. Lower jaw The elements of the lower jaw are not as well preserved as those of the skull (Fig. 2). The tooth-bearing dentary is long and slender, and medially overlapped by the large, plate-like splenial. Posterior to the last tooth of the dentary, there is the small, low, and oblong coronoid that gradually increases in height towards its posterior end. The contact to the surangular is not preserved. The ventral margin of the lower jaw is formed by the angular, which unfortunately is too poorly preserved to permit a description. The surangular only preserves its posterior portion, forming a tubular element that possesses a disc-like posterior facet that contacts the articular. The small prearticular is only represented by a short anterior process, and a posterodorsal facet contacting the articular. The latter bone is a well-preserved, small, yet massive single element, which exposes a deeply emarginated facet receiving the quadrate condyle, as well as an elongated, somewhat tapering retroarticular process. Dentition Twenty-nine teeth are preserved in the upper jaw and 28 in the mandible (Figs. 1, 2). Together with the empty alveoles, the maximum count must have been 36 and 34, respectively. This count disagrees with Currie (1980) who counted 37 and 32, but this difference may be due to working with different casts and the poor preservation. The teeth are conical, apically tapering, and slightly recurved at their tips. More posteriorly, the teeth decrease slightly in size. There are no caniniform teeth. Due to poor preservation, the type of tooth implantation cannot be determined. Postcranium Eighteen vertebrae are preserved in articulation (Figs. 1, 3C–3E) and three more are scattered on the slab (Fig. 1). According to their position in the specimen, all of them can be considered presacral vertebrae. The presacral number of 21 is smaller than in Hovasaurus from the same locality, which possesses 25 presacrals (Currie 1981). However, this difference may be an artifact of the poor preservation. The centra are rather long, amphicoelous, and most likely notochordal, as seen in Figs. 3C and 3E. Laterally, there are small oval pits in the central part of each centrum, which can be identified as subcentral foramina (Fig. 3D). The articulating surfaces of the transverse processes, which are broad and oval in shape, are more or less at the same level as the zygapophyses. The neural spines are tall and straight. Published by NRC Research Press
656
Currie (1981) mentions that the neural spines of Hovasaurus are relatively taller than in Acerosodontosaurus and many other Permo-Triassic diapsids, but the overall shape of the neural spines is similar in these two taxa. There is no evidence of any accessory articulations in Acerosodontosaurus, which Currie (1981) described for at least some vertebrae of Hovasaurus. The same author also mentions accessory articulations for other ‘‘Younginiformes,’’ whereas in our opinion this interpretation requires further scrutiny. For example, the proposed evidence for an accessory articulation in Tangasaurus (Currie 1982, fig. 2) seems to consist only of a taphonomically deformed postzygapophysis, a special type of (misleading) preservation that can also be found in marine reptiles from other deposits, such as the Middle Triassic of Monte San Giorgio, Switzerland (J. Mu¨ller, personal observation, 2002). Currie (1980) also mentions the presence of neurocentral sutures in the posterior vertebrae of Acerosodontosaurus and figures such a suture in his reconstruction of the dorsal vertebrae (Currie 1980, fig. 5d). However, in his in situ figure he shows only the second dorsal vertebra anterior to the sacrum to possess such a suture (Currie 1980, fig. 1). In our opinion, it cannot be ruled out that this suture is merely the result of breakage. In fact, the same vertebra does not show any evidence of a suture on the counterslab (Fig. 1B), which was not pictured by Currie (1980). The centrum and the neural arch of the atlas, as well as the intercentrum of the axis, are partially preserved (Fig. 2). The atlantal centrum is a large and thick element, quadrangular in shape, and with unfinished bone surfaces. The atlantal neural arch bears a concave and somewhat elevated articulation facet anteriorly and two flat posterior processes. The ventral posterior process has a distinct articulating surface, possibly for articulation with the atlantal centrum. The axial intercentrum is tiny and pyramidal in shape, and lacks finished bone surfaces. Two sacral vertebrae are preserved (Fig. 1). However, only the first one is sufficiently preserved to permit description. The neural arch is long and broadened dorsally. The sacral rib is short and broad, with a triangular tip. Preservation does not reveal whether it is fused or sutured to the centrum. All ribs are in articulation, but most of them are only fragmentarily preserved. They are long, slender, gently curved, double-headed, and distally somewhat broadened (Fig. 3B). Towards the sacrum, they decrease in length (Fig. 1). Various gastralia are scattered in the abdominal region (Fig. 1). They are very slender and gracile. Pectoral and pelvic girdle There is one identifiable element of the pectoral girdle, the cleithrum (Fig. 1). It is rather thin and oblong, slightly curved and with tapering ends. Its surface is covered by small, elongate grooves. Overall it is very similar to the cleithrum of Hovasaurus (Currie 1981). All three elements of the pelvic girdle are identifiable (Fig. 1), which is in contrast to Currie (1980), who lists only the presence of an ilium and a pubis. The pubis is massively built, possessing a broad, oval pubic plate that bears a pronounced obturator foramen and a long pubic tubercle extending anteroventrally. The ilium is preserved in medial view, with the iliac blade being long and flattened. Whereas a preservational artifact cannot be excluded, the latter struc-
Can. J. Earth Sci. Vol. 46, 2009
ture appears dorsoventrally wider than the iliac blade of Hovasaurus. The acetabulum is well developed. The ischium is only partly preserved, showing a broad, highly flattened posterior blade. Limb bones Both humeri are only partly preserved. The right humerus is represented by the distal end and part of the shaft (Fig. 3A), whereas the left humerus preserves only a fragmentary distal end. From what can be discerned, the distal shape of the humerus is fairly similar to that of Hovasaurus (Currie 1981). The preaxial edge of the shaft is not as concave as the postaxial edge. The entepicondyle is well developed, as are the entepicondylar foramen and the ectepicondylar groove, the latter being an almost completely closed foramen. The condyles articulating with radius and ulna are only modestly pronounced. The surface of the humerus is covered by small elongate grooves. The radius, best represented by the right element, is curved and strongly twisted, and only slightly smaller than the ulna (Fig. 3A). The proximal part of the bone is thicker and more robust than the flattened distal part, even though both ends are not markedly widened and show unfinished ossifications. The right ulna is well exposed, showing that it is a long element with a constricted central portion (Fig. 3A). The proximal end is thick and robust, whereas the distal end is flattened and broad. Again, both ends show unfinished bone. The olecranon, which in tetrapods is usually present at the proximal end, is not ossified here, as is also the case in Hovasaurus (Currie 1981). The right manus is preserved in both ventral and dorsal views (Figs. 1, 2, 3A). There are eleven carpal elements: ulnare, radiale, intermedium, pisiform, lateral and medial centrale, as well as five distal carpals (Fig. 3A). This is the typical number seen in early diapsids and ‘‘Younginiformes.’’ The polygonal intermedium is the largest of all carpals, and its margins show unfinished bone. The ulnare is quadrangular, with a small concavity preaxially. The pisiform is also quadrangular, tapering preaxially, but smaller than the ulnare. All three elements are thin and slightly depressed in the center. The radiale is quadrangular in shape with rounded margins, not flattened dorsoventrally, but rather thick. The lateral centrale is a quadrangular element with a small concavity proximally. The medial centrale is the smallest of all carpal elements, polygonal in outline and tapering postaxially. Whereas the remaining bones are fairly similar in shape to those of Hovasaurus (Currie 1981), the medial centrale of Acerosodontosaurus is much smaller and shorter and, in contrast to the interpretation given by Currie (1981), does not articulate with the first three distal carpals (dc) but only with dc1 and dc2. All but the 4th distal carpal are rounded elements, all similar in size; dc4 is more quadrangular and much larger than the others. Its center is depressed. All five metacarpals (mtc) are well preserved (Fig. 3A); mtc1 is short and stout, with only slightly broadened proximal and distal ends and mtc2–mtc4 are oblong bones with broadened ends and a constricted shaft, increasing in length from the 2nd to the 4th, whereas mtc4 is only moderately longer than the other elements, which is different from many other early diapsids but similar to Hovasaurus. Also, mtc5 is only slightly longer than mtc1; it is as Published by NRC Research Press
Bickelmann et al.
657
Fig. 4. (A) The 50% majority rule consensus and (B) Adams consensus of 12 equally parsimonious trees obtained from the phylogenetic analysis using TNT 1.1. For further discussion see text.
stout as mtc1, but its shaft is more constricted. Ten phalanges are more or less completely preserved (Figs. 1, 2, 3A). The 4th finger preserves three phalanges in articulation (Fig. 1). All are small and stout elements with a disc-like articulating surface on the proximal end, as well as a rounded articulating surface on the distal end. The right femur is preserved in both ventral and dorsal views, but has suffered from crushing in the dorsoventral plane, whereas only the proximal head of the left femur is present (Fig. 1). The femur is a long bone with moderately expanded ends, its shape being very similar to the femur of Hovasaurus. Overall, the element is gently curved, especially the distal end. Distally, the right femur articulates with the right tibia and fibula (Fig. 1), of which only the proximal heads are preserved. The incompletely preserved tibia has an expanded and rounded proximal head, followed by a constricted shaft. Also the fibula is only fragmentarily preserved, but appears to be more gracile than the tibia. Phylogenetic analysis The phylogenetic analysis for testing the position of Acerosodontosaurus and the monophyly of ‘‘Younginiformes’’ was conducted using both parsimony (TNT 1.1, Goloboff et al. 2008) and Bayesian inference using the Mk model and a gamma shape parameter (MrBayes 3.2, Ronquist and Huelsenbeck 2003; 2 000 000 generations, every 100th generation sampled, burn-in first 1000 trees). In the parsimony run, the 12 most parsimonious trees were obtained after the exclusion of drepanosaurids, which acted like a wildcard taxon due to the inclusion of the additional taxa (TL: 753) (Fig. 4). Except for kuehneosaurids, which now group with Lepidosauromorpha, the overall topology of Sauria (Gauthier 1984) and closely related taxa is similar to that obtained by Mu¨ller (2004). However, there was no support for a monophyletic ‘‘Younginiformes.’’ Only the Adams consensus groups most of the ‘‘younginiform’’ taxa in a single clade, whereas Kenyasaurus and Lanthanolania remain outside and unresolved (Fig. 4B). However, Acerosodontosaurus and Hovasaurus always form a clade (Fig. 4), their monophyly being unequivocally supported by the absence of a quadratojugal (#19-1), the presence of a cleithrum (#51-0), and the relatively short metacarpal IV (#148-1), whereas only one extra step is required to break this monophyly. The Bayesian analysis was unable to recover any well-supported groupings among the Permian taxa, with all posterior probabilities remaining far below 0.5, which basically corresponds to an unresolved relationship (results not shown).
Discussion Our phylogenetic analysis does not support the hypothesis previously proposed by Currie (1981, 1982). ‘‘Younginiformes’’ sensu Currie (1982) do not form a monophyletic group, but rather consist of a paraphyletic assemblage of Permo-Triassic nonsaurian neodiapsids. While we do not want to exclude the possibility that the relationships of these Published by NRC Research Press
658
taxa will be better resolved once new anatomical information becomes available, we nonetheless assume that the previously suggested monophyly of ‘‘Younginiformes’’ is mainly due to the plesiomorphic character of most of their anatomy in relation to other neodiapsids. Further fossil finds and anatomical reinvestigations will be required to resolve this issue in a more satisfying manner. The restudy of Acerosodontosaurus resulted in the reinterpretation of the element designated as quadratojugal by Currie (1980), which was here determined to be a rib fragment. This reidentification as well as the short extension and the appearance of the posterior process of the jugal imply that the lower temporal bar was incomplete in Acerosodontosaurus. In addition, the comparison of the morphology of the jugal in Hovasaurus with the one in Acerosodontosaurus, along with the fact that for Hovasaurus Currie (1980) only tentatively assigned a disarticulated element to be a quadratojugal, lead us to conclude that the lower temporal bar was also incomplete in Hovasaurus. These new findings indicate remarkable anatomical differences among ‘‘younginiform’’ reptiles, further contradicting a monophyletic origin. The loss of the lower temporal bar took place in at least two early neodiapsid taxa, Acerosodontosaurus and Hovasaurus, whereas the lower temporal arcade is closed in Youngina (Broom 1914; Gow 1975). Unfortunately, all other ‘‘Younginiformes’’ do not preserve elements of the lower temporal arcade or even lack the whole skull. Mu¨ller (2003, 2004) suggested that the loss of the lower temporal bar occurred on the branch leading to Claudiosaurus and other derived neodiapsids. Because the unresolved relationships of ‘‘Younginiformes’’ hamper process-related evolutionary interpretations for early Neodiapsida, it is currently impossible to determine whether Acerosodontosaurus and Hovasaurus are most closely related to Claudiosaurus and other neodiapsids, which would make the loss of the bar a unique event, or whether these two taxa lost the lower temporal arcade independently. Currie (1980) tentatively suggested an aquatic lifestyle for Acerosodontosaurus, as the fossil was found in a locality containing other potentially aquatic reptiles, such as Hovasaurus and Barasaurus (Ketchum and Barrett 2004; Smith 2000). At the same time, the author mentions the rarity of Acerosodontosaurus in comparison with Hovasaurus. However, the reasons for this taphonomic bias remain unclear at this point (for example, the two taxa could have had different feeding ecologies preventing them from sharing the same habitat, but any interpretation is currently rendered speculative). Currie (1980) also noted that Acerosodontosaurus has a twisted radius very similar to that of the highly aquatic champsosaurs. We present two additional postcranial features that further support an aquatic lifestyle, namely the lack of an ossified olecranon and the relaxed arrangement of the carpal elements (the term ‘‘relaxed’’ is here defined as a lack of a tight contact between the carpal or tarsal elements, leaving wellvisible gaps between the different bones). In highly aquatic taxa, an ossified olecranon is often reduced or even absent, such as in nothosaurs, ichthyosaurs, and thalattosaurs (McGowan and Motani 2003; Mu¨ller 2005; Rieppel 2000; Romer 1956). This lack of ossification can
Can. J. Earth Sci. Vol. 46, 2009
be referred to the phenomenon of skeletal paedomorphosis, which is frequently seen in secondarily aquatic amniotes (Rieppel 1989). In contrast, an olecranon is well developed in the ulna of terrestrial taxa like Thadeosaurus (Carroll 1981). The arrangement of the elements of the manus is relaxed, which is not due to taphonomy but probably reflects their condition at time of death of the animal. In Hovasaurus, the same condition is present, as opposed to the very tight mosaic in the pes of the terrestrial Kenyasaurus (Harris and Carroll 1977). Also here, skeletal paedomorphosis may have been the reason, as a relaxed arrangement in both manus and pes can be frequently found in a variety of secondarily aquatic amniotes (Rieppel 1989). We, therefore, conclude that an aquatic lifestyle is highly likely for Acerosodontosaurus, and at the same time can also be assumed for Hovasaurus. In addition, an aquatic lifestyle was also proposed for Tangasaurus based in part on the presence of a long, powerful tail adapted for swimming (Currie 1982; Haughton 1924). By contrast, the remaining ‘‘younginiform’’ reptiles, such as Kenyasaurus, Thadeosaurus, and Youngina probably possessed terrestrial life habits. In fact, these remarkable ecological differences provide further circumstantial evidence for the hypothesis of a paraphyletic origin of ‘‘Younginiformes.’’
Acknowledgments The authors thank Daniel Goujet for providing new latex casts from specimen MNHN 1908-32-57. Many helpful comments by L.A. Tsuji and F. Witzmann, as well as by the reviewers S. Modesto and N. Fraser, and by the Associate Editor H.-D. Sues, greatly improved the manuscript. This study was financially supported by the Deutsche Forschungsgemeinschaft (Mu 1760/2-3).
References Benton, M.J. 1985. Classification and phylogeny of the diapsid reptiles. Zoological Journal of the Linnean Society, 84(2): 97–164. doi:10.1111/j.1096-3642.1985.tb01796.x. Broom, R. 1914. A new thecodont reptile. Proceedings of the Zoological Society of London, pp. 1072–1077. Carroll, R.L. 1976. Galesphyrus capensis, a younginid eosuchian from the Cistecephalus zone of South Africa. Annals of the South African Museum, 72(4): 59–68. Carroll, R.L. 1981. Plesiosaur ancestors from the Upper Permian of Madagascar. Philosophical Transactions of the Royal Society of London, 293(1066): 315–383. doi:10.1098/rstb.1981.0079. Currie, P. 1980. A new younginid (Reptilia: Eosuchia) from the Upper Permian of Madagascar. Canadian Journal of Earth Sciences, 17(4): 500–511. doi:10.1139/e80-046. Currie, P. 1981. Hovasaurus boulei, an aquatic eosuchian from the Upper Permian of Madagascar. Palaeontologia Africana, 24: 99– 168. Currie, P. 1982. The osteology and relationships of Tangasaurus mennelli Haughton (Reptilia, Eosuchia). Annals of the South African Museum, 86(8): 247–265. Evans, S.E. 1988. The early history and relationships of the Diapsida. In The phylogeny and classification of the tetrapods, Vol. 1: Amphibians, reptiles, birds. Systematics Association, Special Vol. 35A, Edited by M.J. Benton. Clarendon Press, Oxford, UK., pp. 221–260. Gaffney, E.S. 1980. Phylogenetic relationships of the major groups Published by NRC Research Press
Bickelmann et al. of amniotes. In The terrestrial environment and the origin of land vertebrates. Edited by A.L. Panchen. Academic Press, London, UK., pp. 593–610. Gauthier, J. 1984. A cladistic analysis of the higher systematic categories of the Diapsida. Ph.D. thesis, University of California, Berkeley, Calif. Goloboff, P.A., Farris, J.S., and Nixon, K.C. 2008. TNT, a free program for phylogenetic analysis. Cladistics, 24(5): 774–786. doi:10.1111/j.1096-0031.2008.00217.x. Gow, C.E. 1975. The morphology and relationships of Youngina capensis Broom and Prolacerta broomi Parrington. Palaeontologia Africana, 18: 89–131. Harris, J.M., and Carroll, R.L. 1977. Kenyasaurus, a new eosuchian reptile from the Early Triassic of Kenya. Journal of Paleontology, 51(1): 139–149. Haughton, S.H. 1924. On reptilian remains from the Karroo beds of East Africa. Quarterly Journal of the Geological Society of London, 80: 1–11. Ketchum, H.F., and Barrett, P.M. 2004. New reptile material from the Lower Triassic of Madagascar: implications for the Permian–Triassic extinction event. Canadian Journal of Earth Sciences, 41(1): 1–8. doi:10.1139/e03-084. Laurenti, J.N. 1768. Classis Reptilium. Specimen medicum, exhibens synopsin Reptilium emendatam, cum experimentis circa venena et antidote Reptilium Austriacorum. Joan. Thom. Nob. de Trattern, Vienna. Laurin, M. 1991. The osteology of a Lower Permian eosuchian from Texas and a review of diapsid phylogeny. Zoological Journal of the Linnean Society, 101(1): 59–95. doi:10.1111/j.10963642.1991.tb00886.x. McGowan, C., and Motani, R. 2003. Handbook of paleoherpetology. Vol. 8. Ichthyopterygia. Verlag Dr. Friedrich Pfeil, Munich, Germany. Modesto, S.P., and Reisz, R.R. 2002. An enigmatic new diapsid reptile from the Upper Permian of Eastern Europe. Journal of Vertebrate Paleontology, 22(4): 851–855. doi:10.1671/02724634(2002)022[0851:AENDRF]2.0.CO;2. Mu¨ller, J. 2003. Early loss and multiple return of the lower temporal arcade in diapsid reptiles. Naturwissenschaften, 90: 473– 476. doi:10.1007/s00114-003-0461-0. Mu¨ller, J. 2004. The relationships among diapsid reptiles and the influence of taxon selection. In Recent advances in the origin and early radiation of vertebrates. Edited by G. Arratia, V.H.
659 Wilson, and R. Cloutier. Verlag Dr. Friedrich Pfeil, Munich, Germany, pp. 379–408. Mu¨ller, J. 2005. The anatomy of Askeptosaurus italicus from the Middle Triassic of Monte San Giorgio, and the interrelationships of thalattosaurs (Reptilia, Diapsida). Canadian Journal of Earth Sciences, 42(7): 1347–1367. doi:10.1139/e05-030. Piveteau, J. 1926. Pale´ontologie de Madagascar, XIII — Amphibiens et reptiles Permiens. Annales de Pale´ontologie, 15: 55– 178. Piveteau, J. 1955. Existence d’un reptile du groupe Procolophonides a` Madagascar. Conse´quences stratigraphiques et paleontologiques.. Comptes rendus hebdominaires des se´ances de l’Acade´mie des Sciences, 241: 1325–1327. Reisz, R.R. 1977. Petrolacosaurus, the oldest known diapsid reptile. Science, 196(4294): 1091–1093. doi:10.1126/science.196. 4294.1091. Reisz, R.R. 1981. A diapsid reptile from the Pennsylvanian of Kansas. Ph.D. thesis. University of Kansas, Museum of Natural History, Special publication No.7. Reisz, R.R., Berman, D.S., and Scott, D. 1984. The anatomy and relationships of the Lower Permian reptile Araeoscelis. Journal of Vertebrate Paleontology, 4: 57–67. Rieppel, O. 1989. Helveticosaurus zollingeri Peyer (Reptilia, Diapsida) skeletal paedomorphosis, functional anatomy, and systematic affinities. Paleontographica, 208: 123–152. Rieppel, O. 2000. Encyclopedia of paleoherpetology. Vol. 12A, Sauropterygia I. Verlag Dr. Friedrich Pfeil, Munich, Germany. Romer, A.S. 1945. Vertebrate paleontology. 2nd ed. The University of Chicago Press, Chicago, Ill. Romer, A.S. 1956. Osteology of the reptiles. The University of Chicago Press, Chicago, Ill. Ronquist, F., and Huelsenbeck, J.P. 2003. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics, 19(12): 1572–1574. doi:10.1093/bioinformatics/btg180. Smith, R.M.H. 2000. Sedimentology and taphonomy of Late Permian vertebrate fossil localities in southwestern Madagascar. Palaeontologia Africana, 36: 25–41. Wescott, W.A., and Diggens, J.N. 1998. Depositional history and stratigraphical evolution of the Sakamena Group (Middle Karroo Supergroup) in the southern Morondava Basin, Madagascar. Journal of African Earth Sciences, 27(3–4): 461–479. doi:10. 1016/S0899-5362(98)00073-6.
Appendix A. Description of added characters used in the phylogenetic analyses Postorbital: posterior process contacting squamosal long (0) or short (1). Presacral vertebrae: 25 or more (0) or less than 25 (1). Neck: short with 5 or less vertebrae (0) or long with more than 5 vertebrae (1). Ulnare: wider than long (0), longer than wide (1), or as long as wide (2). Appendices B and C appear on the following pages.
Published by NRC Research Press
660
Can. J. Earth Sci. Vol. 46, 2009
Appendix B. Data matrix of Mu¨ller (2004) with additional taxa 1
2
3
4
Acerosodontosaurus Tangasaurus Hovasaurus Kenyasaurus Youngina Galesphyrus Thadeosaurus Lanthanolania
1234567890 ??????013? ?????????? ??????0?3? ?????????? 0000010131 ?????????? ?????????? 000?01013?
1234567890 ?????0201? ?????????? ?0??0201? ?????????? 0000001000 ?????????? ?????????? ?????02???
1234567890 ?????00?1? ?????????? 00210?011?? ?????????? 0010010?0? ?????????? ?????????? ??????????
1234567890 ???0?1??10 ?????????? ?????????0 ?????????? ?0001???10 ?????????0 ?0???????? ?0?0??????
Acerosodontosaurus Tangasaurus Hovasaurus Kenyasaurus Youngina Galesphyrus Thadeosaurus Lanthanolania
5 1234567890 00?0?00?0? ?????0??0? 00?0?00?00 ?0?0?0??0? 00?000??0? ???0?0??00 00?0?0?001 ??????????
6 1234567890 0????????0 ?????????0 010?1??000 ?????????0 ????10???0 1????????0 ????10?000 ??????????
7 1234567890 ??00020??? 00000????? 0002020??1 ??0??????1 ??000?10?1 ?????2?0?1 00020200?1 ??????????
8 1234567890 ?????????? ?????01??? 01?000?00? ???0?01000 ??????0000 ???0?00000 011???00?0 ??????????
Acerosodontosaurus Tangasaurus Hovasaurus Kenyasaurus Youngina Galesphyrus Thadeosaurus Lanthanolania
9 1234567890 ?0?????0?1 1????????? 10???????1 1????????? 100000?001 ?????????? 1????????? ????????1?
10 1234567890 0????????? ?????????? 0???????00 ?????????? 10010?1?00 ?????????? ??????1??0 0?????????
11 1234567890 ??01?0??0? ???1?0?2?? ?1?102?200 ??010?110? ??0110???? ?10?1????? ??0110??0? ??????????
12 1234567890 ???????0?? ?????????? ??0?0???00 ??0??????? 100?00000? ????????0? ?00?????00 ??????????
Acerosodontosaurus Tangasaurus Hovasaurus Kenyasaurus Youngina Galesphyrus Thadeosaurus Lanthanolania
13 1234567890 ?0????100? ??0??????? ??0??????? ?????????? 1000?00000 ??0??????? ??0??????? ??????00??
14 1234567890 ?0???????? ?????????? 01010????? ?????????? 1?1?????10 ?????????? ?????????? ??????????
15 1234567890 ?????111?? 0???01?0?? ????011110 ????0???00 ????01???? ????0??010 ????011010 ??????????
16 1234567890 ????0????? ?????????? 1??10????? 1????????? ?00?01???? 1????????? 1????????? ??0?0?????
Acerosodontosaurus Tangasaurus Hovasaurus Kenyasaurus Youngina Galesphyrus Thadeosaurus Lanthanolania
17 1234567890 ???????1?? ?????????? ?????????0 ????????11 ??1???0?1? ????????11 ??1??????0 ??2???????
18 1234567890 ???0???0?? ?????????? 0?100?00?? 011??????? ??1?0??0?0 ??100????? 0?100????? ????????0?
1234 0?1? ???? ??1? ???? 00?1 ???? ???? 0?1?
Note: Unknown characters were scored with ?. For character list see Mu¨ller (2004).
Published by NRC Research Press
Bickelmann et al.
661
Appendix C. Characters added to data matrix of Mu¨ller (2004) 18 Seymouriidae Synapsida Parareptilia Captorhinidae Araeoscelidia Rhynchocephalia Squamata Prolacerta Tanystropheus Trilophosaurus Rhynchosauria Archosauriformes Choristodera Pachypleurosaurs Simosaurus Placodus Pistosauridae Claudiosaurus Askeptosaurus Clarazia Thalattosaurus Testudines Palaeagama Saurosternon Coelurosauravus Macrocnemus Helveticosaurus Ichthyopterygia Kuehneosauridae Hupehsuchus Acerosodontosaurus Tangasaurus Hovasaurus Kenyasaurus Youngina Galesphyrus Thadeosaurus Lanthanolania
5678 1A00 A011 AAAA 1000 1011 0011 A01A 1011 1012 ?011 1011 0A1B 0010 1012 0012 1012 0010 0110 0010 00?? 011? A110 1112 ???2 0112 001? ???? 1012 001? 1012 00?0 ?00? ?000 ?1?? 0100 ?0?1 ?0?0 1???
Note: Polymorphism is indicated by letters, such that A = 0&1 and B = 0&1&2. Unknown characters were coded with ?. For character list see Appendix A.
Published by NRC Research Press
