127 Palaeont. afr., 25, 127- 149 (1984) A NEW PROTOSUCHIAN CROCODILE FROM THE UPPER TRIASSIC ELLIOT FORMATION OF SOUTH AFRICA by Arthur B Busbey III 1 and Chris Gow2 1 Department of Geological Sciences, University of Texas at Austz"n, Austin, TX 78713-7909, U.S.A. 2 Bernard Price Institute for Palaeontological Research, University of the Witwatersrand, Johannesburg ABSTRACT A new protosuchian crocodilian, Baroqueosuchus haughtoni from the Upper Triassic or Lower Jurassic Elliot Formation of the Orange Free State is the most primitive proto­ suchian crocodilian known. There are no contacts between the quadrate and opisthotic or below the crania-quadrate canal, the internal carotid arteries were not enclosed in separate foramina and the basicranium was flat, with the basisphenoid being broadly exposed on the base of the skull. The basic diagnosis of the Order Crocodilia is discussed and a new diagnosis is offered based upon cranial anatomy. CONTENTS Page INTRODUCTION. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 SYSTEMATICS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 The Order Crocodilia ...................... : . . . . . . . . . . . . . . . . . . . . . . . . 128 The Family Protosuchidae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132 DESCRIPTION OF NEW PROTOSUCHIAN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132 COMPARISON WITH OTHER PROTOSUCHIANS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143 PROTOSUCHIAN RELATIONSHIPS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147 ACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148 REFERENCES ............ . INTRODUCTION The predominantly Triassic (see below) Elliot Formation of South Africa has produced several genera of protosuchian crocodilians and possibly related thecodonts including a new and primitive protosuchian crocodilian. The most complete and best known protosuchian is Orthosuchus stormber­ gi (Nash 1968, 1975), while Notochampsa istedana (Broom 1904, Haughton 1924), Pedeticosaurus le­ viseuri (van Hoepen 1915, Bonaparte 1972) and Erythrochampsa (Notochampsa) longipes (Broom 1904, Haughton 1924) are poorly preserved or poorly prepared. Sphenosuchus acutus (Haughton 1924, Walker 1972) has a number of synapomor­ phies with some thecodonts (Hesperosuchus and Pseudhesperosuchus, Bonaparte 1971) and may have some crocodilian synapomorphies (see be­ low) though its affinities with the crododilia not yet clear; this problem is beyond the scope of this paper and will be treated in a paper on proto­ suchian systematics and anatomy (Busbey in prep). MS accepted March 1983 148 The new protosuchian described herein reveals de­ tails of the otic region and suspensorium that were previously unknown (and to some extent unrecog­ nised) and permits an evaluation, based upon cra­ nial characters, of some of the controversy surro­ unding protosuchian systematics. The Elliot Formation (formerly the Red Beds of the Stormberg Series) has classically been thought of as being entirely Triassic. Recently Olsen et al (1982), as part of a paper correlating the Newark Supergroup of the Newark Basin, suggested that the entire upper Elliot Formation and the Clarens Formation (formerly the Cave Sandstone) are entirely Jurassic in age. (See also Olsen and Galton, this volume). The correlation is based on radiome­ tric dates of the Drakensberg volcanics, that overlie the Clarens Formation in the east but interfinger with the Clarens Foramtion in the West, and on reptile skeletal and footprint assemblages. A paper referenced in support of this reassignment is Clemens et al. (1979), in which neither reptile skeletal or footprint assemblages are discussed. In 128 fact, the work of Olsen and Galton (1977) is re­ ferred to in Clemens et al. (1979), only to the ~xtent of acknowledging the possibility that the upper Elliot and Clarens Formations might be lower Jurassic. In their paper they provide no evidence to support Olsen and Galton ( 1977) or Olsen et al. (1982). Olsen and Galton (1977) note that radiometric age dates of the Drakensberg volcanics from Fitch and Miller (1971) are said to be 181 and 165 My BP. However, Fitch and Miller (1971) actually re­ port a date of 18 7± 7My. This provides a minimum estimate of the initiation of Drakensberg vulcanism for more southern localities where the samples were collected. Based on their reconstructed rates of deposition they suggest that the Elliot Forma­ tion is 187-193±7My. old. The disparity between the ages reported in the actual article and those mentioned in Olsen and Galton ( 197 7) are diffi­ cult to understand; it is sufficient to note that the estimated dates for the deposition of Elliot sedi­ ments are much earlier based on the original data in Fitch and Miller (1971) than from the re-esti­ mates by Olsen and Galton (1977). Evidence that Olsen and Galton (1977) considered to be decisive for an early Jurassic date of the upper Stormberg is the presence of the prosauropod Anchisaurus from Olsen's Zone 3 of the Newark Basin and from the Clarens Formation of South Africa. Anchisaurus is not found in the Elliot Formation, and thus has no bearing on the age of the deposits from which most of the protosuchians are known. Other problems in correlation between the Newark Basin and South Africa are a lack of latest Triassic semionotid fish and footprint assemblages for comparison. We are merely told in Olsen and Galton (1977) that the fish and footprint assemblages compare favourably with earliest Jurassic assemblages . Until a stronger case is made we acknowledge the possibility of an earliest Jurassic date of the upper Elliot Formation, but will continue to consider it as latest Triassic. The new protosuchian is known from two par­ tial braincases. BP-1-4746 was found by Dr. J.W. Kitching in 1978, while conducting preliminary field collections from the Clarens and Elliot for­ mations on the farm Sunnyside (11 78) near Cla­ rens in the Orange Free State, along the northern border of Lesotho. BP -1-4 7 46 is represented by the posterior half of a skull missing the distal ends of the quadrates, the 'cheek' bones, the lateral edges (to varying degrees) of the squamosals, and the left postorbital. It was heavily encrusted in a haematite­ cemented sandstone that was mechanically remo­ ved. The second specimen, BP-1-4946, was found in 1980 by C.E. Gow on the farm Damplaats (55), in the Orange Free State, in the Upper Elliot For­ mation. It was found with a collection of rolled postcranial bones mixed with a fabrosaurid skele­ ton. This same locality also yielded the advanced cynodont Tritylodon and the ubiquitous prosau­ ropod Massospondylus. BP-1-4946 constitutes the posterior half of the skull with quadrates in posi­ tion. It was encased in soft calcareous red mud- stone with very little haematite adhering to the bone. The nearly complete left and partial right squamosals were heavily encrusted with haema­ tite and had become dissociated prior to burial and fossilized separately. Distortion has pushed the left side of the braincase inwards; concomi­ tant plastic deformation has particularly affec­ ted quadrate morphology. Mechanical prepara­ tion of the braincase was taken as far as possi­ ble. The right quadrate was removed and com­ pletely cleaned with N/10 formic acid. The squa­ mosals were partially cleaned of haematite through prolonged treatment with thioglycollic acid (Rixon 1976). Detailed preparation of the braincase was achieved through the careful use of N/1 0 formic acid dropped onto small areas at a time. Spent acid was withdrawn using tissue paper and replaced with fresh acid until the desired degree of penetration was obtained. BP Institutional abbreviations used are:- Bernard Price Institute for Palaeontolo­ gical Research, University of the Witwa­ tersrand. MCZ Museum of Comparative Zoology, Har­ vard University. SYSTEMATICS The Order Crocodilia Unfortunately there is no clear consensus about what constitutes a crocodilian. We thus offer a dis­ cussion of some current problems and a new diag­ nosis of the crocodilia based on cranial characters primitive for the group. There is some confusion as to the cranial cha­ racters that are unique to the crocodilia (Walker 1970, 1972, 1974, Whetstone and Martin 1979, Tarsitano and Hecht 1980, Crompton and Smith 1980). The problem is also embroiled in debate about the phylogeny of birds and the phylogenetic relationships of primitive archosaurs. In so far as the protosuchians are concerned the confusion is due in part to: 1. Lack of data about definitive cranial cha­ racters of protosuchian crocodilians. 2. Misinterpretation of the morphology of protosuchian crania. 3. The tacit use of modem eusuchian cha­ racter states as being typical of all fossil crocodilians. 4. A lack of outgroup comparisons with non­ crocodilian archosaurs. Langston (1973) presented an extensive sum­ mary list of characters that are presumably primi­ tive for crocodilians. Because of the incomplete na­ ture of most protosuchian material it is not possi­ ble to use all of the characters on any one genus. The cranial characters that he listed that have the potential to be seen in most protosuchians are: 1. The flat ornamented skull deck. 2. Strongly inclined quadrates, bordered an­ teriorly by a long, slender quadratojugal. 3. Akinetic palate (i.e., basipterygoid articu­ lation closed) pterygoids with wide and deep wings (transverse processes). 4. Many cranial bones, including articular, more or less pneumatic. 5. Eustachian passages more or less enclosed in bone. 6. Posttemporal fenestra reduced. 7. Squamosal, quadrate, and paraoccipital process combine to form an otic meatus. Characters 1,3,4, and 5 are taxonomically signi­ ficant but require further clarification, characters 2 and 7 do not appear in early protosuchians, and character 6 is widely distributed amongst various archosaur taxa and may be of little utility. Walker (1970, 1972, 1974) has published clas­ sifications and lists of characters (with subsequent modifications) that we feel significantly broaden and at the same time distort the concept of the crocodilia as a natural group; this was done to in­ clude Sphenosuchus acutus within the crocodilia. Walker's classification thus contains a confusing mixture of characters, some of which are primitive for archosaurs, some which seem to be autapomor­ phic for non-crocodilian lineages, some which may be based on faulty interpretation of the morphology of Sphenosuchus acutus (see discussion below), and some which may indeed be synapomorphic for protosuchians and sphenosuchids. His taxonomic categories are based upon very poorly known pro­ tosuchian genera (Stegomosuchus, Pedeticosaurus, Hallopus) and include diagnoses that contain cha­ racters not known for these genera. The holotype of Stegomosuchus, from the Connecticut Valley, is only minimally prepared (see Lull 1953 and sketch in Olsen 1980) and demonstrates few diagnostic cranial characters. Pedeticosaurus is known only from the holotype, which is an impression in fine sandstone, with a partial steinkern of the right otic meatus, and has only two characters that may sug­ gest it is a crocodilian (shape of the otic region as revealed by the steinkern - Bonaparte 1972; and possibly elongated carpals, but see Nash 1975). Hallopus seems to have elongated carpals, but there is insufficient diagnostic material for specific assign­ ment and no cranial material and for the time being the name should apply only to the type. Walker's broader category, the order Crocodylomorpha, in­ cludes forms that do not possess diagnostic croco­ dilian features. We shall discuss some of these cha­ racters below. Recently Tarsitano and Hecht (1980) have re­ viewed the relationships of Archaeopteryx, and are highly critical of the recent studies of Walker (1970, 1972, 1974) and of Ostrom (1976). They also address the problem of crocodilian relation­ ships and are highly critical of most of the charac­ ters listed by Walker. They dismiss most of the characters Walker used. Though we agree in prin- 129 ciple and to some degree in part with their analy­ sis in regard to the relationships of crocodilians (that Archaeopteryx is not closely related to cro­ codilians and that the affinities of Sphenosuchus are not clear), many of the criticisms are made as unsubstantiated or incorrect assertions. We shall only address some of those points concerning pro­ tosuchian cranial characters in this paper. Tarsitano and Hecht (1980) maintain that the following characters, listed by Walker, are actually primitive for archosaurs; 1. Possession of laterosphenoids. 2. An external mandibular fenestra. 3. Antorbital fenestrae. 4. Crescentic occipital surface. 5. Short paroccipital process projecting be­ hind the quadrate forming the posterior wall of the tympanic cavity. 6. Carotid circulation based upon paired grooves and pneumatic basisphenoid. 7. Elongate cochlear duct. We shall discuss characters 6 and 7 below. La­ terosphenoids are clearly not primitive for theco­ donts. Chasmatosaurus and Euparkeria do not po­ sess laterosphenoids (Cruickshank 1970) and their braincases are not derived over those of more pri­ mitive non-thecodont diapsids. Characters 2 and 3 appear to be primitive for archosaurs, as stated, but character 4 is sufficiently vague to be of little use, since it is the composition of the occipital crescent and not the overall shape that is important. Charac­ ter 5 may be primitive for diapsids and requires bo­ th further elaboration and study. Tarsitano and Hecht ( 1980) provide their own list of primitive cranial characters for the Crocodi­ lia and evaluate the characters deemed important by Walker: 1. crocodilian eustachian (pharyngotympa- nic) tubes; 2. forward sloping quadrate; 3. crocodilian otic notch; 4. sculptured dermal bones; They do not provide more detailed definitions of these four character states. For example, there­ ference on the crocodilian pharyngotympanic sys­ tem (Colbert 1946) is merely a description of the morphology in a few modem crocodilians. They have made no attempt to extract useful characters from their source. The 'forward sloping quadrate' is of little use, since the quadrate slopes forward in many thecodonts (for example, see Proterosuchus ( Chasmatosaurus ), original photographs or sketches, not the reconstruction of Cruickshank 1972, and Proterochampsa, Sill196 7), and in primitive proto­ suchians it is actually fairly vertical (see Hemipro­ tosuchus, Bonaparte 19 71, and the new genus be­ low). No explicit definition is given of the charac­ ters involved in the 'crocodilian otic notch' (based on comments later in their paper they implicitly 130 mean modem morphology and thus concur with the character state as given by Langston); yet the protosuchian otic notch lacks many of the specia­ lizations of more derived crocodilians (Busbey, in prep). In short, without further elaboration this list of undefined characters has little utility. Tarsitano and Hecht (1980) state that the qua­ drate of any true crocodilian must possess all of the contacts of the quadrate of a modem crocodi­ lian; this includes opisthotic contacts above and be­ low the cranioquadrate canal. No currently accep­ ted protosuchian genus possesses all of the contacts present in 'modem' crocodilians and even metrior­ hynchid and teleosaurid 'mesosuchian' crocodilians do not possess a contact between quadrate, squa­ mosal, and opisthotic behind the tympanum (Eu­ des-Deslongchamps 1863, Antunes 1967, de Gas­ parini and Dellape 1976, Wenz 1968, Nash 1975). And furthermore, the quadrate and body of the opisthotic fail to meet below the cranioquadrate canal in several primitive protosuchian genera (see below). Thus, Walker (1972) was correct for primi­ tive protosuchians (and apparently sphenosuchid thecondonts) - the quadrate did not contact the body of the opisthotic or the paroccipital process behind the tympanum; its contacts were exclusive­ ly with the squamosal and side of the braincase. As Tarsitano and Hecht (1980) point out, the supposedly kinetic skull and possible possession of salt glands in Sphenosuchus are based on Walker's interpretations and are hypothetical even for Sphe­ nosuchus. Thus these 'characters' cannot be used with any reliability in crocodilian systematics. Tarsitano and Hecht ( 1980) note that the path of the internal carotid is also hypothetical in Sphe­ nosuchus and therefore has little systematic value (which is true), but in making this assertion they make a sweeping statement about the 'similarity' of the carotid circulation in dinosaurs, thecondonts, and other amniotes without describing the nature of the similarity. They are possibly correct about the pneumatic basisphenoid. This is seen in various camosaurs (Osborn 1912, Russell 1970), coeluro­ saurs (Osmolska 1976, Osmolska et al. 1972, Col­ bert and Russell 1969, Greg Paul, pers comm), and possibly other archosaurs. It does not, however, seem to be a primitive feature (the basisphenoid of primitive thecodonts such as Proterosuchus and Euparkeria is still primitively shallow and is not pneumatized- and may possibly be related to the depth of the basicranium. Tarsitano & Hecht (1980) are correct in noting that many archosaurs (we include birds) have an elongated cochlear recess (see also Langston 1960), Whetstone and Martin 1979, Hopson 1979) and that it might not necessarily indicate a close rela­ tionship between birds and crocodilians. However, the distribution of this character within the Archo­ sauria is currently poorly known. They further maintain that an enlarged cochlear recess occurs in many groups (see below) and is therefore plesio­ morphic for amniotes. The presence of an 'elonga­ ted' cochlea in sea turtles and primitive mammals does not invalidate its use for comparisons within archosaur taxa; there is simply no evidence (see the survey by Hopson 1979) that primitive synapsids ( theropsids) or early sauropsids shared an elongate cochlear recess and that this represents an amniote synapomorphy. In many primitive reptiles the floor of the inner ear is unossified and in taxa where it is ossified, for example, in primitive therapsids (Hop­ son, pers comm.) or Captorhinus (Heaton 1979), the cochlear recess is short. In any case, their re­ reference for cochlear size in the sea turtles (based only on Chelonia midas) is Wever (1979), who pro­ vides data on the length and width of the basilar membrane, auditory papilla, and Organ of Corti, not on the size of cochlear recess itself. While other turtles have basilar membrane lengths of 500 to 1100 microns, Chelonia midas has a basilar mem­ brane length of about 1700 microns. This does not compare with a basilar membrane length of over 3 700 microns in Caiman crocodilus and 4900 mi­ crons for Alligator and Crocodylus which, based on the size of the cochlear recess, may be shorter than that in more derived archosaurs. In fact J in their own paper they are not clear in regard to the signi­ ficance of the size of the cochlear recess; on page 173, paragraph 4, they state: 'It should be noted that an elongate cochlea is also present in the Pro­ totheria .... dinosaurs} and sea turtles .... and is therefore a plesiomorphy', yet on the same page in paragraph 9: 'The distribution of this character sta­ te (the elongate cochlea) among outgroup taxa (see above) indicates that it cannot be used as a synapo­ morphy showing the common ancestory of birds and crocodiles, but rather that it represents an ad­ vanced amniote or primitive archosaurian condi­ tion'. Is it an archosaurian apomorphy, a mamma·· lian apomorphy, or an advanced amniote apomor-· phy (whatever is meant by 'advanced amniote')? The occurrence of an enlarged cochlear recess in derived members of several lineages probably repre­ sents parallel development of these structures. Tarsitano and Hecht (1980) also question the significance of the subcapsular process in demon­ strating a close relationship between crocodilians and birds. This process extends anteroventrally from the opisthotic, subdividing the primitive me­ totic fissure into an anterior recessus scala tympani (containing a lateral projection of the perilympha­ tic duct) and posterior me to tic fissure in crocodi­ lians and birds. An analogous structure in therian mammals, the processus recessus, also subdivides the primitive metotic fissure. Again, they state that the distribution of this character is unkown within the Archosauria and that in any case it is plesio­ morphic for mammals, reptiles and birds. In so far as the subcapsular process can be recognised, Whet­ stone and Martin ( 19 79) are partially correct - it can only be identified easily in crocodilians and birds. But the problem is not as circumscribed as this; the basic question is did other archosaurs have a medial projection of the perilymphatic duct that was separated in some way from the primitive me­ totic fissure? qruisckshank ( 19 7 0) notes a possible separate lateral opening for the perilymphatic sys­ tem (fenestra pseudorotunda?) in the braincase of Euparkeria implying the presence of a subcapsular process, but this requires further inspection of the specimen to verify its presence. A fenestra pseudo­ rotunda has been identified in the coelurosaurs Ga­ limimus bullatus (Osmolska et al. 1972), and Dro­ maeosaurus albertensis (Colbert and Russell 1969) which makes the projection of the opisthotic dor­ sal to the opening, a subcapsular process by default. For the present, after an admittedly limited initial literature survey of the inner ears in primitive the­ codonts and other dinosaurs, we must partially ag­ ree with Tarsitano and Hecht (1980); with the cur­ rent state of preparation and with the available ma­ terial, it is difficult to approach this question since this region is poorly ossified or highly modified in the taxa we have seen (various phytosaurs, ornithi­ schians, and saurischians). It is significant, though, that a fenestra pseudorotunda has been identified in two coelurosaurs . It would appear from Walker's sketches (he has not published photographs) that Sphenosuchus does indeed have a subcapsular pro­ cess similar in morphology to that of crocodilians and different from that of birds; but there is insu­ fficient published information to allow an evalua­ tion of the ear in related thecodonts (such asHes­ perosuchus and Pseudhesperosuchus) or coeluro­ saurs. We take issue with Tarsitano and Hecht's (1980) remarks about the importance of similar embryo­ nic origin of the subcapsular process of processus in mammals, birds, and crocodiles. They maintain that since this process is formed from the basal pla­ te in these groups it represents a plesiomorphy and is therefore primitive for all the groups. The em­ bryonic origin of this process in these groups is of little importance here; it is equivalent to saying that any structure derived from the neural crest is homologous and therefore plesiomorphic for all the groups in which it appears and that it is there­ fore not useful as a character for phylogenetic ana­ lysis. In this instance it is the ultimate expression of the morphogenetic process that is important not the embryonic origin of the tissue. Their point also sidesteps fossil evidence that indicates that such a process was not possessed by therapsids, or even by early mammals (Kermack et al. 1981 ). They cite the study by Kuhn ( 19 71) on the pouch young of the echidna, noting that a transient struc­ ture appears on the anterior face of the metotic fissure separating the metotic fissure and foramen rotundum. They then state that this process is homolgous to the processus recessus of therian ma­ mmals and represents a synapomorphy for all ma­ mmals; it was supposedly lost in the adults of mo­ dem monotremes because of aquatic adaptations. Kuhn ( 1971) is less sure of the homology of this process than are Tarsitano and Hecht (1980); he carefully points out that this transient structure may not be homolgous to the processus recessus of therian mammals and in the text and figures deno­ tes this by enclosing processus recessus within quo- 131 tes. The most detailed and explicit character list for Crocodilia is that of Crompton and Smith (1980), which is partially compiled from previous sources. We will paraphrase their criteria for the skull: 1. Displacement of the anterior margin of the primary head of the quadrate from its contact with the opisthotic (paraoccipital process) and squamosal onto the prootic and laterosphenoid. 2. Development of extensive secondary con­ tacts between the quadrate and the squa­ mosal behind the tympanic membrane and with the opisthotic below the hyo­ mandibular ramus of the facial nerve. 3. Development of an enlarged eustachian system which penetrates basioccipital and basisphenoid and expands into an exten­ sive system of pneumatic spaces in the ba­ sioccipital, basisphenoid, opisthotic, proo­ tic, quadrate, and pterygoid bones. 4. Fusion of the basipterygoid joint with su­ rrounding structures so that the basisphe­ noid, pterygoid, and quadrate are joined as a solid structure. 5. The suturing of the bones forming the oc­ cipital region with the quadrate, squamo­ sal, and parietal into a solid plate. The fu­ sed exoccipital and basioccipital expand to close the posterior temporal openings and also expand down wards to establish a secondary connection with the basis­ phenoid. This latter expansion traps the internal carotid and IX and X nerves in secondary foramina 6. Change in the primitive position of the external opening of the perilymphatic cis­ tern from the fissura metotica to a new opening, the fenestra pseudorotunda, se­ parated by a broad bony septum from the fissura metotica. Character 1 is significant, but requires further elaboration and clarification (see new diagnosis be­ low). Character 2 does not draw an important distin­ ction between two separate regions of contact of the quadrate with surrounding bones. A contact between the body of the opisthotic and quadrate is developed only in some advanced protosuchians and contacts between the quadrate and paroccipital pro­ cess are not even developed in some longirostrine 'mesosuchians'. (The term mesosuchia is used here in an informal sense, since it refers to a paraphyle­ tic grade of crocodilians). Thus, these are advanced features for crocodilians and do not constitute pri­ mitive features for the group (Busbey, in prep). Character 3 recognises the importance of cra­ nial pneumaticity but it is more specific than this character as listed in Langston (1973). Unfortuna­ tely many of the bones mentioned are also pneuma­ tic in birds, coelurosaurs, and camosaurs (see abo- 132 ve). Based upon the newprotosuchian specimen and study of previously published work, the only uni­ que pneumatic cranial elements of crocodilians are the supraoccipital and parietal. With the addition of these bones to the list in Crompton and Smith ( 1980), the degree of pneumaticity in crocodilians can be seen to surpass that of birds and saurischian dinosaurs. Character 4 is also significant, because it recog­ nises the closure of the basipterygoid articulation and involvement of the quadrates, which does seem to be unique for crocodilians. Character 5 is partially contingent upon charac­ ter 2 and is incorrect in its description of how the occiput formed. Since the quadrate and squamosal do not meet behind the tympanum, and in primi­ tive forms (see below) the quadrate and body of the opisthotic are not in contact, the condition they list here is actually derived beyond protosu­ chians, teleosaurs, and metriorhynchids. The clo­ sure of the postemporal fenestra was brought abo­ ut by the posterior extension of the squamosals and their contact along the entire length of the pa­ roccipital processes. It would also appear (Busbey, in prep) that it was the rotation of the quadrate and flexure of the basicranium that closed up the space between the quadrate and opisthotic, not a downward expansion of the exoccipital and basio­ ccipital. In the most primitive protosuchians the internal carotid was not enclosed in bone, thus the presence of a carotid foramen is derived for croco­ dilians. We discussed the subcapsular process above and we shall not use presence herein. The critical ques­ tion here might not be the possession of this pro­ cess but its detailed morphology. This has yet to be determined in non-crocodilians. Based upon the characters discussed above we would like to offer an improved diagnosis for the Crocodilia based only on cranial characters (these are primitive features for crocodilians): 1. Quadrate akinetic, head is V-shaped with lateral and medial contacts - 'V' points posteriorly. Lateral quadrate contact is with the squamosal, medial contact with the prootic, laterosphenoid, and parietal. Posterior portions of medial and lateral walls of superior temporal fossa are for­ med by quadrate. Quadrate has numerous lateral "pneumatic" fenestrae. 2. Flat ornamented skull deck, with squamo­ sal arched down behind the otic notch, attaching to the dorsal edge of the paro­ ccipital process along its entire length. Squamosal with lateral smooth shelf for attachment of otic flap. 3. Pneumatic supraoccipital and parietal, with transverse connections between mid- dl~ ears. Medial and lateral eustachian op­ enings. 4. Broad, sheet-like quadratojugal oriented parasagittally. Primary dorsal contact with postorbital. 5. Fused basipterygoid articulation. The Family Protosuchidae In the last 15 years the number of known pro­ to suchian crocodilians has increased considerably. Nash (1975) provided a review ofprotosuchian taxonomy from Broom ( 1904) through Romer (1972). She rejected attempts to broaden the diag­ nosis of the Crocodilia (Walker 1970, Romer 1972) and agreed with Bonaparte (1971) that crocodiles are distinct and should not include forms which may only be crocodilian sister groups - specifically, Sp­ henosuchus acutus and related thecodonts. She also rejected (correctly, we feel) the attempts by Walker and Romer to sub-divide the protosuchia at the family level and questioned the validity of assignments of various genera to their proposed fa­ milies. We accept her conservative classification re­ cognizing a single family (especially since many of the genera are poorly known and are hardly the ba­ sis for new families): this is the family Protosuchi­ dae as proposed by Brown (1934). We propose that two subfamilies be recognised: the Protosuchinae, to include all brevirostrine genera normally associa­ ted with the Protosuchidae; and the Eopneumato­ suchinae , to include the longirostrine Eopneumato­ suchis colberti, which is distinctly different from all other protosuchians and warrants inclusion in its own separate sub-family. Th.e subfamily Proto­ suchinae includes those genera listed by Nash (1975): Pedeticosaurus, Notochampsa, Stegomosu­ chus, Erythrochampsa, Protosuchus, Orthosuchus, Platyognathus, Hemiprotosuchus, and possibly Ho­ plosuchus. DESCRIPTION OF NEW PROTOSUCIDAN Order Crocodylia Gmelin, 1788 Suborder Protosuchia Mook, 1934 Family Protosuchidae Brown, 1934 Subfamily Protosuchinae Busbey and Gow, 1982 BAROQUEOSUCHUS n. gen. Etymology. The name emphasises the ornate achitecture of the braincase and quadrate, reminiscent of the baro­ que period in art and architecture. Diagnosis. Protosuchian crocodilian with broad flat basicranium meeting occiput at the level of the basisphenoid-basioccipital suture. Internal carotid arteries shielded medially by basisphe­ noid-basioccipital flanges, unenclosed laterally. Quadrate large and dorsal surface broadly ex­ posed in lateral view. Distribution, Elliot Formation, Southern Afri­ ca. Type Species. Baroqueosuchus haughtoni. Baroqueosuchus haughtoni n. sp. Figures 1-12 and Table 1 Etymology The new species is named in honour of Dr. S.H. Haughton (1888-1982), doyen of South African geology, whose work on the Red Beds published in 1924 set the scene for many important subsequent discoveries. Holotype. BP-1-4 746, partial braincase. Housed in the collections of the Bernard Price Institute for Palaeontological Research, University of the Wit­ watersrand, Johannesburg. Type locality. Farm Sunnyside, near Clarens, Ora­ nge Free State (O.F .S.), South Africa. Locality co­ ordinates 28°31' 40"S; 28°30'06''£ (1:50 000 seri­ es, map 2828 DA Golden Gate). Paratype. BP-1-4946., partial braincase with disar­ ticulated squamosals. Housed in the Bernard Price Institute collections, as for the holotype. Paratype locality. Farm Damplaats near Lady brand, O.F .S., South Africa. Locality co-ordinates 29°13' 12"S;27° 20'30"E (1: 50 000 series, map 2927 AB Ladybrand). Diagnosis. Same as for genus. DESCRIPTION The description is based upon the holotype and paratype. The descriptions of the otic bones are primarily based upon the paratype. BASIOCCIPITAL (Figures 1 and 6). As in mod­ efl! crocodilians, the basioccipital is a nearly verti­ cal elliptical element lying between the basisphenoid and the foramen magnum that is bordered late­ rally by the exoccipitals. It contributes the central portion of the occipital condyle. The ventrolateral comers bear small basal tubera. There is a midven­ tral foramen for the passage of the median pharyn­ gotympanic (eustachian) canal and notches for the lateral pharyngotympanic canals ventrolaterally be­ tween basioccipital and basisphenoid. EXOCCIPIT AL AND OPISTHOTIC. (Figures 1,6,7,8,9 and 10). These elements cannot be sepa­ rated in occipital view but are, unlike those of modem crocodilians, clearly distinguishable in late­ ral aspect. The lateral extremities of the paroccipi­ tal processes are somewhat damaged and distorted. A laterally open cranioquadrate passage runs from a position lateral to the foramen vagi on the occipi­ tal surface, up along the anterior face of the paroc­ cipital process. The paroccipital processes are verti­ cal, their ventral margins forming ridges that over­ hang the openings of the cranioquadrate canals. 133 Medial to the posterior openings of these canals the exoccipital is pierced by the large, ventrally direc­ ted foramen vagi for the passage of nerves IX and X; internally foramina for these nerves pierce the floor of the metotic fissure. The course of the ra­ mus communicans between VII and IX runs down the anterior border of the exoccipital, then turns in under a raised portion of the floor of the metotic fissure towards the internal lateral branch of the foramen vagi, as in modem crocodilians. The exoc­ cipital contributes a very small, lateral portion of the occipital condyle and lateral to this is pierced by a single small foramen for the Xllth nerve. The exoccipital contributes to the roof of the lateral pharyngotympanic canal from where its contact with the basisphenoid runs to the ventrolateral comer of the basicranium; at this point it forms an open channel laterally for the passage of the inter­ nal carotid artery. In lateral aspect the course of the cranioquadrate canal is apparent. Dorsally there is a distinct sutural contact of the opisthotic and prootic where these elements border the postero­ dorsal transverse pneumatic sinus. From here the anterior border of the opisthotic can be traced ac­ ross the floor of the otic capsule. The subcapsular process is contributed by the opisthotic and over­ lain anteriorly by an anterodorsal projection of the exoccipital. The median border of the metotic fis­ sure consists of the opisthotic; the posterior wall is pic;rced dorsally by a foramen communicating with the pneumatic spaces within the opisthotic. The floor of the fissure is pierced by foramina for the IXth and Xth nerves and is raised for the pas­ sage beneath it of the ramus communicans. Ante­ rior to the distal portion of the cranioquadrate ca­ nal the exoccipital forms a cup shaped recess pier­ ced by a semicircular foramen communicating via a short broad canal with the metotic fissure. The rim of this recess represents the site of the lower quadrate-opisthotic contact in more advanced pro­ tosuchians and all other crocodilians. Anteriorly a broad concave surface runs up from the internal border of the lateral pharyngotympanic canal, flan­ ks the ventral pneumatic sinus and terminates in an anterodorsally directed portion hollowed ventrally and pierced posteriorly to transmit the ramus communicans. A groove across the top of the opis­ thotic marks the course of the occipital vein throu­ gh a much reduced posttemporal fenestra. Lateral to this there is a distinct facet situated dorsally on the opisthotic for the reception of a lappet of the squamosal. SUPRAOCCIPITAL. (Figures 1, 6, and 8). The supraoccipital is a triangular bone apparently con­ fined to the occiput by the parietal and just exclu­ ded from the border of the foramen magnum by the exoccipitalfopisthotic. Vertical median and lat­ eral ridges flanked by depressions constitute the site of attachment of neck muscles (M. spinalis cap­ itis). The suture with the parietal appears to run in a groove above the lateral ridges and through the central ridge beneath a posteriorly directed pointed extension of the parietal. In lateral aspect the reg- 134 Figure 1. Baroqueosuchus haughtoni BP-1-4946. Stereophotographs of (from top to bottom):­ Occiput. Skull roof with squamosals remo\'ed. Skull roof with squamosals in position. Ventral surface of left squamosal. Scale= 1 em Figure 2. Baroqueosuchus haughtoni BP-1-4946. Stereophotographs of (from top to bottom):­ Right side of braincase with quadrate in position. Right side of braincase with quadrate removed. Dorsal surface of right quadrate. Ventral surface of right quadrate. Scale= 1 em. 135 136 Figure 3. Baroqueosuchus haughtoni BP-1-4946. Stereophotographs of (from top to bottom):­ Left side of braincase. Ventral surface of braincase. Scale= 1 em. ion of overlap between parietal and supraoccipital is obscured by the foot of a buttress of prootic. Breakage in BP-1-4 746 shows that the supraoccipi­ tal was fully pneumatic as in modem crocodilians; small spicules of bone can be seen between the in­ ner and outer tables of bone. PROOTIC. (Figures 2, 4, 8, 10, and 11). The right prootic is fully exposed in BP-1-4946. This is the first time such detail has been displayed in a primitive protosuchian crocodilian (not all the re­ lationships of the braincase of the more derived Eopneumatosuchus are fully understood since no quadrates are known and the regions of quadrate contact are eroded) because this bone is normally obscured by the overlying quadrate. The prootic has a large dorsal facet against which the head of the quadrate was firmly applied. This is buttressed fore and aft on the mesial surface by struts having robust contacts with the lateroventral edge of the parietal, these struts being separated by a large cir­ cular opening for the anterior transverse pneumatic canal. The posterior strut is pierced by a foramen that transmitted the occipital vein. This strut par­ tially overlies the supraoccipital and abuts against the opisthotic. Antero-dorsally the prootic conta­ cts the parietal and laterosphenoid. The region be­ hind the anterior transverse pneumatic sinus is a complex of struts and foramina (for which func- tions cannot be assigned) . The prootic contributes the anterior and dorsal border of the foramen ova­ le, where it is formed into a low boss which corres­ ponds with a shallow depression on the mesial sur­ face of the quadrate; this boss stands free of the wall of the prootic supported by a complex of de­ licate struts. The ventral relationships of the pro­ otic are not clear. There seems to be a long sutural contact with the basisphenoid dorsal to the carotid pillar of the latter and clearly confluent with it. A prominent cochlear eminence supports opisthotic and exoccipital where they meet at the subcapsular process. Numerous foramina pierce the prootic. Posteriorly the bone contributes the anterior wall of the posterodorsal transverse pneumatic sinus. Below this is the otic capsule, a deep, vertically oriented, anteroposteriorly elongated pit with a bony floor which communicates mesially via the fenestra pseudorotunda with another pit that hou­ ses the fenestra ovalis deep within it. The bony rim surrounding the fenestra ovalis is pierced anterod­ orsally by a small foramen from whence a groove runs along the anterior rim. Anterior to the ear re­ gion is a small foramen for the transmission of the Vllth nerve. Running ventrally from this the course of palatine branch of VII has left its impression in the surface of the bone, while a dorsal impression marks the course of the hyomandibular branch for a short distance. Anterior to the transverse pneu­ matic sinus are two foramina connected by a mar­ ked depression on the surface of the prootic; the dorsal foramen has the form of a narrow slit, so that this is probably the course of a nerve rather than a blood vessel. Below the dorsal quadrate ar­ ticulation of the prootic lies a group of three fora­ mina linked to the pneumatic sinus within. BASISPHENOID. (Figures 3, 5, 6, 7 and 8). The bone is exposed as a broad plate on the ventral and occipital surfaces. Sutures with the pterygoids are distinct on the ventral surface where a medial projection of basisphenoid reaches a point level with the posterior edges of the transverse pterygoid flanges. Posteriorly there is a low midventral ridge 137 flanked by low lateral ridges. The bone sweeps up abruptly onto the occipital surface where it meets the exoccipitals and basioccipital. The basisphe­ noid has a ventrolateral contact with the quadrate. Detailed outlines of the basisphenoid rostrum are difficult to interpret due to cracking and distor­ tion. The pila antotica is firmly sutured to the laterosphenoid. Foramina emerging laterally from the region of the hypophysial fossa transmitted the internal carotid artery above and the Vlth nerve below. The rostrum is enveloped below by the pterygoids which rise anteriorly to the full height of the rostrum and continue forward as a median septum. The basipterygoid articulations are en­ tirely obscured by the pterygoids. Internally, in the Abbreviations Bo Bs b.t. c.Q.c. d.l.Q. d.t.p.s. Ect e.f.g. Eo f.o. fen. o f.p.r. i.c.a. L l.p.c. m.f. m.p.c. o.c. Op Op a.Sq o.v. p Pal P.a. Q poc.r. prc.r. Pt Ptf p.t.f. Q r.. f.lO.V. ro.Bs r.s. s.c.p. So Sq v.l.Q v.p.s. basioccipital basisphenoid basal tubera cranioquadrate canal dorsal lappet of quadrate dorsal transverse pneumatic sinus ectopterygoid ear flap groove ex occipital foramen ovale fenestra ovalis fenestra pseudorotundum interal carotid artery laterosphenoid lateral pharyngotympanic canal metotic fissure median pharyngotympanic canal occipital condyle opisthotic opisthotic articulation for squamosal occipital vein parietal palatine prootic articulation for quadrate postcarotid recess precarotid recess pterygoid pterygoid flange post temporal fenestra quadrate ramus communicans rim of infraorbital vacuity basisphenoid rostrum rhomboid sinus subcapsular process supraoccipital squamosal ventral lappet of quadrate ventral pneumatic sinus 138 L Pa.Q Pt Op a. Sq Op Bo 1cm Figure 4. Baroqueosuchus haughtoni. Dorsal view of BP-1-4 746 (above) and BP-1-4946 (below). region occupied by the basisphenoidal pneumatic sinuses, the basisphenoid consists of a delicate filigree of interconnecting struts. LATEROSPHENOID. (Figures 4, 8 and 9). The laterosphenoid has a straight nearly horizontal su­ tural contact with the parietal just above the level of the upper contact between prootic and quadrate. In both specimens the delicate postorbital process­ es of the laterosphenoids are damaged, but are pre­ sent as delicate lateral flanges directed out towards the postorbital bar. The bone contributes the pillar anterior to the foramen ovale and meets the basis­ phenoid below this where the contact is obscured by a sheet of the pterygoid reaching to the same height. The posterior edge of the laterosphenoid, as far as the top of the foramen ovale, is overlapped slightly by the quadrate. Laterally the bone is sepa­ rated by a rounded ridge into temporal and orbital surfaces. The laterosphenoids meet in the dorsal midline, though there is some damage in this area. These bones are exposed dorsally anterior to the parietal where each is pierced by a dorsally facing foramen for the passage of the anterior transverse pneumatic sinus; the area between them housed the olfactory lobe of the brain. The ventral orbi­ tal surface of the laterosphenoid contains several foramina and received the pila antotica of the basis­ phenoid. V3 emerges medial to the pillar in front of the foramen ovale. Medial to this is a smaller fo­ ramen for the levator bulbi branch of V. Above these latter foramina is the tiny exit of IV. The an­ terior edge of the laterosphenoid bears two identa­ tions marking the points of exist of nerves II and III. PARIETAL. (Figures 1, 4, 6, 7 and 8). The sing­ Le fused parietal has a pitted dorsal surface. A slight ridge separates dorsal and temporal parts of the pa­ rietal anteriorly and becomes a deep recess poster- 139 Q Q m.p.c. Bo l.p.c. 1cm Figure 5. Baroqueosuchus haughtoni. Ventral view of BP-1-4 746 (above and BP-1-4946 (below). iorly. The anterior edge of the parietal bears a small median raised 'nib' with grooves lateral to it for the reception of the frontals. The contact with the supraoccipital is at the top of the occiput, from whence the supraoccipital runs under the parietal. Sutures with the squamosals are slightly concave. The sutural contacts with the prootic and latero­ sphenoid are described above. SQUAMOSAL. (Figures 4, 6, 7 and 10). The disarticulated squamosals of BP-1-4946 allow a complete description of their contacts. The squa­ mosal has an ornamentation of grooves and pits, with long grooves near the attachment site for the ear flaps. The ledge for the ear flap can be traced along its lateral edge. Most of the facet which con­ tacts the parietal is preserved, as also is part of the lateral and posterior border of the upper temporal fenestra. Posteriorly there is a wide groove for the attachment of neck musculature. Below this is a facet with a rounded distal edge, occupying the proximal two thirds of the width, for the recep­ tion of the paroccipital process of the opisthotic. A small raised flap of bone near its base rests on a facet of the dorsal surface of the opisthotic. On the ventral surface the attachment for the quadrate runs from the outer rim of the upper termporal opening 140 p 1cm Figure 6. Baroqueosuchus haughtoni. Occipital views of BP-1-4 746 (above} and BP-1-4946 (below). back to abut against the process of th~ squa?Iosal that lies on the dorsal surface of the opisthotic. Al­ ong the posterior half of its length as preserved this contact has the form of a smooth concave channel. Between the quadrate and parietal c~ntacts a sm~­ oth surface region roofs the space which communi­ cates with the occiput via the reduced postemporal fenestra. The squamosal forms a deep lateral recess covering the ear region. QUADRATOJUGAL. The quadratojugals are absent from both specimens. The broad dorso­ lateral edge of the post<;>rbital bar indicat.es that the quadratojugal was a thm broad sheet onente~ par­ asgittally, as in Hemiprotosuchus. The pnmary contact of the quadratojugal seems to have been with the postorbital bar. PALATINES. Portions of the palatines may be present but, as no sutures are discernible on eit?er specimen, they will be treated with the pterygmds. PTERYGOIDS. (Figures 3, 5, 7, and 9). The relationship of the ptery~oids t? the b~isphenoid in ventral view is shown In the Illustrations. From the point where the pterygoids me~t in the midline the surface is marked by several ndges ra- diating out and forward from a midventral ridge; these probably mark the position of the internal choanae. In BP-1-4946 much of the left pterygoid flange is present, though the distal surface has been eroded away, and the right pterygoid flange is inta­ ct. The flanges are heavily eroded in BP-1-4746. In lateral aspect the pterygoid is slightly overlapped by the anteroventral surface of the quadrate. The pterygoid rises to the level of the basisphenoidfla­ terosphenoid contact, then dips anteriorly expos­ ing much of the basisphenoid rostrum. The ptery­ goids then meet . to form a vertical septum which enfolds the anterior portion of the rostrum. The quadrate ramus of the pterygoid consists of a thin sheet in rigid overlapping contact with the ba_sis­ phenoid ventrally; it curves up onto the ventrola­ teral surface of the braincase where it is overlapped anteriorly by the ventral quadrate lappet. The in­ ternal contact with the pterygoid ramus of the qua­ drate is not displayed in either specimen. ECTOPTERYGOID. (Figures 5 and 7). The left ectopterygoid of BP-1-4946 was preserved intact lying on the dorsal surface of the root of the ptery­ goid flange. The proximal portion of the right ele- 1cm . I I . I I ., · I 141 Figure 7. HJaroqueosuchus haughtoni. Right side of BP-1-4 746 {above) and BP-1-4946 {below). ment was found loose. The ectopterygoid is a slen­ der bone having an extensive conttact along the anterior edge of the pterygoid flange, a narrow free portion and a short jugal contact. QUADRATE. (Figures 6, 7 and 9). The quad­ rates were better revealed in BP-1-4946 because of the removal of one, and the descripttion below will specifically refer to that specimen. The left quad­ rate is the more complete, lacking only most of the articulatory condyle. This quadrate is somewhat distorted which makes precise reco:nstruction and comparison with the right quadrate, itself probably distorted, difficult. The quadrates were easily dis­ torted because they are thin and delicate. Both bones are illustrated as preserved. The quadrate has extensive contacts with the side wall! of the brain­ case. Mesial contacts are with the prootic , latera­ sphenoid and parietal above, and basisphenoid be­ low. Anteriorly two delicate sigmoid lappets of the quadrate overlap several bones. The dorsal lappet lies on laterosphenoid and partially walls the fora- men ovale. Well behind the edge of this lappet is a slight depression in the mesial surface of the quad­ rate which. fits over a boss of prootic supported by several delicate bony struts. The lower lappet lies alon~ the most superficial of these structures, ex­ tendmg on to the laterosphenoid and pterygoid. Be­ tween these lappets and the anterolateral border of t~e quadrate is a smooth notch forming the paste­ nor border of the dorsal portion of the temporal fossa. The dorsal two thirds at least of this anterior border of the quadrate lateral surface presents a broad grooved area for contact with the quadra­ tojugal. More ventrally the border narrows so that one cannot be certain of the full extent of the quad­ ratojucal contact. The posterior edge of the quad­ rate forms a posteriorly concave otic notch. This edge nowhere contacts the braincase or squamosal so that the cranioquadrate canal and otic region are exposed laterally and do not form a closed otic meatus posteriorly. The quadrate is pierced by se­ veral large foramina the most prominent of which 142 y--~~ main branch o.v. So f. p. r. c. Q c. s.c.p. m.f. r.c. Op Pt 1cm 1cm Figure 8. Baroqueosuchus haughtoni. Right side of BP-1-4946. ~'1. levator bulbi Pt y_ Above: Anterior oblique viewshowinganterior cranial nerve exits . Below: Braincase detail. is situated middorsally. Within it a rim of bone for­ ms a deep pocket anteriorly; there are short ventral struts supporting an anteroposteriorly arranged br­ ace continuous with the top of the ventral lappet. Below and anterior to this foramen is a smaller one bisected by a deep strut of bone. Behind this is a large, dorsoventraly oriented groove floored in the dorsal portion by part of the pterygoid ramus of the quadrate. The top of the groove is pierced by a small foramen which emerges in the prominent oval depression on the mesial surface, that com­ municates with the other foramen previously de· scribed. The pterygoid ramus is crushed and incom­ plete on the right side and it is not exposed on the left. The ventral portion of the left quadrate is complete anterior to the condylar region where it is folded into a laterally facing channel held away from the ventral border of the oval foramen by two delicate internal struts. The anterior edge of this fold represents the position of the insertion of the medial sheet of the cranial adductor tendon. The left quadrate has three smaller foramina ante­ roposteriorly aligned below the groove. The expo­ sed ventromesial portion of the left quadrate is probably part of the condyle but it is possible that some of this is missing along with the ventrolateral portion of the quadrate condyle. FRONTAL. (Figure 4). The frontal is know11 only from the holotype. Erosion has destroyed the anterior portion of the frontal along a diagnoalline from just anterior to the frontal-postorbital suture back to the levef of the front of the supratemporal fenestra. The frontal participates in the supratem­ poral fenestra, as in thecodonts. The superficial or­ namentation consists primarily of grooves radiating away anteriorly from the fronto-parietal suture. The fronto-parietal suture is 'W' -shaped. There is a central process on the parietal that extends for­ ward into the frontal. The posteriorly pointing lim­ bs of the 'W' occur at the edge of the supratempo­ ral fenestrae, and the anterolateral limbs run for­ ward obliquely towards the frontal-postorbital su­ ture. A small portion of the frontal rim of the or­ bit is present and it shows no raised rim. The con­ tact with the postorbital is similar in appearance to that of modem brevirostrine crocodilians. POSTORBITAL. (Figures 4 and 7). The right postorbital is preserved in BP-1-4 746. The orbital edge of the postorbital continues as the front edge without an upraised rim, at an angle of about 45 uto the midline. The postorbital forms a slight saddle­ shaped surface into which the frontal inserts, with a long process extending under the frontal. A slight 1cm 143 ridge extends along the top of the postorbital bar about two thirds of its Width. There is a slight de­ pression on the skull deck at the level of the ante­ rior edge of the postorbital bar. The bottom of the depression is ornamented with pits. Only the base and a very short piece of the shaft of the postorbi­ tal bar is present, but the outer surface is slightly ornamented distally, implying that the bar was su­ perficial. The bar where broken, is triangular in cross section, with the central axis of the bar poin­ ting anteriorly at an angle of about 3 0° from the midline and a deflection of only about 20°from ho­ rizontal. The ventral apex of the triangle gives rise to a thin laterally curved sheet of bone that exten­ ds along the lower edge of the bar back to the level of the quadrate-squamosal suture. The sheet is bro­ ken ventraily but must have contacted a broad qua­ dratojugal along its ventral edge. This forms an an­ terior blind pocket for the dorsal space surrounded by squamosal laterally and quadrate mesially, whi­ ch is similar in shape to the steinkem described by Bonaparte (1972) for Pedeticosaurus. The posteri­ or contact with the squamosal is formed by a flat process that extends over the squamosal back to the level of the frontoparietal suture. This process forms a slightly raised region on the skull deck or­ namented with anteroposteriorly oriented grooves. COMPARISON WITH OTHER PROT OS UCHIANS Only Eopneumatosuchus colberti from the Ka­ yenta Formation of North America has the side of the brain case under the quadrate exposed. Recent­ ly collected Protosuchus material and the newly re­ prepared type were not available for study. Only those protosuchian crocodilians with cranial mater­ ial will be considered here. Southern Africa Nash (1975) presents a summary of the taxon- Figure 9. Baroqueosuchus haughtoni. BP-1-4946 Left side of braincase and posterior oblique aspect of dorsal surface of left quadrate. 144 Baroqueosuchus 1cm Alligator omy and localities of Southern African protosuchi­ ans; Baroqueosuchus haughtoni is the first new southern African genus to be found since her semi­ nal paper. It is not possible to compare directly Notochampsa istedana (Broom 1904, Haughton 1924), Erythrochampsa longipes (Broom 1904, Haughton 1924 ), or Pedeticosaurus leviseuri (Van Hoepen 1915) with B. haughton£. The holotype of Erythrochampsa is only partly prepared and as yet no cranial material, if present, is exposed. The ho­ lotype of Notochampsa istedana represents a par­ tial mold of the undersurface of the skull deck and does not exhibit diagnostic characters. Many com­ parisons have been made with Notochampsa (for example see Nash 1975), but since this specimen is a steinkern the dorsal surface and so-called sutures are not directly comparable to dorsal views of oth- Figure 1 O.Baroqueosuchus above. Alligator below. Braincase with prootic and squamosal folded out. Contacts for prootic - heavy stipple. Contacts for braincase -light stipple. Contacts for squamosal- V. pattern. er protosuchians. We suggest that the name Noto­ champsa should be applied only to the type; thus the so-called Notochampsa specimen at the British Museum of Natural History (R8503) (Whetstone & Martin 1979), while it may be a crocodilian, should not be referred to this genus. Likewise, the partial steinkern of the otic region of Pedeticosaurus str­ ongly suggests it is a crocodilian (Bonaparte 19 7 2), but it lacks diagnostic characters; hence, this name should probably also be applied to the type only. There are differences in the shape of the otic stein­ kerns in Notochampsa and Pedeticosaurus, which may be partially due to preservation. Both have an otic region more similar to Barqueosuchus than Orthosuchus. 145 Pt Alligator p I. ):;:::iil·ilii • ·I pro Baroqueosuchus llflso ~~~~~~~~~~~~~ 0 p -So [\\\\\\\\\\\\\\\\\\\\\\\\\1 8 s Figure 11. Braincase of Baroqueosuchus and Alligator compared. Orthosuchus stormbergi (Nash 1968, 1975). upper Triassic Los Colarados Formation of La Rio­ ja, Argentina. The type material represents a smal- The braincase of both the type and paratype of Baroqueosuchus is larger, deeper, and broader than that of Orthosuchus. Orthosuchus demonstrates two major features that are derived over Baroqueo­ suchus: the internal carotids are enclosed in carotid foramina and the opisthotic and quadrate are in contact below the cranioquadrate canal. In addi­ tion, the entire basicranium is flexed up at the ba­ ck edge of the pterygoid flanges, producing a more characteristic crocodilian condition as well as cau­ sing the medial scars for the cranial adductor ten,­ don to become more prominent. The long axis of the quadrate of Orthosuchus is oriented both more medially and more subparallel to the tooth row, which foreshadows the condition seen in more de­ rived 'mesosuchians'. It would appear that there are fewer fenestrae exposed on the lateral (dorsal) surface of the quadrate and that their size and arr­ angement is fundamentally different from that seen in Baroqueosuchus, Hemiprotosuchus (Bona­ parte 1971 ), or the Berkeley Protosuchus material. Nash (1975) in several illustrations shows different numbers and different arrangements of these fenes­ trae in Orthosuchus, but it would appear that there are actually fewer and less ornate fenestrae in Or­ thosaurus than in Baroqueosuchus. South America Hemiprotosuchus leali (Bonaparte 1971) is the only protosuchian known from South America. The holotype and paratype material are from the ler animal than Baroqueosuchus. The number and arrangement of lateral quadrate fenestrae appear to be quite similar to those of Baroqueosuchus which in fact compare favourably with those seen in Pro­ tosuchus (based on the unpublished Berkeley mate­ rial). It should be noted that the position of the tympanum, the attachments of the quadrate to the skull, and the occiput are incorrectly restored in Bonaparte's original description (Busbey, in prep.); suffice to say that the otic region is similar to that of Baroqueosuchus. The quadrate, relative to that of Baroquesuchus, is rotated further to the rear so ~hat i~ is in partial cont~ct with the body of the op­ Isthotic below the crantoquadrate canal. The basic­ ranium is flexed upward, but not as much as in Or­ thosuchus, causing the medial tendon ridges of the cranial adductor tendon to be more robust than those of Baroqueosuchus. This also results in a de­ eper basisphenoid keel in Hemiprotosuchus. The skull deck is damaged, though it would appear that the ornamented skull deck is less extensive in He­ miprotosuchus than in Baroqueosuchus because of the relatively larger supratemporal fenestrae of the former. North America Protosuchus r£chardsoni Brown (1933) The holotype of Protosuchus r£chardson£ (Br­ own 1933, Colbert and Mook 1951) is currently unavailable for study (Max Hecht pers. comm. to I46 I I I I I / / ., ...... _ --,..;:..,., // f I (,._.J ( / / I I I / { I Figure I2.Baroqueosuchus haughtoni. Table 1 : Measurements of type skulls (in mm.). 1. Width of skull table at fronto-parietal suture 2. Minimum width skull table 3. Width of skull table at posterior edge of supratemporal fenestra 4. Length of parietal 5. Height of skull 6. External length supratemporal fenestra 7. External width supratemporal fenestra 8. Internal length supratemporal fenestra 9. Internal width supratemporal fenestra IO. Maximum width of supraoccipital (at posttemporal fenestrae) II. Minimum width of postorbital behind orbit I2. Length of head of quadrate at squamosal suture I3. Width across basicranium (at level of carotid notches) I4. Width across basicranium (at lateral pharyngotympanic openings) I5. Width foramen magnum Dorsal and occipital views of BP-I-4 746 showing measurements taken for Table I. BP-I BP-I 4746 4946 6,6 7,0 3,6 3,8 I3,2 I2,9 28,3 26,2 35,7 34,4 2I,6 II, 7 I2,2 I2,2 4,7 4·,6 I3,2 I3,I 6,6 26,0 23,0 24,8 I9,0 I6,0 I2,6 6,5 6,3 James Clark), though it has apparently been re-pre­ pared and is being redescribed. The new Berkley material from the Kayenta Formation is currently being prepared and studied by James Clark and thus is also unavailable for detailed study. Until this material is made available to other research­ ers, only cursory observations are possible; these are made mostly from the descriptions of Protosu­ chus in Crompton and Smith ( 1980) (in which the­ re are mistakes in interpretation of quadrate rela­ tionships) and inspection of some of the Berkely material. The general morphology of the quadrate con­ tacts appears to be quite similar in Protosuchus and Baroqueosuchus haughtoni. One partially prepared specimen demonstrates a lateral -quadrate surface with quadrate fenestrae very similar to those of B. haughtoni and Hemiprotosuchus in number, mor­ phology and position. The line drawing of Proto­ suchus (Fig. 11.1) in Crompton and Smith (1980) shows the large dorsal opening (labelled 'tm ') and only a single more ventral fenestra; the differences are perhaps due to specimen damage. The partially prepared specimen and MCZ 6727 also demonstra­ te the large plate-like quadratojugal in primary con­ tact with the postorbital. The degree of contact be­ tween the opisthotic below the cranioquadrate ca­ nal is at this point not clear. In the partially prepa­ red specimen the lower edges of the quadrate and body of the opisthotic are not exposed, but at lea­ st two thirds of the body of the opisthotic is not in contact with the quadrate. In other, more badly crushed Berkely material, it would appear that the quadrate and opisthotic are in slight contact ven­ trally but this could be due to crushing. In any case, the physical separation of the quadrate and opisthotic in Protosuchus and Baroqueosuchus se­ ems to be greater than in Hemiprotosuchus and as such they represent less derived animals. The basi­ cranium of Protosuchus is, as in Hemiprotosuchus, flexed up at the back and as a result has the more robust basisphenoid keel and medial cranial adduc­ tor tendon scars. The supratemporal fossae of Pro­ tosuchus are relatively smaller than those of Baro­ queosuchus (based on the restoration in Crompton and Smith (1980) and Colbert and Mook (1951) and are essentially 'juvenile' in shape and orientat­ ion. This results in a very broad skull deck lateral to the supratemporal fenestra and between the su­ pratemporal fenestrae, unlike Baroqueosuchus. Eopneumatosuchus colberti Crompton and Smith (1980). Eopneumatosuchus is the first longirostrine pr­ otosuchian crocodilian that has been found. Becau­ se of this the bones of the braincase are rather ex­ panded anteriorly, quite unlike Baroqueosuchus. Unfortunately the quadrates and regions of quad­ rate contact on the braincase proper are eroded and thus do not preserve any evidence of an opis­ thotic-quadrate contact below the cranioquadrate canal (contra Crompton and Smith 1980). The bra- 147 incase of Eopneumatosuchus is much shallower and much broader than in Baroqueosuchus. The su­ pratemporal fenestrae of this longirostrine form are, of course, quite large and are comparable in shape to those of teleosaurs and not to any other proto­ suchians. It appears that there are carotid forami­ na, though they are incorrectly identified in Crom­ pton and Smith ( 1980) as simple vascular openings (the supposed carotid foramen labelled in their Fig. 11.3 as 'i.e.' is actually the lateral branch of the fo­ ramen vagi, see Busbey in prep.). As in Baroqueo­ suchus, the internal carotid arteries were carried forward in enclosed tubes in the basisphenoid (la­ belled as the vidian canal in Crompton and Smith) to the vicinity of the sella turcica, though the ba­ sisphenoid is greatly elongated in Eopneumatosuc­ hus. The exact position of the subcapsular process in Eopneumatosuchus is equivocal. Crompton and Smith (1980) picture a dotted (and therefore hypo­ thetical) lower rim of the foramen ovale, deep with­ in a small pocket. I (Busbey) cannot find any evi­ dence of this bony spicule on inspection of the holotype, though the sides seem to have been da­ maged during preparation (which must have been unavoidable because of the extremely delicate bone). The large bony plate labelled as the subcap­ sular process in Eopneumatosuchus may thus be the back edge of the foramen ovale, but we are in­ clined to accept Crompton and Smith's (1980) in­ terpretation based upon general size and position of the large process. A similar large plate in Baro­ queosuchus is similarly interpreted as being the subcapsular process; the delicate bony lower edge of the foramen ovale is preserved. Both Baroqueo­ suchus and Eopneumatosuchus demonstrate a re­ markable degree of pneumaticity, especially in the basisphenoid. This is assumed to be primitive for crocodilians, and represents the morphology of the basisphenoid before its anteroposterior shortening and flexure in the region of the opening of the me­ dial pharyngotympanic canal. Eopneumatosuchus does have greater pneumaticity than Baroqueosu­ chus, due to the anterior and lateral expansion of the bones of the braincase and skull roof. PROTOSUCHIAN RELATIONSHIPS The known protosuchian crocodilians occur within a fairly narrow time span, probably no ear­ lier than Rhaetian to no later than Late Liassic. Ne­ vertheless, they demonstrate a fairly rapid change from the condition of Baroqueosuchus which we consider to be the most primitive crocodilian, to that of Orthosuchus where the quadrate has ro­ tated up to contact the opisthotic and form a ra­ ther 'modem' basicranium. With the preparation of the new Berkeley material the similarities betw­ een Baroqueosuchus and Protosuchus are quite evi­ dent, mainly the orientation, relationships and fe­ nestrae of the quadrate. The only southern African forms which seem to have the right sizes and shap­ es to be related to Baroqueosuchus are not suffi­ ciently well preserved to find important diagnostic 148 characters. Orthosuchus stormbergi is dissimilar to Baroqueosuchus in its size, proportions, and details of sutural contacts. ACKNOWLEDGEMENTS Travel funds to South Africa for the first au thor and partial support for illustrations were made possible by Na­ tional Science Foundation Research Grants DEB 77-25339 and DEB 80-22183 to Dr. James A. Hopson. The first aut­ hor also wishes to thank the South African Museum and Fi­ eld Museum of Natural History for allowing him to use the­ ir preparation facilities. Drs. Mike Raath andJames Kitchi­ ng, BPI in Johannesburg; Jacques van Heerden, formerly of the National Museum in Bloemfontein; and Mike Cluver of the South African Museum kindly permitted one or both of us to study and borrow protosuchian material. The first au­ thor would like to acknowledge the aid of the Geology Fo­ undation of the Department of Geological Sciences the Un­ iversity of Texas at Austin; he also thanks Dr. A.W.Cromp­ ton, Harvard University, who provided stereophotographs and the holotype of Eopneumatosuchus colberti for study, and Mr. James Clark for allowing him to look at Protosuc­ hus material belonging to Berkeley that he is currently stu­ dying. Dr. Jaime Powell, Tucuman, kindly provided a cast of the recently re-prepared holotype of Hemiprotosuchus lealii. The shaded illustrations of BP-1-4 746 were prepared by Ms. Claire Vanderslice. REFERENCES ANTUNES, M.T. (196 7). Sur quelques caracteres archaiques des crocodiliens a propos d' un mesosuchian du Lias supe­ rieur de Tomer (Portugal). Colloq. Int. du Cent. Nat. Rech. Scient., 163, 409-414 BONAPARTE, J.F. (1971). Los tetrapodos del sector superior de la formacion Los Colorados, La Rioja, Argentina. (Triasico Superior) I Parte. Op. Lilloana, 22:1-183. ----------- (1972). Pedeticosaurus leviseuri Van Hoepen, a probable protosuchian, Navors. Res. Nas. Mus. 2, 301-305. BROOM, R. (1904). On a new crocodilian genus (Notochampsa) from the upper Stormberg beds of South Africa. Geol. Mag., 5, 582 - 584. BROWN, B. (1934). A change of names. Science, 79:80. CLEMENS, W.A., LILLEGRAVEN, J.A., LINDSAY, E.H., SIMPSON, G.G.S. (1979). Where, When and What- A sur­ vey of known Mesozoic Mammal Distribution. In: Lillegraven, J.A., Kielan-Jaworowska, Z., and Clemens W.A., eds. Mesozoic Mammals : the First Two-Thirds of Mammalian History, Berkeley, Univ. of California Press, p. 5-58. COLBERT, E.H. {1946). The eustachian tubes in the crocodilia. Copeia, 1946., 12-14. COLBERT, E.H. and RUSSELL, D.A. {1969). The small Cretaceous dinosaur Dromaeosaurus. Amer. Mus. Novit., 2380, 1-49. CROMPTON, A.W. and SMITH, K.K. {1980). A new genus and species of crocodilian from the Kayenta Formation (Late Triassic?) of Northern Arizona. In: L. Jacobs {ed), Aspects of Vertebrate History, Flagstaff Museum of No­ rthern Arizona. p. 193-217. CRUICKSHANK, A.R.I. (1970). Early thecodont braincases. Proc. Gond. Symp., 1970; 683-685. -------------- {1972). The proterosuchian thecodonts. In: Joysey, K.A. and Kemp, T.S. (eds.), Studies in Vertebrate Evolution. Edinburgh, Oliver and Boyd, p. 89-119. EUDES-DESLONGCHAMPS, J.-A., (1864). Memoires sur les teleosauriens de 1' epoque Jurassique. Mem. Soc. Linn. Normandie, 13, 1-138. FITCH, F.J., MILLER, J.A. (1971). Potassium-argon radioages of Karoo volcanic rocks from Lesotho. Bull.volcano­ loqique., 35, 64-85. GASPARINI, Z.B. de., DELLAPE, D. {1976). Un nuevo crocodilo marino (Thalattosuchia, Metriorhynchidae) de la Formacio Vaca Muerta Uurasico, Tithoniano) de la Provincia de Neuquen {Republica Argentina). Primer. Conq. Geol. Chilena, C1-C21. HAUGHTON, S.H. {1924). The fauna and stratigraphy of the Stormberg series. Ann. S. Afr. Mus., 12,323-497. HEATON, M.j. {1979). Cranial anatomy of primitive captorhinid reptiles from the late Pennysylvanian and early Permian; Oklahoma and Texas. Okl. Geol. Surv. Bull., 127, 1-84. HOPSON, J.A. {1979). Paleoneurology, In: Gans, C, Northcutt, R.G., and Ulinkski, P., (eds), Biology of the Reptilia. Vol. 9 New York, Academic Press, pp. 39-146. KERMACK, K.A., MUSSETT, F., and RIGNEY, H.W. (1981). The skull of Morganucodon. Zool.]. Linn. Soc., 71, 1-158. KUHN, H.]. ( 19 71). Die En twicklung und Morphologie des Sch 'adels von Tachyglossus aculeatus. A bh. Senck. Natur­ forsch. Ges., 528, 1-224. LANGSTON, W. (1960). The vertebrate fauna of the Selma Formation of Alabama, Part VI. The Dinosaurs., Fieldiana (Geology), 3 (6), 315-361. ---------- {1973). The crocodilian skull in historial perspective. In: Cans, C., and Parsons, T.S. (eds.) Biology of the Reptilia. Vol. 4. New York Academic Press, pp. 263-284. LULL, R.S. {1953). Triassic Life of the Connecticut Valley. Bull. Connecticut Ceo. & Nat. Hist. Survey., 81, 1-331. MOOK, C.C. {1934). The evolution and classification of the Crocodilia. ]. Geol., 42, 295-304. NASH, D., {1968). A crocodile from the upper Triassic of Lesotho, j.Zool., 156, 163-179. ------ {1975). The morphology and relationships of a crocodilian, Orthosuchus stormbergi, from the upper Trias­ sic of Lesotho. Ann. S. Afr. Mus., 67,227-329. OLSEN, P.E. (1980). A comparison of the vertebrate assemblages from the Newark and Hartford Basins (early Meso­ zoic, Newark Supergroup) of Eastern North America. In: Jacobs, L. (ed.), Aspects of Vertebrate History, Flagstaff, Museum, off Northern Arizona. p. 35-53. --------and GALTON, P.M. (1977). Triassic-Jurassic tetrapod extinctions: Are they real? Science, 197:983-986. ------,Me CUNE, A.M. and THOMPSON, K.S. (1982). Correlation of the early Mesozoic Newark Supergroup by vertebrates, principally fishes. Amer. Jour. Sci., 282, 1-44. OSBORN, H.F. (1912). Crania of Tyrannosaurus andAllosaurus. Mem. Amer. Mus. Nat. Hist. (new Ser), 1, 1-30. OSTROM, ].H., (1976). Archaeopteryx and the origin of birds. Biol. jour. Linn. Soc. Lond., 8, 91-182. OSMOLSKA, H. (1976). New light on the skull anatomy and systematic position of Oviraptor. Nature, 262 (5570), 683-684. --------- , RONIEWICZ, E., and BARSBOLD, R. (1972). A new dinosaur Gallimimus bullatus n. gen., n.sp. (Ornithomimidae) from the Upper Cretaceous of Mongolia, Results of the Polish-Mongolian Palaeontological Expe­ ditions Part IV. 2 7, 1 04-143. RIXON, A. E. (1976). Fossil animal remains: Thei1· preparation and conservation. London, Athlone Press. 149 ROMER, A.S. (1972). The Chanares (Argentina) Triassic reptile fauna. 16. Thecodont classification. Brevzora, :$95, 1-24. RUSSELL, D.A. (1970). Tyrannosaurs frorn the late Cretaceous of western Canada. Can. Nat. Mus. Pub!. Pal., 1, 34 pp. SILL, W.D. (196 7). Proterochampsa barrionuevoi and the early evolution of the Crocodilia. Bull. Mus camp. Zoo!. Harv., 135,415-446. TARSITANO, S. and HECHT, M.K. (1980). A reconsideration of the reptilian relationships of Archaeopteryx. Zoo!. Jour. Linn. Soc. Land., 69, 149-182. · VAN HOEPEN, E.C.N. (1915). Contributions to the knowledge of the reptiles of the Karroo formation. 4. A new pseudosuchian from the Orange Free State. Ann. Trans. Mus., 5, 83-87. WALKER, A.D., (1970). A revision of the Jurassic reptile Hallopus victor with remarks on the classification of croco- diles. Phil. Trans. R. Soc. Lon d., B, 25 7, 3 23-3 72. - '-------- (1972}. New light on the origin of birds and crocodiles. Nature, 237, 257-263. --------- (1974}. Evolution, organic. McGraw-Hill Yearbook of Science and Technology, 177-179. WENZ, S. (1968}. Contribution a I' etude du genere Metriorhynchus - crane et moulage endocranien de Metriorhyn­ chus superciliosus. Ann. de Paleontologie (verts.), 54, 149-183. WEVER, E.G. (1979). The Reptilian Ear. Princeton University Press. WHETSTONE, K.N., and MARTIN, L.D. (1979}. New look at the origin of birds and crocodiles. Nature, 279, 234-236.