A small new basal crocodyliform, Fruitachampsa callisoni gen. nov., sp. nov., is represented by several partial skeletons from the Morrison Formation at the Fruita Paleontological Area near Grand Junction, Colorado. It is placed in the Family Shartegosuchidae Efimov, 1988, previously comprising three genera from the Late Jurassic locality of Shar Teeg in western Mongolia and possibly a fourth genus from the Early Cretaceous of Siberia. Shartegosuchids share a sculpted palatal surface of the pterygoids, the absence of a mandibular fenestra, and posterior maxillary teeth and post-caniniform dentary teeth with a flat and horizontal apical region and vertical crenulations extending basally from it. Fruitachampsa and Shartegosuchus form a clade supported by ventral half of the lacrimal tapering ventroposteriorly, sculpturing on palatines, and lower teeth absent anterior to caniniforms. The shartegosuchids are most parsimoniously considered to be outside of the mesoeucrocodylian clade and are possibly allied with the Asian taxa Shantungosuchus, Sichuanosuchus, and Zosuchus. Fruitachampsa is unusual in possessing a series of small protuberances along the occipital margin of the parietal and squamosal and procoelous vertebrae, and lacking an antorbital fenestra or fossa. This is the first occurrence of a shartegosuchid in North America, and the close relationship of Fruitachampsa with Shartegosuchus nested among other Asian taxa indicates it dispersed to North America from Asia.
An unusual new crocodyliform from the Late Jurassic Morrison Formation of western Colorado is known from several partial skeletons and articulated skulls collected from a single locality, making it one of the best known Mesozoic terrestrial crocodyliforms from North America. Although it has several features indicating a derived position within crocodyliforms – the absence of an antorbital fenestra, a palatine secondary palate, and procoelous vertebrae – its affinities clearly lie with more basal taxa, especially the Asian Shartegosuchidae. Below I present a detailed description of this new taxon and discuss its relationships with other crocodyliforms.
Elsewhere I presented phylogenetic analyses of the Crocodyliformes (Clark, 1986; Clark in Benton & Clark, 1988; Clark, 1994), including the taxon described below (referred to in previous publications as the ‘Fruita form’). These analyses indicated that it is one of the most primitive members of the Mesoeucrocodylia. More recently, the relationships of basal mesoeucrocodylians have been explored by Diego Pol (Pol, 2003; Pol & Norell, 2004a, b; Pol & Gasparini, 2009), whose analyses place it in a slightly more basal position outside of Mesoeucrocodylia, due to the inclusion of Hsisosuchus (see also Wu, Li & Li, 1994; Wu, Sues & Dong, 1997), a taxon not included in my analyses. In their description of Neuquensuchus universitas from the Late Cretaceous of Argentina, Fiorelli & Calvo (2007) used a modification of Pol's matrix in an analysis that found the Fruita form to be slightly more basal, grouped with several Asian taxa and Neuquensuchus. None of these analyses included the Shartegosuchidae of Asia, a poorly known group of Late Jurassic small crocodyliforms (Efimov, 1988). Below I present a new analysis including these forms, which places the new taxon with this family and places the family in a position similar to the one Fiorelli & Calvo (2007) found for the Fruita form.
The fossils described below are from the Fruita Paleontological Area (FPA) in Mesa County, Colorado (Clark, 1985). Large dinosaurs, most notably the holotype of Brachiosaurus altithorax, have been collected from the Morrison Formation in this area since the turn of the last century (Riggs & Farrington, 1901). Small vertebrate fossils were first found in the FPA by G. Callison and the author in 1975. The initial discovery, an articulated series of cervical vertebrae of the new crocodyliform (LACM 115737), is described below along with further, more complete remains collected during 1976, 1977, and 1979 under Callison's supervision. The localities and their immediate surroundings were set aside as a protected area by the Bureau of Land Management in 1978 (Armstrong & Kihm, 1980). Excavations later continued under the auspices of Dinamation and the Carnegie Museum of Natural History (see Kirkland, 1997, 2006; Luo & Wible, 2005). This taxon has been referred to informally in several publications (e.g. Chure et al., 1998; Foster, 2007) but has never been formally named and described. Other small crocodylomorph remains from the FPA were identified as belonging to the enigmatic taxon Macelognathus and referred to the Sphenosuchidae (Göhlich et al., 2005).
Terminology, except where noted, follows Iordansky (1973) for cranial features. Positional terms used in limb descriptions, such as lateral and ventral, refer to the limb of a crocodylian in its position at the beginning of a step cycle when the humerus is perpendicular to the body and the femur is extended anteriorly nearly parallel to the body.
Abbreviations: AMNH, American Museum of Natural History; FPA, Fruita Paleontological Area, Mesa County, Colorado; IVPP, Institute of Vertebrate Paleontology and Paleoanthropology, Beijing; LACM, Los Angeles County Museum of Natural History, California; MCZ, Museum of Comparative Zoology, Harvard University, Cambridge, MA; PIN, Paleontological Institute of the Russian Academy of Sciences, Moscow; UCMP, University of California Museum of Paleontology, Berkeley; USNM, National Museum of Natural History.
Discussion: Crocodyliformes are diagnosed by a large suite of characters (Clark, 1994), some of which were recently discovered in Junggarsuchus sloani, hypothesized to be the sister taxon of Crocodyliformes (Clark et al., 2004). All non-mesoeucrocodylian crocodyliforms are sometimes united in the taxon Protosuchia, and the monophyly of this group has been supported by some analyses (e.g. Wu et al., 1997) but not others (Clark, 1994; Pol & Norell, 2004a, b; Pol & Gasparini, 2009), including the one below.
Distribution:Shartegosuchus, Nominosuchus (two species), and Adzhosuchus are from the Tsagaantsav and Ulan Malgaint beds of Shar Teeg, western Mongolia, considered to be Tithonian in age (Gubin & Sinitza, 1996). Kyasuchus saevi (Efimov & Leschinskii, 2000) from the Aptian–Albian Ilek Formation at Shestakovo, Kemerovo Province, western Siberia, may be referable to this family. Fruitachampsa is from the Morrison Formation of Fruita, western Colorado, from a part of the formation considered to be Kimmeridgean or possibly early Tithonian or even Oxfordian in age (Kowallis et al., 1998; Trujillo, Chamberlain & Strickland, 2006). Shartegosuchids are also present in the Oxfordian upper part of the Shishugou Formation of Xinjiang, China (Clark & Xu, 2008).
Revised diagnosis: Very small crocodyliforms with palatal surface of the pterygoids sculpted, mandibular fenestra absent, posterior maxillary teeth and post-caniniform dentary teeth with flat, horizontal cusp and vertical crenulations extending proximally.
Etymology: From Fruita, Colorado, the town nearest the type locality; and champsi, Greek for crocodile (derived from the Egyptian language).
Type species:F. callisoni.
Diagnosis: As for species.
F. callisonisp. nov.
Etymology: For George Callison, who led the expedition that discovered the Fruita microvertebrate locality and directed the recovery of its fossils.
Holotype: LACM 120455a, a skull lacking the ventral braincase and portions of the occiput and left temporal regions; and the anteriormost four cervical vertebrae. A postcranial skeleton is associated with the holotype but is intermixed with bones of a smaller individual, numbered LACM 120492. The holotype is restricted to the skull and the vertebrae in articulation with it in order to ensure its integrity.
Type locality: Quarry 4, FPA, south of Fruita, Colorado.
Age: Samples for radiometric dating from the FPA were inconclusive but 40Ar/39Ar dates from the upper part of the Morrison Formation elsewhere range from 148.1 ± 0.5 to 150.3 ± 0.3 Ma (Kowallis et al., 1998). These dates correspond to the Kimmeridgean to the early part of the Tithonian marine stages (Gradstein, Ogg & Smith, 2005). Recent 206Pb/238U dating of detrital sanidines from a fossiliferous stratum in the upper Morrison Formation of south-east Wyoming at 156.3 ± 2 Ma place it in the Oxfordian (Trujillo et al., 2006), older than indicated by biostratigraphy.
Referred material: LACM 115726, associated skull and skeletal fragments; LACM 115737, articulated cervical vertebrae, limb material, and posterior part of skull; LACM 115745, maxilla, premaxilla, and limb bone fragments; LACM 120439, 1/2 vertebra; LACM 120456, skeletal and skull fragments; LACM 120480, osteoderm; LACM 120481, metapodials; LACM 120483, skull and skeletal fragments; LACM 120486, vertebra; LACM 120487, maxilla; LACM 120491, premaxilla and dentary fragments; LACM 120492, associated skeleton and skull fragments on the same blocks with LACM 120455; LACM 120494, partial skull; LACM 128218, dentaries and limb fragments; LACM 128284, posterior part of a palate; and 128306, nearly complete but dorsoventrally crushed skull with mandible, associated with postcranial fragments.
Diagnosis: A relatively large shartegosuchid (midline length of largest skull approximately 9 cm) distinguished by mandibular symphysis with a transverse groove on its ventral surface, no antorbital fenestra or fossa, an anteriorly reduced lacrimal that contacts a lateral extension of the nasal, a horizontal longitudinal ridge on the medial surface of the jugal articulating with the ectopterygoid, four small protuberances along the occipital margin of the parietal and squamosal, and all post-atlantal vertebrae procoelous.
DESCRIPTIONS AND COMPARISONS
General features of the skull
In dorsal view (Figs 1, 4A), the post-orbital part of the skull is nearly square, converging slightly posteriorly, and the anterior part of the skull is roughly triangular (the narial region is completely preserved). The specimens are all smaller than the adults of modern crocodylians – the average skull size suggested by the specimens is 7–9 cm in length (see discussion under ‘Ontogenetic stage’).
The preorbital region is short and high, not long and vertically compressed as in most living crocodylians. It is not as high as in Protosuchus (Crompton & Smith, 1980; Clark, 1986), however. The rostrum accounts for approximately one-third of the total length the skull. Unlike the condition in many primitive crocodylomorphs, the preorbital region gradually widens posteriorly without a distinct change in breadth at the anterior edge of the orbits. The shape of the preorbital region (and the large palpebral bones) in Fruitachampsa are reminiscent of the extant alligatorid Paleosuchus, except for the flatter narial region in the latter.
The post-temporal fenestrae are not preserved. Antorbital fenestrae are not present because the maxilla contacts the lacrimal broadly on the type specimen.
The entire extent of the external narial openings is not preserved on the holotype, but the preserved posterior border of the nares on LACM 120483d (Fig. 2) and LACM 128306 (Fig. 4A) clearly demonstrates that the openings were anterodorsally oriented, being higher posteriorly than anteriorly. The nares are separated by a process from the anterior end of the nasal, preserved on LACM 128306 (Fig. 4A), but the dorsal process of the premaxilla that presumably met them is not preserved.
The orbits are laterally directed (Fig. 2), covered dorsally by two large palpebrals, as in most basal crocodyliforms. Compared with the typical crocodylian condition in which the orbits are dorsolaterally directed, the lateral orientation in Fruitachampsa is due to a more vertical orientation of the maxilla and lacrimal anteriorly and of the quadrate, quadratojugal, and jugal posteriorly. The frontal is not laterally expanded in comparison with forms with dorsally directed orbits (Crocodylus, for example).
The internal choana is incompletely known, as the holotype preserves only its anterior edge (Fig. 3) and LACM 128306 preserves its anterior and lateral edges (Fig. 4C). The anterior border is formed by the palatine without any participation of the pterygoid. The choana was not unusually large or small, being approximately half the width of the palatal secondary palate. Posteriorly it is confluent with a wide groove in the pterygoids, and the posterior end of this groove does not reach as far ventrally as the remainder of the opening. This groove appears to be slightly deeper than that found in the shartegosuchid Nominosuchus matutinus, and similar to the groove in an Edentosuchus-like protosuchid from the Kayenta Formation, Arizona (Sues, Clark & Jenkins, 1994). There is no evidence of a bony septum of the pterygoid dividing it along the midline, as in many crocodylians and in Shartegosuchus (Efimov, 1988). The straight lateral edge of the palatine does not extend dorsomedially to enclose the narial passage as in crocodylians. A dorsal sheet of bone within the internal choanal opening of the holotype is probably part of the pterygoid, and in the anterior part of the orbit the pterygoid can be seen lying dorsal to the palatine and, presumably, the narial passage.
An incisive foramen was apparently present between the premaxillae and the palatal portions of the maxillae, as the premaxillae do not meet posteriorly. However, the anterior part of the premaxillae is not preserved on the holotype and is not exposed on LACM 128306.
A large, circular palatal fenestra is present in the anterior part of the palate in the holotype (Fig. 3) and its posterior border is evident on LACM 128306 (Fig. 4C). A similar fenestra is present in the shartegosuchids Shartegosuchus and Nominosuchus matutinus and in Zosuchus, and the fenestra of Fruitachampsa is nearly circular, like that of Zosuchus, unlike the more anteroposteriorly elongate fenestra of the other taxa and the much smaller fenestrae in this position in some notosuchids. A palatal vacuity that is not bordered posteriorly is found in Sichuanosuchus and in an Edentosuchus-like protosuchid from the Kayenta Formation (Sues et al., 1994). In Fruitachampsa the fenestra is bordered laterally by the posteromedial borders of the maxillae, posteriorly by the anterior border of the palatines, and anteriorly by the secondary palatal portion of the maxillae. It lies between the positions of the orbits and the notch between the maxillae and premaxillae. It is separated from the maxillary tooth row laterally by a medial shelf on the maxilla, about equal in breadth to the dentigerous portion, also preserved on an isolated maxilla (LACM 120487) with a smooth medial edge on its palatal portion.
The suborbital fenestra on the holotype is bordered anteriorly by the palatine, posteriorly by the ectopterygoid, and laterally by the maxilla and ectopterygoid, although the latter is incomplete (Fig. 3). The posterior border of the fenestra is slightly more ventral than the anterior end, giving the fenestra a slight anterior orientation. The fenestra is shorter than the orbit, extending from the orbit's posterior edge to the level of the descending process of the prefrontal, just short of the anterior edge of the orbit.
The supratemporal fenestra passes ventrally from a wider fossa. The diameter of the supratemporal fenestra lateral to the braincase is approximately one-third that of the orbit in the holotype and LACM 120494 (Fig. 1). It is relatively round in dorsal view but is slightly wider than long. It is bordered posteriorly by the quadrate, medially by the laterosphenoid and the parietal, and laterally by the postorbital and a small anterior part of the squamosal. The fossa is roughly teardrop shaped; it narrows posteriorly as the squamosal broadens and the lateral part of the postorbital narrows, and its anterior breadth is due to the transverse narrowness of the parietal and the part of the laterosphenoid posterior to the capitate process, which brings these bones within a parasagittal plane. A shelf in the posterior part of the fossa formed by the parietal and squamosal is broadly exposed and is not covered dorsally by a shelf from the squamosal.
The infratemporal fenestra of the holotype is misleading because the quadratojugal is only partially preserved. As preserved it is triangular with a horizontal base, a vertical anterior edge formed by the postorbital bar, and a hypotenuse formed by the quadratojugal (Fig. 2). However, the quadratojugal of LACM 120494 is broader, and although it is incomplete anteroventrally and dorsally it probably constricted the fenestra to a small opening, as in Protosuchus. The lateral temporal region is vertically oriented so that the fenestra opens laterally. Because the quadratojugal contacts the postorbital, the quadrate does not border on the fenestra.
Bones of the skull
A vomer was not found on any specimen, but LACM 128306 is not prepared in this area. In comparison with other crocodylomorphs it should be situated immediately dorsal to the palatal fenestra of Fruitachampsa, but no bone was found in this region of the holotype. A separate dermo-supraoccipital bone is absent from the three skulls preserving this region. As in all other known crocodylomorphs a postfrontal is absent.
The vertically oriented premaxilla is sculptured on its external surface, gradually becoming smoother ventrally (Figs 2, 4A). The tooth-bearing lateral portion of the bone descends lower than the medial palatal portion. The palatal portion of the bone is apparently unsculptured. At least four tiny foramina lie medial to the bases of the teeth on the palatal portion. On the lateral skull surface, the posterior edge forms with the maxilla a laterally open notch shaped like an inverted ‘U’, and a dorsal lip overhangs the edges of this notch. A foramen between the premaxilla and maxilla connects the notch to the nasal cavity dorsomedially. An anterior process of the maxilla separates the premaxillae from meeting on the palate; the premaxillae almost certainly met at the anteriormost end of the palate although this region is not preserved. The premaxilla contacts the nasal dorsomedially and the maxilla posteriorly and forms the entire posterior border of the nares. The anteroventral portion of the bone is only preserved in LACM 128306, but it is not completely exposed; it is very slender beneath the lateral part of the narial opening and tapers medially.
The maxilla is vertically oriented and curves dorsomedially. The two bones meet anteriorly along the midline of the palate but not posteriorly, and together they form the anterior edge of the palatal fenestra described above. The maxilla is sculptured heavily on its lateral surface and faintly on its palatal surface. The sculpturing is exceptionally rugose on the dorsal part of the lateral surface, especially immediately lateral to the suture with the nasal. The maxilla meets the palatine posteriorly lateral to the posterior part of the palatal fenestra, and the palatines overlap the maxillae dorsally in this articulation. This suture is straight ventrally, extending posterolaterally from the fenestra. The lateral portion of the maxilla meets the premaxilla anteriorly to form the laterally open notch described above. Within this notch open two small fenestrae, presumably neurovascular, that extend posterodorsally to the narial passage. The articulation with the premaxilla extends dorsally from the notch to meet the nasals. A bulge is present over the larger teeth in the middle of the maxillary tooth row on large specimens (LACM 120487, 128306). The ventral border of the lateral edge is straight in a moderate-sized specimen (the holotype) but is gently convex downward in larger specimens (LACM 120487, 128306). The maxilla narrows abruptly in the vertical plane at the anterior edge of the orbit, where it forms the anterior half of the ventral border of the orbit and articulates with the lacrimal. The jugal broadly overlaps the posterior end of the maxilla although both bones are very narrow dorsoventrally in this region. The posterior end of the maxilla beneath the orbit is apparently edentulous. The dorsomedial articulation with the nasals is straight and parasagittal except for a slight extension of the maxilla medially anteriorly. The maxilla does not meet the prefrontal posterodorsally. Between the palatal fenestra and the tooth row is a narrow palatal shelf that the tooth row overhangs laterally, so that the tooth row is situated somewhat ventral to the palatal portion. An anterior midline process of the maxilla separates the posterior parts of the premaxillae on the palate. There may be a posteriorly open pocket, for a paranasal diverticulum (Witmer, 1997), in the posterior part of the bone medially along the narial passage (LACM 120487) similar to those seen in crocodylians.
The nasal extends posteriorly to meet the frontal at the level of the anterior edge of the orbit. The sutures are obscured on both the holotype and LACM 128306 but are preserved on LACM 120494. The posterior part of the nasal broadens towards the anterior end of the orbit, then narrows for a short distance between the orbits. It extends laterally to contact the medial half of the anterior end of the lacrimal and the entire anterior edge of the prefrontal. The nasal is sculptured over its dorsal surface, and the sculpturing is longitudinally oriented. The nasal narrows anteriorly at the level of the suture between the premaxilla and maxilla. The anterior end of the nasal, preserved only on LACM 128306, narrows to one-third of its former breadth to form an unsculpted anterior process dividing the nares along the midline.
The prefrontal articulates with the frontal posteriorly and medially, the lacrimal ventrolaterally, and the nasal anteromedially. Its dorsal surface is sculptured evenly, without any ridges or a ‘spectacle’ as in many caimans. It is relatively short anteriorly, with a smoothly arched anterior edge in articulation with the nasal medially. It extends ventrally a short distance in a manner similar to the prefrontal pillar of crocodylians, but this ventral projection does not reach the palatine in any specimen. The ventral edge of this ‘pillar’ is horizontal and straight and has a small process extending posteriorly from it. This ventral portion is transversely flattened and lies in a parasagittal plane, unlike the crocodylian condition in which the pillar is transversely oriented. Although it may be an artifact of preservation, a similarly shaped ‘pillar’ is found in Protosuchus (Clark, 1986).
The lacrimal is vertically oriented ventral to the lacrimal canal, but the bone becomes horizontal dorsally (Figs 1, 2). It contacts the nasal anteriorly, the maxilla anteriorly and ventrally, and the prefrontal dorsomedially. It does not contact the jugal posteriorly. A large, vertically oval lacrimal canal is enclosed within the lacrimal; the canal is somewhat rounder in a large specimen (LACM 120494) than in the holotype. There are tiny pits on the floor of the canal in one specimen (LACM 115745a) but these may be due to weathering. A bulge in the lacrimal lies directly over the lacrimal canal dorsolaterally, forming the base of a shelf that the palpebral is preserved resting on in the holotype. The lacrimal extends posteriorly along the ventral margin of the orbit a very short distance to the level of the sixth maxillary tooth in this specimen. The bone is short anteriorly (Figs 2, 3) and the anterior edge is smoothly convex. The bone is sculptured externally with pits but not grooves.
Separate anterior and posterior palpebral bones are present on the holotype and LACM 128306 and lie above the orbits (Figs 1, 4A). They lie horizontally and are sculptured dorsally. The anterior palpebral is the larger of the two and is teardrop shaped with a posterolateral ‘tail’ on the holotype, but the tail area is much broader on the larger LACM 128306. The entire bone is arched dorsally slightly on the holotype. Its lateral border is parallel to the lateral edge of the maxilla, its anterior and medial edges follow the outline of the orbit, and it is posteromedially concave. The posterior palpebral is flat and triangular with its apex directly posterior to the ‘tail’ of the anterior palpebral and its base resting on the postorbital in the posterior dorsolateral corner of the orbit.
An isolated osteoderm may be a left anterior palpebral, although its anterior edge is not as smoothly convex as those of the holotype and LACM 128306. It is unlikely to be a dorsal osteoderm as it has no area for imbrication anteriorly and has a complex ventral surface. LACM 120480 (Fig. 5) is finely pitted over the entire area of its slightly concave dorsal surface. A process is present on the presumed anterolateral edge of the bone, and it is sculptured dorsally. A transverse ridge evenly divides the ventral surface into lateral and medial halves. The lateral half has many mediolateral grooves. The medial half is smooth except for a ridge directly beneath the anterior process that meets the transverse ridge at a right angle. The posterior and lateral edges of the osteoderm are straight, and the incomplete medial edge is concave.
The jugal articulates dorsally with the postorbital and forms the ventral half of the postorbital bar (Fig. 2). It is sculptured on its lateral surface, including the ascending process. The dorsal ascending process is much longer anteroposteriorly than it is transversely wide; it is slightly concave laterally but is not distinctly set off from the ventral part of the bone. It is dorsally oriented with little, if any, posterodorsal inclination. This process originates broadly and tapers slightly dorsally, and its anterior edge is slightly thicker than its posterior edge. The articulation with the postorbital is overlapping, with the jugal lateral to the postorbital as in crocodylians (vertical striations are present on an articulating surface of an isolated jugal, LACM 120456c).
The ventral border of the jugal is horizontal. The posterior end may descend ventrally with the quadratojugal, but this area is poorly preserved. In dorsal view the lateral border is slightly convex laterally. The horizontal ventral portion is very thin transversely, and in cross-section it is slightly convex laterally along its entire length but does not form a sharp ridge as in Sichuanosuchus. There is a ridge on the medial surface of this horizontal portion that divides it into roughly equal dorsal and ventral halves. This ridge thickens anteriorly and articulates medially with the ectopterygoid; this suture is strictly horizontal and the ectopterygoid does not ascend the postorbital bar. Anterior to this contact the ridge tapers until it is absent at the articulation with the maxilla.
The jugal does not extend to the anterior end of the orbit. It tapers gradually anteriorly and does not expand vertically as in sebecosuchians and crocodylians. The jugal articulates with the quadratojugal posteriorly, but details of this region are poorly preserved. The anterior process of the horizontal portion is approximately 50% longer than the posterior process.
The frontals are fused in larger individuals but paired in the smallest specimen (LACM 120492d). In the latter the midline articulation is straight and the dorsal surfaces of the bones are flat. In the holotype (Fig. 1) and the largest cranial specimen (LACM 120494 and 128306; Fig. 4A) the frontals are fused and are slightly concave dorsally, with their lateral edges raised above the orbits. In LACM 128306 a low, narrow ridge is present along the midline of the skull posteriorly but not anteriorly. The frontal articulates with the nasal anteriorly by a strongly interdigitating suture at the level of the narrowest interorbital width. It is sculptured dorsally in a very fine pattern, much less coarsely than on the palate or on crocodylian frontals of similar size. The interorbital width is relatively narrow, similar to that of Osteolaemus (Iordansky, 1973). The fronto-parietal articulation is strongly interdigitating, lying at the level of the anteriormost extent of the supratemporal fenestra; the frontal participates in this fenestra as a tiny segment in the anteromedial corner. The frontal does not extend far laterally (less than one-quarter of the way across the bar separating the orbit from the supratemporal fenestra), but the exact position of the suture with the postorbital is obscure. The frontal articulates ventrolaterally with the laterosphenoid along the posterior part of the crista cranii frontalis.
The parietals are fused, and the combined bone is sculptured over its dorsal surface except in the supratemporal fossa (Figs 1, 4A). The sculptured portion is relatively broad and lies along the midline with roughly parallel lateral edges; this region narrows in the middle and broadens at a 45 ° angle in the posteriormost part of the fossa. This portion is approximately one-third to half as broad as the entire parietal. The parietal articulates with the squamosal laterally, the supraoccipital posteroventrally, the frontal anteriorly, and the quadrate and laterosphenoid anteroventrally; it does not appear to articulate with the postorbital. Delicate posteroventral struts dorsal to the transverse canal through the supraoccipital may be present but are not exposed. The parietal forms the medial edge of the supratemporal fossa and the medial half of a broad shelf in the posterior part of the fossa.
The postorbital articulates with the frontal medially, the squamosal posteriorly, the quadratojuqal and jugal ventrally, and the laterosphenoid ventromedially (Figs 1, 2, 4A). It is sculptured dorsally and probably laterally (on the descending process), although this is poorly preserved. It overhangs the postorbital bar laterally and anteriorly. The medial process, which separates the orbit from the supratemporal fenestra, is unusually narrow laterally, as in atoposaurids, Sichuanosuchus, Shartegosuchus, Nominosuchus matutinus, and Zosuchus; this process expands medially in LACM 120494 to meet the frontal. The posterior edge of this medial process is apparently very straight. The descending process of the postorbital forms the dorsal part of the postorbital bar; it is transversely flattened and does not meet the ectopterygoid ventrally. There is no evidence for a foramen on the dorsolateral surface of the postorbital bar (as in many eusuchians). A short descending process extends from the dorsal end of the bar posteroventrally to meet the quadratojugal so that the ventrolateral part of the postorbital is shaped as an inverted ‘V’ in lateral aspect. The posterodorsal part of the bone is poorly preserved but the contact with the squamosal is relatively narrow. The postorbital probably contacted the quadrate posteroventrally, but if so the contact is not as broad as in other crocodylomorphs due to the large fenestra in the dorsal part of the quadrate.
The squamosal articulates with the parietal medially, the postorbital anteriorly, the exoccipital posteroventrally, the supraoccipital ventromedially, and the quadrate anteroventrally and posterolateroventrally (Figs 1, 2, 4A); a possible contact with the opisthotic ventromedially is not exposed. The articulations with the parietal and postorbital are obscured, but they do not appear to be unusual. The articulation with the quadrate anteriorly is apparently not as solid as in crocodylians as there is a fenestra in the dorsolateral edge of the quadrate that extends to the squamosal.
The posterior edge of the left squamosal and parietal on the largest cranial specimen, LACM 128306, preserves a series of five small, sculptured protuberances along the occipital margin (Fig. 4D). This area is incompletely preserved in other specimens, and they are absent from the right side, suggesting they are not firmly attached, possibly osteoderms that become fused during ontogeny. The medialmost protuberance is near the midline on the parietal and has a smooth posterior edge, and a protuberance from the lateral part of the parietal's occipital border is nearly triangular. Two small protuberances lie on the medial part of the squamosal's occipital border, and a larger process projects laterally from the posterolateral edge of the squamosal. The lateralmost of these protuberances is comparable in position to a larger process on the squamosal of a few, much larger, crocodylians, such as Voay robustus from Madagascar (Brochu, 2007), but the latter is clearly an extension of the squamosal.
The posteroventrolateral articulation of the squamosal with the paroccipital process and quadrate is apparently broad, although it is not completely preserved. The squamosal meets the quadrate on both sides of an opening in a position similar to that of the cranioquadrate passage of Mesoeucrocodylia, as inferred from the isolated distal portion of the left quadrate of the holotype. The presence of an enclosed cranioquadrate passage cannot be confirmed with the present material, but it is possible that this mesoeucrocodylian synapomorphy has a broader distribution. The dorsal surface of the squamosal is horizontal with a slightly raised lateral edge (the squamosal of the holotype is deformed in that the lateral edge has been moved ventrally). The dorsal part of the bone is very delicately constructed internally; it is thicker than a single layer of bone and is composed of thin vertical laminae. The squamosal forms a roof extending laterally over the area dorsal to the quadrate as in other crocodylomorphs. It is sculptured dorsally except in the supratemporal fossa, and the sculptured portion roughly forms a right triangle; the base of the triangle lies laterally, the posterior edge is perpendicular to it, and the remaining edge lying along the supratemporal fossa is oblique. The sculptured portion of the posteromedial part of the squamosal is narrower anteroposteriorly than the unsculptured part of the bone in the supratemporal fossa. The articulation with the paroccipital process of the exoccipital on the occiput is relatively straight, extending transversely across the occiput, but the squamosal extends slightly further laterally than does the paroccipital process and so is much broader vertically at its lateral edge. The lateral edge of the squamosal has a groove similar to that found in crocodylians for the attachment of the external ear musculature (Shute & Bellairs, 1955).
The quadratojugal is not completely preserved on any specimen, but most of the right element is preserved on LACM 120494. It is very broad throughout most of its extent, and its articulation with the postorbital dorsally is broad (Fig. 2). If it was like Protosuchus it would have articulated along the entire length of the descending process of the postorbital, but this is not preserved. It is inclined posteriorly along with the quadrate but is otherwise nearly vertical without extending laterally. The ventral portion is poorly preserved but on LACM 120455 it is visible flaring laterally, similar to Sichuanosuchus shuhanensis and Zosuchus. Unlike the quadrate, which faces posterolaterally, it faces laterally and meets the quadrate at an angle along their contact. It forms the anterolateral border of a large pocket in the dorsal surface of the quadrate and therefore does not meet the quadrate dorsally in this region. It is sculptured near the articulation with the jugal but not elsewhere. It does not appear to participate in the articulation with the mandible. It articulates with the jugal and the quadrate posteroventrally; this region is poorly preserved but does not appear unusual in comparison with crocodylians.
The quadrate is typically crocodylian in being extensively sutured to the squamosal, parietal, laterosphenoid, and pterygoid and in buttressing the lateral part of the braincase. It differs from adult crocodylians in being virtually hollow except for delicate internal struts, and in extending ventrally so that its ventral surface is roughly horizontal. The dorsolateral surface of the bone is highly fenestrated, possibly homologous to a small siphonial foramen in the dorsal part of the quadrate of crocodylians.
The quadrate fenestrae, best preserved on LACM 120494 (Fig. 6), are similar in many respects to those described for Protosuchus (Hecht & Tarsitano, 1983; Clark, 1986) and Shantungosuchus (Wu et al., 1994). The quadrate of Protosuchus has five fenestrae on the dorsal surface: a large vertically oriented fenestra located posteriorly and distally (fenestra A in the terminology of Hecht & Tarsitano, 1983), three smaller fenestrae located in a vertical line anterior to the first fenestra (fenestrae B–D, in ascending order), and a large fenestra situated dorsally near the squamosal suture that contains several pockets within it. In Fruitachampsa the fenestrae are similar except that fenestrae B–D are apparently coalesced into a single large, subdivided fenestra and fenestra E is somewhat reduced in size.
Fenestra A in Fruitachampsa is vertically elongate, extending dorsally from immediately above the level of the condyle, and is subdivided by at least two horizontal struts. It is somewhat narrower than the fenestra illustrated for the holotype of Protosuchus richardsoni (Hecht & Tarsitano, 1983), which has only a single strut, but other specimens of Protosuchus have a similarly shaped fenestra with multiple struts (Clark, 1986).
The large fenestra equivalent to fenestrae B–D lies anterodorsal to fenestra A and may extend dorsally to the squamosal. It is roughly oval in posterodorsal view with a vertical long axis. It is bordered anteriorly by the quadratojugal so that most of the posterior border of this bone does not meet the quadrate. This fenestra is subdivided by several struts into dorsal and ventral parts of equal size. The dorsal part is connected with fenestra E through at least one foramen, and there is probably similar communication with fenestra A. Fenestra E is situated at the dorsal border of the quadrate at the squamosal contact. It is roughly triangular in shape with equilateral sides and a ventral apex. There are no obvious pockets within it but it has not been completely prepared.
Because the fenestrae are relatively large the quadrate itself is extremely delicate. The three fenestrae and the middle ear cavity posterior to the quadrate are situated in quadrants such that the dorsal surface of the bone consists of struts shaped like a narrow ‘X’.
The position of the tympanum in primitive crocodylomorphs such as Fruitachampsa and Protosuchus may have been much larger than in crocodylians. The fenestrae on the dorsal surface of the quadrate in crocodylians open into the middle ear cavity medial to the tympanum; were this the case in these fossil taxa the tympanum would be unusually large, and in Fruitachampsa it would attach to the quadratojugal anteriorly. The tympanum may therefore have been supported by non-osseous tissue in places.
The trochlear surface of the quadrate resembles those of Sebecus and Pristichampsus in being triangular with a dorsal apex (Langston, 1973). An isolated distal quadrate from the holotype demonstrates that the quadrate was sutured to the squamosal and paroccipital process at the dorsal apex of this surface (i.e. the cranioquadrate passage was closed laterally). There is a small opening corresponding to that for the siphonial system of crocodylians medial to this suture on the dorsal surface of the quadrate. The orientation of the condyles on the trochlear surface is generally similar to that of crocodylians. The lateral condyle is larger and faces posteriorly (as determined with the isolated specimen). The lateral condyle faces slightly laterally and is roughly twice the size of the medial condyle.
The sutures between the quadrate and the pterygoid, parietal, and laterosphenoid are not well preserved but are similar in position to those in crocodylians. The quadrate forms the posterior border of the supratemporal fenestra. A ridge is developed on the anterolateral edge of its ventral surface, but there is no evidence of the mandibular adductor muscle scars typical of crocodylians (Iordansky, 1973).
The epiotic is very similar to that of crocodylians in being fused to the supraoccipital and hollow with vertical struts that rise dorsally to meet the parietal. In extant crocodylians a pneumatic diverticulum passes through this transverse pneumatic canal connecting the otic regions. The struts are remarkably delicate; they are much thinner than are their crocodylian homologues. The epiotic articulates laterally with the prootic in front and the opisthotic behind, and with them it forms the roof of the otic capsule. The opisthotic contact continues posteromedially from the capsule to the occiput. The opisthotic and prootic articulations form ridges on the dorsal surface of the capsule. The ridge formed with the prootic gives rise anteriorly to one of the vertical struts dorsally. Medial to this ridge is a depression in the roof of the otic capsule that lies between the ridges previously mentioned. A ridge that contained the anterior semicircular canal runs anteroposteriorly in the anterior part of the depression. The posterior part of the depression is divided into a medial and a lateral part by a posteroventral-to-anterodorsal ridge that anteriorly joins the base of a stout process projecting posterolaterally. This process arches over the posteromedial roof of the otic capsule and joins the opisthotic. The base of this process gives rise to an anterodorsal process that flattens transversely and broadens longitudinally dorsally, forming a vertical septum dorsally. A small process also extends dorsolaterally from the base of the large process, but its distal end is not preserved. Medial to the otic capsule, the epiotic forms the roof of the posterior part of the braincase, which is also the floor of the transverse canal between the otic regions. In this region the epiotic forms a broad longitudinal ridge along the midline that rises gradually anteriorly so that the depressions on the dorsal surface of the horizontal otic capsule are deeper anteriorly than posteriorly. An extremely slender process arises from the dorsal surface of the lateral edge of this broad ridge in the anterior part of the epiotic. The articulation of the epiotic with the posteroventral part of the parietal is not visible, so it is impossible to determine if the parietal forms part of this latter process.
The occiput is poorly preserved on the holotype and LACM 128306 but the occipital surface of the supraoccipital was apparently flat and faced somewhat posterodorsally. The supraoccipital apparently does not extend onto the dorsal surface of the skull.
The opisthotic forms the posterior part of the otic capsule. In general it is similar to this element in crocodylians except that the ridges on the dorsal surface of the otic capsule are formed with the epiotic. A space separates the ventral part of the opisthotic and the anterior surface of the occiput, but much of this region is obscure.
The exoccipital is fused with the opisthotic. It extends laterally onto the occiput to meet the quadrate ventrolaterally. Dorsally and laterally it meets the squamosal along very straight contacts. The dorsal part of the paroccipital process lacks the posterior twisting present in some crocodylians. The squamosal slightly overhangs this process along its dorsal surface. At least one canal through the exoccipital (that for nerve XII) is present immediately anterior to the lateral edge of the foramen magnum.
Only the condylar portion of the basioccipital is preserved on the holotype, and only the anterior part of the bone is preserved on LACM 128306. It is so pneumatized that it appears to be virtually hollow. It articulates with the exoccipitals dorsolaterally and forms the ventral border of the foramen magnum.
The prootic is similar to that of crocodylians. The portion anterior to the lagenar recess is absent on the specimen in which this area is exposed, but it is otherwise complete. It forms the anterior part of the otic capsule; the portion forming the anterolateral roof of the capsule has a distinct depression on its dorsal and ventral surfaces formed by ridges around its periphery. The prootic is longer than this bone in crocodylians (but as long as that of protosuchids); there is a 5-mm-long lamina of bone between the trigeminal foramen and the otic capsule that is not seen in crocodylians. The prootic articulates with the opisthotic and supraoccipital, and a possible articulation with the quadrate is not exposed. The otic capsule is recessed relative to the inferred position of the tympanum on the posterodorsal edge of the quadrate, similar to its position in crocodylians but different from its more superficial position in protosuchids (Clark, 1986).
The basisphenoid is poorly preserved in both skulls. The ventral surface apparently faces posteroventrally, but it is neither as horizontal as in protosuchids nor as vertical as in crocodylians. Although its ventral extent is not entirely preserved, the preserved portion on LACM 120455a and 128306 (Fig. 5E) strongly suggests that it was not broadly exposed ventrally. Comparison with Nominosuchus matutinus suggests it could be almost completely covered by the pterygoids. Fragments of what appear to be the basisphenoid rostrum are preserved in the holotype but indicate only its probable presence.
The laterosphenoid is slightly longer than that of crocodylians, but not as long as in thalattosuchians. It forms the anterolateral wall of the braincase and the anteromedial border of the supratemporal fenestra. It articulates with the frontal anterodorsally, the postorbital dorsolaterally, and the parietal posterodorsally. The capitate process is not transversely elongate and does not extend very far laterally beneath the postorbital bar. The posterior portion of the laterosphenoid is in a parasagittal plane; it is not oblique as in juvenile stages of most crocodyliforms. The cotylar crest descends from the capitate process as in crocodylians but it is much more slender. The trigeminal foramen is not well preserved. The laterosphenoid articulates with the pterygoid ventrally but the suture is unclear. The contacts with the quadrate, basisphenoid, and prootic are not preserved or exposed.
The pterygoid is very poorly preserved on the holotype (Fig. 3) but is well preserved on LACM 128306 (Fig. 4C). The ventral surface of the bone is sculptured near the midline, where it meets the palatine, both on the palatal surface and, to a lesser degree, within the choana. The palatal portion of the bone is continuous with the sculptured portion of the palatine and is set off from the choana by a ridge. Bones that probably represent the anterior processes of the pterygoid are present above the palatines of LACM 120494; they are semicircular in cross-section and concave dorsolaterally. Their dorsal edges are horizontal in lateral aspect and extend at least the length of the orbit. They do not meet the descending process of the prefrontal, as they do in crocodylians. The pterygoids roof at least that part of the narial passage immediately dorsal to the choana. The lateral flange of the pterygoid is similar to that of crocodylians in being strongly deflected ventrally, but it differs in being relatively flat instead of curling anteriorly ventrally and smaller anteroposteriorly. It meets the ectopterygoid along the lateral and anterior edges of the flange. On the holotype, in which the mandible is articulated, the flange extends ventrally to the level of the ventral edge of the mandible and laterally nearly meets them. The quadrate ramus of the pterygoid is broad with a convex ventral surface, suggesting it is pneumatized.
The ectopterygoid is a small bone that meets the jugal laterally and the pterygoid medially and forms the posterior edge of the suborbital fenestra (Fig. 3). It is lightly sculptured ventrally at its ventromedial extreme. It is triangular in dorsomedial aspect, with the base of the triangle in articulation with the jugal laterally; it does not meet the maxilla. It articulates with the medial horizontal ridge of the jugal and does not take part in the postorbital bar. It is convex dorsally, the posterior edge is concave and the anterior edge is straight. It articulates with the pterygoid medially, and lies dorsal to it in the articulation. It does not extend the full length of the lateral edge of the pterygoid flange. The medial end is triangular in ventromedial aspect with the base along the anterior edge of the pterygoid flange.
The paired palatines are heavily sculptured over their entire ventral surface, and a suture is present along the midline (Figs 3, 4C). The ventral surface is flat, and unlike in crocodylians its lateral edge does not curve dorsally and its posterior edge does not form a ventral lip. On LACM 128306 the lateral edges curl ventrally somewhat. An expansion for the base of the prefrontal pillar is not present dorsally. The lateral edges are parallel and straight, not concave laterally as in crocodylians. The palatines articulate anteriorly with the maxillae on either side of the palatal fenestra, forming the posterior border of this fenestra. The sculpturing in the region immediately posterior to the fenestra is longitudinally oriented. The palatines form the anterior edge of the internal choana and articulate with the pterygoid lateral to it. In this articulation the palatines lie ventral to the pterygoids. The articulations are obliquely oriented and extend anterolaterally from the anterolateral margin of the choana, but the sculpturing obscures details. The palatines articulate laterally with the lacrimals anterior to the suborbital fenestrae.
Bones of the mandible
All of the mandibular bones except the coronoid and prearticular are preserved and exposed. The dentary articulates with the surangular posterodorsally, the splenial medially along its length, and the angular posteroventrally (Figs 2, 3, 4B). The symphysial region of the mandible is wide and long; its width approximates its length. The symphysial suture is straight at its anterior end and becomes increasingly deeply interdigitating posteriorly, as best seen on LACM 128218. A midline horizontal process projects anteriorly from each dentary immediately lateral to the symphysis, and there appear not to have been teeth anterior to two large caniniform teeth in the symphysial region. The midline process is bordered posteriorly by a transverse groove on the ventral surface of the dentary. Except for the caniniforms and the first two teeth posterior to them, dentary teeth are not separated by interalveolar septa. A bulge lies opposite the caniniform teeth on the ventral surface. The dentary is sculptured over its ventral and lateral surfaces. An unsculptured groove on its lateral surface lies immediately below the tooth row near its anterior end. Meckel's canal is open and relatively large, extending from the symphysis to the mandibular adductor fossa. The dentary is straight in ventral aspect posterior to the symphysis. It forms approximately half the total length of the mandible.
The surangular is strongly arched dorsally (Fig. 2). It articulates with the dentary anteriorly, the articular posteriorly, and the angular ventrally. It is sculptured on its lateral surface. The presence of a mandibular fenestra cannot be ascertained because this region is poorly preserved, but on LACM 128306 the dentary, angular, and surangular nearly close off the opening between them (Fig. 4B). This is one of the largest specimens, so closure may occur during ontogeny. The surangular apparently does not take part in the articulation with the quadrate to a significant degree. It forms the lateral and dorsal edges of the mandibular adductor fossa. A flat area lies at the anterior part of its dorsal surface, probably the site of attachment for the m. adductor mandibulae externus. The surangular appears to continue to the posterior edge of the mandible, where it descends to the ventral edge of the mandible. The posterior end of the mandible lacks a distinct retroarticular process, and instead forms a surface facing posterodorsally immediately behind the quadrate articulation.
The angular has straight ventral, lateral, and medial edges (Figs 2, 4B). It is sculptured laterally and ventrally. It forms the ventral border of the mandibular adductor fossa. Its ventral edge faces slightly medially as well as ventrally. The angular meets the dentary obliquely so that the posterior part of the mandibular rami are parallel whereas anteriorly they converge, so that the mandible is lyre shaped in ventral aspect. The posteriormost portion of the angular meets the articular posterodorsally but its ventral margin does not rise posteriorly as does the angular of crocodylians. The angular meets the surangular and articular dorsally, but the articular contact is not well exposed. The angular–surangular contact is preserved on LACM 128306 where the contact is visible approximately mid-height on the posterior mandible (Fig. 4B), unlike Zosuchus, in which the surangular is expanded ventrally and the angular is reduced. The lateral surface is sculptured, and the sculpturing is set off from the ventral edge of the bone by a longitudinal ridge. The posterior end of the angular curves posteromedially behind the quadrate articulation.
The splenial does not participate in the ventral part of the symphysis, but it may participate dorsally. It deepens vertically behind the symphysial region to form the medial wall of the mandible anterior to the adductor fossa (Fig. 2). It articulates with the dentary laterally, the angular posteroventrally, and the surangular posterodorsally. It does not form the medial border of the tooth row posteriorly as it does in some crocodylians and notosuchians. Ventrally it extends almost to the ventral surface of the mandible but does not form a significant part of this surface.
The articular is represented on the holotype only by the articulating surface hidden beneath the distal end of the quadrate and is only partially exposed on LACM 128306. The articulating surface is relatively wide for a skull of this size compared with the width of the articulating surface of crocodylians. It lies at a level approximately two-thirds the total height of the mandible because the surangular arches dorsally. The articulating surface appears to be relatively narrow anteroposteriorly as in most crocodylomorphs except some notosuchids, in which it is elongate (Clark, Jacobs & Downs, 1989). The exposed portion of the posterior mandible of LACM 128306 indicates the retroarticular process was short and ventrally directed, unlike the longer process of most neosuchians.
The anterior part of the premaxilla is not well known but four teeth are present on the preserved portion of the holotype. The anterior tooth is smallest, the second and fourth teeth are equal in size and somewhat larger, and the third tooth is somewhat larger (it is twice as long as the second and fourth on the holotype). The second tooth is slightly recurved and the others are straight. Their surfaces are smooth and lack the longitudinal ridges seen in goniopholidids, but the second and third teeth have keels on their anterior and posterior edges (LACM 120483). Serrations are not present on any tooth. In the holotype the mandible is present in articulation, and the third tooth extends ventrally below the level of the ventral surface of the dentaries. In all of the teeth of Fruitachampsa the bases are not constricted, unlike those of crocodylians.
Nine teeth are present in the complete maxilla of the holotype with the following relative diameters: 2 > 3 > 4 > 5 > 6 > 7 = 1 = 8 = 9. The second tooth is particularly large on LACM 128306, where it is approximately three times the anteroposterior breadth of the third tooth. The anterior two teeth are round in cross-section but beginning with the third tooth they are identical to the post-caniniform teeth of the dentary described below in being laterally compressed and rectangular in lateral aspect. All are essentially straight and vertical, and none appears to be recurved. The anterior maxillary teeth lack carinae or ridges. Except for the first tooth, the teeth are not set in individual sockets (i.e. interdental septae are absent) in the relatively large individual (LACM 120487).
Paired caniniform teeth are present in each dentary in the symphysial region. They are round to oval in cross-section. The anterior tooth extends dorsolaterally from the dentary and immediately becomes vertical by curving dorsally; the posterior arises vertically. Although there is space for additional teeth anterior to them, no teeth or alveoli are present; if present, they arose horizontally. The anterior caniniform is keeled on its anterior surface (LACM 120483) but keels may be lacking on the posterior edge and the posterior caniniform; ridges are not present. The teeth posterior to the caniniform (Fig. 7) are of uniform size and shape (the missing first post-caniniform may have been slightly larger than teeth posterior to it). These teeth are all rectangular in lateral aspect (they are twice as high as long) and transversely flattened. The dorsal edge (i.e. the crown) is horizontal and fine vertical ridges extend ventrally from it on the lateral and medial sides half way down the tooth.
All preserved vertebrae are procoelous and have anteroposteriorly elongate neural spines (Fig. 8). The articulating surfaces on all centra extend to the peripheral margins of the centra. Unlike crocodylian cervical and caudal vertebrae, the convex posterior articulating surfaces do not originate from the centre of flat discs, and the anterior concavities have a correspondingly reversed shape lacking a broad margin around the depression. There is no evidence in most vertebrae for sutures between the neural arches and their centra (except in the smallest specimens). ‘Spine tables’ (distal expansions of the neural spines) are not present.
The fragmentary proatlas of the holotype was shaped as an inverted ‘V’, but it is unclear whether it was entirely fused along the midline as in modern crocodylians. It articulates ventrolaterally with the prezygapophyses of the atlas. The atlas neural arches (Fig. 1) are not fused along the dorsal midline but have a midline contact that is longer than the ventral part of the arch. The postzygapophyses are well developed, horizontal, and articulate with the axis prezygapophyses. Diapophyses are not present. The atlas intercentrum (Fig. 3) is anteroposteriorly short and articulates with the neural arch dorsally along its entire length. The prezygapophyses are small and horizontal.
The odontoid process is completely co-ossified to the axis centrum and bears well-developed diapophyses and parapophyses similar to those of crocodylians. There is no diapophysis on the neural arch. A slight keel lies posteriorly on the ventral surface of the centrum, but there is no evidence of a hypapophysis. The neural arch extends the length of the centrum but its shape and orientation are not preserved. The prezygapophyses are horizontal and weakly developed, and the stout postzygapophyses are oriented at a 45 ° angle to the horizontal.
There are at least eight cervicals. There is a distinct change in orientation between the ribs of the 6th and 7th preserved vertebrae on LACM 115737, and the scapula preserved in place lies immediately lateral to this area as it does to the cervical–trunk transition in crocodylians. The anteriormost vertebra preserved on this specimen is not the axis, so there are at least eight cervical vertebrae if the above comparison is accurate. All known crocodylomorphs have nine cervical vertebrae, and this could have been the number in Fruitachampsa.
Hypapophyses are not developed on the ventral surface of cervical vertebrae, but anterior cervical vertebrae have low keels. The third vertebra preserved on LACM 115737 has an extremely long process on its ventral surface, but it appears to be pathological. The rib of this vertebra articulates with the ventral process, and its shape is distorted in comparison with other cervical ribs. Zygapophyses are uniformly oriented; they are slightly more vertical than 45 ° to the horizontal and lie dorsal to the level of the roof of the neural canal. Anteriorly situated diapophyses and parapophyses are well developed and separate, and the parapophyses remain at a constant level throughout the series. The centra are of consistent length, equal to the length of the anterior trunk vertebrae but shorter than the axis centrum. The neural spines are taller than the height of the centrum and are slightly longer than the spines of the trunk vertebrae. They are anteroposteriorly elongate throughout the series, and those of posterior cervical vertebrae are not cylindrical as in crocodylians (except Gavialis; Hoffstetter & Gasc, 1969).
The transition between the cervical and trunk vertebrae is not entirely preserved but does not appear to differ from the crocodylian pattern; the parapophyses on posterior vertebrae rise onto the transverse process with the diapophyses, and both move to the distal end of the processes on posterior vertebrae. As in crocodylians, the transverse processes of trunk vertebrae are long, wide, and flat in posterior vertebrae. They arise from near the level of the top of the neural canal. The neural spines are low and anteroposteriorly broad. The centra are slightly concave ventrally and the few specimens available are relatively uniform in size. The zygapophyses are flat in posterior vertebrae but they are inclined somewhat in anterior vertebrae. However, the postzygapophyses of the posteriormost trunk vertebrae were apparently inclined as in the anterior trunk vertebra, as indicated by the anterior sacral vertebra.
There are two sacral vertebrae and their centra are tightly articulated. The posterior sacral has a posterior condyle and the anterior has an anterior depression. The neural arches of a small specimen (LACM 120492) are not fused to the centra but the arches of other specimens are. The neural arch and spine of the posterior sacral are longer than those of the anterior. The neural spine originates along the entire length of the neural arch in both and does not taper dorsally; their dorsal edge is flat. Zygapophyses are oriented at slightly less than 45 ° to the horizontal and lie at the level of the roof of the neural canal. Both centra are of equal size and are equivalent in length to the last trunk vertebra, but are slightly narrower. They are longer and broader than the anteriormost caudal vertebra.
The caudal vertebrae are similar to those of crocodylians in being transversely compressed and bearing low ridges on the lateral edges of the ventral surfaces. Five are present in one articulated series (LACM 120455), but the total number was certainly much greater than this. The second caudal is distinctly smaller than the first and the size decreases gradually posteriorly. An isolated series of five vertebrae (LACM 120455g) are from a region posterior to the first five; they are much longer than wide and their small zygapophyses are horizontal and extend beyond the end of the centrum. The neural spines in these and other posterior caudals are restricted to the posterior end of the neural arch (LACM 120455 for example), but their height is not known. The transverse processes are situated in the middle of the centrum on anterior caudal vertebrae, they move to the posterior end by the fifth, and they are absent from isolated posterior vertebrae. Only fragments of chevrons (hemal arches) are preserved.
The proximal end of the first cervical rib is not preserved, but because the atlas intercentrum has a single articulating surface as in crocodylians it was probably single headed. The proximal end of the second rib is also not preserved; its shaft extends posteriorly past the posterior end of the second vertebra, if the isolated distal end of this rib of the holotype is in its natural position. The third to ?ninth ribs are double headed with short anterior and long posterior processes parallelling the centra. The third preserved vertebra on LACM 115737 articulates with a process from the ventral surface of the third centrum, but this appears to be pathological.
Forelimb and girdle
The scapula of LACM 120492 is incomplete, preserving only the ventral portion articulated with the humerus and coracoid. What is preserved is a flat bone that expands anteroposteriorly at its top so that it is roughly an inverted triangle in lateral aspect. The incompletely preserved dorsal edge may have continued much further dorsally. The anterior edge of the scapula is weakly concave and the posterior edge is strongly concave, especially ventrally. The glenoid fossa is taller than wide and faces generally posteriorly but with a slight lateral component. The surface of the glenoid fossa is roughly rectangular in shape; it is slightly taller than long. The coracoid articulated with it is very poorly preserved but appears to be relatively short dorsoventrally.
The humerus is exceptionally slender in comparison with this element in crocodylians. On the small skeleton associated with the holotype specimen (LACM 120492) the humerus is 51.5 mm long and 3.0 mm in diameter; for an Alligator mississippiensis specimen with a humerus 50 mm long (USNM 167541; datum from P. Dodson) the diameter is 4.0 mm. The humerus of Fruitachampsa is relatively elongate because this specimen of Alligator has a foramen magnum width of 8.7 mm compared with 4.5 mm for LACM 120455, which is larger than LACM 120492. The shaft is cylindrical except at its ends. The deltopectoral crest is well developed, and its proximal end is convex dorsally. The shaft is slightly convex dorsally midway along its length. The distal articulating condyles are well developed; the posterior condyle has an oblique posterior edge that faces posterodorsally, and the anterior condyle has an anterior edge that is vertical. The distal surface of the condyles is approximately twice the width of the shaft. The broad intercondylar groove on the dorsal surface extends medially from the distal end to the medial edge of the condyles.
The ulna is relatively slender. On LACM 120455 the ulna is approximately 62 mm long and 2.7 mm in diameter; for an Alligator mississippiensis specimen with an ulna 61.1 mm long (AMNH 66373) the element is 2.9 mm in diameter. Its proximal articulating surface includes a distinct process on what is presumably the anterior side. As in crocodylians, this articulating surface is oriented obliquely because a low olecranon process is present laterally. The cylindrical shaft is very straight and is preserved very close to the radius. The poorly preserved distal end has a distinct longitudinal ridge on its posterior surface that distally becomes as high as the distal end is wide; the distal end is not turned in medially (to meet the ulnare, which is not preserved). The length of the ulna is slightly greater than that of the humerus, if the incomplete ulna of LACM 120492f is undistorted.
The radius is very slender and its shaft is straight and cylindrical. The diameter of the shaft is approximately half that of the ulna. The proximal end of the radius is flattened so that it is broadest along the long axis of the humeral articulating surface. The distal end is not preserved.
Hindlimb and girdle
The ilium is preserved only on a small, presumably juvenile specimen (LACM 120492d). It is tall and lacks a supra-acetabular crest. There is no anterior process but the posterior process is well developed, and its distal end curves laterally. The medial surface of the posterior process bears coarse longitudinal ridges. The ventral (acetabular) border of the ilium is concave and the posterior ventral process meeting the ischium is stout. The pubis and ischium are not preserved.
The femur is relatively slender; accurate measurements are impossible, however. The head is continuous with the shaft and is only slightly inturned. Trochanters are not visible but preservation is poor in this area on all available specimens. The distal condyles are well developed and equal in size. The cylindrical shaft is gently curved sigmoidally in a vertical plane, but not as strongly as in Alligator specimens of the same size.
The tibia is very slender. On LACM 120492 it is 59.0 mm long and 2.8 mm in diameter; for an Alligator mississippiensis specimen with a tibia 58.3 mm long (USNM 107730) the shaft is 4.3 mm in diameter. The proximal end is slightly broader posteriorly and the femoral articulating surface is relatively flat. The posterior edge of the proximal end bears tuberosities on its lateral and medial edges. The distal end articulates with the astragalus, but details are obscure because preservation is poor. The shaft is straight and its posterior surface is flattened, especially the distal end. The tibia is slightly longer than the femur.
The fibula is extremely slender. On LACM 120492 it is approximately 59 mm long and is 1.0 mm in diameter; on an unnumbered specimen of Alligator mississippiensis in the University of Chicago Anatomy Department the diameter of a fibula 55.0 mm long is 2.8 mm. The fibula articulates with the calcaneum and a lateral process of the astragalus, as in crocodylians, but details are obscure. The shaft is straight and closely situated to the tibia as preserved. The femoral articulation is not preserved. The shaft is approximately one-quarter the diameter of the tibia and the fibula is approximately the same length as the tibia.
Most of the lateral tuber of the calcaneum is missing and most exposed surfaces have been damaged by preparation. The astragalus is also poorly preserved, but it clearly has a process that meets the fibula. A distal tarsal may be present distomedial to the astragalus, but precise identification is impossible.
Several metatarsals are preserved, but the material is insufficient to provide details. None of the bones is enlarged, and what is preserved is similar to the crocodylian condition.
Very few osteoderms are preserved and all appear to be dorsal. Two articulated osteoderms in LACM 115737d span the gaps between cervical neural spines and indicate that there was one pair of osteoderms per vertebral segment, at least in this region. The osteoderms are imbricated, with the posterior edge overlapping the anterior edge of a posterior element. The osteoderms on this specimen are very thin, gently convex dorsally, and finely pitted on their dorsal surfaces. The edges of these osteoderms are not exposed and it is not possible to examine the anterior part of the dorsal surface for the presence of an anterior process, as is typical in basal crocodyliforms.
An isolated osteoderm (LACM 120456) is much larger than those in LACM 115737d, indicating that it is from the trunk region, a larger animal, or both. LACM 120456 is the lateral or medial half of an osteoderm. It is sculptured dorsally like the others except that deeper pitting creates a rugose surface near the preserved edge. Its sculptured surface is gently convex. Its smooth surface has fine longitudinal grooves and a transverse ridge. This ridge lies posteriorly and broadens at the lateral or medial edge. This latter edge is convex, and the osteoderm was apparently wider posteriorly than anteriorly.
A number of features present in this material suggest that the sample reflects the size range of the species, and the species did not reach far larger size. No bones were sectioned to reveal lines of arrested growth, but several features correlated with sexual maturity in living crocodylians offer circumstantial evidence of its maturity. First, the axial skeleton is fused in many places in which most adult modern crocodylians have open sutures (Brochu, 1996) – the neural arches of most vertebrae, the odontoid process of the axis, and the sacral ribs. Second, and of particular phylogenetic interest, the frontals apparently fuse during post-embryonic ontogeny as suggested by their separation in a small specimen (LACM 120492) and fusion in larger specimens (LACM 120494 and 128306). Third, the cranial sculpturing of the Fruitachampsa material is more extensive than that of even adult crocodylians. Sculpturing is almost absent in juveniles of extant crocodylians and becomes accentuated with age. Alternatively, it should be noted that the delicate construction of the cranial bones and the absence of interalveolar septa are features found only in juveniles of living crocodylians (Ferguson, 1985), although septa are absent in adults of some basal mesoeucrocodylians (e.g. Lecuona & Pol, 2008).
The fossils from a single horizon fall into at least three distinct size classes represented by the following specimens: small, LACM 120492; medium, LACM 120455 (holotype) and 120483; large, LACM 115726, 120492, 120494, 120456, and 128306.
The relationships of Fruitachampsa were determined using the data matrix of Pol & Gasparini (2009), which incorporates characters used by Clark (1994). The matrix includes 262 characters scored for 62 taxa, with characters 258–262 added in this study (see Supplementary Information). The codings by Pol & Gasparini for Fruitachampsa were revised, so that 14 characters with question marks were coded and the codings for 11 characters were altered (see Supplementary Information). The coding of Sichuanosuchus shuhanensis for characters 67 (size of antorbital fenestra, state 1 to 2), 70 (dentary beneath mandibular fenestra, ? to state 1), and 201 (lateral exposure of the angular, state 1 to 0) were changed based upon observation of the type specimen.
Three shartegosuchids were included in the analysis, Nominosuchus matutinus, Shartegosuchus asperopalatum, and Adzhosuchus fuscus. Adzhosuchus and Shartegosuchus differ mainly in size and possibly in the morphology of the palate, but the latter requires further study. For some analyses these two taxa were therefore combined. The fragmentary holotype of Nominosuchus arcanus was not included, and it may be synonymous with Shartegosuchus asperopalatum. The fragmentary Argentinean taxon Neuquensuchus (Fiorelli & Calvo, 2007) was not included in the analysis pending a detailed study of the specimen; a brief examination suggested that the holotype is a composite of two individuals.
The analysis was performed using PAUP* (Swofford, 2000), using 2000 replicates of random stepwise addition. Following Pol & Gasparini (2009) character 5 was excluded. The strict consensus of 162 trees 881 steps long (CI = 0.364, RI = 0.711) resolves the relationships of Fruitachampsa as a member of a group including the shartegosuchids Nominosuchus, Adzhosuchus, and Shartegosuchus (Fig. 9). These same relationships hold when Adzhosuchus and Shartegosuchus are united into a single taxon. Unambiguous synapomorphies of this group in all fundamental trees are 40(1, palatal surface of pterygoids sculpted), 75(1, mandibular fenestra absent), and 262(1, posterior maxillary teeth and post-caniniform dentary teeth with flat and horizontal cusp and vertical crenulations extending proximally from it). When Zosuchus is sister taxon, additional unambiguous synapomorphies are 12(1, lacrimal contacts nasal along medial and anterior edges), 55(0, basisphenoid wide and similar in length to, or longer than, basioccipital), 127(0, nasal lateral border posterior to external nares laterally concave), and 201(0, angular posterior to mandibular fenestra: widely exposed on lateral surface of mandible). Fruitachampsa forms a clade within this group with an Adzhosuchus–Shartegosuchus clade, supported by the following unambiguous synapomorphies: 229(1, ventral half of the lacrimal tapering ventroposteriorly, does not contact or contacts the jugal only slightly), 258(1, sculpturing on palatines), and 259(1, lower teeth absent anterior to caniniforms). The monophyly of Shartegosuchidae collapses in trees one step longer, along with much of the tree's resolution, except for the monophyly of the Fruitachampsa–Shartegosuchus–Adzhosuchus clade, which collapses two steps out.
The ambiguity in relationships among Sichuanosuchus, Shantungosuchus, Zosuchus, and shartegosuchids is due to conflicting relationships among all four taxa. Among the fundamental trees the sister taxon to shartegosuchids is variously Zosuchus, a Shantungosuchus–Sichuanosuchus clade, a Zosuchus–Shantungosuchus–Sichuanosuchus clade, or the Hsisosuchus–Mesoeucrocodylia clade. In the trees in which shartegosuchids are closer to Hsisosuchus and Mesoeucrocodylia than is a clade comprising the other three taxa, it is supported by the following unambiguous characters: 55(0, basisphenoid shorter in length than basioccipital), 70(1, dentary does not extend beneath mandibular fenestra), and 140(1, maxillary teeth laterally compressed).
The relationships among basal crocodyliforms are not well resolved in this analysis and are poorly supported. However, as in Pol & Gasparini (2009) Protosuchia are paraphyletic, with shartegosuchids, Zosuchus, Sichuanosuchus, Shantungosuchus, and Hsisosuchus closer to Mesoeucrocodylia than other ‘protosuchians’, and Hsisosuchus is the sister taxon of Mesoeucrocodylia.
Fruitachampsa callisoni is the first shartegosuchid discovered outside of Asia and the youngest member of this family with the exception of Kyasuchus from the Early Cretaceous of Siberia. Until now this family has only been reported from three localities, one in Mongolia (Gubin & Sinitza, 1996), another in Siberia (Efimov & Leschinskii, 2000), and the third in China (Clark & Xu, 2009). Although the relationships among basal crocodyliforms are poorly resolved in this analysis, the position of Fruitachampsa with Shartegosuchus within the Asian Shartegosuchidae and their phylogenetic position surrounded exclusively by Asian taxa (Zosuchus, Sichuanosuchus, Shantungosuchus, Gobiosuchidae, and Hsisosuchus), with the possible exception of the Argentinean Neuquensuchus (Fiorelli & Calvo, 2007), indicates that Fruitachampsa dispersed to North America from Asia. This suggests that shartegosuchids should also be found in Europe in the Late Jurassic, but none has yet been reported.
Although the relationships of shartegosuchids to Shantungosuchus, Sichuanosuchus, and Zosuchus are poorly resolved, there is evidence that they form a clade. Sichuanosuchus (Peng, 1996; Wu et al., 1997) and Nominosuchus (Fig. 10F) preserve an unusual morphology in which the pterygoid and maxilla contact and separate the palatines from the suborbital fenestra (character 200). The palate of Zosuchus is incomplete but has a large pterygoid and small palatine (Pol & Norell, 2004a), and it is likely the palatine was excluded from the suborbital fenestra. In the descriptions of Shartegosuchus and Adzhosuchus, Efimov (1996), Efimov et al. (2000), and Storrs & Efimov (2000) identify the bone forming much of the secondary palate in these taxa as the pterygoid, implying a eusuchian-type palate and possibly a lateral contact of the pterygoid and maxilla. Examination of the type specimens (Fig. 10A–D) indicates that further preparation is needed to reveal the precise extent of the palatine and pterygoid in these two taxa, but it does not appear that the pterygoid contacted the maxilla lateral to the palatine, nor does it do so in Fruitachampsa.
A palatine secondary palate, such as is present in Fruitachampsa, Shartegosuchus, Adzhosuchus, Zosuchus, and incipiently in Nominosuchus, is a mesoeucrocodylian synapomorphy. The absence of a palatine secondary palate in Hsisosuchus and the unresolved relationships among Sichuanosuchus, Shantungosuchus, Zosuchus, and shartegosuchids makes the evolution of this feature ambiguous. However, the palatine secondary palate in shartegosuchids differs from that of Mesoeucrocodylia in being flat and not arching dorsolaterally, as in meoseucrocodylians. A further mesoeucrocodylian feature may be present – the enclosure of the cranioquadrate passage – but further study of the anatomy of this region in shartegosuchids is necessary to clarify this.
The absence of an antorbital fenestra and the presence of procoelous vertebrae are derived features within Neosuchia (Pol & Norell, 2004b), but their occurrence in Fruitachampsa is most parsimoniously interpreted as convergent. Procoelous vertebrae are otherwise known in only a few crocodylomorphs outside the Eusuchia, including the ‘sphenosuchian’Junggarsuchus (Clark et al., 2004), the atoposaurid Theriosuchus (Joffe, 1967), and the atoposaurid relative Pachycheilosuchus (Rogers, 2003) (see also Molnar, 1980).
Fruitachampsa is the first crocodylomorph described with a condyle on the posterior end of the sacrum. In Pachycheilosuchus, the first caudal is biconvex (Rogers, 2003), as in living crocodylians, and the same may be true of Theriosuchus (Joffe, 1967; Clark, 1986). A specimen of another taxon with this condition exists, but unfortunately the specimen, a few uncatalogued associated vertebrae in the collections of the American Museum of Natural History, lacks data. The bone of this specimen is similar to bone from the Bridgerian of Wyoming, which has a diverse crocodylian fauna, but unfortunately bears no features that would serve to identify it below the level of Crocodyliformes.
G. Callison started me on the road to palaeontology, led the expedition that discovered small vertebrates at Fruita, and allowed me to study the material described here. J. T. Gregory and K. Padian diligently supervised numerous drafts of the original manuscript, and H. Greene was also very helpful. E. Buffetaut and the late A. Walker gave exceedingly helpful comments on the manuscript for which I am very grateful. I benefited greatly from conversations and correspondence with A. Busbey, J. Gauthier, J. Hopson, M. Norell, W. Langston, and especially D. Pol. The exquisite illustrations are the work of Ms Claire Vanderslice (Figs 1–3, 5, 7), who also finished the preparation on the illustrated specimens, and Mick Ellison (Fig. 7). For access to specimens in their care I am grateful to the following: D. Berman (Carnegie Museum, Pittsburgh), A. Charig (British Museum of Natural History), M. Efimov (Paleontological Institute, Moscow), E. Gaffney and M. Norell (American Museum, New York), J. H. Hutchison (University of California), W. Langston (University of Texas, Austin), M. Phillippe (Musee Guimet de Lyon), G. Viohl (Jura Museum, Eichstatt, Germany), H. Voris (Field Museum, Chicago), P. Wellnhofer (Bayerische Staatssammlung fur Palaontologie und hist. Geologie, Munich), and D. Whistler and L. Chiappe (Los Angeles County Museum). P. Dodson (University of Pennsylvania) provided measurements of Alligator. Financial support was provided by the Society of Sigma Xi and a Dissertation Improvement grant from the National Science Foundation. Some preparation of the material was started at the Los Angeles County Museum. I would especially like to thank my parents for their support over the years.