Neural and endocranial anatomy of Triassic phytosaurian reptiles and convergence with fossil and modern crocodylians

Specimens

The studied specimens consist of the holotypes of Ebrachosuchus neukami (BSPG 1931 X 501; Bayerische Staatssammlung für Paläontologie und Geologie, Munich, Germany) and Parasuchus angustifrons (BSPG 1931 X 502) (Kuhn, 1936; Butler et al., 2014). The skull of Ebrachosuchus neukami is fully articulated and complete and preparation work has removed most of the sandstone matrix, with the exception of most internal cavities, which remain filled with matrix. The skull of Parasuchus angustifrons is articulated and mostly complete apart from the premaxilla, of which only a small portion immediately rostral to the external nares is preserved. Sandstone matrix remains within most of the internal cavities. Both of the skulls have undergone some plastic deformation, primarily in the form of dorsoventral compression, but are otherwise remarkably well preserved, with minimum transverse distortion. Brittle deformation and fracturing are largely absent.

CT scanning and digital reconstruction

The holotypes of Ebrachosuchus neukami (BSPG 1931 X 501) and Parasuchus angustifrons (BSPG 1931 X 502) were scanned at the Klinikum rechts der Isar (Munich) using a Siemens SOMATOM Sensation 64 CT scanner. Datasets consisted of 1,634 slices (512 × 512 × 1,634 pixels, 0.6 mm voxel size) for Ebrachosuchus neukami and 809 slices (512 × 512 × 1,634 pixels, 0.6 mm voxel size) for Parasuchus angustifrons. CT data sets are deposited with the specimens in the BSPG collections and on Figshare (10.6084/m9.figshare.3443963; 10.6084/m9.figshare.3443960).

The respective CT data files were imported into Avizo 7.0 (Visualisation Science Group) for image segmentation and digital reconstruction. Anatomical structures of interest (endocasts, endosseous labyrinths, neurovascular and sinus structures) were labelled using Avizo’s segmentation editor. The magic wand tool was used where possible to perform the segmentation semi-automatically. In regions with poor contrast between matrix, bone and structures of interest the paintbrush tool was used for manual segmentation. 3D surface models and volumes were created to visualize the segmented structures and to illustrate this article with traditional figures. In addition, surface models of the individual structures were downsampled to a degree that allowed for small file sizes but preserved all details, and were exported as separate OBJ files for the creation of the interactive 3D figures in the supplementary material as outlined in Lautenschlager (2014) using Adobe 3D reviewer (Adobe Systems Inc.).

As both taxa have been compressed dorsoventrally to a moderate amount, the resulting endocasts were retrodeformed. For the retrodeformation process, the digital skull and endocast models were scaled in dorsoventral direction using the “transform editor” in Avizo. BSPG 1931 X 502 was scaled to the same dorsoventral dimensions as a less compressed skull of Parasuchus hislopi (ISI R42, Indian Statistical Institute, Kolkata, India), corresponding to a scaling factor of approximately 40%. The same scaling factor was assumed for BSPG 1931 X 501 based on the fact that both specimens were found in close proximity to one another on a single bedding plane and likely had a similar diagenetic history (Butler et al., 2014).

Endocranial anatomy

The endocasts of Ebrachosuchus neukami and Parasuchus angustifrons are very similar in their morphology. Both endocasts are elongate, straight (i.e. arranged horizontally) and mediolaterally narrow (Figs. 2 and 3). Long olfactory tracts extend rostrally and are as long as the main portion of the endocasts in each taxon. Fossae for olfactory bulbs are preserved in both taxa, but only in Parasuchus angustifrons is a rostral separation into two olfactory bulbs visible (Fig. 2B). The ventral extent could not be reconstructed as no bony structures cover this region. The cerebrum and the cerebral hemispheres are only weakly demarcated and form the widest part of the endocast in each taxon. Again, in Parasuchus angustifrons the cerebral hemispheres are slightly more prominent than in Ebrachosuchus neukami. The mid- and hindbrain region, including the cerebellum, is mediolaterally compressed between the endosseous labyrinths in both taxa. Caudally, the endocasts increase in width towards the foramen magnum. The floccular lobes, extending from the cerebellum, are prominent but short. In comparison to the more oval-shaped morphology in Parasuchus angustifrons, the floccular lobes are slightly dorsoventrally flattened in Ebrachosuchus neukami (Fig. 3A). In both taxa, the floccular lobes enter the vestibular apparatus of the endosseous labyrinth, but do not extend beyond the rostral semicircular canal. Despite the elongate morphology of the endocasts, the cephalic flexure (between the fore- and mid-brain) and the pontine flexure (between the mid- and hindbrain) are pronounced in the endocasts. The cast of the pituitary fossa is prominent and extends ventrally from the ventral surface of the cerebrum. The pituitary fossae have equal dimensions and positions in both taxa.

The canals of the neurovascular structures were reconstructed for as far as indicated by osteological correlates. Their arrangement and dimensions are similar in Ebrachosuchus neukami and Parasuchus angustifrons. The optic nerve (CN II) and the oculomotor nerve (CN III) canals could not be traced due to the lack of preserved bony structures on the rostroventral part of the endocasts. The trochlear nerve canal (CN IV) is small and originates from the ventral surface of the cerebrum, rostral to the trigeminal nerve (CN V). The latter is prominent in both taxa and originates from the ventrolateral surface of the midbrain region. A split into a rostrally directed ophthalmic branch (CN V₁) and a laterally projecting combined canal for the maxillary (CN V₂) and mandibular (CN V₃) branches is evident in both endocasts. A subdivision of the latter two is not visible in the CT scans, but most likely occurred further outside of the endocranial cavity as is the plesiomorphic archosaurian condition (Witmer et al., 2008). The canal for the abducens nerve (CN VI) originated from the ventral surface of the endocast below the trigeminal nerve canal. The canal for the facial nerve (CN VII) is situated caudal to that of the abducens nerve. The vestibulocochlear nerve canal (CN VIII) could not be reconstructed in either taxon as the resolution of the CT scans is not clear enough in this region to identify the nerve canal confidently. A large metotic fissure is present in both taxa, transmitting the glossopharyngeal (CN IX), the vagus (CN X) and the spinal accessory nerves (CN XI). The hypoglossal nerve (CN XII) exits the braincase via a single nerve canal. A further foramen located dorsal to the hypoglossal nerve foramen has a blind ending and likely represents a diverticulum of the longitudinal sinus (Witmer & Ridgely, 2008; Witmer & Ridgely, 2009).

Due to the resolution of the CT data set only the larger vascular structures could be reconstructed. The roots of the caudal middle cerebral vein are prominent and originate from the cerebellum rostrodorsally to the floccular lobes in Ebrachosuchus neukami and Parasuchus angustifrons. They can be traced caudally through the bone for a short extent exiting the braincase near the supraoccipital-parietal suture. Ventrally, the canals for the carotid artery originate from the pituitary fossa and exit the basisphenoid ventrolaterally.

In comparison to the other endocranial components, the endosseous labyrinths of Ebrachosuchus neukami and Parasuchus angustifrons show more prominent differences (Fig. 6). In general, the labyrinths are dorsoventrally short and compact. The vestibular apparatus approaches a rectangular outline and is rostrocaudally elongate in both taxa, but more pronounced in Ebrachosuchus neukami. This may partly due to the preservation of Parasuchus angustifrons, which seems to have been dorsoventrally compacted to a moderate extent (Figs. 6A and 6B). In Ebrachosuchus neukami, the rostral semicircular canal is the longest and describes a somewhat quadrangular shape, whereas the caudal semicircular canal is more oval-shaped (Figs. 6C and 6D). The lateral semicircular canal is short and compact. The semicircular canals in Parasuchus angustifrons appear, as far as preserved, dorsoventrally compressed and with more uniform dimension than in Ebrachosuchus neukami, although this is partly a preservational artefact. The cochlear ducts are short in Ebrachosuchus neukami and Parasuchus angustifrons and extend largely ventrally, with only a slight medial component. The fenestra vestibuli were reconstructed in Ebrachosuchus neukami and Parasuchus angustifrons.

As with the endocranial cavities, the arrangement and extent of the various sinuses is similar in Ebrachosuchus neukami and Parasuchus angustifrons (Figs. 4 and 5). The antorbital sinus is large and fills the antorbital fenestra, as well as the space between the palate and the palatal shelf of the maxilla. A small diverticulum also appears to enter the jugal via a foramen near the ectopterygoid-jugal contact (Butler et al., 2014) in both taxa. Rostrally, a large canal is present, which opens into the antorbital cavity. The canal likely transmitted neurovascular structures, including the maxillary branch of the trigeminal nerve. The region rostral to the external nares comprises a large air-filled space for the entire length of the premaxilla medial to the neurovascular canal supplying the alveolar cavities. This region may have housed a premaxillary sinus (most likely as extension of the antorbital sinus) and/or neurovascular bundles (Butler et al., 2014) as in extant crocodiles (Leitch & Catania, 2012). The airway is simple and unbranched in Ebrachosuchus neukami and Parasuchus angustifrons and connects the external nares with the choanae and the olfactory region.

In Parasuchus angustifrons, the ectopterygoid is pierced medially by a single oval foramen opening into a large cavity within the bone (Fig. 4). No further foramina are identifiable, suggesting the foramen connected to a pneumatic recess rather than transmitting neurovascular structures. The pneumatic recess is either part of the antorbital sinus or a separate ectopterygoid sinus of unknown source (Witmer, 1997). The respective region is only partly preserved and damaged, but a similarly large cavity appears to be absent in Ebrachosuchus neukami. A further sinus located dorsal to the brain endocast is present in Ebrachosuchus neukami and Parasuchus angustifrons. While this structure could be interpreted as part of the paratympanic sinus, there is no clear connection to the middle ear visible in the datasets. In both taxa, it covers the cerebellum dorsally. In Ebrachosuchus neukami two small diverticula extend rostrally covering the cerebrum dorsolaterally. These diverticula are not visible in Parasuchus angustifrons. Laterally, subsidiary canals of the tympanic sinus are present in both taxa, but more pronounced in Parasuchus angustifrons, in which they exit the braincase via a foramen between the parietal and the prootic and connect to the caudal tympanic recess. This sinus possibly had a further connection to the quadrate foramen, but the pathway for this canal is not indicated by osteological correlates. The sinus is therefore most likely a combination of the endocranium and the dural venous sinuses.

Comparison with other phytosaurs

A comparison with other phytosaurs shows that, while similar to each other, the endocranial anatomy of Ebrachosuchus neukami and Parasuchus angustifrons differs in several aspects from that of more derived taxa (Fig. 7). However, it should be noted that accurate comparisons are exacerbated by the scarcity of detailed reconstructions of endocasts. Existing reconstructions are mostly based on physical casts (natural and artificial) or interpretive drawings (Cope, 1888; Case, 1928; Mehl, 1928; Camp, 1930; Chatterjee, 1978). All phytosaur endocasts appear to share a basic bauplan with the individual brain regions arranged longitudinally (in contrast to a more vertical arrangement such as seen in birds) and a mediolaterally narrow morphology. The olfactory tracts are significantly elongate in all taxa (as far as preserved/reconstructed), making up approximately half the length of the entire endocasts. Caudal to the olfactory tracts, the various taxa show large differences in the orientation of the individual brain portions. Cephalic and pontine flexure is very variable. While Ebrachosuchus neukami and Parasuchus angustifrons share very large flexure angles (following Lautenschlager & Hübner, 2013) with derived taxa, such as Machaeroprosopus pristinus and Machaeroprosopus buceros, the fore- and mid-brain and the mid- and hind-brain appear to be almost perpendicular to each other in Smilosuchus gregori and Parasuchus hislopi. However, although the studied specimens of Parasuchus angustifrons (BSPG 1931 X 502) and Ebrachosuchus neukami (BSPG 1931 X 501) are well preserved and mostly complete, they show signs of moderate dorsoventral compaction. Retrodeformed endocast reconstructions exhibit cephalic and pontine flexures more similar to Machaeroprosopus mccauleyi and Parasuchus hislopi (Fig. 8). A significant difference is found in the presence of a pineal organ or epiphysis dorsal to the cerebrum. A pineal organ has been suggested to be present (Jaekel, 1910; Langston, 1949) and been reconstructed for the majority of phytosaurs, but is absent in Ebrachosuchus neukami and Parasuchus angustifrons. The dorsal expansion in the respective region in these taxa is interpreted in this study to represent parts of the dural venous sinus or alternatively the paratympanic sinus, due to the rostral and lateral expansion of this structure into parts of the braincase. Hopson (1979) similarly considered a pineal organ in phytosaurs unlikely and suggested that the respective region in the endocranial cavity housed a cartilaginous portion of the supraoccipital. Although reconstructed by Mehl (1928), an enlarged epiphysis was reported to be absent in Machaeroprosopus pristinus in an as-yet-unpublished recent study (Smith et al., 2010). Existing endocast reconstructions provide ambiguous results regarding the presence of the epiphysis, but suggest that it may have elaborated through phytosaur evolution (Fig. 7). However, considering its absence in modern crocodilians, the epiphysis must have been lost at some stage prior to the origin of the crocodilian crown group.

Similar to the actual brain endocast, the endosseous labyrinths show, as far as reconstructed, subtle differences between different phytosaurian taxa. The vestibular part of the labyrinth of Ebrachosuchus neukami and Parasuchus angustifrons is rostrocaudally longer than dorsoventrally high, whereas it seems to have more uniform dimensions in Machaeroprosopus mccauleyi and Leptosuchus sp. Again, this is to some extent a result of dorsoventral compression. The retrodeformed endosseous labyrinths of Ebrachosuchus neukami and Parasuchus angustifrons are more similar to the labyrinth of Leptosuchus sp. It should be noted, though, that the retrodeformation is based on dorsoventral scaling of the complete skull and endocast, respectively. However, the opisthotic and the paroccipital, which house the endosseous labyrinth, might not have suffered the same amount of compression as the complete skull. Furthermore, the scarcity of reconstructed and preserved natural labyrinthine endocasts confounds wider comparisons.

Comparison with Crocodyliformes

As in comparison with other phytosaurs, the endocranial anatomy of Ebrachosuchus neukami and Parasuchus angustifrons shares a basic bauplan with most Crocodyliformes in the form of a longitudinally arranged and elongate brain architecture thought to be plesiomorphic for the whole lineage (Hopson, 1979). As in phytosaurs, the olfactory tracts are elongate in extant crocodylians, including Alligator mississippiensis (Witmer & Ridgely, 2008), Crocodylus johnstoni (Witmer et al., 2008) and Crocodylus moreleti (Franzosa, 2004), as well as in several phylogenetically distinct Mesozoic longirostrine crocodylomorphs, such as the neosuchian Pholidosaurus (Edinger, 1938; Hopson, 1979) and the metriorhynchid Cricosaurus araucanensis (Herrera, Fernandez & Gasparini, 2013). In contrast to phytosaurs, the cerebral hemispheres are prominent and mediolaterally enlarged in most Crocodyliformes (Wharton, 2000; Franzosa, 2004; George & Holliday, 2013). Extant crocodylians possess an enlarged dural venous sinus covering the endocast dorsally (Witmer et al., 2008), which has been interpreted to be present in fossil Mesoeucrocodylia (Hopson, 1979; Wharton, 2000). Where preserved or reconstructed, the endosseous labyrinths show a dorsoventrally compressed vestibular region and short cochlear ducts in Crocodyliformes (Franzosa, 2004; Witmer et al., 2008), similar to Ebrachosuchus neukami and Parasuchus angustifrons.

Paranasal sinuses have been reconstructed only for a handful of extant and extinct Crocodyliformes (e.g., Alligator mississippiensis, Cricosaurus araucanensis), which limits comparisons of these structures. A clear difference is found in the size of the antorbital sinus. In Ebrachosuchus neukami and Parasuchus angustifrons the antorbital sinus is enlarged, but it is considerably smaller in Crocodyliformes (Witmer & Ridgely, 2008; Herrera, Fernandez & Gasparini, 2013). Due to the position of the external nares, the airway is short in the studied phytosaurs. In the longirostrine metriorhynchid Cricosaurus araucanensis, the rostrum comprises the airway for its entire length (Herrera, Fernandez & Gasparini, 2013), whereas the comparable region was likely filled by a premaxillary sinus in phytosaurs.