Paleoneuroanatomy of the aetosaur Neoaetosauroides engaeus (Archosauria: Pseudosuchia) and its paleobiological implications among archosauriforms

The paleoneuroanatomy of pseudosuchian archosaurs is poorly known, based on direct examination of the internal morphology of braincases and a few artificial endocasts. Among aetosaurs, only one endocast has been described almost a century ago by Case (1921) corresponding to Desmatosuchus spurensis from the Chinle Formation (Norian) of Texas, US, based on a resin cast. Here, we describe the first natural endocast of an aetosaur, Neoaetosauroides engaeus from the Los Colorados Formation (Norian) of NW Argentina, and also developed the first digital endocast of this taxon including the encephalon, cranial nerves, inner ear, and middle ear sinuses. The neuroanatomy of Neoaetosauroides engaeus exhibits several differences from that of Desmatosuchus spurensis despite their phylogenetic proximity, which may be a reflection of their different habits. The information provided by the endocasts of Neoaetosauroides engaeus about its olfactory region and the orientation of its head, based on the inclination of the inner ear, could support the proposal for an animalivorous diet, instead of an herbivorous one as in most aetosaurs. The new information here obtained contributes to the knowledge of the neuroanatomical diversity of archosauriforms and more specifically among pseudosuchians and their paleobiological roles in the Triassic continental communities.

Even though the external anatomy of aetosaurian skulls is well known, only one endocast was described in detail almost a century ago by Case (1921). This description of the encephalon of the aetosaur Desmatosuchus spurensis was based on an artificial endocast of the cranial cavity of the holotype and represents not only the encephalon, but also the soft tissues surrounding it, such as the dura mater and venous sinuses (Hopson, 1979;Gans & Billett, 1985). In living pseudosuchians, these surrounding tissues represent approximately 50% of the endocranial cavity (Jirak & Janacek, 2017), differing from that of other living archosaurs such as birds in which the encephalon essentially fills the endocranial cavity (Hopson, 1979;Witmer & Ridgely, 2008). The study of the internal cavities of the skull (encephalon, inner ear, paranasal sinuses, cranial nerves, muscles) by applying computed tomography to fossils allowed the collection of crucial anatomical information for morphological, ontogenetic, and functional analyses (Sereno et al., 2007;Balanoff et al., 2013;Paulina-Carabajal, Carballido & Currie, 2014;Jirak & Janacek, 2017). These new studies allow a better understanding of the paleobiological roles of the different pseudosuchian groups in the Triassic continental communities worldwide.
During the redescription of the cranial anatomy of Neoaetosauroides engaeus, based on three skulls, the first known natural endocast of an aetosaur was discovered during the preparation of the material (Desojo, 2005;Desojo et al., 2013). This exceptionally preserved material, PULR 108, consists of the mould of the forebrain, midbrain, and hindbrain and both partially preserved inner ears. Additionally, the partial skull of Neoaetosauroides engaeus PVL 4363 bears the natural moulds of the olfactory tracts and bulbs in situ. Complementarily, the digital cast of the endocranial cavity of the most complete braincase was generated recovering the encephalon, cranial nerves, and also the first known middle ear sinuses and inner ears of an aetosaur (PVL 5698). The aim of the present contribution is to describe, analyse, and compare the endocranial anatomy of the aetosaur N. engaeus to comprehend the evolution of these structures within archosauriforms and its implications for paleobiology. ARCHOSAURIA Cope, 1869sensu Gauthier & Padian, 1985PSEUDOSUCHIA Zittel, 1887-1890sensu Gauthier & Padian, 1985AETOSAURIA Marsh, 1884sensu Parker, 2007NEOAETOSAUROIDES Bonaparte, 1969Neoaetosauroides engaeus Bonaparte, 1969

Materials
PULR 108: natural casts of inner surface of the skull, lower jaw, and natural cast of the endocranial cavity. PVL 5698: almost complete skull lacking the snout region. PVL 4363: incomplete skull with natural moulds of the skull roof, left side, and left jaw.

CT scanning and digital reconstruction
The referred specimen PVL 5698 was CTscanned at the Clínica La Sagrada Familia (Buenos Aires, Argentina) using a medical 64-channel Phillips Multislice CT scanner. The dataset consists of 413 slices with the following settings: field of view 289 mm, penetration power of 120.0 Kv and 313 mA, slice thickness of 0.9 mm and 0.45 mm of overlap. Analysis of the images and 3D reconstructions were developed with the open source software 3D Slicer v4.1.1 (Fedorov et al., 2012). The terminology used for the description of the digital endocast does not refer strictly to the soft tissue regions of the brain because, as seen in modern archosaurs, the endocast also includes the volume occupied by other tissues surrounding the brain (i.e., vascular tissue) (e.g., Paulina-Carabajal & Currie, 2012;Paulina-Carabajal, Carballido & Currie, 2014).

DESCRIPTION Encephalon
The endocast of N. engaeus is elongate, narrow and has a strong angle between the forebrain and the midbrain (135 • ). The anterior margin of the olfactory bulbs is located at the level of the posterior margin of the antorbital fenestra and the posterior end is located at the level of the anteroposterior midline of the orbit (PVL 4363). The olfactory tracts are short and the bulbs are elliptic, being anteroposteriorly longer than wide. The olfactory bulbs do not contact each other at the sagittal plane (Fig. 1). The cerebral hemispheres are elongate, slightly expanded, and 1.13 times longer than wide (PULR 108). The cerebral hemispheres are the widest region of the forebrain which is 1.5 times wider than their posterior end at the medial vestibular constriction (MVC). The hypophysis should be on the ventral surface of the cerebral hemispheres, just anterior to the MVC, but it is broken on the natural cast and could not be reconstructed on the digital cast, therefore, its ventral extension

Cranial nerves
Most of the cranial nerves (CN) have been identified in the natural endocast of N. engaeus (PULR 108) and the digital reconstruction based in PVL 5698 (Figs. 2 and 3). The cranial nerves II to VI were recognized on both endocasts, but CNs VII to XII were only seen on the digital reconstruction. The CN I corresponds to the olfactory nerve indicated at the level of the olfactory tracts previously described. The optic nerve (CN II) is recognized at the level of the optic chiasma as a prominent structure located at the midline of the ventral (CN VII) also projects from the ventral surface of the hindbrain but is located posterior to the CN VI and ventromedially to the endosseous labyrinth (Fig. 3). The vestibulocochlear nerve (CN VIII) is situated posteroventrally to the CN VI and medially to the endosseous labyrinth and projects laterally. The metotic passage can be identified posterior to the semicircular canals and would probably have accommodated the glossopharyngeal nerve (CN IX), the vagus nerve (CN X) and the spinal accessory nerve (CN XI). Two branches are recognized splitting from the metotic passage (Figs. 3B-3D). This passage can also be recognized externally as a foramen on the braincase of PVL 5698 between the ventral process of the opistotic and the lateral ridge of the exoccipital and basioccipital (sensu Gower & Walker, 2002). The hypoglossal nerve (CN XII) exits laterally through a single passage on the posteriormost part of the hindbrain.

Inner ear
The inner ears of N. engaeus are incompletely preserved in the natural endocast of PULR 108, but were able to be reconstructed almost entirely on the right side of the digital endocast of PVL 5698 (Figs. 2 and 3A). Therefore, the following description is based only on the right inner ear of N. engaeus.
The endosseous labyrinth is as high as anteroposteriorly long, quadrangular in lateral view. The anterior and posterior semicircular canals are apparently equivalent in diameter but the anterior semicircular canal has a wider diameter of the tube. The anterior canal describes a quadrangular shape while the posterior one is more circular. These two canals form an angle of approximately 85 • and connect each other dorsomedially forming a common crus that projects ventrally reaching the vestibulum (Fig. 4: ac, pc, cc, ve). The lateral semicircular canal is oval-shaped, mediolaterally compressed, and has a swelling on its contact with the anterior canal which probably represents the anterior ampula ( Fig. 4: hc, amp). The posterior ampula cannot be identified because the posterior end of the lateral semicircular canal could not be reconstructed on the digital endocast. The vestibulum and the lagena were identified only on the digital reconstruction (PVL 5698). The vestibulum is located between the vestibular apparatus and the fenestra pseudorotunda; it is slightly higher than the lagena and equally long as high in lateral view ( Fig. 4: fps, ve, la). The fenestra pseudorotunda is placed in the posteroventral corner of the vestibulum and the lagena is located ventral to this structure. The lagena is short, mediolaterally compressed, it maintains its anteroposterior width along its length, and is rounded distally.

Pneumatic sinuses
The middle ear sinus system could be identified on the left side of the digital endocast of N. engaeus, but no structures corresponding to the paratympanic and pharyngotympanic sinuses could be recognized (Fig. 5).
The middle ear sinus system pierces through the paroccipital process; it is a thin, straight, and tubular canal that is posterolaterally directed. It expands dorsoventrally on its distal end being four times higher than the proximal region (Figs. 5A-5C). This expansion corresponds to the cranioquadrate passage that would house the stapedial artery and a branch of the facial nerve (Bona, Paulina-Carabajal & Gasparini, 2017). The siphonium should be located ventral to this structure, but it could not be identified on this digital endocast, despite recognizing its distal exit on the quadrate foramen of PVL 5698.

COMPARISON WITH OTHER ARCHOSAURIFORM ENDOCASTS
Comparing the endocast of N. engaeus with D. spurensis there are some noticeable differences between these aetosaurs, among archosauriforms (Fig. 6). The olfactory tracts of N. engaeus (PVL 5698, PULR 108) are narrow being half the width of the cerebral hemispheres, whereas the olfactory tracts of D. spurensis (UMMP 7476;Hopson, 1979) are about the same width of the cerebral hemispheres as it also happens in the erpetosuchid Parringtonia gracilis (Nesbitt et al., 2018). This condition of narrow olfactory tracts seen in N. engaeus resembles more that of archosauriforms (Tropidosuchus romeri: Trotteyn & Paulina-Carabajal, 2016;Triopticus primus: Stocker et al., 2016)    The cerebral hemispheres are clearly recognized in N. engaeus (PVL 5698). They are approximately one-third (1/3) wider than the MVC, similar to the pseudosuchians R. tenuisceps, S. clarki, and S. icaeorhinus, and lambeosaurine ornithischians (H. altispinus, Corythosaurus sp., A. ardevoli) differing from the condition seen in phytosaurs (P. angustifrons, Ebrachosuchus neukami), the crocodylomorph Gryposuchus neogaeus, the theropods G. carolinii and Acrocanthosaurus atokensis, the ankylosaurids Euoplocephalus tutus, Tarchia teresae, and Talaurus plicatospineus, in which the cerebral hemispheres are almost as wide as the MVC. The aetosaur D. spurensis shows a condition that is intermediate between that seen in the previous taxa and N. engaeus.
The flexure between the major axis of the forebrain and that of the midbrain measured for N. engaeus (135 • ) resembles that of Riojasuchus tenuisceps (130 • : Baczko & Desojo, 2016) and D. spurensis (120 • : UMMP 7476), being slightly lower in this last one. This flexure has a lower angle than that of crocodylomorphs (G. neogaeus, G. gangeticus, S. clarki) and phytosaurs (P. angustifrons, E. neukami, W. scurriensis) which is around the 150 • , contrasting with the more acute angles registered in some dinosaurs such as D. longus, A. atokensis, and Tyrannosaurus rex (110 • ) (Fig. 6). This variability in the angle between the forebrain and midbrain could be related to the height of the braincase which is apparently higher in the later than in the pseudosuchians mentioned above. Moreover, within pseudosuchians the lowest values were registered in phytosaurs and extant crocodiles which have the flattest skulls. The hypophysis of N. engaeus was not preserved in the natural endocast (PULR 108) but the base of this structure can be identified in the digital reconstruction of PVL 5698, based on its topology and size. However, the hypophysis is still incomplete in PVL 5698 and therefore it cannot be determined whether it is vertical as in phytosaurs (e.g., Ebrachosaurus neukami, P. angustifrons) and ornithischians (e.g., H. altispinus, Corythosaurus sp., A. ardevoli, Euoplocephalus tutus, Talaurus plicatospineus, Panoplosaurus mirus) or posteroventrally directed as in most archosaurs (e.g., A. mississippiensis, G. gangeticus, S. icaeorhinus, S. clarki, R. tenuisceps, T. rex, D. longus) (Fig. 6). The artificial endocast of the aetosaur D. spurensis (UMMP 7476) preserved only the posterior part of the hypophysis and the anterior region is obscured by leakage of the resin between the bones of the braincase used during its preparation.
The location of the CNs VII and VIII of N. engaeus differs from that proposed by Case (1921) for D. spurensis. The exits for these two cranial nerves are identified separately on the digital endocast of N. engaeus (PVL 5698), whereas in D. spurensis they were identified by Case (1921) as a single foramen posterior to the exit of the CN VI and anterior to a putative exit for the CN IX-XI.
The size and location of the CNs II, III, IV and VI do not exhibit any particular difference between N. engaeus and D. spurensis.

DISCUSSION
When analyzing the olfactory region, two general patterns could be recognized. The first corresponding to short olfactory tracts and wide, rounded bulbs that was seen in archosaurs associated to herbivorous habits such as D. spurensis, S. clarki, Stegosaurus armatus, E. tutus, S. validum, H. altispinus, Corythosaurus sp., and D. longus. On the other hand, the other pattern with elongated tracts and narrow, elliptic bulbs was present in archosaurs associated to carnivorous habits like P. angustifrons, P. gracilis, S. icaeorhinus, C. yacare, Sinraptor dongi, Carcharodontosaurus saharicus, A. atokensis, and T. rex (Fig. 6). N. engaeus has a morphology that resembles the second pattern corresponding to carnivorous habits, which is also congruent with the dental morphology of this aetosaur (Desojo & Báez, 2007).
Several authors discussed the orientation of the lateral semicircular canal of the inner ear proposing that in an 'alert' or 'neutral' position this canal should be parallel to the ground (e.g. Witmer et al., 2003;Sereno et al., 2007;Witmer & Ridgely, 2008). When orienting the skull of N. engaeus with the lateral semicircular canal parallel to the ground, the snout is anteroventrally directed and its ventral margin forms an angle of approximately 30 degrees with the horizontal plane. This position favours both a more binocular visibility, with the snout being less obstructive for the visual field, as well as the nares positioned towards the ground for tracking. However, different studies propose that the lateral semicircular canal should not be used as a reference for the skull orientation based on documented evidence on living tetrapods whose semicircular canals tend to be misaligned with the Earth's axes (Hullar, 2006;Taylor, Wedel & Naish, 2009;Marugán-Lobón, Chiappe & Farke, 2013). The misalignment of the semicircular canals is apparently physiologically advantageous to perceive the angular acceleration in all canals during horizontal head rotation (Cohen & Raphan, 2004). Discarding the lateral semicircular canal as a reference system, other structures need to be considered when orienting the skull. If the maxillary tooth row parallel to the ground is used to orient the skull (Marugán-Lobón, Chiappe & Farke, 2013), the occipital region of N. engaeus faces posteroventrally and it adopts an antinatural position when trying to articulate the skull with the cervical osteoderms of the dorsal armour of this aetosaur as well as when articulating the occipital condyle with the atlas. Following the discussion presented by Kley et al. (2010) for the pseudosuchian S. clarki, the central portion of the palate and ventral surface of the braincase can be used as reference to orient the posture of the head. When orienting the skull of N. engaeus with the palate and ventral surface of the braincase, the lateral semicircular canal is almost horizontal, tilted slightly anterodorsally at five degrees from the ground, whereas the snout is oriented anteroventrally at approximately 27 degrees from the horizontal (Fig. 3A).
Holding the skull in this position and the morphology of the olfactory region, different from that of an herbivorous aetosaur (D. spurensis), would support the proposal of animalivorous habits for this aetosaur. These animalivorous habits for N. engaeus were previously proposed by Desojo & Vizcaíno (2009) based on its jaw biomechanics and dental morphology. The conical teeth of Neoaetosauroides engaeus do no exhibit wear facets, differing from the leaf-shaped, serrated teeth with wear facets of several aetosaurs that were better adapted for crushing, chopping, and slicing and evidenced some degree of food processing capacities. The jaw biomechanics of N. engaeus allowed inferring bite movements faster than those of herbivorous aetosaurs (Desojo & Vizcaíno, 2009). These fast bite movements suggested for N. engaeus could allow catching prey such as invertebrates and microvertebrates, a condition consistent with the neuroanatomical features discussed above. Therefore, N. would occupy a different role within Triassic continental communities, representing animalivorous habits probably alongside with other aetosaurs such as Longosuchus meadei, Aetosaurus ferratus, Aetosauroides scagliai and Paratypothorax sp. based on their cranial morphology (Desojo & Báez, 2005;Desojo & Báez, 2007). This differs from the traditional herbivorous condition proposed for aetosaurs, which is generally based on the northern hemisphere taxa, such as D. spurensis and Stagonolepis robertsoni (Walker, 1961;Small, 2002;Desojo et al., 2013).

CONCLUSION
Here we carried out the first description of a natural endocast of an aetosaur, N. engaeus, and also developed the first digital endocast which complemented the former including delicate structures such as the inner ear and the middle ear sinus. The anatomy of the endocast of N. engaeus exhibited remarkable differences (e.g., location of CNs VII, VIII, XII) when compared with the neuroanatomical features previously known for aetosaurs, based only in D. spurensis (Case, 1921;Hopson, 1979;Stocker et al., 2016). The new neuroanatomical information provided by this contribution allowed us to make an incursion about the paleobiology of N. engaeus supporting this taxon as an animalivorous aetosaur, which differs from the traditional interpretation of aetosaurs as exclusively herbivorous pseudosuchians.
These results contribute to the knowledge of aetosaurs and their paleoneurology, a topic poorly studied among pseudosuchians but crucial for the understanding of archosaur paleobiology and evolution.