On the presence of the subnarial foramen in Prestosuchus chiniquensis (Pseudosuchia: Loricata) with remarks on its phylogenetic distribution

Many authors have discussed the subnarial foramen in Archosauriformes. Here presence among Archosauriformes, shape, and position of this structure is reported and its phylogenetic importance is investigated. Based on distribution and the phylogenetic tree, it probably arose independently in Erythrosuchus, Herrerasaurus, and Paracrocodylomorpha. In Paracrocodylomorpha the subnarial foramen is oval-shaped, placed in the middle height of the main body of the maxilla, and does not reach the height of ascending process. In basal loricatans from South America (Prestosuchus chiniquensis and Saurosuchus galilei) the subnarial foramen is ‘drop-like’ shaped, the subnarial foramen is located above the middle height of the main body of the maxilla, reaching the height of ascending process, a condition also present in Herrerasaurus ischigualastensis. These results suggest that this structure might be phylogenetically important and further investigation with a large set of valid taxa is necessary to properly evaluate its importance among Archosauria.

pretation due the supposed mobile joint between the premaxilla and maxilla (Mastrantônio 2010). Additionally, in cranial material of "Rauisuchia", this joint may have suffered taphonomic distortions (Nesbitt 2011, Lacerda 2012. The aims of this contribution are the evaluation of new evidence regarding presence, size, shape, and position of the subnarial foramen in Prestosuchus chiniquensis and discuss its implication for archosaurian phylogeny. The study is based upon a new nearly complete individual of P. chiniquensis from the municipality of Dona Francisca (central region of Rio Grande do Sul State, southern Brazil). The specimen (
The premaxilla possesses two long dorsoposteriorly oriented projections: the anterodorsal and posterodorsal (= postnarial) processes. In ULBRA-PVT-281, both processes meet the nasal, forming both anterior and posterior margins of the external naris. Unlike some "Rauisuchia" (e.g. Batrachotomus kupferzellensis), in ULBRA-PVT-281 the maxilla does not participate in the external naris. The posteroventral margin of the posterodorsal process contacts the ascending process of the maxilla and forms an angle of 55°, in relation to the body of premaxilla, contrasting with 45° observed in Decuriasuchus quartacolonia (França et al. 2013). The anterodorsal process is semicircular in crosssection. The anterior margin of the main body of the premaxilla is straight and vertical (not sinuous). The contact between premaxilla and maxilla is slightly sigmoidal. The ventral margin of the premaxilla is horizontal, bearing four teeth. This number of teeth is shared with several loricatans, such as Saurosuchus, Fasolasuchus, Batrachotomus, Postosuchus, Polonosuchus, and other specimens of Prestosuchus (UFRGS-PV-0629-T; UFRGS-PV-0156-T) (Barberena 1978, Bonaparte 1981, Long and Murry 1995, Gower 1999, Weinbaum 2002, Sulej 2005, Mastrantônio 2010). The medial surface is not visible due to lower jaw occlusion, precluding morphological description of the medial surface, interdental plates and palatal process of premaxilla. Subnarial foramen of Prestosuchus chiniquensis. In the first description of a complete skull of Prestosuchus chiniquensis (UFRGS-PV0156T) by Barberena (1978), the subnarial foramen was not recognized and it was acknowledged as absent. Nesbitt (2011), in his extensive phylogenetic analysis of archosaurs, also argued that this structure is absent in Prestosuchus chiniquensis and Saurosuchus galilei, drawing attention to taphonomic distortion in both species. Based upon new material, this structure was recently mentioned as present in Prestosuchus chiniquensis (UFRGS-PV-0629-T; Mastrantônio 2010). However, França et al. (2013) suggested that the subnarial foramen was also present in the specimen described by Barberena (1978 -UFRGS-PV-0156-T) regardless the unquestionable fact that this particular specimen is clearly taphonomically distorted. Accordingly, Mastrantônio (2010) sustained that the shape of this structure should be considered with caution as a probable movable joint between maxilla and premaxilla that could have been distorted in UFRGS-PV-0156T during taphonomic processes. The present specimen (ULBRA-PVT-281, ascribed to P. chiniquensis) does not present any visible sign of taphonomic distortion and clearly possess the subnarial foramen, which is well evident between the maxillary/premaxillary suture (Fig. 5a). Both shape and position of the subnarial fenestra varies in rauisuchians as relative medium-sized foramina occur in Prestosuchus and Postosuchus, whereas a large fenestra is observable in Luperosuchus fractus (Fig. 5c). In the latter taxon, a taphonomic or disarticulated origin of fenestra is questionable, added to uncommon position of fenestra. Other remarkable feature of ULBRA-PVT-281 is the presence of a well-marked groove, which starts at the premaxilla-maxilla suture and runs anteroventraly to the premaxillary body. Previously, this premaxillary groove was not mentioned to any other known specimen of Prestosuchus. However, at this point taphonomic artifact cannot be ruled out regarding this structure, as both taphonomically distorted specimens UFRGS-PV-0156-T and 0629-T (ascribed to P. chiniquensis) presents this feature in one side of the skull, whereas, in the   Weinbaum 2002, Lautenschlager 2008, Desojo and Arcucci 2009, França et al. 2011. Therefore, further examination of more specimens attributable to P. chiniquensis is necessary in order to solve this question. If it is not taphonomic, the premaxillary groove would represent an autapomorphic character of P. chiniquensis. This groove differs from those described by Nesbitt (2011) for Revueltosaurus and Erythrosuchus, because, in these taxa, the groove is posteriorly directed on the maxilla. Conversely, in Prestosuchus chiniquensis (ULBRA-PVT-281) this groove is anteriorly directed on the premaxilla. In Prestosuchus chiniquensis the subnarial foramen is a drop-like shaped opening located between premaxilla and maxilla, both equally contributing to its borders in ULBRA-PVT-281. The height of the foramen occupies part of the main bodies and posterodorsal processes of maxilla/premaxilla, as in Saurosuchus galilei (Fig.  5b) (Alcober 2000). In Saurosuchus galilei, the subnarial foramen is slit-like in juvenile specimens (Alcober 2000). According to Nesbitt (2011), this structure would be absent in this taxon and those alleged openings would be artifact of preservation. However, in the holotype of S. galilei (PVL 2068), considered an adult individual, there is a subnarial foramen, in spite of the fact that this specimen shows signs of taphonomic distortion in some degree. Its morphology and position is quite similar to that of ULBRA-PVT-281. In Batrachotomus kupferzellensis, Fasolasuchus tenax, Effigia okeeffeae, Shuvosaurus inexpectatus, Qianosuchus mixtus, Lotosaurus adentus, and Postosuchus kirkpatricki the subnarial foramen is oval, located at the average height regarding the main body of maxilla, not reaching the height of the posterodorsal process (Chatterjee 1985, Bonaparte 1981, Gower 1999, Nesbitt and Norell 2006, Weinbaum 2011. Additionally, in Decuriasuchus quartacolonia the subnarial foramen occupies part of main body of maxilla and reaches the base of the ascending process, resembling Prestosuchus chiniquensis and Saurosuchus galilei (Alcober 2000, França et al. 2013. Despite the vast literature regarding "Rauisuchia", occurrence/description of the subnarial foramen is dubious in some taxa. For instance, in Polonosuchus silesiacus (Sulej 2005) the presence/absence of such feature is doubtful due the lack of a preserved premaxilla and maxilla in articulation.

PHYLOGenetic anaLYSiS
Phylogenetic status and inner relationships of "Rauisuchia" are still open to debate, in spite of many new descriptions and revisions (Brusatte et al. 2010, Nesbitt 2011, Lautenschlager and Rauhut 2014. Different results created many challenges to provide a satisfactory phylogenetic definition to this group (Nesbitt et al. 2013). Presence/absence of the subnarial foramen has been previously considered useful for taxonomical purposes; however, its phylogenetic utility was also challenged. Gower (2000) pointed out that using this structure is problematic due to its highly homoplastic distribution among Archosauria. Conversely, previous contributions claimed that the presence of the subnarial foramen is phylogenetically important to some rauisuchians (Benton and Clark 1988, Parrish 1993, Juul 1994). Nesbitt (2011) provided a comprehensive phylogenetic analysis of Archosauria, scoring the subnarial foramen (either absent or present and the contribution of maxilla and premaxilla to its borders) in his data-matrix (character 12), but did not recognize its presence in Prestosuchus chiniquensis. In this contribution, we change the scores of some operational taxonomic units (OTUs) regarding such character in the dataset of Butler et al. (2014), which is a modified version of the matrix presented by Nesbitt (2011). Originally, Prestosuchus chiniquensis and Saurosuchus galilei were scored as lacking the subnarial foramen (0). In this contribution, in both species the subnarial foramen is scored as present and limited by both maxilla and premaxilla (character state 1). The herrerasaurid Herrerasaurus ischigualastensis bears two openings along the suture between premaxilla and maxilla (Sereno and Novas 1993), in which the more ventral is topologically similar to saurischian dinosaurs (e.g. Efraasia minor, Plateosaurus engelhardti, and Eoraptor lunensis), whereas the larger and more dorsal opening is topologically equivalent to that present in pseudosuchians (Fig.  5h). Therefore, among evaluated saurischians, just Herrerasaurus ischigualastensis is scored here with the condition (1). In the remaining saurischians, this structure is considered absent (character state 0). Another modification in the dataset of Butler et al. (2014) includes the addition of a new character (414) related to the position of the subnarial foramen. Such character includes two states: (0) in the middle average height regarding the maxillary body, not reaching the ascending process; (1) above the middle height of the main body of the maxilla, reaching at least the base of the ascending process. Those taxa without subnarial foramen are encoded as inapplicable "-". The subnarial foramen of Crocodylomorpha is considered inapplicable, as it is an artifact resulting from the insertion of the caniniform teeth in the maxilla, not being homologous to the subnarial foramen (the full codification of the new character is in Appendix I).
The analysis was performed under equally weighted parsimony using the software TNT 1.1 (Goloboff et al. 2003(Goloboff et al. , 2008. A traditional search with 100 replicates of Wagner trees (with random addition sequence) followed by TBR branch-swapping (holding 10 trees per replicate) was performed. The analysis recovered 81 most parsimonious trees (MPTs) of 1312 steps each (consistence index=0.369; retention index=0.771). Bremer and bootstraps values (1000 replicates) were also obtained from TNT v 1.1 (Goloboff et al. 2008). Decuriasuchus quartacolonia (Fig. 5d) was not included in this dataset because it is currently under review by M.A.G.F.

DiScUSSiOn
In spite of the discussion regarding the subnarial fenestra/foramen in archosaurs, there is no consensus about its function. According to Gower (2000), two hypotheses are in dispute, the first would be related to the air sinus system and the second to either blood vessels or nerve transmission. However, no evidence has been provided to corroborate either. In  (Desojo and Arcucci 2009), the subnarial foramen is only mentioned as present and no additional information is provided for this structure.
Despite the possible phylogenetic importance of the subnarial foramen, recent contributions discuss inconsistencies and alternative interpretations regarding this structure (Gower 2000, Mastrantônio 2010, Lacerda 2012. According to Gower (2000), its presence is homoplastic amongst "Rauisuchia". Additionally, a similar opening occurs in other nonrauisuchian archosaurs, such as the erythrosuchid Shansisuchus shansisuchus, the herrerasaurid dinosaur Herrerasaurus ischigualastensis (Sereno and Novas 1993), and pterosaurs (e.g. Dorygnathus banthensis) (Ösi 2010). Moreover, in Shansisuchus shansisuchus, for instance, the subnarial foramen is formed by a bifurcation of the anteroventral process of the nasal and extends ventrally interrupting the premaxilar/maxilar contact (Fig. 5g). Therefore, this configures an analogous condition in relation to the feature present in "rauisuchians". In addition, this structure was recognized as an autapomorphy for the saurischian dinosaur Herrerasaurus ischigualastensis (Sereno and Novas 1993). So, it would have been convergently developed in taxa other than rauisuchians Arcucci 2009, Lacerda 2012). On the other hand, the subnarial foramen was recognized as a diagnostic character of "Rauisuchia" (Chatterjee 1985, Gower 2000, Lautenschlager 2008, Lautenschlager and Rauhut 2014. Both presence and shape of the subnarial foramen is uncertain in many rauisuchians, as the majority of specimens are incomplete or with disarticulated bones. Additionally, the shape of the subnarial foramen would be subject to both taphonomic distortion and ontogenetic variation as well (Alcober 2000, Nesbitt 2011). Based on specimens of Saurosuchus galilei, Alcober (2000) suggested that the subnarial fenestra is only present in juvenile individuals, fully closing in adults. However, this statement is dubious as the ontogenetic state of S. galilei specimens is uncertain. Actually, the two specimens of Saurosuchus galilei (PVL 2062 and PVSJ 32, respectively putative adult and young individuals) show an evident opening between premaxilla and maxilla. In addition, putative adult specimens of Prestosuchus chiniquensis (UFRGS-PV-0156-T; UFRGS-PV-0629-T; ULBRA-PVT-281) possess this structure. There are also differences between UFRGS-PV-0629-T and ULBRA-PVT-281. In the first, the subnarial foramen is slit-like and elongated (Mastrantônio 2010), whereas in ULBRA-PVT-281 it is "drop-like" shaped. According to Mastrantônio (2010), these differences can be explained by the presence of kinesis between maxilla and premaxilla (a mobile joint between these two bones). In both cases, Mastrantônio (2010) argued that this joint would show distortions due to diagenesis, resulting in different shapes and sizes. Given the accentuated lateral deformation of UFRGS-PV-0629-T, we consider that the elongated shape of its subnarial foramen is an artifact of preservation (Fig. 5f).
The presence of relative movement among intracranial joints is discussed by many paleontologists. Different types of kinesis were defined according to the joint location in dorsal parts of the skull (Versluys 1910, 1912, 1936, Holliday and Witmer 2008. In extant taxa, as in most lizards, there are three types of cranial kinesis: mesokinesis (between frontal/parietal), metakinesis (quadrate/paroccipital) and pleurokinesis (occipital condyle laterally) (Rieppel and Zaher 2001). So far, extant taxa lack mobility between maxilla and premaxilla. Additionally, in several extant lineages, such as turtles, crocodilians, and Sphenodon, cranial kinesis is totally absent (Iordansky 1990, Holliday andWitmer 2008). In neither case, a kinetic joint between premaxilla and maxilla was mentioned by the aforementioned authors. Moreover, according to Holliday and Witmer (2008), an animal may exhibit morphological features suggestive of intracranial mobility but not necessarily demonstrate significant functional mobility in vivo. Based upon these arguments, we state that a kinetic joint between premaxilla and maxilla is possible only if new data were obtained.
In Postosuchus kirkipatricki there is an excavation (visible in medial view) that starts in the premaxilla and ends in the maxilla. According to Chatterjee (1985) it would serve to accommodate the Jacobson's organ (also called vomeronasal organ), a chemoreceptor which is part of the olfactory system of amphibians, reptiles, and mammals, although it does not occur in all tetrapod groups (Rehorek et al. 2000). Probably, due to the presence of a Jacobson's organ there would be a movable premaxillary/maxillary joint in this taxon. However, other authors (Sill 1974, Bonaparte 1981, Witmer 1995, Gower 1999, Sulej 2005, Weinbaum 2011) do not agree with the presence of this organ in Postosuchus. In some basal archosauriforms, like Euparkeria capensis (Ewer 1965), the anterior region of lateral surface of maxilla bears a foramen. Most rauisuchians have a rostrolateral foramen on the anterior surface of maxilla, as Decuriasuchus quartacolonia (França et al. 2013), whereas others also have a rostromedial foramen, as Polonosuchus and Teratosaurus (Sulej 2005). The presence of these structures in taxa with distinct affinities indicate that this region would be the end extremity for blood vessels or nerve transmission, independently if subnarial foramen is present.
The inclusion of the new character in the phylogenetic analysis did not change the topology of "Rauisuchia" presented by Butler et al. (2014) and the results allow the proposal of the likely evolutionary pathway for the subnarial foramen among paracrocodylomorphs (Fig. 6). The presence of the subnarial foramen probably begins in the common ancestor of Paracrocodylomorpha, in which the small subnarial foramen is oval shaped in average height regarding the main body of maxilla, not reaching the base of the ascending process [414 (0)  tenax, and Postosuchus kirkpatricki (Fig. 5e). In basal loricatans from South America, Prestosuchus chiniquensis and Saurosuchus galilei, the subnarial foramen presents a new condition, related to its position: located above the middle height of the main body of the maxilla, reaching the base of the ascending process [414(1)]. Although not revealed in this analysis, this could suggest the presence of a less inclusive monophyletic group composed of some rauiuchians from Gondwana, as Prestosuchus, Saurosuchus and, possibly, Decuriasuchus, a hypothesis to be further evaluated. Considering only the pseudosuchian lineage, the subnarial foramen is shared by several pseudosuchian taxa other than Aetosauria, Crocodylomorpha, Gracilisuchidae or Ornithosuchidae (e.g. Nesbitt 2011). In this context, the presence of the subnarial foramen is a possible synapomorphy of Paracrocodylomorpha, being small, oval-shaped and restricted to main body of maxilla (Fig. 6). This condition is retained in Poposauroidea, but modified to drop-like, occupying the main body and ascending process of maxilla in basal Loricata, as Prestosuchus and Saurosuchus. Subsequently, the condition is reversed to a small oval-shaped fenestra in the clade including Batrachotomus, Fasolasuchus and Rauisuchidae.
Despite the close geographic distribution of Prestosuchus and Saurosuchus, as well as several shared characteristics including the presence and shape of subnarial foramen, our phylogenetic analysis did not recover them as sister taxa . Concluding, the results presented here suggest that there is a phylogenetic signal linked to the presence and position of the subnarial foramen in Archosauria. However, the lack of knowledge regarding its presence in several fragmentary and badly-preserved taxa is an obstacle to the understanding of the importance of this structure among this large group of Archosaurs, stressing the necessity of discover of more complete and better preserved specimens to clarify this issue.

acKnOWLeDGMentS
We thank the Fundação de Apoio à Tecnologia e Ciência (FATEC -process 3.01.0046) for the financial support to LRS; the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) for the research grant to SDS (process 301801/2012-6); the Coordenação de Aperfeiçoamento de Pessoal de Nivel Superior (CAPES) for the scholarship to RTM; and the Fundação de Amparo à Ciência e Tecnologia do Estado de Pernambuco (FACEP) (process APQ-0165-2.04/14) for the financial support to MAGF. We thank the Willi Hennig Society, for the gratuity of TNT software. We also thank Martin Ezcurra and an anonymous referee for their valuable suggestions that greatly improved the manuscript.