The monophyly of Susisuchidae (Crocodyliformes) and its phylogenetic placement in Neosuchia

Specimens examined

Comparative sample

Fossil neosuchian material used for comparisons is listed in Table 1. The table lists the taxon name, the collection number of the most informative specimen, and the reference for that taxon. Comparisons for the listed taxa are based on these specimens unless otherwise noted. The terminology for the major crocodyliform clade names Mesoeucrocodylia and Neosuchia follow Clark (in Benton & Clark, 1988). The term “advanced neosuchians” is used here for a group of deeply nested neosuchians more crownward than Goniopholididae and includes such taxa as Shamosuchus, Paralligator, Acynodon, Allodaposuchus, Borealosuchus, and Crocodylia.

Phylogenetic analysis

Dataset and analysis

The dataset used here is identical to that of Turner (in press). Details of the character set are in Appendix S2. Reference specimens and literature consulted for information on the ingroup taxa are available in Appendix S1 and the full dataset, as well as files related to the sensitivity analyses, are available on MorphoBank (O’Leary & Kaufman, 2007) at www.morphobank.org/permalink/?P1200. The phylogenetic dataset was analyzed with equally weighted parsimony using TNT v.1.0 (Goloboff, Farris & Nixon, 2008a; Goloboff, Farris & Nixon, 2008b). A heuristic tree search strategy was conducted performing 10,000 replicates of Wagner trees (using random addition sequences, RAS) followed by TBR branch swapping (holding 10 trees per replicate). The best trees obtained at the end of the replicates were subjected to a final round of TBR branch swapping. Zero-length branches were collapsed if they lacked support under any of the most parsimonious reconstructions (i.e., rule 1 of Coddington & Scharff, 1994).

The character support of the nodes present in the most parsimonious reconstructions was calculated using two different methods. The first technique is the jackknife applied to character resampling (Farris et al., 1996). The second method used is Bremer support (Bremer, 1988; Bremer, 1994), which evaluates node stability/sensitivity by exploring suboptimal tree solutions in order to determine how many additional steps must be allowed in searching for topologies before the hypothesized clade is no longer recovered. The jackknife support analysis was calculated using TNT (Goloboff, Farris & Nixon, 2008a; Goloboff, Farris & Nixon, 2008b). The analysis was performed using 1,000 replicates for which the probability of independent character removal was set to 0.20. Each jackknife replicate was analyzed using a tree search strategy consisting of 10 replicates of RAS followed by TBR branch swapping (saving 10 trees per replicate). The topologies obtained during the jackknife replicates are summarized using GC frequencies (Goloboff et al., 2003). Bremer support was calculated using the BREMER.RUN script provided with TNT.

Morphological observations and dataset changes

Secondary palate

The paratype skull of Isisfordia (and only specimen of a susisuchid preserving the choana or with an exposed palate) presents a challenge for interpreting the sutural relationships between the pterygoid and palatines. A fracture split the skull diagonally into two pieces, one containing the snout and the other the braincase, posterior portion of the palate, and quadrates. The two pieces were skillfully fit together but the fracture resulted in damage to the nasopharyngeal passage and anterior margin of the pterygoid where it contacts the palatine (Fig. 1). Fortunately, the left pterygoid and palatine are not as extensively damaged as their contralateral pair. Thus, whereas some morphological information is indeed present in its current state, the interpretation of the sutural relationships of the palatine with the pterygoid is a bit more ambiguous than one would like. Further complicating interpretation of sutural relationships is the presence of heavy black mineralization in the form of splotches and veins. The pterygoid has numerous micro-fractures that are also often associated with black mineralization. Lastly, small ridges and grooves frame the lateral margin of the choana, adding further complexity to the region where one would expect the palatine/pterygoid contact to be.

Salisbury et al. (2006) reconstructed the palatine-pterygoid contact in an irregular suture slightly anterior to the choana. Thus, Isisfordia was considered to have the derived eusuchian condition of an entirely pterygoid-bound choana. The contact identified by Salisbury et al. (2006) as sutural we interpret here as a thin, partially transmitting fracture on the palatine. We interpret a more posterior line, lapping on the pterygoid in ventral view and extending distally before curving back anteriorly to form the anterior margin of the choana, to be the sutural contact between the palatine and pterygoid (Figs. 1–3). Therefore we view Isisfordia as lacking the derived pterygoid-bound choana.

A number of lines of evidence support this interpretation. If our interpretation of Isisfordia is correct, it suggests that the posterior ends of the palatines expand laterally just anterior to the pterygoid contact (Fig. 1). Similar palatine expansions occur in many other advanced neosuchians (Pol, Turner & Norell, 2009; Turner, in press). This is evident in paralligatorids like Shamosuchus djadochtaensis, Rugosuchus nonganensis, and Paralligator gradilifrons, as well as the putative paralligatorid Batrachomimus pastosbonensis. It also appears to be the condition in Theriosuchus pusillus. This contrasts with palatines that remain narrow throughout their lengths (e.g., Elosuchus cherifiensis, Eutretauranosuchus delfsi, Wannchampsus, Glen Rose Form). In no eusuchians that possess laterally expanded palatines just anterior to the pterygoid contact (e.g., Allodaposuchus subjuniperus, Allodaposuchus cf. precedens, Acynodon iberoccitanus, Alligator mississippiensis, Crocodylus acutus) does the pterygoid/palatine suture cut across this posterior expansion of the palatines. Instead, the expanded area of the nasopharyngeal passage is always formed by the palatines, with the palatines contacting the pterygoid posterior to the expansion.

In most advanced neosuchians outside of Eusuchia, each palatine is overlapped by the pterygoid in a shallow interdigitating suture before curving back to the midline (e.g., Goniopholis simus, Wannchampsus, Bernissartia, Paralligator). This results in the posterior margin of the joined palatines forming a curved or V-shaped anterior margin to the choana. The pterygoid extends anteriorly at a very shallow angle along the dorsal surface of the palatines to frame the dorsolateral portions of the nasopharyngeal passage. We see a similar conformation of palatine and pterygoid in Isisfordia. Tracing the lateral margin of the palatine along its expanded posterior end, we see it extend onto the ventral surface of the pterygoid and form a short suture before curving back anteriorly (Figs. 2 and 3). Thus the anterior margin of the choana is palatine-formed and takes a gentle curved shape like that seen in paralligatorids (Turner, in press), Bernissartia, and some dyrosaurids (e.g., Rhabdognathus) and unlike the V-shape margin of Theriosuchus pusillus or the narrow straight margin of Wannchampsus (Adams, 2014) or the Glen Rose Form (Langston, 1973; Langston, 1974).

An oblique lateral view of the nasopharyngeal passage on the left side reveals the overlapping contact of the pterygoid on the dorsal surface of the palatine (Fig. 2). Tracing this contact posteriorly, it appears continuous with the line on the pterygoid we have interpreted to be the palatine/pterygoid contact. Thus, it is our view that the choana of Isisfordia is not pterygoid-bound but instead shows an intermediate neosuchian condition wherein the palatines contribute extensively to the posteriormost floor of the nasopharyngeal passage, flare laterally at their contact with the pterygoid, but still form a gently curved anterior margin to the choana.

Vertebrae

Vertebral material has been described from two specimens of Susisuchus anatoceps (see Salisbury et al. (2003) and Figueiredo et al. (2011)). The holotype (SMNK 3804 PAL) vertebral series consists of the axial column up to the mid-caudal region. This is preserved on a limestone slab and only exposed in dorsal view. The cervical series is partially obscured by the nuchal osteoderms. The cervical and dorsal vertebrae referred to Susisuchus anatoceps by Figueiredo et al. (2011) are three-dimensionally preserved, prepared in articulation, and described in detail.

Vertebral material for Isisfordia duncani is likewise abundant. The holotype QM F36211 preserves articulated portions of the cervical, dorsal, and caudal regions of the column. The preserved segment of the cervical column in QM F36211 contains the first nine cervical vertebrae and the first dorsal vertebra in articulation (Fig. 4A). None of the vertebrae are complete as a portion of the right half of each has been sheared off. The vertebrae are visible in right lateral view, such that one observes the sheared surface of the elements. Cervicals 6 through 9 (C6–C9) are represented mostly by the neural arch. Vertebral centra are better preserved on C2 through C5. The axial centrum appears to lack a hypapophysis, a condition shared with a specimen referred to Susisuchus anatoceps (Figueiredo et al., 2011). A subtle convexity is apparent on the posterior surface of the centra of C2, C3, and especially C5 (Fig. 4B). Further posteriorly, the centra are too poorly preserved to assess this morphology. This subtle convexity is what Salisbury et al. (2006) characterized as “incipient procoely”, and is similar to the weak procoely observed by Figueiredo et al. (2011) in a referred specimen of Susisuchus anatoceps. This minor development of procoely is in contrast to the well-developed procoely present in many other advanced neosuchians (e.g., Shamosuchus, Pachycheilosuchus, Acynodon adriaticus) (Fig. 5), and as such appears transitional from the plesiomorphic amphicoelous condition.

The preserved segment of the dorsal region in QM F36211 preserves nine partial dorsal vertebrae in articulation and fragments of a tenth posteriorly (Fig. 4C). As in the cervical region, these have been sheared through vertically, although these are visible in section from a left lateral perspective. In each preserved centrum, a subtle posterior convexity is apparent.

Phylogenetic analysis

Maximum parsimony analysis recovered 108 optimal trees with a length of 1,662 steps (CI = 0.239, RI = 0.700). A reduced strict consensus of the trees (Fig. 6) unites Isisfordia and Susisuchus as a monophyletic Susisuchidae. This clade is the secondmost basal divergence within Neosuchia. A pholidosaurid + dyrosaurid clade is at the base of Neosuchia. Susisuchidae is separated from Eusuchia by Paluxysuchus and a goniopholidid clade. Eusuchia is comprised of three large clades: Crocodylia, a Paralligatoridae + Theriosuchus clade, and a speciose Hylaeochampsidae. The derived position of Shamosuchus and other paralligatorids, not recovered in most analyses of mesoeucrocodylian phylogeny, is discussed in detail by Turner (in press). Support metrics are generally low for most neosuchian clades. Susisuchidae has a Bremer value of 4 and GC of 57, representing moderate clade support and some of the higher values among neosuchians. The susisuchid + derived neosuchian node also has a Bremer value of 4 but was not recovered in the jackknife analysis. The discussion that follows will concentrate on character support for the basal position of Susisuchidae and the phylogenetic sensitivity of the clade to the morphological observations made above.

Phylogenetic position of Susisuchidae

Reanalysis of the phylogenetic position of putative  susisuchids incorporating character state changes based on the observations discussed above results in Isisfordia and Susisuchus no longer being eusuchians but instead occupying a more basal position within Neosuchia compared to previous analyses (e.g., Salisbury et al., 2006; Pol, Turner & Norell, 2009; Andrade et al., 2011). Six synapomorphies unite Isisfordia and Susisuchus. Citation of characters from the phylogenetic dataset follows the convention of (character.character state). Characters uniting Susisuchidae include nasals that contribute to the border of the naris (13.0), dorsal osteoderms that lack an anterior articular process (96.0), symmetrical lateral compression of maxillary teeth (140.2), exposure of the supraoccipital on the skull roof (171.1) and the absence of a shallow fossa at the anteromedial corner of the supratemporal fenestra (265.1). One character, the presence of a pear-shaped external naris (309.1) is unique to susisuchids.

A seventh character may also diagnose this clade. Both Isisfordia and Susisuchus anatoceps each have an extremely large incisive foramen. This morphology is currently represented in the dataset by character 7, which codes for whether the palatal parts of the premaxillae meet behind the incisive foramen. They do not meet in Isisfordia due to the massive size of the foramen. In Susisuchus, because of preservation, it is uncertain if the massive foramen similarly interrupts the premaxillae posteriorly, although it seems likely that it does.

Susisuchidae is united with neosuchians more derived than the dyrosaurid + pholidosaurid clade based on six synapomorphies. These include changes to the skull like a platyrostral snout (3.3), a cylindrical postorbital bar (26.1), and the jaw joint placed at the level of the occipital condyle (105.0). Additionally, hypapophyses are only present on the cervical vertebrae (91.1) and the posterior half of the axial neural spine is narrow (258.1). Members of this clade of neosuchians are also characterized by a radiale with the proximal end more expanded proximolaterally than proximomedially (117.1).

Susisuchids occupy a basal divergence within Neosuchia due to the absence of a number of features present in more advanced forms. Susisuchids have no variation in maxillary tooth size (79.0), whereas Paluxysuchus and the majority of sampled advanced neosuchians have two waves of maxillary tooth enlargement (79.2). This trait is further derived to a single wave of tooth enlargement (79.1) in hylaeochampsids and in most atoposaurids (not Theriosuchus pusillus) and Shamosuchus djadochtaensis. In dorsal view, susisuchids have a straight lateral margin of the snout (178.0) and thus lack the sinusoidal lateral contour seen in nearly all advanced neosuchians (178.1). Likewise, Isisfordia has transversely expanded prefrontal pillars (182.0) whereas all other advanced neosuchians have prefrontal pillars that are transversely expanded in their dorsal part and columnar ventrally (182.1) or longitudinally expanded in their dorsal part (182.3) as in brevirostrine crocodylians. Lastly, susisuchids have a straight ventral margin of the maxilla in lateral view (183.0), whereas all other advanced neosuchians have a sinusoidal central margin of the maxilla in lateral view (183.1—reversed in Calsoyasuchus, Theriosuchus sympiestodon, Iharkutosuchus, Acynodon, and Argochampsa).

Sensitivity to interpretation of palate and vertebral morphology

To examine how sensitive the phylogenetic placement of susisuchids  are to the interpretation of the palate and vertebral morphology, alternate scorings for Isisfordia were made and phylogenetic trees were estimated from these alternate datasets. Scoring Isisfordia as derived for a eusuchian-style palate (43.1) but plesiomorphic for vertebral centrum morphology (92.0; 93.0) results in trees 1,663 steps long (Fig. 7A). This is one step longer than the most parsimonious trees from the primary analysis but, in this case, there is significant reduction in phylogenetic resolution among neosuchians. The clade containing Theriosuchus + paralligatoridae is monophyletic and resolved, as is Crocodylia, Susisuchidae, and most of Hylaeochampsidae.

Scoring Isisfordia as derived for procoely (92.1; 93.1) but not the palate (43.0) results in trees 1,664 steps long and with even less resolution among neosuchians (Fig. 7B). Only Crocodylia and the Theriosuchus + Paralligatoridae clade remain resolved. Isisfordia is in a large polytomy at the base of advanced neosuchians.

Scoring of Isisfordia as derived for a eusuchian-style palate (43.1) and for procoely (92.1; 93.1) results in trees 1,663 steps long (Fig. 7C). This is one step longer than the most parsimonious trees from the primary analysis. Tree topology among advanced neosuchians is broadly similar to that present in the primary phylogenetic analysis. The trees differ in that Goniopholididae is paraphyletic with respect to the rest of Neosuchia, Bernissartia is the sister taxon to the Theriosuchus + Paralligatoridae clade, and bizarrely, Isisfordia and Susisuchus nest within hylaeochampsids but do not always form a susisuchid clade.

Whereas our reinterpretation of the palate construction of Isisfordia unquestionably changes how it is scored within our dataset (i.e., it should not be scored as having a pterygoid-bound choana), we do agree with the previous characterization of the vertebral morphology as being “weakly procoelous” or “incipient” in the degree of procoely present. Our primary departure therefore lies in our choice not to homologize the weak procoely of Isisfordia and Susisuchus (at least MPSC-R1136) with the well-developed procoely in other advanced neosuchians, as has been done by various authors subsequent to its description (Salisbury et al., 2006; Andrade et al., 2011). If the morphological variation present in the vertebrae of Susisuchidae is to be distinguished from the plesiomorphic condition, that it is best done as a separate character state coding for the incipient procoely exhibited by these forms. We explored this option by including an additional incipient procoely state to character 92. Scoring Isisfordia and Susisuchus for this character state did not change the phylogenetic position of the two taxa or deviate in any other way from the primary analysis topology (Fig. 7D). The only difference was in the number of steps for the most parsimonious tree (in this case 1,659). Figueiredo et al. (2011) scored Susisuchus for a similar “semi-procoelous” state in their re-analysis of the phylogenetic position of the taxon based on the dataset of Jouve (2009). They recovered Susisuchus outside of a Shamosuchus + Crocodylia node, a topology differing from that presented here only in the placement of Goniopholididae.