Phylogenetic relationships of Necrosuchus ionensis Simpson, 1937 and the early history of caimanines




Cranial fragments associated with the holotype of Necrosuchus ionensis reveal a dorsally shifted foramen aereum on the quadrate and a long, slender descending process of the exoccipital lateral to the basioccipital and approaching the basioccipital tubera. The former suggests that Necrosuchus is an alligatoroid and not a crocodylid, as first suggested; and the latter that it is a caiman. The scapulocoracoid shows evidence of early closure of the synchondrosis, further supporting a caiman affinity. Although we cannot yet pinpoint the phylogenetic placement of Necrosuchus amongst caimans, it nevertheless establishes a caimanine presence in South America by the Early Palaeocene. A review of other Palaeocene–Eocene caimans reveals a complex biogeographical history suggesting multiple dispersal events between North and South America, even if the modern caiman assemblage is monophyletic.

© 2011 The Linnean Society of London, Zoological Journal of the Linnean Society, 2011, 163, S228–S256.

We generally recognize six living caiman species: the spectacled caiman (Caiman crocodilus), yacaré (Caiman yacare, sometimes treated as a subspecies of C. crocodilus), broad-snouted caiman (Caiman latirostris), black caiman (Melanosuchus niger), and two species of dwarf or smooth-fronted caiman (Paleosuchus palpebrosus and Paleosuchus trigonatus). Some of these – especially C. crocodilus– may actually be cryptic species complexes (Amato & Gatesy, 1994; Venegas-Anaya et al., 2008). These are the living representatives of Caimaninae, a group including C. crocodilus and all crocodylians closer to it than to Alligator mississippiensis. Monophyly of modern caimanines is broadly supported by morphological and molecular data (e.g. Densmore, 1983; Norell, 1988; Brochu, 1999, 2004a, Gatesy et al., 2003; Harshman et al., 2003; Aguilera, Riff & Bocquetin-Villanueva, 2006; Hrbek et al., 2007; Willis et al., 2007; Willis, 2009).

The question of caiman origins is simultaneously tantalizingly simple and frustratingly complex. On first sight, it looks straightforward: contemporary Mesoamerican Caiman appears to be a recent range extension from South America, where all other living caimans are found (Estes & Báez, 1985; Vanzolini & Heyer, 1985; Venegas-Anaya et al., 2008). Alligator mississippiensis and its closest relatives (Alligatorinae) are exclusively Laurasian, as are the immediate outgroups to crown Alligatoridae (Norell, Clark & Hutchison, 1994; Brochu, 1999; Martin, 2007). For vicariance to explain this distribution, alligatorines and caimanines (as well as their fossil outgroups) should have diverged by the end of the Jurassic; but alligatorines first appear in the Early Palaeocene, and molecular divergence estimates generally put the alligatorine–caimanine divergence at or near the Cretaceous–Palaeogene boundary (Hass et al., 1992; Roos, Aggarwal & Janke, 2007). A single dispersal event in the Late Cretaceous or Early Palaeogene seems sufficient to explain the origin of the caiman radiation (Sill, 1968).

At first, the fossil record of caimans and other crocodyliforms in South America appears consistent with patterns seen around the world through the Cenozoic. Crocodyliform diversity since the Campanian has been bimodal, with peaks in the Palaeocene–Eocene and Miocene that track global climatic patterns – peaks in crocodyliform diversity coincide with peaks in estimated mean annual temperature (Berg, 1964; Hutchison, 1982; Taplin, 1984; Russell & Wu, 1997; Markwick, 1998; Vasse & Hua, 1998; Böhme, 2003). As the planet warmed, more of the surface became compatible with crocodyliform physiology and, hence, there were more crocodyliforms – but the peaks have different phylogenetic structures. Many crocodyliforms in the Palaeogene peak come from geographically widespread lineages, but those in the Neogene peak represent endemic radiations (Brochu, 2003). Neogene crocodyliform assemblages in the Neotropics are dominated by sebecids, gryposuchine gavialoids, and caimanines (Langston, 1965; Gasparini, 1996; Langston & Gasparini, 1997; Aguilera et al., 2006; Paolillo & Linares, 2007; Riff & Aguilera, 2008; Riff et al., 2010; Scheyer & Moreno-Bernal, 2010) – groups that, by the Neogene, were found nowhere else.

However, this simplicity masks ambiguity. Some North American fossils appear to represent Palaeogene caimans (Busbey, 1989). Phylogenetic analyses suggest that one of these, Orthogenysuchus olseniMook 1924, from the Early Eocene of Wyoming, is closely related to Mourasuchus and not basal to all living caimans (Brochu, 1999). Another Early Eocene species from Wyoming, Tsoabichi greenriverensis, appears to be a basal caiman unrelated to Orthogenysuchus olseni (Brochu, 2010). This implies multiple dispersal events between North and South America.

Unfortunately, nearly everything we know of the caimanine fossil record is of Miocene age or younger. Most published fossils appear to represent extinct lineages and are not demonstrably closer to one living species than to another, and these are amongst the most bizarre crocodyliforms of all time –Mourasuchus, resembling a surfboard with lots of small teeth, and the gigantic, robust Purussaurus (Price, 1964; Langston, 1965; Bocquetin et al., 1991; Aguilera et al., 2006). Is the apparent increase in Neotropical crocodyliform endemism real or the artefact of an incomplete fossil record?

Until recently, unambiguous caimans in the South American Palaeogene were limited to EocaimanSimpson 1933, based on the type species (Eocaiman cavernensis) from the Early Eocene but subsequently reported from the Palaeocene (Eocaiman palaeocenicusBona, 2007) of Argentina. Eocaiman falls out as the basal-most caimanine in phylogenetic analyses (Brochu, 1999; Aguilera et al., 2006; Bona, 2007; Martin, 2007). Two additional putative caimans – Palaeocene Notocaiman stromeriRusconi 1937 from Argentina and Oligocene Caiman tremembensisChiappe 1988 from Brazil – are known primarily from partial dentaries and isolated postcranial material, and neither has been included in a phylogenetic analysis.

A species that might prove pivotal is Necrosuchus ionensisSimpson 1937 from the Early Palaeocene Salamanca Formation of Chubut Province, Argentina. The holotype consists of a right dentary, a partial postcranial skeleton, and scraps of bone referable to a single individual collected during the First Scarritt Expedition to Patagonia in 1931. Only the dentary was figured and described. Based on alveolar size patterns, Simpson (1937) thought Necrosuchus was a close relative of Leidyosuchus or Borealosuchus, which at the time were considered congeneric and allied with crocodylids.

In a review of Leidyosuchus, Brochu (1997) indicated that cranial fragments identified from the bone scrap, along with postcranial features, suggested a relationship with caimans. Leidyosuchus and Borealosuchus (neither of which is a crocodyloid) are known only from North America, and unambiguous crocodyloids are otherwise absent from South America until late in the Neogene (Gasparini, 1996). A relationship between Necrosuchus and living caimans would thus make biogeographical sense. However, the evidence in support of this assertion was neither formally tested with a phylogenetic analysis, nor was the morphological evidence in favour of a caimanine affinity described in detail.

The purpose of this paper is twofold: first, to describe more fully the holotype of Necrosuchus ionensis; and second, to review the evidence for its phylogenetic relationships and explore its implications for the early phylogenetic and biogeographical history of one of the major living crocodylian radiations.

Institutional abbreviations: AMNH, American Museum of Natural History, New York; FMNH, Field Museum of Natural History, Chicago; MACN, Museo Argentino de Ciencias Naturales, Buenos Aires; UF, Florida Museum of Natural History, University of Florida, Gainesville.


Crocodylia Gmelin, 1789 (sensuBenton & Clark, 1988)

Alligatoridae Cuvier, 1807 (sensuNorell etal., 1994)

CaimaninaeBrochu, 2003 (followingNorell, 1988)

N ecrosuchus ionensis Simpson, 1937

Holotype: AMNH 3219, right dentary with associated cranial fragments and partial postcranial skeleton.

Occurrence: Salamanca Formation, Estancia Las Violetas, Chubut Province, Argentina. Palaeocene, Peligran South American Land Mammal Age (SALMA). Further stratigraphical and locality information are provided by Simpson (1937).

Emended diagnosis: Alligatorid crocodylian with long, slender descending processes of the exoccipitals extending ventrally nearly to the basioccipital tubera (shared with caimanines); slender dentary with at least 18 alveoli; first four dentary alveoli widely spaced; dentary symphysis extends back to a level just behind the fourth dentary alveolus (shared with crown caimanines); splenial bears slender anterior process extending almost to dentary symphysis.

Description: Fragments of the skull were identified in boxes of bone fragments associated with the jaw. These include a partial right quadrate and the ventral-most tip of the braincase. However incomplete these are, they nevertheless preserve derived features that allow assessment of its phylogenetic position.

The preserved quadrate fragment (Fig. 1C–E) includes the distal end of the ramus. The medial hemicondyle is depressed ventrally relative to the lateral hemicondyle and is not dorsoventrally expanded. The foramen aereum is located on the dorsal, and not the dorsomedial, surface of the ramus.

Figure 1.

AMNH 3219, holotype, Necrosuchus ionensisSimpson, 1937, cranial fragments. A, basioccipital, posterior view. B, basioccipital, anterior view. C, right quadrate ramus, ventral view. D, right quadrate ramus, dorsal view. E, right quadrate ramus, condylar view. Abbreviations: exo, exoccipital; fae, foramen aereum; leu, lateral eustachian foramen; lhc, lateral hemicondyle of quadrate; meu, median eustachian foramen; mhc, medial hemicondyle of quadrate; n, notch on dorsal surface of quadrate ramus. Scale bar = 1 cm.

The braincase fragment (Fig. 1A, B) includes the basioccipital tubera and the base of the occipital condyle. The basisphenoid is not preserved, but its sutural surface with the basioccipital is preserved. On the left side, the ventral-most tip of the descending process of the exoccipital is preserved, revealing a slender structure that reaches nearly to the basioccipital tubera. The Eustachian foramina themselves are not preserved, but discontinuities in the ventral margin of the basioccipital indicate a median foramen at the ventral-most tip and lateral foramina dorsolateral to the median foramen.

The right dentary (Fig. 2) is largely complete, although its posterior end is distorted and the anterior margin of the external mandibular fenestra is not preserved. At least 18 alveoli are preserved. The posterior-most alveolus is incomplete, as its medial wall would have been the unpreserved splenial. The third and fourth alveoli are not confluent. The dorsal surface of the dentary is concave between the enlarged fourth and 13th alveoli.

Figure 2.

AMNH 3219, holotype, Necrosuchus ionensisSimpson, 1937, right dentary in dorsal (A), lateral (B), and medial (C) views. D and E are close-up images and line interpretations of the symphyseal region in medial view. Abbreviations: d4, fourth dentary alveolus; d13, 13th dentary alveolus; ds, dentary symphysis; mg, Meckelian groove; sp, articulation surface for splenial on dentary. Scale bars = 1 cm.

The dentary symphysis extended posteriorly to just behind the level of the fourth alveolus (Fig. 2A, D). The first four alveoli are spaced relatively widely apart, and the anterior margin of the dentary is orientated anteromedially, which would have given the articulated jaw an acute anterior end. The dorsal surface of the dentary is broad and smooth between the symphyseal surface and the tooth row.

The splenial is not preserved, but we can nonetheless see its attachment scar on the medial surface of the dentary (Fig. 2D, E). Simpson (1937) stated that the splenial entered the symphysis, but the anterior end of the splenial appears to have passed along the dorsal rim of the Meckelian groove and approached, but not actually reached, the symphysis itself. The anterior tip of the splenial was dorsal to the Meckelian groove.

Four procoelous dorsal vertebrae are preserved in articulation (Fig. 3A). The neural spines are rectangular in lateral view, and the transverse processes are missing. Neurocentral sutures are visible (Fig. 3B). At some point, someone numbered these vertebrae ‘D-5’ to ‘D-8.’ The basis for these identifications is not known, but the anterior-most of these preserves the base of a hypapophyseal keel. In living crocodylians, hypapophyses typically extend no further back than the tenth or 11th postaxial vertebra, and the first five or six dorsal vertebrae (as defined by the position of the parapophysis on the neural arch rather than the centrum) have hypapophyses that diminish in dorsoventral height posteriorly. The anterior-most vertebra (‘D-5’) may, indeed, be the fifth dorsal vertebra.

Figure 3.

AMNH 3219, holotype, Necrosuchus ionensisSimpson, 1937, axial skeletal elements. A, anterior dorsal vertebrae, right lateral view; B, close-up of neurocentral suture of ‘D-7,’ left lateral view; C, posterior-most two dorsal and sacral vertebrae, dorsal view (anterior to right); D, posterior sacral vertebra and rib, left lateral view. Abbreviations: h, base of hypapophysis; if, iliac facet on sacral rib; ncs, neurocentral suture; sr, sacral rib. Scale bars = 1 cm.

The last two dorsal vertebrae are also preserved in articulation with the anterior sacral vertebra (Fig. 3C). The sacral ribs have posterolaterally orientated distal facets for the ilium, and the anterior margins of the capitula lie sufficiently forward of the anterior margins of the tubercula to make the capitula visible in dorsal view. The posterior surface of the centrum lacks a hemispherical cotyle.

The intact left rib remains in articulation with the posterior sacral vertebra (Fig. 3D). Its articulation facet for the ilium is orientated laterally and slightly anteriorly. The anterior one-fifth of the facet is approximately circular in lateral view, and the remaining surface is dorsoventrally very thin. The anterior surface of the centrum is flat, and the posterior surface is deeply concave. As with the anterior sacral vertebra, all neurocentral and vertebrocostal sutures are visible.

The pectoral girdle is represented by parts of both scapulocoracoids, both humeri, both radii and ulnae, and the right radiale (Fig. 4A–J). The left scapulocoracoid is best preserved, and the coracoid is nearly complete. It bears a circular foramen anterior to the glenoid fossa and a mediolaterally thin, anteroposteriorly flared ventral blade. Neither scapular blade is preserved, and the remaining scapular bases preserve rugose attachment scars for the triceps longus lateralis muscle (Meers, 2003). The deltoid crest along the anterolateral side of the base is thin. On both scapulae, thin laminae extend from the ventrolateral and ventromedial margins across the joint toward the coracoid (Fig. 5A, B). This is very similar to the condition found in extant crocodylians as the scapulocoracoid synchondrosis closes (Brochu, 1995; Fig. 5C), and I interpret this as evidence that the synchondrosis was in the process of closing in AMNH 3219.

Figure 4.

AMNH 3219, holotype, Necrosuchus ionensisSimpson, 1937, postcranial appendicular elements. A, left scapulocoracoid, lateral view. B, right scapulocoracoid, lateral view. C, right radiale, dorsal view (proximal at top). D, right humerus, dorsal view. E, right humerus, ventral view. F, right ulna, proximal view. G, right ulna, dorsal view. H, right ulna, ventral view. I, right radius, medial view. J, right radius, lateral view. K, ?interclavicle, ventral view, anterior at right. L, right ilium, lateral view. M, left ilium, lateral view. N, right ischium, lateral view. O, right ischium, medial view. P, right pubis, ventral view. Q, right pubis, dorsal view. R, proximal half of left femur, dorsal view. S, left femur, ventral view. T, proximal end of left fibula, lateral view. U, right tibia, anterior view. V, articulated right foot and tarsus, dorsal view. Abbreviations: 4t, fourth trochanter; anterior expansion of interclavicle; as, astragalus; ca, calcaneum; co, coracoid; dc, deltoid crest; dpc, deltopectoral crest; dt3–4, third to fourth dentary alveolus; gf, glenoid fossa of scapulocoracoid; mt2–4, second to fourth metatarsal; op, olecranon process; pb, posterior iliac blade; pf, facet for pubis on ischium; sc, scapula. Scale bar = 1 cm.

Figure 5.

Left scapulocoracoid of AMNH 3219, holotype, Necrosuchus ionensisSimpson 1937, in lateral (A) and medial (B) view. C, UF 39062, Caiman crocodilus, left scapulocoracoid, medial view. Abbreviations: cf, coracoid foramen; sc, scapula. Scale bars = 1 cm.

The preserved forelimb bones are largely congruent with homologues in most extant crocodylians. The ventromedial margins of both humeral deltopectoral crests are damaged (Fig. 4D, E), and whether they were concave or emerged smoothly from the humeral head cannot be determined. The olecranon process of the ulna is mediolaterally broad (Fig. 4K).

A thin, rectangular bone is interpreted as an incomplete interclavicle (Fig. 4K). It is slightly thicker along the midline, but because the specimen is compressed, whether it was flat along its length or flexed anteriorly is unknown. It bears an expansion at what I interpret to be its anterior end. Remarkably, the sternal ribs appear to have been preserved, suggesting that they were calcified prior to the animal's death.

Both pelvic girdles are preserved (Fig. 4L–Q). The right ilium is distorted, but the shape of the iliac blade appears to be preserved on the left side (Fig. 4M), revealing a dorsoventrally low structure. This is consistent with the unusually thin articulation facet on the posterior sacral rib. The blade shows no constrictions toward its posterior tip. The pubic blades are less symmetrical than in most living crocodylians and extend further medially than laterally.

Preserved hindlimb elements consist of a complete left femur and tibia, the proximal end of the left fibula, and an articulated right foot preserving the astragalus, calcaneum, distal tarsals, and the second, third, and fourth metatarsals (Fig. 4R–V). These are consistent with corresponding structures in most extant crocodylians.

Several osteoderms are preserved with the holotype (Fig. 6). They were not found in articulation, and so the arrangement of osteoderms on the dorsal and nuchal shields is unknown. Most have thin, smooth articulation facets along the anterior margin and sutural margins medially and laterally. None has a sutural margin perpendicular to these, suggesting that none comes from a ventral osteoderm comprised of paired anterior–posterior ossifications. Most have midline keels, and most are square or nearly square in outline. At least one – a lateral osteoderm with only one sutural margin, presumably from the dorsal shield – has what appears to be a compound keel (Fig. 6B), but most keels are single thin crests (Fig. 6A).

Figure 6.

Osteoderms associated with AMNH 3219, holotype, Necrosuchus ionensisSimpson 1937. Scale bar = 1 cm.


The dorsal position of the foramen aereum on the quadrate ramus (Fig. 1D) is a derived condition found in alligatoroids (Norell et al., 1994; Brochu, 1999). In the plesiomorphic condition, found in nearly all other crocodylians for which the quadrate is known, the foramen aereum is located on the dorsomedial surface of the quadrate ramus. This strongly suggests that Necrosuchus is an alligatoroid.

A long, slender descending process of the exoccipital lateral to the basioccipital (Fig. 1a) is typical of caimanines. The descending process typically extends to a point about half of the distance from the occipital condyle and the ventral-most tip of the basioccipital, but the process extends further in caimanines, sometimes contacting (and forming part of) the basioccipital tubera (Brochu, 1999). Gavialoids also have relatively long exoccipital descending processes, but in this case the process is anteroposteriorly broad (Brochu, 2004b) and dissimilar to the structure seen in Necrosuchus.

The symphyseal region of the dentary is reminiscent of those of extant caimans (Fig. 7a, b). The dentaries of most globidontans curve broadly as they approach the symphysis and meet each other at a high angle, sometimes nearly perpendicular to the sagittal plane. As a result, the articulated lower jaw adopts the typical U-shaped outline in dorsal view that is seen in modern Alligator. In contrast, the dentaries of most caimans intersect each other at a lower angle acute to the sagittal plane. Simpson (1937: 1) was probably referring to this by including ‘mandible pointed anteriorly, narrow across symphysis, and not noticeably expanded at fourth tooth’ in the original diagnosis for Necrosuchus. This is outwardly similar to the condition found in several crocodyloids, including most living Crocodylus. This is not true for all caimans; the mandibles have a broader U-shape in Mourasuchus (Langston, 1965; Bocquetin & Filho, 1990) and Tsoabichi (Brochu, 2010). The anterior ends of the dentaries are not known in Eocaiman cavernensis (Simpson, 1933), but the preserved portions are suggestive of a broader intersection.

Figure 7.

Symphyseal region of right dentary, dorsal view. A, AMNH 3219, holotype, Necrosuchus ionensisSimpson 1937. B, FMNH 69871, Paleosuchus trigonatus (left dentary, image reversed). C, Notocaiman stromeri, line interpretation based on illustration in Rusconi (1937). Scale bars = 1 cm.

Simpson (1937) included the presence of 18 dentary alveoli in his diagnosis of Necrosuchus. I cannot reject the possibility that additional alveoli were present because the dentary is damaged toward the posterior end of the tooth row, but the space available for additional alveoli is short, and a count greater than 20 is unlikely. This is within the range of variation for E. cavernensis and extant Caiman and Melanosuchus, but it distinguishes Necrosuchus from Paleosuchus, which uniformly has more than 20 dentary alveoli (Wermuth, 1953).

Simpson (1937) relied on variation in alveolar diameter along the dentary to draw his conclusion that Necrosuchus was related to ‘Leidyosuchus’. (In fact, most of the comparisons are between Necrosuchus and species now regarded as Borealosuchus.) However, Simpson also noted similarities between Necrosuchus and caimans, in particular with the relative enlargement of the 13th or 14th dentary alveolus. This was included in his diagnosis for Necrosuchus. Alligatorines and stem alligatoroids, by contrast, share the plesiomorphic condition of a relative enlargement of the 11th or 12th alveolus (Brochu, 2004a). Some of the specimens that Simpson measured appear to have been atypical; the pattern reported for Jacare sclerops (= Caiman crocodilus) shows an enlarged 11th alveolus, but the 13th or 14th alveoli are enlarged in that species (pers. observ.). Simpson also reported a specimen of Borealosuchus sternbergii with an enlarged 13th alveolus, but the 11th or 12th is enlarged in Borealosuchus (pers. observ.).

There are additional similarities between Necrosuchus and other caimans. The alveoli in the concave portion of the dentary between the fourth and tenth alveoli are comparatively larger and more widely spaced than in basal alligatorines. The anterior-most four alveoli are more widely spaced from each other in Necrosuchus and caimans than in most other alligatoroids.

The postcranial skeleton of Necrosuchus argues against a relationship with Borealosuchus. The limb bones of Necrosuchus are not as long and slender as in Borealosuchus (Brochu, 1997), and the ilium of Necrosuchus lacks the discrete prominent anterior process found plesiomorphically in Borealosuchus. Langston (2008) reports a prominent anterior process on the ilium of Mourasuchus, but the condition in Mourasuchus is dissimilar from that in Borealosuchus and more like what is seen in other alligatoroids.

The scapulocoracoid synchondrosis of AMNH 3219 appears to be closing (Fig. 5). Closure is seen in other living crocodylians, but usually very late in ontogeny. This condition is rarely observed in most crocodylian species, but it occurs at earlier ontogenetic stages in extant caimans (Brochu, 1995). That the sacral and dorsal neurocentral sutures are still open (Fig. 3B) suggests that AMNH 3219 was not a fully mature individual, and incipient closure of the scapulocoracoid synchondrosis at a relatively early ontogenetic stage is a similarity shared between Necrosuchus and living caimans.

The posterior blade of the left ilium (Fig. 4M) is comparatively low compared with the condition seen in most alligatoroids. A low iliac blade with a modest posterior constriction is, however, seen in living caimans (Brochu, 1999).

Another caimanine –E. palaeocenicusBona, 2007– has been described from the Salamanca Formation of Chubut Province, Argentina. The holotype is a partial lower jaw from a locality south-west of Estancia Las Violetas that can be clearly distinguished from N. ionensis. The articulated rami form a broad U, and the fifth to ninth alveoli are comparatively smaller, more closely spaced, and set within a deeper concavity on the dorsal surface of the dentary. Bona (2007) stated that the dentary symphysis extends to a level just behind the fifth alveolus, but the published figures suggest the symphysis might have been somewhat longer.

Another putative caiman, Notocaiman stromeriRusconi, 1937, has also been described from the Palaeocene of Patagonia. In this case, the material is limited to the anterior end of a right dentary. Derived states linking it with Caimaninae, or even Alligatoroidea, are lacking; but there is a broad resemblance to the anterior end of the dentary of Necrosuchus (Fig. 7C), although the symphysis is comparatively longer, extending to the level of the sixth alveolus.

Material possibly pertaining to Necrosuchus

Kuhn (1933) described fragmentary crocodyliform material, including cranial remains and vertebrae, from Punta Peligro, about 50 km south-east of the type locality of N. ionensis. The specimens derive from approximately correlative levels of the Salamanca Formation, and although Simpson (1937) suspected a close affinity with Necrosuchus, he stopped short of making a formal referral. Additional specimens from Punta Peligro have since been found and are housed in the MACN collections.

A partial dentary figured by Kuhn (1933: plate 1, fig. 1A) suggests a more broadly concave anterior end, more closely resembling that of Eocaiman than Necrosuchus, but direct comparisons are needed. Bona (2007) referred additional fragmentary mandibular material from Punta Peligro to E. palaeocenicus, but one of these specimens, MACN CH1916 (Fig. 8), is a fragment of a right dentary that resembles N. ionensis in three respects – the lateral margin suggests an acute intersection between dentaries, the dentary symphysis extends no further back than the fourth alveolus, and the scar for the splenial indicates a slender anterior terminus dorsal to the Meckelian groove approaching, but not quite reaching, the dentary symphysis. The anterior end is not preserved, so the actual configuration of the symphysis is unknown; moreover, the extent of the symphysis in living crocodylians can vary somewhat in living species (pers. observ.).

Figure 8.

MACN CH 1916, fragment of right dentary, Punta Peligro locality, Salamanca Formation (Lower Palaeocene), Argentina. Abbreviations: d4, fourth dentary alveolus; ds, dentary symphysis; mg, Meckelian groove; sp, articulation surface for splenial on dentary. Scale bar = 1 cm.

Isolated frontals (Kuhn, 1933) suggest exclusion of the frontal from the supratemporal fenestrae, but differ from those of living caimans in lacking upturned orbital margins and being nearly planar between the orbits. There is no evidence for a ridge (‘spectacle’) anterior to the orbits, but the specimens are incomplete. The frontoparietal suture on these specimens is anteriorly convex, not linear (as in modern caimanines). More intriguing are incomplete parietals, one of which Kuhn (1933) referred to Leidyosuchus, that suggest little, if any, dorsal exposure of the supraoccipital on the skull table.

Unfortunately, one cannot refer any of the isolated frontals or parietals to either Necrosuchus or Eocaiman. Both procoelous and amphicoelous presacral vertebrae are known from the Salamanca Formation at Punta Peligro (pers. observ.), demonstrating that at least one non-eusuchian crocodyliform – perhaps a sebecid – was also present in the fauna. We do not know which cranial remains correspond to which vertebrae.



Necrosuchus ionensis was included in a maximum parsimony analysis to test its phylogenetic relationships. Heuristic searches based on 100 random addition sequence replicates were implemented using TNT (version 1.1; Goloboff, Farris & Nixon, 2008). The analysis included 181 morphological characters and 80 ingroup taxa (Appendix), and used Bernissartia fagesii as an outgroup. Multistate characters were left unordered and all characters had equal weight.


The analysis recovered 814 equally optimal trees (length = 574, consistency index without uninformative characters = 0.387, retention index = 0.818). The strict and Adams consensus trees (generated with PAUP*, version 4.0b10; Swofford, 2002) derived from them are consistent with previous results using similar data (Brochu, 1999, 2004b, 2010; Salisbury et al., 2006; Martin, 2007, 2010; Ösi, Clark & Weishampel, 2007; Delfino, Martin & Buffetaut, 2008a; Delfino et al., 2008b), although resolution is diminished relative to earlier analyses. In particular, the present analysis was unable to recover a monophyletic Alligatoridae exclusive of Cretaceous globidontans (Brachychampsa, Stangerochampsa, Albertochampsa), although monophyly of Globidonta as a whole remains stable.

Based on this analysis, neither Acynodon nor Allodaposuchus are crocodylians (Fig. 9). This is relevant because Acynodon has generally been regarded as a basal globidontan (Buscalioni, Ortega & Vasse, 1997; Martin, 2007; Delfino et al., 2008a), and a recent analysis by Martin (2010) suggested that Allodaposuchus precedens, which was previously seen as a stem eusuchian (Buscalioni et al., 2001; Delfino et al., 2008b), might also be a globidontan. Acynodon and Allodaposuchus are from the Late Cretaceous of Europe, which would have profound implications for the palaeobiogeography of Alligatoroidea. Based on this analysis, Acynodon (which may not be monophyletic) and Allodaposuchus are part of an endemic European radiation (Hylaeochampsidae). Moving them to Alligatoroidea increases tree length by at least 15 steps.

Figure 9.

Strict (solid) and Adams (dashed) consensus of 814 equally optimal trees (length = 574, consistency index without uninformative characters = 0.387, retention index = 0.818).

Necrosuchus is most parsimoniously seen as a basal caimanine. Owing to its incompleteness, it acts as a wildcard and causes some loss of resolution within caimanines. An Adams consensus draws Necrosuchus into a polytomy along with Tsoabichi and crown caimanines (Fig. 9).

This is not entirely robust. Bootstrap support is non-existent for any node involving Necrosuchus. Any other position within Caimaninae, or as the sister lineage to all other caimanines, increases tree length by only one step. This reflects the incomplete nature of the known material of Necrosuchus. Other positions within Alligatoroidea increase tree length by four or five steps, and one of the character states drawing Necrosuchus amongst the caimanines (early scapulocoracoid closure) is open to interpretation. Recoding the character expressing scapulocoracoid heterochrony, either to unknown or the plesiomorphic state (late ontogenetic closure), has no impact on the resulting tree. The long, slender exoccipital descending process is a unique feature found only amongst caimans, and the alveolar patterns likewise support a caimanine affinity.

A single character state places Necrosuchus closer to crown caimans than E. cavernensis: the dentary symphysis extends no further back than the level of the fourth or fifth alveolus. The symphysis is relatively longer in E. cavernensis and basal alligatorines, extending to at least the level of the sixth alveolus. Short symphyses have arisen multiple times not only within Crocodylia, but within Alligatoroidea – they extend no further than the fifth alveolus in Procaimanoidea, some Diplocynodon, some Alligator, and Arambourgia amongst noncaimanine alligatoroids. Moreover, Bona (2007) stated that the dentary symphysis extends only to behind the fifth alveolus in E. palaeocenicus, although mandibular material figured in the description suggests a somewhat longer symphysis.

Trees in which Necrosuchus and Leidyosuchus are close relatives are only four steps longer, but a close relationship with Borealosuchus requires ten steps beyond optimal. Most of the characters in the matrix reflect cranial variation, and taxa known primarily from mandibular or postcranial information (such as Necrosuchus) are typically labile when included in the analysis (e.g. Hill & Lucas, 2006). Necrosuchus lacks derived (e.g. long, slender limb bones) and plesiomorphic (e.g. prominent anterior process on the ilium) states found in Borealosuchus, and Borealosuchus lacks derived states linking Necrosuchus with caimanines.


Antiquity of the alligatorinecaimanine divergence

The holotype of Necrosuchus was collected from the upper portion of the Salamanca Formation of Chubut Province, Argentina (Simpson, 1937). The uppermost levels of the Salamanca Formation, now regarded as the Hansen Member, represent fluviolacustrine deposits overlying shallow marine and estuarine strata (Andreis, Mazzoni & Spalletti, 1975). The Salamanca Formation has long been thought to be of Danian age (Méndez, 1966; Bertels, 1975; Andreis, 1977; Marshall, Sempere & Butler, 1997; Iglesias et al., 2007), but the upper part of the unit (and of the Peligran SALMA) may instead extend into the early part of the Selandian (Gelfo et al., 2009). In either case, Necrosuchus helps establish the presence of caimanines in southern South America early in the Cenozoic.

The oldest unambiguous alligatorines are likewise from the Palaeocene. The oldest is Navajosuchus mooki from the Early Palaeocene (Puercan) of New Mexico. Some optimal trees in this analysis draw some Late Cretaceous globidontans within crown Alligatoridae. This reinforces the comparatively deep genetic split between alligatorines and caimanines (e.g. Merchant et al., 2006) and a minimum divergence time of at least 60 million years for the two groups (Brochu, 1999, 2004a).

Early caimanine biogeography

By itself, the Patagonian record is consistent with a simple single-dispersal model for caimanine biogeography. Although alligatorids are generally intolerant of salt water (Taplin et al., 1982; Taplin & Grigg, 1989; Jackson, Butler & Brooks, 1996), dispersal could have been enabled by a chain of islands or a short-lived land connection between North and South America during the Maastrichtian or early Palaeogene (Krause, Kielan-Jaworowska & Bonaparte, 1992; Marshall et al., 1997; Iturralde-Vinent & MacPhee, 1999). However, the situation grows more complicated when enigmatic alligatorids from North America are considered.

Orthogenysuchus olseni Mook 1924 is based on a nearly complete but poorly preserved skull from the Early Eocene (Wasatchian NALMA) of Wyoming. It has sometimes been interpreted as a crocodylid (e.g. Steel, 1973) or a dorsoventrally crushed pristichampsine (e.g. Rossmann, 1998), but the quadrate of Orthogenysuchus preserves a dorsally shifted foramen aereum, strongly suggesting a phylogenetic position within Alligatoroidea, and constricted supratemporal fenestrae and heart-shaped palatine process (one of the few sutures preserved on the specimen) suggest a relationship amongst caimanines. The specimen bears more than 18 maxillary alveoli and the external naris is mediolaterally elongate, features shared with the bizarre Mourasuchus from the Miocene of South America. Phylogenetic analyses support a relationship between Mourasuchus and Orthogenysuchus (Brochu, 1999, 2004a; Aguilera et al., 2006; Bona, 2007), although this view has not been universally accepted (Langston, 2008).

Inclusion of Necrosuchus causes a close relationship between Mourasuchus and Orthogenysuchus to collapse in strict consensus trees. Although an additional alveolus or two may have existed on a complete Necrosuchus dentary, the tooth count would not have approached that of Mourasuchus, which exceeded 30 (Langston, 1965). Although the number of dentary alveoli for Orthogenysuchus is unknown, the minimum number of maxillary alveoli strongly suggests a dentary with many more than 20 alveoli. A placement for Necrosuchus closer to either Orthogenysuchus or Mourasuchus thus appears unlikely –Orthogenysuchus and Mourasuchus share an elongated rostrum with an increased number of small teeth, a feature not directly expressed in the matrix.

Although the precise phylogenetic relationships of Tsoabichi greenriverensis are problematic, there is no evidence for a close relationship to Orthogenysuchus. Depending on how the presence or absence of a splenial symphysis is coded, Tsoabichi is either a basal caimanine (and nearly as labile as Necrosuchus, in spite of being more completely known) or related to Paleosuchus (Brochu, 2010). Even if Tsoabichi is a stem nettosuchid, Orthogenysuchus is most parsimoniously seen as more closely related to Mourasuchus than to Tsoabichi.

Assuming that the relationships of these North American fossils have been estimated correctly, neither a simple single-dispersal scenario nor vicariance are sufficient to explain the observed distribution of fossil caimanines. The biogeographical relationships between North and South American crocodyliforms are more complex than previously supposed. More thorough sampling of latest Cretaceous and early Palaeogene alligatoroid fossils is needed to clarify the situation.


Necrosuchus demonstrates that some of the features found in extant caimanines arose comparatively early in caimanine history. The slender descending processes of the exoccipital are powerful evidence against any other phylogenetic position than within Caimaninae. That the scapulocoracoid synchondrosis may be closing early in ontogeny may also support this hypothesis, although more work is needed to determine that the features supporting the interpretation favoured here are not the result of post-mortem distortion or damage.

However, because Necrosuchus remains so poorly known, many questions about the early morphological evolution of the group remain unanswered. We are unable to resolve the polarity of character states expressing variation in dorsal exposure of the supraoccipital; the supraoccipital is separated from the squamosals by the parietal in Paleosuchus and Tsoabichi, but the supraoccipital contacts the squamosals dorsally in jacareans, Mourasuchus, and Purussaurus, excluding the parietal from the posterior margin of the skull table. The condition in Eocaiman is unknown. Whether these are expressions of a continuously evolving character (gradual expansion of the supraoccipital until it joins with the squamosals) is unclear at present. And although nothing resembling a compound ventral osteoderm has been found in Palaeocene deposits of South America, we cannot definitively state that structures comprised of paired ossifications were absent from basal caimanines.

Lack of information on the skull of Necrosuchus limits our ability to resolve its phylogenetic relationships, and this in turn limits its value in reconstructing the early biogeographical history of caimanines. Necrosuchus does, however, further demonstrate the presence of Caimaninae in the Early Palaeocene of South America. This constrains the minimum timing of at least some crocodyliform faunal exchange between North and South America to the earliest part of the Cenozoic. Whether additional exchange occurred, and when, remains an open question that can best be resolved through the collection of more fossils in the Palaeogene of South America.


I am grateful to Diego Pol, L. Fiorelli, and the organizers of the third Latin American Congress of Vertebrate Paleontology for the invitation to participate in an international symposium on crocodyliform evolution, and to Hans Larsson for co-editing the special volume that arose from it. M. Norell, M. Méchin, Y. Laurent, T. Tortosa, D. Arbolla, A. Baez, A. Kramarz, O. Rieppel, and Z. Gasparini granted access to collections that proved critical to this study. Two reviewers provided comments that greatly improved this paper. (That smacking sound everyone heard in February 2010 was me knocking myself in the head for mislabelling the exoccipital as the epipterygoid in Figure 1. Both reviewers noticed, for which I am thankful. To those who complain about peer review, I can only rework what Winston Churchill said – peer review is the worst form of scientific quality control, except for all those other forms that have been tried from time to time.) The Willi Hennig Society is acknowledged for making TNT version 1.1 available. I acknowledge discussions with P. Bona, D. Fortier, W. Langston, A. Turner, and S. Jouve. Funding was provided by the US NSF (DEB 0444133) and the University of Iowa.


Characters and character codings used in this analysis. Some codings are based in part on the literature (Aguilera et al., 2006, for Purussaurus mirandai; Piras & Buscalioni, 2006, for Diplocynodon tormis and Diplocynodon muelleri; Ösi et al., 2007, for Iharkutosuchus makadii). All others are based on direct observation of specimens.

This is based on matrices used previously by this author (Brochu, 1999, 2004a, b). The order in which characters appear has been revised to make more anatomical sense, a few characters have been added, and a few have been removed. Many of these are figured in these earlier publications; character states relevant to relationships amongst alligatoroids, in particular, are figured in Brochu (1999).

An extract of this matrix was used in the recent description of Tsoabichi greenriverensis (Brochu, 2010), but whereas the analysis (and character list) in that paper excluded parsimony-uninformative characters, the matrix itself included a larger sample of characters. This obviously complicates efforts to use that matrix. The matrix presented here replaces that used in Brochu (2010). The author has self-flagellated to atone for the error.

Bernissartia fagesii


Allodaposuchus precedens


Acynodon adriaticus


Acynodon iberoccitanus


Iharkutosuchus makadii


Hylaeochampsa vectiana


Borealosuchus formidabilis


Borealosuchus wilsoni


Borealosuchus acutidentatus


Borealosuchus sternbergii


Eothoracosaurus mississippiensis


Thoracosaurus neocesariensis


Eosuchus minor


Eogavialis africanum


Gryposuchus colombianus


Gavialis gangeticus


Pristichampsus vorax


Pristichampsus geiseltalensis


Planocrania hengdongensis


Planocrania datangensis


Leidyosuchus canadensis


Diplocynodon ratelii


Diplocynodon hantoniensis


Diplocynodon muelleri


Diplocynodon tormis


Diplocynodon darwini


Baryphracta deponiae


Stangerochampsa mccabei


Albertochampsa langstoni


Brachychampsa montana


Brachychampsa sealeyi


Alligator sinensis


Alligator mississippiensis


Alligator mefferdi


Alligator thomsoni


Alligator olseni


Alligator mcgrewi


Alligator prenasalis


Ceratosuchus burdoshi


Hassiacosuchus haupti


Navajosuchus mooki


Allognathosuchus polyodon


Allognathosuchus wartheni (‘Wilwood alligatorid’)


Wannaganosuchus brachymanus


Procaimanoidea kayi


Procaimanoidea utahensis


Arambourgia gaudryi


Necrosuchus ionensis


Tsoabichi greenriverensis


Purussaurus mirandai


Purussaurus neivensis


Orthogenysuchus olseni


Mourasuchus spp.


Eocaiman cavernensis


Caiman yacare


Caiman crocodilus


Caiman latirostris


Caiman lutescens


Melanosuchus fisheri


Melanosuchus niger


Paleosuchus trigonatus


Paleosuchus palpebrosus


Mecistops cataphractus


Crocodylus niloticus


Crocodylus porosus


Crocosylus acutus


Osteolaemus tetraspis


Osteolaemus osborni


Voay robustus


Rimasuchus lloydi


Crocodylus megarhinus


Australosuchus clarkae


Kambara implexidens


Trilophosuchus rackhami


Tomistoma schlegelii


Thecachampsa americana


Kentisuchus spenceri


Crocodylus acer


Crocodylus affinis


Asiatosuchus germanicus


Prodiplocynodon langi


  • 1Ventral tubercle of proatlas more than one half (0) or no more than one half (1) the width of the dorsal crest [Brochu, 1999, character 1]
  • 2Fused proatlas boomerang-shaped (0), strap-shaped (1), or massive and block-shaped (2) [Brochu, 1999, character 2]
  • 3Proatlas with prominent anterior process (0) or lacks anterior process (1) [Brochu, 1999, character 10]
  • 4Proatlas has tall dorsal keel (0) or lacks tall dorsal keel; dorsal side smooth (1) [Brochu, 1999, character 17]
  • 5Atlas intercentrum wedge-shaped in lateral view, with insignificant parapophyseal processes (0), or plate-shaped in lateral view, with prominent parapophyseal processes at maturity (1) [Brochu, 1999, character 5; modified from Clark, 1994, character 89]
  • 6Dorsal margin of atlantal rib generally smooth with modest dorsal process (0) or with prominent process (1) [Brochu, 1999, character 14]
  • 7Atlantal ribs without (0) or with (1) very thin medial laminae at anterior end [Brochu, 1999, character 16]
  • 8Atlantal ribs lack (0) or possess (1) large articular facets at anterior ends for each other [Brochu, 1999, character 15]
  • 9Axial rib tuberculum wide, with broad dorsal tip (0) or narrow, with acute dorsal tip (1) [Brochu, 1999, character 20]
  • 10Axial rib tuberculum contacts diapophysis late in ontogeny, if at all (0) or early in ontogeny (1) [Brochu, 1999, character 21]
  • 11Anterior half of axis neural spine orientated horizontally (0) or slopes anteriorly (1) [Brochu, 1999, character 11]
  • 12Axis neural spine crested (0) or not crested (1) [Brochu, 1999, character 12]
  • 13Posterior half of axis neural spine wide (0) or narrow (1) [Brochu, 1999, character 3]
  • 14Axis neural arch lacks (0) or possesses (1) a lateral process (diapophysis) [Brochu, 1999, character 4; adapted from Norell 1989, character 7]
  • 15Axial hypapophysis located toward the centre of centrum (0) or toward the anterior end of centrum (1) [Brochu, 1999, character 6]
  • 16Axial hypapophysis without (0) or with (1) deep fork [Brochu, 1999, character 19; character states reversed from original]
  • 17Hypapophyseal keels present on 11th vertebra behind atlas (0), 12th vertebra behind atlas (1), or tenth vertebra behind atlas (2) [Brochu, 1999, character 7]
  • 18Third cervical vertebra (first postaxial) with prominent hypapophysis (0) or lacks prominent hypapophysis (1) [Brochu, 1999, character 8; adapted from Norell, 1989, character 12; Norell & Clark, 1990, character 11; Clark, 1994, character 91]
  • 19Neural spine on third cervical long, dorsal tip at least half the length of the centrum without the cotyle (0) or short, dorsal tip acute and less than half the length of the centrum without the cotyle (1) [Brochu, 1999, character 9]
  • 20Cervical and anterior dorsal centra lack (0) or bear (1) deep pits on the ventral surface of the centrum
  • 21Presacral centra amphicoelous (0) or procoelous (1) [Brochu, 1999, character 18; adapted from several previous analyses, e.g. Benton & Clark, 1988; Norell & Clark, 1990, characters 8 and 10; Clark, 1994, characters 92 and 93]
  • 22Anterior sacral rib capitulum projects far anteriorly of tuberculum and is broadly visible in dorsal view (0), or anterior margins of tuberculum and capitulum nearly in same plane, and capitulum largely obscured dorsally (1) [Brochu, 1999, character 13]
  • 23Scapular blade flares dorsally at maturity (0) or sides of scapular blade subparallel; minimal dorsal flare at maturity (1) [Brochu, 1999, character 22; adapted from Benton & Clark, 1988]
  • 24Deltoid crest of scapula very thin at maturity, with sharp margin (0) or very wide at maturity, with broad margin (1) [Brochu, 1999, character 23]
  • 25Scapulocoracoid synchondrosis closes very late in ontogeny (0) or relatively early in ontogeny (1) [Brochu, 1999, character 24]
  • 26Scapulocoracoid facet anterior to glenoid fossa uniformly narrow (0) or broad immediately anterior to glenoid fossa, and tapering anteriorly (1) [Brochu, 1999, character 25]
  • 27Proximal edge of deltopectoral crest emerges smoothly from proximal end of humerus and is not obviously concave (0) or emerges abruptly from proximal end of humerus and is obviously concave (1) [Brochu, 1999, character 26]
  • 28M. teres major and M. dorsalis scapulae insert separately on humerus; scars can be distinguished dorsal to deltopectoral crest (0) or insert with common tendon; single insertion scar (1) [Brochu, 1999, character 29]
  • 29Olecranon process of ulna narrow and subangular (0) or wide and rounded (1) [Brochu, 1999, character 27]
  • 30Distal extremity of ulna expanded transversely with respect to long axis of bone; maximum width equivalent to that of proximal extremity (0) or proximal extremity considerably wider than distal extremity (1) [Salisbury et al., (2006), character 173]
  • 31Interclavicle flat along length, without dorsoventral flexure (0) or with moderate dorsoventral flexure (1) or with severe dorsoventral flexure (2) [Brochu, 1999, character 30]
  • 32Anterior end of interclavicle flat (0) or rod-like (1) [Brochu, 1999, character 31]
  • 33Iliac anterior process prominent (0) or virtually absent (1) [Brochu, 1999, character 34; adapted from Benton & Clark, 1988; Clark, 1994, character 84; although transformation here is different]
  • 34Dorsal margin of iliac blade rounded with smooth border (0) or rounded, with modest dorsal indentation (1) or rounded, with strong dorsal indentation (wasp-waisted; 2) or narrow, with dorsal indentation (3) or rounded with smooth border; posterior tip of blade very deep (4) [Brochu, 1999, character 28]
  • 35Supraacetabular crest narrow (0) or broad (1) [Brochu, 1999, character 32]
  • 36Limb bones relatively robust, and hindlimb much longer than forelimb at maturity (0) or limb bones very long and slender (1) [Brochu, 1999, character 33]
  • 37M. caudofemoralis with single head (0) or with double head (1) [Brochu, 1999, character 160]
  • 38Dorsal osteoderms not keeled (0) or keeled (1) [Brochu, 1999, character 35; adapted from Buscalioni, Sanz & Casanovas, 1992, character 22]
  • 39Dorsal midline osteoderms rectangular (0) or nearly square (1) [Brochu, 1999, character 36; adapted from Norell & Clark, 1990, character 16; Clark, 1994, character 95]
  • 40Four (0), six (1), eight (2), or ten (3) contiguous dorsal osteoderms per row at maturity [Brochu, 1999, character 37; adapted from Norell & Clark, 1990, character 12; Clark, 1994, character 97]
  • 41Nuchal shield grades continuously into dorsal shield (0) or differentiated from dorsal shield; four nuchal osteoderms (1) or differentiated from dorsal shield; six nuchal osteoderms with four central and two lateral (2) or differentiated from dorsal shield; eight nuchal osteoderms in two parallel rows (3) [Brochu, 1999, character 38]
  • 42Ventral armour absent (0) or single ventral osteoderms (1) or paired ventral ossifications that suture together (2) [Brochu, 1999, character 39; adapted from Buscalioni et al., 1992, character 21]
  • 43Anterior margin of dorsal midline osteoderms with anterior process (0) or smooth, without process (1) [Brochu, 1999, character 40; adapted from Norell & Clark, 1990, character 13; Clark, 1994, character 96]
  • 44Ventral scales have (0) or lack (1) follicle gland pores [Brochu, 1999, character 155; Poe, 1997]
  • 45Ventral collar scales not enlarged relative to other ventral scales (0) or in a single enlarged row (1) or in two parallel enlarged rows (2) [Brochu, 1999, character 156; Poe, 1997]
  • 46Median pelvic keel scales form two parallel rows along most of tail length (0) or form single row along tail (1) or merge with lateral keel scales (2) [Brochu, 1999, character 157; Poe, 1997]
  • 47Alveoli for dentary teeth 3 and 4 nearly same size and confluent (0) or fourth alveolus larger than third, and alveoli are separated (1) [Brochu, 1999, character 52]
  • 48Anterior dentary teeth strongly procumbent (0) or project anterodorsally (1) [Brochu, 1999, character 53]
  • 49Dentary symphysis extends to fourth or fifth alveolus (0) or sixth to eighth alveolus (1) or behind eighth alveolus (2) [Brochu, 2004a, 166]
  • 50Dentary gently curved (0), deeply curved (1), or linear (2) between fourth and tenth alveoli [Brochu, 1999, character 68]
  • 51Largest dentary alveolus immediately caudal to fourth is (0) 13 or 14, (1) 13 or 14 and a series behind it, (2) 11 or 12, or (3) no differentiation, or (4) behind14 [Brochu, 2004a, character 167]
  • 52Splenial with anterior perforation for mandibular ramus of cranial nerve V (0) or lacks anterior perforation for mandibular ramus of cranial nerve V (1) [Brochu, 1999, character 41; adapted in part from Norell, 1988, character 15 and 1989, character 8]
  • 53Mandibular ramus of cranial nerve V exits splenial anteriorly only (0) or splenial has singular perforation for mandibular ramus of cranial nerve V posteriorly (1) or splenial has double perforation for mandibular ramus of cranial nerve V posteriorly (2) [Brochu, 1999, character 42; adapted in part from Norell, 1988, character 15 and 1989, character 8]
  • 54Splenial participates in mandibular symphysis; splenial symphysis adjacent to no more than five dentary alveoli (0) or splenial excluded from mandibular symphysis; anterior tip of splenial passes ventral to Meckelian groove (1) or splenial excluded from mandibular symphysis; anterior tip of splenial passes dorsal to Meckelian groove (2) or deep splenial symphysis, longer than five dentary alveoli; splenial forms wide V within symphysis (3) or deep splenial symphysis, longer than five dentary alveoli; splenial constricted within symphysis and forms narrow V (4) [Brochu, 1999, character 43; adapted from Clark, 1994, character 77]
  • 55Coronoid bounds posterior half of foramen intermandibularis medius (0) or completely surrounds foramen intermandibularis medius at maturity (1) or obliterates foramen intermandibularis medius at maturity (2) [Brochu, 1999, character 46; adapted from Norell, 1988, character 12]
  • 56Superior edge of coronoid slopes strongly anteriorly (0) or almost horizontal (1) [Brochu, 1999, character 54]
  • 57Inferior process of coronoid laps strongly over inner surface of Meckelian fossa (0) or remains largely on medial surface of mandible (1) [Brochu, 1999, character 55]
  • 58Coronoid imperforate (0) or with perforation posterior to foramen intermandibularis medius (1) [Brochu, 1999, character 56]
  • 59Process of splenial separates angular and coronoid (0) or no splenial process between angular and coronoid (1) [Brochu, 1999, character 59]
  • 60Angular–surangular suture contacts external mandibular fenestra at posterior angle at maturity (0) or passes broadly along ventral margin of external mandibular fenestra late in ontogeny (1) [Brochu, 1999, character 47; adapted from Norell, 1988, character 40]
  • 61Anterior processes of surangular unequal (0) or subequal to equal (1) [Brochu, 1999, character 48]
  • 62Surangular with spur bordering the dentary tooth row lingually for at least one alveolus length (0) or lacking such spur (1) [Brochu, 1999, character 61]
  • 63External mandibular fenestra absent (0) or present (1) or present and very large; most of foramen intermandibularis caudalis visible in lateral view (2) [Brochu, 1999, character 62; Clark, 1994, character 75; incorporates character 64 from earlier matrix, adapted from Norell, 1988, character 14]
  • 64Surangular–dentary suture intersects external mandibular fenestra anterior to posterodorsal corner (0) or at posterodorsal corner (1) [Brochu, 1999, character 65]
  • 65Angular extends dorsally toward or beyond anterior end of foramen intermandibularis caudalis; anterior tip acute (0) or does not extend dorsally beyond anterior end of foramen intermandibularis caudalis; anterior tip very blunt (1) [Brochu, 1999, character 66]
  • 66Surangular–angular suture lingually meets articular at ventral tip (0) or dorsal to tip (1) [Brochu, 1999, character 67]
  • 67Surangular continues to dorsal tip of lateral wall of glenoid fossa (0) or truncated and not continuing dorsally (1) [Brochu, 1999, character 106]
  • 68Articular–surangular suture simple (0) or articular bears anterior lamina dorsal to lingual foramen (1) or articular bears anterior lamina ventral to lingual foramen (2) or bears laminae above and below foramen (3) [modified from Brochu, 1999, character 44]
  • 69Lingual foramen for articular artery and alveolar nerve perforates surangular entirely (0) or perforates surangular/angular suture (1) [Brochu, 1999, character 45; state 2 deleted 2007]
  • 70Foramen aereum at extreme lingual margin of retroarticular process (0) or set in from margin of retroarticular process (1) [Brochu, 1999, character 49; adapted from Norell, 1988, character 16]
  • 71Retroarticular process projects posteriorly (0) or projects posterodorsally (1) [Brochu, 1999, character 50; adapted from Benton & Clark, 1988; Clark, 1994, character 71; Norell & Clark, 1990, character 7]
  • 72Surangular extends to posterior end of retroarticular process (0) or pinched off anterior to tip of retroarticular process (1) [Brochu, 1999, character 51; adapted from Norell, 1988, character 42]
  • 73Surangular–articular suture orientated anteroposteriorly (0) or bowed strongly laterally (1) within glenoid fossa [Brochu, 1999, character 162]
  • 74Sulcus between articular and surangular (0) or articular flush against surangular (1) [Brochu, 1999, character 60]
  • 75Dorsal projection of hyoid cornu flat (0) or rod-like (1) [Brochu, 1999, character 57]
  • 76Dorsal projection of hyoid cornu narrow, with parallel sides (0) or flared (1) [Brochu, 1999, character 58]
  • 77Lingual osmoregulatory pores small (0) or large (1) [Brochu, 1999, character 158]
  • 78Tongue with (0) or without (1) keratinized surface [Brochu, 1999, character 159]
  • 79Teeth and alveoli of maxilla and/or dentary circular in cross-section (0), or posterior teeth laterally compressed (1), or all teeth compressed (2) [Brochu, 2004a, 165]
  • 80Maxillary and dentary teeth with smooth carinae (0) or serrated (1)
  • 81Naris projects anterodorsally (0) or dorsally (1) [Brochu, 1999, character 79]
  • 82External naris bisected by nasals (0) or nasals contact external naris, but do not bisect it (1) or nasals excluded, at least externally, from naris; nasals and premaxillae still in contact (2) or nasals and premaxillae not in contact (3) [Brochu, 1999, character 95; adapted from Norell, 1988, character 3; Clark, 1994, characters 13 and 14]
  • 83Naris circular or keyhole-shaped (0) or wider than long (1) or anteroposteriorly long and prominently teardrop-shaped (2) [Brochu, 1999, character 161; modified]
  • 84External naris of reproductively mature males (0) remains similar to that of females or (1) develops bony excrescence (ghara)
  • 85External naris (0) opens flush with dorsal surface of premaxillae or (1) circumscribed by thin crest
  • 86Premaxillary surface lateral to naris smooth (0) or with deep notch lateral to naris (1) [Brochu, 1999, character 142]
  • 87Premaxilla has five teeth (0) or four teeth (1) early in posthatching ontogeny [Brochu, 1999, character 97; Norell, 1988, character 17]
  • 88Incisive foramen small, less than half the greatest width of premaxillae (0) or large, more than half the greatest width of premaxillae (1) or large, and intersects premaxillary–maxillary suture (2) [Brochu, 1999, character 124]
  • 89Incisive foramen completely situated far from premaxillary tooth row, at the level of the second or third alveolus (0) or abuts premaxillary tooth row (1) or projects between first premaxillary teeth (2) [Brochu, 1999, character 153]
  • 90Dorsal premaxillary processes short, not extending beyond third maxillary alveolus (0) or long, extending beyond third maxillary alveolus (1) [Brochu, 1999, character 145]
  • 91Dentary tooth 4 occludes in notch between premaxilla and maxilla early in ontogeny (0) or occludes in a pit between premaxilla and maxilla; no notch early in ontogeny (1) (Norell, 1988, character 29) [Brochu, 1999, character 77]
  • 92All dentary teeth occlude lingual to maxillary teeth (0) or occlusion pit between seventh and eighth maxillary teeth; all other dentary teeth occlude lingually (1) or dentary teeth occlude in line with maxillary tooth row (2) [modified from Brochu, 1999, character 78; Norell, 1988, character 5; Willis, 1993, character 1]
  • 93Largest maxillary alveolus is no. 3 (0), no. 5 (1), no. 4 (2), nos 4 and 5 are same size (3), no. 6 (4), or maxillary teeth homodont (5), or maxillary alveoli gradually increase in diameter posteriorly toward penultimate alveolus (6) [modified from Brochu, 1999, character 89; Norell, 1988, character 1]
  • 94Maxillary tooth row curved medially or linear (0) or curves laterally broadly (1) posterior to first six maxillary alveoli [Brochu, 1999, character 135; adapted from Clark, 1994, character 79]
  • 95Dorsal surface of rostrum curves smoothly (0) or bears medial dorsal boss (1) [Brochu, 1999, character 101]
  • 96Canthi rostralii absent or very modest (0) or very prominent (1) at maturity [Brochu, 1999, character 143; Norell, 1988, character 34]
  • 97Preorbital ridges absent or very modest (0) or very prominent (1) at maturity [Brochu, 1999, character 144]
  • 98Antorbital fenestra present (0) or absent (1) [Norell & Clark, 1990, character 2; Salisbury et al., 2006, character 176]
  • 99Vomer entirely obscured by premaxilla and maxilla (0) or exposed on palate at premaxillary–maxillary suture (1) [Brochu, 1999, character 125; adapted from Norell, 1988, character 22]
  • 100Vomer entirely obscured by maxillae and palatines (0) or exposed on palate between palatines (1) [Brochu, 1999, character 126]
  • 101Surface of maxilla within narial canal imperforate (0) or with a linear array of pits (1) [Brochu, 1999, character 148]
  • 102Medial jugal foramen small (0) or very large (1) [Brochu, 1999, character 120]
  • 103Maxillary foramen for palatine ramus of cranial nerve V small or not present (0) or very large (1) [Brochu, 1999, character 111]
  • 104Ectopterygoid abuts maxillary tooth row (0) or maxilla broadly separates ectopterygoid from maxillary tooth row (1) [Brochu, 1999, character 91; Norell, 1988, character 19]
  • 105Maxilla terminates in palatal view anterior to lower temporal bar (0) or comprises part of the lower temporal bar (1)
  • 106Penultimate maxillary alveolus less than (0) or more than (1) twice the diameter of the last maxillary alveolus
  • 107Prefrontal dorsal surface smooth adjacent to orbital rim (0) or bearing discrete knob-like processes (1)
  • 108Dorsal half of prefrontal pillar narrow (0) or expanded anteroposteriorly (1) [Brochu, 1999, character 137; adapted from Norell, 1988, character 41]
  • 109Medial process of prefrontal pillar expanded dorsoventrally (0) or anteroposteriorly (1) [Brochu, 1999, character 136]
  • 110Prefrontal pillar solid (0) or with large pneumatic recess (1) [Brochu, 1999, character 99]
  • 111Medial process of prefrontal pillar wide (0) or constricted (1) at base [Brochu, 1999, character 138]
  • 112Maxilla has linear medial margin adjacent to suborbital fenestra (0) or bears broad shelf extending into fenestra, making lateral margin concave (1) [Brochu, 1999, character 105; rephrased]
  • 113Anterior face of palatine process rounded or pointed anteriorly (0) or notched anteriorly (1) [Brochu, 1999, character 108]
  • 114Anterior ectopterygoid process tapers to a point (0) or forked (1) [Brochu, 1999, character 109]
  • 115Palatine process extends (0) or does not extend (1) significantly beyond anterior end of suborbital fenestra [Brochu, 1999, character 110; adapted from Willis, 1993, character 2]
  • 116Palatine process generally broad anteriorly (0) or in form of thin wedge (1) [Brochu, 1999, character 118]
  • 117Lateral edges of palatines smooth anteriorly (0) or with lateral process projecting from palatines into suborbital fenestrae (1) [Brochu, 1999, character 94]
  • 118Palatine–pterygoid suture nearly at (0) or far from (1) posterior angle of suborbital fenestra [Brochu, 1999, character 85]
  • 119Pterygoid ramus of ectopterygoid straight, posterolateral margin of suborbital fenestra linear (0) or ramus bowed, posterolateral margin of fenestra concave (1) [Brochu, 1999, character 88; rephrased]
  • 120Lateral edges of palatines parallel posteriorly (0) or flare posteriorly, producing shelf (1) [Brochu, 1999, character 90; adapted from Norell, 1988, character 2]
  • 121Anterior border of the choana is comprised of the palatines (0) or choana entirely surrounded by pterygoids (1) [Brochu, 1999, character 71; Benton & Clark, 1988; Clark, 1994, character 43; Norell & Clark, 1990, character 1]
  • 122Choana projects posteroventrally (0) or anteroventrally (1) at maturity [Brochu, 1999, character 72]
  • 123Pterygoid surface lateral and anterior to internal choana flush with choanal margin (0) or pushed inward anterolateral to choanal aperture (1) or pushed inward around choana to form neck surrounding aperture (2) or everted from flat surface to form neck surrounding aperture (3) [Brochu, 1999, character 73]
  • 124Posterior rim of internal choana not deeply notched (0) or deeply notched (1) [Brochu, 1999, character 107]
  • 125Internal choana not septate (0) or with septum that remains recessed within choana (1) or with septum that projects out of choana (2) [Brochu, 1999, character 152]
  • 126Ectopterygoid–pterygoid flexure disappears during ontogeny (0) or remains throughout ontogeny (1) [Brochu, 1999, character 116]
  • 127Ectopterygoid extends (0) or does not extend (1) to posterior tip of lateral pterygoid flange at maturity [Brochu, 1999, character 149; adapted from Norell, 1988, character 32]
  • 128Lacrimal makes broad contact with nasal; no posterior process of maxilla (0) or maxilla with posterior process within lacrimal (1) or maxilla with posterior process between lacrimal and prefrontal (2) [Brochu, 1999, character 93]
  • 129Prefrontals separated by frontals and nasals (0) or prefrontals meet medially (1) [Brochu, 1999, character 100; Norell, 1988, character 27]
  • 130Lacrimal longer than prefrontal (0), or prefrontal longer than lacrimal (1), or lacrimal and prefrontal both elongate and nearly the same length (2) [Brochu, 1999, character 117; modified from Norell, 1988, character 7]
  • 131Anterior tip of frontal (0) forms simple acute point or (1) forms broad, complex sutural contact with the nasals
  • 132Ectopterygoid extends along medial face of postorbital bar (0) or stops abruptly ventral to postorbital bar (1) [Brochu, 1999, character 133]
  • 133Postorbital bar massive (0) or slender (1) [Brochu, 1999, character 70; Norell, 1989, character 3]
  • 134Postorbital bar bears process that is prominent, dorsoventrally broad, and divisible into two spines (0) or bears process that is short and generally not prominent (1) (adapted from Norell, 1989, character 2) [Brochu, 1999, character 134]
  • 135Ventral margin of postorbital bar flush with lateral jugal surface (0) or inset from lateral jugal surface (1) [Brochu, 1999, character 146; adapted from Benton & Clark, 1988; Norell & Clark, 1990, character 3]
  • 136Postorbital bar continuous with anterolateral edge of skull table (0) or inset (1) [Norell & Clark, 1990, character 3; Salisbury et al., 2006, character 175]
  • 137Margin of orbit flush with skull surface (0) or dorsal edges of orbits upturned (1) or orbital margin telescoped (2) [Brochu, 1999, character 103]
  • 138Ventral margin of orbit circular (0) or with prominent notch (1) [Brochu, 1999, character 139]
  • 139Palpebral forms from single ossification (0) or from multiple ossifications (1) [Brochu, 1999, character 96; adapted from Norell, 1988, character 8; Clark, 1994, character 65]
  • 140Quadratojugal spine prominent at maturity (0) or greatly reduced or absent at maturity (1) [Brochu, 1999, character 69; adapted from Norell, 1989, character 1]
  • 141Quadratojugal spine low, near posterior angle of infratemporal fenestra (0) or high, between posterior and superior angles of infratemporal fenestra (1) [Brochu, 1999, character 114]
  • 142Quadratojugal forms posterior angle of infratemporal fenestra (0) or jugal forms posterior angle of infratemporal fenestra (1) or quadratojugal–jugal suture lies at posterior angle of infratemporal fenestra (2) [Brochu, 1999, character 75; adapted from Norell, 1989, character 10]
  • 143Postorbital neither contacts quadrate nor quadratojugal medially (0) or contacts quadratojugal, but not quadrate, medially (1) or contacts quadrate and quadratojugal at dorsal angle of infratemporal fenestra (2) or contacts quadratojugal with significant descending process (3) [Brochu, 1999, character 76]
  • 144Quadratojugal bears long anterior process along lower temporal bar (0) or bears modest process, or none at all, along lower temporal bar (1) [Brochu, 1999, character 83]
  • 145Quadratojugal extends to superior angle of infratemporal fenestra (0) or does not extend to superior angle of infratemporal fenestra; quadrate participates in fenestra (1) [Brochu, 1999, character 80; adapted from Buscalioni et al., 1992, character 6]
  • 146Postorbital–squamosal suture orientated ventrally (0) or passes medially (1) ventral to skull table [Brochu, 1999, character 163]
  • 147Dorsal and ventral rims of squamosal groove for external ear valve musculature parallel (0) or squamosal groove flares anteriorly (1) [Brochu, 1999, character 84]
  • 148Squamosal–quadrate suture extends dorsally along posterior margin of external auditory meatus (0) or extends only to posteroventral corner of external auditory meatus (1) [Brochu, 1999, character 132]
  • 149Posterior margin of otic aperture smooth (0) or bowed (1) [Brochu, 1999, character 102]
  • 150Frontoparietal suture deeply within supratemporal fenestra; frontal prevents broad contact between postorbital and parietal (0) or suture makes modest entry into supratemporal fenestra at maturity; postorbital and parietal in broad contact (1) or suture on skull table entirely (2) [Brochu, 1999, character 81]
  • 151Frontoparietal suture concavoconvex (0) or linear (1) between supratemporal fenestrae [Brochu, 1999, character 86]
  • 152Supratemporal fenestra with fossa; dermal bones of skull roof do not overhang rim at maturity (0) or dermal bones of skull roof overhang rim of supratemporal fenestra near maturity (1) or supratemporal fenestra closes during ontogeny (2) [Brochu, 1999, character 87; adapted from Norell, 1988, character 9]
  • 153Shallow fossa at anteromedial corner of supratemporal fenestra (0) or no such fossa; anteromedial corner of supratemporal fenestra smooth (1) [Brochu, 1999, character 92]
  • 154Medial parietal wall of supratemporal fenestra imperforate (0) or bearing foramina (1) [Brochu, 1999, character 104; Norell, 1988, character 51]
  • 155Parietal and squamosal widely separated by quadrate on posterior wall of supratemporal fenestra (0) or parietal and squamosal approach each other on posterior wall of supratemporal fenestra without actually making contact (1) or parietal and squamosal meet along posterior wall of supratemporal fenestra (2) [Brochu, 1999, character 131]
  • 156Skull table surface slopes ventrally from sagittal axis (0) or planar (1) at maturity [Brochu, 1999, character 123]
  • 157Posterolateral margin of squamosal horizontal or nearly so (0) or upturned to form a discrete horn (1)
  • 158Mature skull table with broad curvature; short posterolateral squamosal rami along paroccipital process (0) or with nearly horizontal sides; significant posterolateral squamosal rami along paroccipital process (1) [Brochu, 1999, character 140]
  • 159Squamosal does not extend (0) or extends (1) ventrolaterally to lateral extent of paraoccipital process [Brochu, 1999, character 150]
  • 160Supraoccipital exposure on dorsal skull table small (0), absent (1), large (2), or large such that parietal is excluded from posterior edge of table (3) [Brochu, 1999, character 82; Norell, 1988, character 11]
  • 161Anterior foramen for palatine ramus of cranial nerve VII ventrolateral (0) or ventral (1) to basisphenoid rostrum [Brochu, 1999, character 164]
  • 162Sulcus on anterior braincase wall lateral to basisphenoid rostrum (0) or braincase wall lateral to basisphenoid rostrum smooth; no sulcus (1) [Brochu, 1999, character 122]
  • 163Basisphenoid not exposed extensively (0) or exposed extensively (1) on braincase wall anterior to trigeminal foramen [Brochu, 1999, character 129; adapted from Norell, 1989, character 5]
  • 164Extensive exposure of prootic on external braincase wall (0) or prootic largely obscured by quadrate and laterosphenoid externally (1) [Brochu, 1999, character 74; adapted from Norell, 1989, character 5]
  • 165Laterosphenoid bridge comprised entirely of laterosphenoid (0) or with ascending process or palatine (1) [Brochu, 1999, character 115]
  • 166Capitate process of laterosphenoid orientated laterally (0) or anteroposteriorly (1) toward midline [Brochu, 1999, character 130]
  • 167Parietal with recess communicating with pneumatic system (0) or solid, without recess (1) [Brochu, 1999, character 154]
  • 168Significant ventral quadrate process on lateral braincase wall (0) or quadrate–pterygoid suture linear from basisphenoid exposure to trigeminal foramen (1) [Brochu, 1999, character 127]
  • 169Lateral carotid foramen opens lateral (0) or dorsal (1) to basisphenoid at maturity [Brochu, 1999, character 128]
  • 170External surface of basioccipital ventral to occipital condyle orientated posteroventrally (0) or posteriorly (1) at maturity [modified from Hua & Jouve, 2004, character 167 and Salisbury et al., 2006, character 174]
  • 171Posterior pterygoid processes tall and prominent (0) or small and project posteroventrally (1) or small and project posteriorly (2) [Brochu, 1999, character 98]
  • 172Basisphenoid thin (0) or anteroposteriorly wide (1) ventral to basioccipital [Brochu, 1999, character 113]
  • 173Basisphenoid not broadly exposed ventral to basioccipital at maturity; pterygoid short ventral to median eustachian opening (0) or basisphenoid exposed as broad sheet ventral to basioccipital at maturity; pterygoid tall ventral to median eustachian opening (1) [Brochu, 1999, character 119]
  • 174Exoccipital with very prominent boss on paroccipital process; process lateral to cranioquadrate opening short (0) or exoccipital with small or no boss on paroccipital process; process lateral to cranioquadrate opening long (1) [Brochu, 1999, character 141]
  • 175Lateral eustachian canals open dorsal (0) or lateral (1) to medial eustachian canal [Brochu, 1999, character 147; adapted from Norell, 1988, character 46]
  • 176Exoccipitals terminate dorsal to basioccipital tubera (0) or send robust process ventrally and participate in basioccipital tubera (1) or send slender process ventrally to basioccipital tubera (2) [Brochu, 1999, character 151; adapted from Norell, 1988, character 20; Clark, 1994, characters 57 and 60]
  • 177Quadrate foramen aereum on mediodorsal angle (0) or on dorsal surface (1) of quadrate [Brochu, 1999, character 121]
  • 178Quadrate foramen aereum is small (0), comparatively large (1), or absent (2) at maturity [Brochu, 2004b, character 165]
  • 179Quadrate lacks (0) or bears (1) prominent, mediolaterally thin crest on dorsal surface of ramus
  • 180Attachment scar for posterior mandibular adductor muscle on ventral surface of quadrate ramus forms modest crests (0) or prominent knob (1) [Ösi et al., 2007, character 165]
  • 181Quadrate with small, ventrally reflected medial hemicondyle (0) or with small medial hemicondyle; dorsal notch for foramen aereum (1) or with prominent dorsal projection between hemicondyles (2) or with expanded medial hemicondyle (3) [Brochu, 1999, character 112]