Dyrosauridae

Jouve, Stéphane, Muizon, Christian De, Cespedes-Paz, Ricardo, Sossa-Soruco, Víctor & Knoll, Stephane, 2021, The longirostrine crocodyliforms from Bolivia and their evolution through the Cretaceous-Palaeogene boundary, Zoological Journal of the Linnean Society 192, pp. 475-509 : 500-501

Dyrosauridae

Known Maastrichtian South American crocodyliforms are restricted to terrestrial sebecosuchians and aquatic crocodyliforms. Two groups have been described in the aquatic crocodyliforms from North and South America: dyrosaurids and crocodylians. The Late Cretaceous dyrosaurids are scarce on the American continents. Two possible dyrosaurids have been described, one from the Maastrichtian of northern Bolivia (see above) and six vertebrae from Colombia. Only one vertebra was illustrated by Langston (1965). It bears a large hypapophysis and strongly resembles dyrosaurid vertebrae in the shape of its centrum, i.e. higher than wide and heart-shaped ( Langston, 1965). Nevertheless, the possibility cannot be completely excluded that these vertebrae might belong to a sebecosuchian, even if this hypothesis is doubtful to us. The stratigraphic ages of the remains have been debated and should be confirmed, and it can therefore not be excluded that both were Palaeocene in age (see above).

The dyrosaurid H. derbianus has also been described from the Pernambuco Basin of Brazil, but as stated by Barbosa et al. (2008), no exact location has been provided in the original description for the type specimen of H. derbianus ( Cope, 1885, 1886), only equivalence to the Fox Hills Formation of the USA. Contrary to what is usually considered, it has not been stated clearly that the Hyposaurus remains described by Cope come from the Palaeocene Maria Farinha Formation of the Pernambuco Province. In this province, the Maastrichtian Gramame Formation is also present. Cope (1885, 1886) noted from the same locality as H. derbianus the presence of Apocodon sericeus Cope, (1886), which is absent from the Gramame Formation but found in the Danian Maria Farinha Formation ( de Santana et al., 2011), but also the shark Squalicorax pristodontus (Agassiz, 1843) , an elasmobranch taxon that occurs in the Campanian– Maastrichtian Gramame Formation. Therefore, Cope might have examined a mixed assortment of fossils that lacked clear stratigraphic provenance ( de Santana et al., 2011). Numerous marine reptiles have been described from the Gramame Formation, such as mosasaurs and plesiosaurs ( Price, 1953, 1957; Azevedo & Carvalho, 1997; Kellner & Campos, 1999), but no crocodyliforms. However, several crocodyliform remains have been described from the Danian Maria Farinha Formation, such as Guarinisuchus munizi ( Barbosa et al., 2008) and isolated material ( de Carvalho & de Azevedo, 1997; Gallo et al., 2001). Examination of the holotype of H. derbianus by one of us (S.J.) did not find any morphological differences with G. munizi , and no comparison with this species has been provided by Barbosa et al. (2008). Both species have moderately long symphyses, reaching the level of the 14 th –15 th dentary tooth, and are slightly wider than high. Therefore, awaiting formal revision of the type material of H. derbianus , we consider G. munizi as a probable junior synonym of H. derbianus . All these arguments also suggest that the type material of H. derbianus probably comes from the Danian Maria Farinha Formation, like all other dyrosaurid remains found in Pernambuco province. The attribution of H. derbianus to the genus Hyposaurus was suggested as doubtful (Jouve, 2004), but the present phylogenetic analysis confirms the original attribution, because this species is the sister taxon to H. natator ( Fig. 18).

In North America, two valid dyrosaurid species have been described: Hyposaurus rogersii Owen, 1849 and H. natator . Hyposaurus rogersii was described from two isolated vertebrae. The vertebrae are so poorly preserved that no diagnostic character can be established from its holotype ( Norell & Storrs, 1989). Moreover, the remains described from North America suggest the possible presence of two dyrosaurid species in North America. NJSM 10416, illustrated by Denton et al. (1997), has a lateromedially straight anterior margin of its supratemporal fenestra, a narrow interorbital distance, and the anteriormost portion of the frontal in the supratemporal fenestra is wider than the interorbital distance. These characters are found in the holotype of H. natator described by Troxell (1925) and illustrated by Callahan et al. (2015). Therefore, the three specimens are probably from the same species. NJSM 10861, illustrated by Denton et al. (1997), differs in having a curved anterior margin of its supratemporal fenestra, a flatter and wider interorbital distance (wider than nasal), fused nasals (unfused in H. natator ) and a posterior margin of the parietal that is not concave. Thus, awaiting a formal revision of the North American dyrosaurid material and possible intraspecific variability, we consider H. rogersii as a nomen dubium (following Norell & Storrs, 1989), H. natator as the only valid North American species, and NJSM 10861 as Hyposaurus sp. , a possible different species. This also questions the validity of the genus Hyposaurus , because its type species is the dubious H. rogersii , with a holotype that does not have any diagnostic characters at the generic level, but this is beyond the scope of the present paper.

North American Hyposaurus was previously described as a Maastrichtian–Palaeocene dyrosaurid. Maastrichtian remains come from the main fossiliferous layer and upper fossiliferous layer of the Hornerstown Formation in southern New Jersey and the Pine Barren Member of the Clayton Formation from Alabama. Recent re-evaluation stated that these formations were from the early Danian strata ( Parris, 1986; Landman et al., 2007; Callahan et al., 2015; Wiest et al., 2016). Hyposaurus natator is thus unknown from Maastrichtian layers, and clearly identifiable specimens are restricted to the Danian ( Callahan et al., 2015). Given that the Thanetian North American remains are formed only by unidentifiable postcranial remains, remains from this stage should be considered as Dyrosauridae indet. Several dyrosaurid remains have been reported from the Maastrichtian Navesink Formation, but many of the earliest records of dyrosaurids are from localities that were poorly described and recorded, and because both the Navesink and Hornerstown formations appear similar in outcrop composition, there was a great deal of confusion regarding the exact horizon that yielded these fossils. A careful review should indicate that none of the New Jersey specimens can be placed accurately in the Navesink or New Egypt Formations, and therefore, there should be no unequivocal evidence of the taxon below the Cretaceous–Palaeogene boundary (Wayne Callahan, personal communication, 2017). Therefore, the stratigraphic origin of the North American dyrosaurid remains needs to be re-examined.

A recent paper also described a new dyrosaurid from the Maastrichtian of Mexico, Sabinosuchus coahuilensis Shiller et al. (2016) , but the material is poorly preserved, and numerous characters make the attribution of the specimen to a dyrosaurid particularly doubtful. Sabinosuchus coahuilensis is found to be a pholidosaurid in the present phylogenetic analysis (Supporting Information, Fig. S1; Jouve & Jalil, 2020). Thus, the presence of Dyrosauridae is not known with certainty in Maastrichtian North American Formations, and only poor remains with uncertain determinations have been found in South America.

In summary, the presence of dyrosaurids during the Late Cretaceous in American continents is still unclear and is represented by only fragmentary material from Colombia and Bolivia, if these remains are confirmed as being dyrosaurids, and in North America, if the stratigraphic origin of North American remains is clarified.