Description and phylogenetic relationships of a new species of Torvoneustes (Crocodylomorpha, Thalattosuchia) from the Kimmeridgian of Switzerland

Material

MJSN BSY008-465 is a disarticulated metriorhynchid skeleton (Fig. 2). The specimen was initially collected by the Pal A16 on a large block of limestone. Bones were prepared directly on the surface and kept on pedestals of rock. This initial phase of preparation, especially the acid preparation, was poorly controlled and resulted in damages of the more fragile bony elements such as some cranial and mandibular elements. More recently, all bones were completely removed from the block to facilitate scientific study. The preserved remains of MJSN BS008-465 include cranial and mandibular elements as well as material from the axial and appendicular skeleton. Many elements are fragmented and show evidence of deformation.

The preserved elements of the cranium include the frontal, prefrontals, right postorbital, nasals, maxillae, right premaxillae. The mandible is almost complete and preserve both angulars, surangulars, articulars, splenials, and dentaries. Many isolated teeth are also preserved. The postcranial elements include cervical, dorsal, and caudal vertebrae, ribs, the left ischium, the right femur, and the right fibula. Numerous bone fragments are not identifiable.

Geological setting

Between 2000 and 2011, controlled paleontological excavations were conducted by the Pal A16 before the construction of the A16 Transjurane highway in the canton of Jura (Fig. 1). They revealed the presence of several rich fossiliferous horizons of marls and limestones, as well as many dinosaurs tracksites in the Ajoie region around the city of Porrentruy (Marty et al., 2003). During the Late Jurassic, this region was part of a carbonate platform with diversified shallow depositional environments such as lagoons, reefs, channels and littoral zones forming layers rich in marine fossils (Colombié & Strasser, 2005; Comment et al., 2015).

MJSN BSY008-465 was found in 2008 on the hardground level 4,000 of the Lower Virgula Marls (Fig. 3) in the locality Courtedoux-Bois de Sylleux, Switzerland (Fig. 1). The Lower Virgula Marls belong to the Chevenez Member of the Reuchenette Formation. They date to the late Kimmeridgian and correspond to the end of the Mutabilis ammonite zone and beginning of the Eudoxus ammonite zone (Comment et al., 2015). These marls are notably characterized by the abundance of the small oyster Nanogyra virgula Koppka, 2015, which gives them their name. The hardground level 4,000 is rich in invertebrates, notably encrusted and benthic bivalves and brachiopods. Vertebrates are mostly represented by isolated material, with the exception of the metriorhynchid MJSN BSY008-465 described herein, a partial teleosauroid skeleton provisionally referred to Steneosaurus cf. bouchardi Sauvage, 1872, now combined as Proexochokefalos cf. bouchardi (Schaefer et al., 2018; Johnson, Young & Brusatte, 2020), and a relatively complete shell of the thalassochelydian turtle Thalassemys bruntrutana Puntener, Anquetin, and Billon-Bruyat, 2015 (Püntener, Anquetin & Billon-Bruyat, 2015).

Phylogenetic analyses

The phylogenetic analyses were conducted using the data matrix and procedure of Young et al. (2020a), which were derived from Young et al. (2020b). The phylogenetic analysis has been performed before the publication of Sachs et al. (2021) which present an updated matrix from Young et al. (2020a). While we acknowledge this more recent matrix, the changes compared with the matrix of Young et al. (2020a) mainly focus on pneumaticity of cranial bones, acquired on CT data or broken bones. These characters are not applicable to the described specimen. Therefore using the revised matrix would not significantly improve the results within the frame of this study. The original matrix includes 179 taxa coded for 574 characters. The outgroup is Batrachotomus kupferzellensis Gower, 1999. The matrix was modified with Mesquite 3.61 (Maddison & Maddison, 2019) to include MJSN BSY008-465 as a new operational taxonomic unit (see Material S1). The latter was scored for 204 characters. The analytical procedure strictly follows the one described by Young et al. (2020a). The parsimony analyses were conducted using TNT 1.5 (No taxon limit) (Goloboff, Farris & Nixon, 2008; Goloboff & Catalano, 2016) with the RAM increased to 900 Mb. The analysis used the scripts provided by Young et al. (2020a) and consisted of an unweighted analysis followed by seven different extended implied weighting analyses (k = 1, 3, 7, 10, 15, 20 and 50). The main script (EIW.run) runs an initial “new technology” search (xmult:hits 10 replications 100 rss css xss fuse 5 gfuse 10 ratchet 20 drift 20; sec:drift 10 rounds 10 fuse 3; ratchet:numsubs 40 nogiveup; drift:numsubs 40 nogiveup) holding 20,000 trees per analysis, then runs a “traditional methods” search (bbreak:TBR) on the saved trees. The script then computes the descriptive statistics, the strict consensus tree, the majority rule consensus tree, and the maximum agreement subtree (for more details, see Young et al., 2020a).

The Bayesian analysis was conducted using MrBayes3.2.7 (Ronquist et al., 2012), again following the procedure described by Young et al. (2020a). The sampling model is a Markov Chain Monte Carlo with only variable characters scored and a Gama distribution (Mkv+G model). Five independent analyses are run, each with 10 chains for 10 million generations with a sampling every 5,000 generations and a burn-in of 40% (for more details, see Young et al., 2020a). At the end of the analysis, the consensus tree is saved.

Imaging

Each element of the cranium of MJSN BSY008-465 was individually scanned with a portable surface scanner (Artec Space Spider). The scans were treated with the software Artec Studio 13 to produce textured 3D models. The elements were assembled in Blender 2.8 in order to reconstruct the skull of MJSN-BSY008-465 in 3D (De Sousa Oliveira et al., 2023). This technique helped the description of the specimen and highlighted lost contacts between the bones. The 3D models of each individual element, as well as the reconstructed skull and mandible are made openly available in De Sousa Oliveira et al. (2023) and in Material S2.

The microscopic observation and detailed photographs of the teeth of MJSN BSY008-465 were realized with a digital microscope (Keyence VHX-970F).

The electronic version of this article in Portable Document Format (PDF) will represent a published work according to the International Commission on Zoological Nomenclature (ICZN), and hence the new names contained in the electronic version are effectively published under that Code from the electronic edition alone. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix http://zoobank.org/. The LSID for this publication is: act:5DEFCF6F-D7EF-4711-9CB6-A7219F612ECB. The online version of this work is archived and available from the following digital repositories: PeerJ, PubMed Central SCIE and CLOCKSS.

Systematic paleontology

CROCODYLOMORPHA Hay, 1930

THALATTOSUCHIA Fraas, 1901

METRIORHYNCHIDAE Fitzinger, 1843

GEOSAURINAE Lydekker, 1889

GEOSAURINI Lydekker, 1889

Torvoneustes Andrade et al., 2010

Diagnosis. see in Young et al. (2013b)

Type species. Dakosaurus carpenteri Wilkinson, Young & Benton, 2008

Included valid species. Torvoneustes coryphaeus Young et al., 2013b; Torvoneustes mexicanus Wieland, 1910; Torvoneustes jurensis sp. nov.

Occurrence. Kimmeridgian of Mexico (Barrientos-Lara et al., 2016); Kimmeridgian of Dorset, UK (Grange & Benton, 1996; Wilkinson, Young & Benton, 2008; Young et al., 2013b); middle Oxfordian to Tithonian of Oxfordshire, UK (Young, 2014); middle Oxfordian of Yorkshire (Foffa, Young & Brusatte, 2018); late Kimmeridgian of canton of Jura, Switzerland (Schaefer et al., 2018); upper Valanginian of Moravian-Silesian Region, Czech Republic (Madzia et al., 2021).

Torvoneustes jurensis sp. nov.

urn:lsid:zoobank.org:act:5DEFCF6F-D7EF-4711-9CB6-A7219F612ECB

Diagnosis. Torvoneustes jurensis sp. nov. is identified as a member of Torvoneustes by the following combination of characters: robust teeth lingually curved, conical crown, bicarinate with a prominent keel; enamel ornamentation made of conspicuous subparallel apico-basal ridges on the basal two thirds of the crown shifting to short, low-relief ridges forming an anastomosed pattern on the apex; carinae formed by a keel and true microscopic denticles forming a continuous row on the distal and mesial carinae; enamel ornamentation extending up to the carinae near the crown apex; inflexion point on the lateral margin of the prefrontals directed posterolaterally in dorsal view, at an angle of approximately 70° from the anteroposterior axis of the skull; acute angle (close to 60°) between the posteromedial and the lateral processes of the frontal. Torvoneustes jurensis differs from the other species of Torvoneustes by the following combination of characters: presence of a distinct angle between the anteromedial and the lateral processes of the frontal; lack of cranium ornamentation aside from faint ornamentation on the anterior part of the maxilla. Torvoneustes jurensis also differs from To. carpenteri in: lacking “finger-like” projections on the posterior margin of the prefrontal; having an angle of 142° in average between the anterior and lateral processes of the frontal; and having a long anteromedial process of the frontal reaching the same level as the anterior margin of the prefrontals (shorter process in To. carpenteri). Torvoneustes jurensis is also clearly distinct from To. coryphaeus in having: a tooth enamel ornamentation extending up to the carinae; prefrontals with a rounded distal margin (forming an acute angle in To. coryphaeus); and a supraorbital notch of 90° (45° in To. coryphaeus). The slender morphology, acuteness, and great number of teeth of Torvoneustes jurensis nov. sp. differs from both To. coryphaeus and To. carpenteri. Torvoneustes jurensis differs from To. mexicanus in having irregular denticle basal length (120–200 µm, mean: 160 µm) and distribution (density up to 40 denticles/5 mm in the upper middle part of the crown and down to 30 denticles/5 mm in the base and apex) on the carinae (regular basal length of 142 µm and density of 30 denticles/5 mm in To. mexicanus).

Etymology. The species is named after the Jura, which corresponds to both the mountain range and the Swiss canton where the holotype was found. The complete name could therefore be translated from Latin as “savage swimmer from Jura”.

Holotype. MJSN BSY008-465, a relatively complete disarticulated skeleton including the frontal, prefrontals, right postorbital, nasals, maxillae, right premaxillae, angulars, surangulars, articulars, splenials, dentaries, several isolated teeth, cervical, dorsal, and caudal vertebrae, ribs, the left ischium, the right femur, and the right fibula.

Type horizon and locality. Courtedoux-Bois de Sylleux, Ajoie, canton of Jura, Switzerland (Fig. 1). Lower Virgula Marls, Chevenez Member, Reuchenette Formation, late Kimmeridgian (Eudoxus ammonite zone), Late Jurassic (Fig. 3; Comment et al., 2015).

Cranium

Cranial elements are disarticulated except for the frontal preserving its contact with the left prefrontal. The preserved elements of the cranium are exclusively part of the skull roof and snout. Many pieces are broken, incomplete, and/or deformed. As a result, the general shape and size of most of the skull fenestrae and apertures, such as the preorbital fossae, orbits, and supratemporal fenestrae cannot be clearly determined. Based on the skull reconstruction (Fig. 4), the skull total length is estimated at 88 cm, and the rostrum length (nasals and maxillae) is about 49 cm (55.7% of the total skull length), corresponding to the mesorostrine condition as defined by Young et al. (2010). The skull width-to-length ratio is 0.28. The supratemporal fenestrae are greatly enlarged and the orbits are facing laterally (Fig. 4).

Premaxillae

An incomplete right premaxilla is preserved. Its anterior part is severely damaged and the posterior part, including the contact with the maxilla, is missing. The premaxilla seemingly suffered deformation as the location of the alveoli relative to the narial opening seems unnatural, looking more laterally placed than it should. The external nares aperture is partly conserved, oblong, moderate in width and formed entirely by the premaxilla (Figs. 5D, 5E). A complete alveolus, likely P2, is visible in ventral view. In dorsal view, the external surface of the bone is smooth except for a few superficial pits. The bone thickens on the external edge of the narial opening and close to the alveolus, but it thins posteriorly. The alveolar orientation suggests a tooth implantation directed anteroventrally (Fig. 4C).

Maxillae

Both maxillae are preserved. The right maxilla is almost complete, only missing its anteriormost part and the posterior end of the tooth row (Fig. 6). This element is heavily fractured and deformed, especially in its anterior part where the bone was broken in several parts and glued back together. This anterior part is deformed as the alveoli show extreme anteroposterior orientation that does not seems natural compared to the alveoli posterior to this major break. The remaining matrix prevents the observation of the medial side of the right maxilla. The left maxilla is incomplete and broken in four parts. Some of the fragments are severely damaged. Both maxillae contact with a posteromedial bone fragment which is probably formed by the palatine. As preserved, the right maxilla is 49.8 cm long. In ventral view, it widens in its posterior region. In lateral and dorsal view, the bone surface is smooth with only pitting ornamentation on its anterior half (Fig. 6). Shallow grooves extend a few millimeters posterior to some of the foramina but are barely visible and can only be spotted on the 3D model (see Material S2). Most other members of the Geosaurini subclade with known maxillae present conspicuous surface bone ornamentation made of grooves and ridges, often with a unique pattern (Young et al., 2013b). Plesiosuchus manselii and To. coryphaeus both have maxillae ornamented with grooves and raised ridges, while Dakosaurus maximus has additional pits to this ornamentation pattern (Young et al., 2012a, 2013b). The incomplete maxilla of cf. Torvoneustes (MANCH L6459) shows a similar pattern to To. coryphaeus with moderate to strong grooves and a raised edge aligned with the sagittal axis of the skull but lack posteroventral foramina (Young, 2014). Torvoneustes carpenteri has a pattern of pits and grooves on the lateral edge of its maxilla (Grange & Benton, 1996; Wilkinson, Young & Benton, 2008). In contrast, MJSN BSY008-465 has an ornamentation pattern closer to the one seen in Geosaurus giganteus, or Purranisaurus potens (Young & Andrade, 2009; Young et al., 2013b; Herrera, Gasparini & Fernández, 2015).

The alveoli are set on a slightly more dorsal plane than the palatal surface of the maxilla. The medial margin of the palatal plate of the right maxilla bears a longitudinal furrow that likely corresponds to the maxillary palatal groove. Each maxilla preserves at least 15 alveoli, which is more than the estimated number of 14 (with 11 strictly preserved alveoli) for To. carpenteri (Grange & Benton, 1996; Wilkinson, Young & Benton, 2008; Young et al., 2013b) and less than the estimated alveolar count of 17–19 for To. coryphaeus (Young et al., 2013b). However, there is at least one or two missing alveoli on MJSN BSY008-465 considering the dentary alveolar count. None of the specimens referred to the genus Torvoneustes preserves a complete maxilla, therefore this character should be treated carefully (see Discussion). Among the derived Geosaurini, maxillary alveolar count is usually estimated to be lower than 16 (Table 1; Wilkinson, Young & Benton, 2008; Young & Andrade, 2009; Young et al., 2012b). The alveoli are large, overall homogeneous in size, slightly anteriorly oriented and subcircular with a slightly longer anteroposterior axis than their mediolateral axis, which differentiates MJSN BSY008-465 from the members of the informal ‘E-clade’ taxa whose members show the opposite condition with alveoli wider than long (Abel, Sachs & Young, 2020). The interalveolar space is homogeneously narrow, being less than a quarter the length of the adjacent alveoli. Tooth enlargement and interalveolar space reduction are characteristics shared among several members of the Geosaurini subclade (Herrera et al., 2015) like Dakosaurus, Plesiosuchus (Young et al., 2012b), P. potens (Herrera, Gasparini & Fernández, 2015) some members of the ‘E-clade’ (Abel, Sachs & Young, 2020) and Torvoneustes (Grange & Benton, 1996; Wilkinson, Young & Benton, 2008; Young et al., 2013b; Young, 2014). The maxillary sutural surfaces with the nasal and premaxilla are visible in lateral and dorsal views. In medial view, the maxilla is concave to accommodate the nasal cavity and thicker around the tooth row. A notch between the 5^th and 6^th alveoli (Fig. 6A) on the right maxilla is a reception pit but is the only one on the fossil and therefore does not support a tooth-on-tooth vertical interlocking as seen in Dakosaurus (Young et al., 2012a) or Tyrannoneustes lythrodectikos (Foffa & Young, 2014). In addition, there is no evidence for a maxillary overbite as seen in Geosaurus giganteus (Young & Andrade, 2009). It is likely that the tooth occlusion of MJSN BSY008-465 follows the same interdigitated pattern as in To. mexicanus (Barrientos-Lara et al., 2016).

Table 1:

Overview of the Kimmeridgian-Lower Tithonian metriorhynchids species and a few of their dental characteristics.

		Species	Age	Country	Teeth ornamentation	Carinae	Denticules	Ziphodonty	Denticules density (number of denticle/5 mm)	Maxillary tooth count (absolute)	Maxillary tooth count (estimated)	Dentary tooth count (absolute)	Dentary tooth count (estimated)	Specimen description
Metriorhynchinae	Rhacheosaurini	Cricosaurus suevicus (Fraas, 1901, 1902, SMNS 9808)	Upper Kimmeridgian	Germany	Smooth	Yes, faint	No	/	/	26		24		Complete specimen in limestone
		Cricosaurus albersdoerferi (Sachs et al., 2021, BMMS-BK 1-2)	Upper Kimmeridgian	Germany	Smooth	Yes, faint	No	/	/	23	23+	22	22+	Complete specimen in limestone
		Cricosaurus elegans (Wagner, 1852, BSPG AS I 504)	Lower Tithonian	Germany	Smooth	Yes, faint	No							Skull in limestone
		Cricosaurus bambergensis (Sachs et al., 2019, NKMB-P-Watt14/274)	Upper Kimmeridgian	Germany	Smooth	Yes, faint	No	/	/	23	23+	18+	18+	Complete specimen in limestone
		Cricosaurus rauhuti (Herrera, Aiglstorfer & Bronzati, 2021, SNSB-BSPG 1973 I 195)	Lower Tithonian	Germany	Well-spaced, low longitudinal ridges	Yes, at least unicarenate	No	/	/	/	/	/	/	Incomplete skull
		Maledictosuchus nuyiviianan (Barrientos-Lara, Alvarado-Ortega & Fernández, 2018, IGM 4863)	Kimmeridgian	Mexico	Smooth labial side. Discontinuous low apicobasal ridges on the lingual surface	Yes	No	/	/	17	26?	/	/	Incomplete skull
		‘Cricosaurus’ saltillensis (Buchy, Young & Andrade, 2013, MUDE CPC 487)	Lower Tithonian	Mexico	Faint apicobasally aligned subparallel ridges	Yes, faint	No	/	/	12	17	15	~15?	Disarticulated skull
		Metriorhynchus palpebrosus (Phillips, 1871; Grange & Benton, 1996, (OUMNH J.29823)	Lower Tithonian	United Kingdom	?	?	?	?	?	25 (OUMNH J.29823)-27	/	14	14+	Skull. No teeth remaining
		Metriorhynchus geoffroyii (Metriorhynchus brevirostris) (Young et al., 2020a; MHNG V-2232)	Lower Kimmeridgian	France	?	?	?	?	?	14	20+? (half of the rostrum is easily missing)	/	/	Anterior half of the rostrum, no teeth remaining
Geosaurinae	Geosaurini	‘Metriorhynchus’ cf hastifer (Chouquet cf ‘hastifer’, Lepage et al., 2008; Eudes-Deslongchamps, 1867–1869)	Kimmeridgian	France	Conspicuous apicobasal ridges	Yes	?	?		20?	20+?	/	/	complete skull
		Dakosaurus maximus (Young et al., 2012b, SMNS 8203)	Upper Kimmeridgian	Germany	Overall smooth	Yes, prominent	Yes	Macro	16–18	13		12		Incomplete skull
		Plesiosuchus manselii (Young et al., 2012a, NHMUK PV OR40103)	Upper Kimmeridgian	United Kingdom	Low relief apicobasal ridges	Yes, prominent	Yes	Micro	?	14	14 to 18	13		Incomplete skull
		Geosaurus giganteus (Young & Andrade, 2009, NHM R.1229, NHM 37020)	Lower Tithonian	Germany	Overall smooth	Yes, prominent	Yes	Micro	?	12 (NHM 37020)	12+	7 (NHM 37020)	7+	NHM R.1229 : middle portion of the skull and mandible, deformed. NHM 37020 : skull and mandible in limestone.
		Geosaurus grandis (Young et al., 2012a, BSPG AS-VI-1)	Lower Tithonian	Germany	Smooth	Yes, prominent	Yes	Micro	28,1	14
		Torvoneustes carpenteri (Grange & Benton, 1996, BRSMG Ce17365)	Upper Kimmeridgian	United Kingdom	Conspicuous apicobasal ridges	Yes, prominent	Yes	Micro	?	11	14	/	/	Heavily crushed, incomplete skull
		Torvoneustes coryphaeus (Young et al., 2013b, MJML K1863)	Lower Kimmeridgian	United Kingdom	Conspicuous apicobasal ridges	Yes, prominent	Yes	Micro	?	11	17-19	/	/	Incomplete skull, half of the rostrum missing
		Torvoneustes mexicanus (Barrientos-Lara et al., 2016, IGM 9026)	Kimmeridgian	Mexico	Conspicuous apicobasal ridges	Yes, prominent	Yes	Micro	30	5		4		Fragmentary rostrum
		MJSN BSY008-465	Upper Kimmeridgian	Switzerland	Conspicuous apicobasal ridges	Yes, prominent	Yes	Micro	30–40	15	Up to 21	16	Up to 17	Disarticulated skull

DOI: 10.7717/peerj.15512/table-1

Note:

Gracilineustes acutus is excluded from the table due to the lack of information, the specimen was lost during WW2. The same goes for Rhacheosaurus gracilis NHMUK PV R 3948 for whom the teeth or alveoli are indistinguishable, while Rhacheosaurus cf gracilis LF 2426 skull is not entirely preserved. Therein the distinction between “true ziphodont” and “false ziphodont” condition as defined by Young et al. (2010) is not specified. The estimated tooth count is from the referred articles; the denticle density for D. maximus and Geo. grandis from Andrade et al. (2010) and for To. mexicanus from Barrientos-Lara et al. (2016).

Prefrontal

Both prefrontals are preserved, but the anterior parts and the descending processes are missing. The left prefrontal is still connected with the frontal (Fig. 7) while the right one is found apart (Fig. 5C). The right prefrontal seemingly suffered more damage than the left one, with the bone surface being partially flaked off and the bone flattened and stretched. It has also been broken and glued back together. The left prefrontal shows little deformation but is raised in its anterior part above the level of the frontal during taphonomy. The posterior part of the descending process on both prefrontals is barely preserved as a thin, concave structure. In dorsal view, the prefrontal is large, laterally extended, and it partially overhangs the orbit (Fig. 4A), a feature common within Metriorhynchidae (Fraas, 1902; Andrews, 1913; Herrera, Gasparini & Fernández, 2015; Young et al., 2010). The bone surface is smooth with numerous round or elliptical pits, as in P. potens (Herrera, Gasparini & Fernández, 2015), more densely distributed on the anteromedial part (Figs. 5C, 7). As in most metriorhynchids, the posterolateral corner of the prefrontal is rounded in MJSN BSY008-465 (Andrews, 1913; Lepage et al., 2008), which differs from the geosaurines To. coryphaeus (Young et al., 2013b) and D. maximus (Young et al., 2012a; Pol & Gasparini, 2009) in which the posterolateral corner is angular. The inflection point of this corner is directed posteriorly and forms with the midline of the skull an angle of about 70°, which is found in other Geosaurinae such as To. carpenteri, D. maximus and D. andiniensis. A posteriorly directed inflection point of the prefrontal of 70° or less is an apomorphy of the tribe Geosaurini (Cau & Fanti, 2011). If they share a similar shape, the posterior edge of the prefrontals in MJSN BSY0008-465 lacks the “finger-like” projections described in To. carpenteri (Young et al., 2013b).

Postorbital

Only the part forming the anterolateral margin of the supratemporal fossa of the right postorbital is preserved (Figs. 8A–8C). It is broken around its middle and was glued back together. The ventrolateral part, which should make a major contribution to the postorbital bar, is missing. The posterior portion of the postorbital curves slightly laterally but this is probably the result of postmortem deformation. Two processes are distinguishable on what is preserved of the postorbital. The anterior process is anteromedially oriented in dorsal view, making most of the curve of the lateral edge of the supratemporal fossa, and was in contact with the posterolateral process of the frontal. The posterior process is almost straight, slightly posteromedially oriented, and was in contact with the squamosal. The contact with the lateral process of the frontal is V-shaped, as described in several other metriorhynchids like Ty. lithrodectikos (Andrews, 1913; Foffa & Young, 2014). In the posterior part, the postorbital becomes a thin raised ridge that forms the anterior part of the postorbital-squamosal ridge. Posteriorly, the postorbital-squamosal suture is not readily visible but a change in the bone texture on the medial surface of the bone could match with a similar change on the anterior part of the right squamosal. In medial view, a deep incision in the postorbital marks the contact with the frontal (Fig. 8B), which agrees with observations made on complete skulls of other Metriorhynchidae (Andrews, 1913; Young et al., 2013b). This contact marks the point where the postorbital extends medioventrally to take part in the intratemporal flange. Based on the skull reconstruction, the postorbital likely extended further laterally than the prefrontal, resulting in an enlarged supratemporal fossa as usually seen in metriorhynchids (Wilkinson, Young & Benton, 2008; Foffa & Young, 2014). The bone surface is unornamented except for one elliptical foramen on the anterior process. There might be another foramen at the corner of the “V-shaped” suture but it is hard to discern due to poor bone preservation.

Mandible

Both mandibular rami are preserved (Fig. 4B). They are relatively complete, but some parts are fractured, eroded, and deformed. Some elements of the mandible are not articulated anymore. The mandible is represented by several main parts: two ensembles made of the angular, surangular, articular and prearticular; the disarticulated splenials; and the dentaries (preserved in several pieces). The coronoids are lost. The notable absence of a mandibular fenestra is an apomorphy of the Metriorhynchidae (Fraas, 1902; Andrews, 1913; Vignaud, 1995; Young et al., 2010). The total length of the mandible is about 815 mm with the anterior end missing. The general shape of the mandible is similar to the one described for Ty. lythrodectikos and Geosaurini with the coronoid process located higher than the plan of the tooth row and lower than the retroarticular process, indicating an increase in gape (Young et al., 2012a, 2012b, 2013a, Foffa & Young, 2014).

Dentary

Both dentaries are preserved but not equally well. The left dentary is almost complete but broken into three parts (Fig. 9), only missing its anteriormost portion. The right dentary is broken into two pieces but is highly damaged and deformed. Its ventral and medial parts are completely lost, whereas the posterior part is crushed and deformed. The latter now lies more dorsally and medially compared to the anterior part. The damages are partly due to taphonomic conditions, but also to the poorly controlled acid treatment of the fossil. The left dentary is about 45 cm long, 4.5 cm high and 2.6 cm wide. The right dentary preserves two teeth, including one on its most deformed part with the alveolus deformed and projected inward (Fig. 4B). The left dentary preserves one tooth in the middle of the rostrum (Fig. 9). In dorsal view, the dentary is a long and thick bone. It narrows posteriorly where it was in contact with the angular and surangular. The dentary widens anteriorly starting from the posteriormost alveolus, which is smaller than the one immediately in front. The anterior part of the dentary, including the three anteriormost visible alveoli, shortly curves inward, this bending of the dentary seems natural. There are at least 16 clearly identifiable alveoli, potentially 17, that are overall homogeneous in size. The alveoli are enlarged, subcircular, and slightly longer than wide. The three anteriormost alveoli are larger than the others and lie slightly more dorsally. Overall, the more anterior are the alveoli the more anterodorsally they are oriented. The remaining tooth on the left dentary is anterodorsally directed while the one on the undeformed part of the right dentary is dorsally oriented. As in the maxillae, the interalveolar space is uniform and greatly reduced, being less than half of the anteroposterior length of the adjacent alveoli (see above). In lateral view, the lateral margin of the alveoli lies slightly lower than the medial margin except for the three anteriormost alveoli where the lateral and medial margins are on the same plane (Figs. 9B, 9D). The surangular-dentary groove is visible (Fig. 9B), and does not end anteriorly with an enlarged foramen as it does in Dakosaurus (Pol & Gasparini, 2009). A pitting pattern is present on the lateral and ventral surfaces of the dentaries. However, there is no heavy grooving as described in To. carpenteri (Wilkinson, Young & Benton, 2008). At least eight alveoli are adjacent to the symphysis. This count is higher than the four observed in Dakosaurus maximus (Young et al., 2012b) but close to estimated height in the indeterminate Geosaurini (SMNS 80149) from the informal E-clade (Abel, Sachs & Young, 2020; Young et al., 2020a) and Geosaurus (Young et al., 2012b); as well as the nine of Pl. manselii (Young et al., 2012b), but lower than the 10–13 of the early Geosaurini Ty. lythrodectikos (Young et al., 2013a; Waskow, Grzegorczyk & Sander, 2018). No reception pits for maxillary teeth are observed on the dentaries, indicating that there was no overbite creating a scissor-like occlusion mechanism like in G. giganteus nor any tooth-on-tooth interlocking as seen in D. maximus (Young et al., 2012a; Young & Andrade, 2009).

Splenial

The left splenial is better preserved than the right one. The two splenials have been flattened and both their anterior and posterior ends are broken off. The left splenial is about 37 cm long. In dorsal view, each splenial is narrow on its anterior and posterior ends and thicker in the middle. The lateral edge is straight, but the medial edge is slightly convex (Fig. 10C). The lateral aspect of the splenial is overall concave, with raised ridges forming the sutural contacts with the angular and dentary ventrally, and with the dentary and surangular dorsally (Fig. 10A). Anteriorly, the splenial likely reaches the fifth or sixth alveoli, but this observation might be obscured by the poor preservation of the specimen in this area. In Pl. manselii (Young et al., 2012b), the splenial anteriorly reaches alveoli six anteriorly and alveoli seven or eight in Ty. lythrodectikos (Foffa & Young, 2014; Waskow, Grzegorczyk & Sander, 2018) and in Rhacheosaurini such as Cricosaurus bambergensis or C. albersdoerferi it is clearly posterior to the 10^th alveoli (Sachs et al., 2019, 2021; L. Girard, 2021, personal observation). The medial surface of the splenial is slightly convex. The ventral edge of the bone is thicker and participates to the ventral edge of the mandible itself. The bone surface is smooth. On the medial surface, a foramen is visible posterior to the thickened part of the bone. The splenials would contact one another along the symphysis and the remains of the contact is visible on the anterior third of the splenial, marked by a rougher surface of the bone (Figs. 10B, 10D).

Angular and surangular

The angular and surangular are preserved on both sides. They are missing their anterior part and show signs of crushing, erosion, and deformation, especially on the left mandibular ramus. The angular and surangular are strongly sutured along their entire length. They respectively form the ventral and dorsal halves of the posterior part of the mandibular ramus (Fig. 11). In lateral view, the surangular-dentary groove is well expressed. This groove is present in all Metriorhynchidae, albeit not always visible due to deformation, and it is especially deep in the members of the Geosaurini tribe (Pol & Gasparini, 2009; Young & Andrade, 2009; Young et al., 2012b, 2013b). This groove is associated with the passage of the mandibular nerve (Holliday & Witmer, 2007; Young & Andrade, 2009; George & Holliday, 2013) and its posterior end is marked by a foramen. Two other smaller foramina, more or less aligned with the one on the surangular-dentary groove, are found on the posterior part of the surangular, following the upward curve of the bone. The dorsal margin of the surangular rises slightly posteriorly before sloping down after reaching the coronoid process. The coronoid process is located on the anterodorsal half of the surangular. The coronoid process is higher than the tooth row, but lower than the retroarticular process. In dorsal view, the coronoid process narrows posteriorly and forms a ridge. In medial view, the surangular medial ridge for the coronoid is visible. The medial surface of the angular and surangular is concave, especially on their anterior part where they would form the lateral wall of the Meckelian groove and contact the splenial and the coronoid. The large foramen that connects with the surangular-dentary groove in the lateral surface of the ramus is present below the coronoid process. The angular is thicker than the surangular and forms the ventral margin of the mandible. The posterior part of the angular curves upward towards the retroarticular process at an angle of approximately 30° with the ventral surface of the angular. Posteriorly, the angular extends beyond and rises higher than the glenoid fossa to form the ventral part of the retroarticular process as on other Geosaurini (Young et al., 2012b; Herrera et al., 2015).

Dentition

At least fifteen isolated teeth were found closely associated with the skeleton MJSN BSY008-465. Nine of these isolated teeth are complete, or almost complete, and preserve both the crown and the root. Three additional teeth are still in place on the dentaries and two more on the left maxilla (see above).

Teeth

Based on macroscopic observation, the teeth correspond to the typical metriorhynchid morphology. They are caniniform, conical, and single-cusped (Massare, 1987; Vignaud, 1997). The teeth are large, robust and bicarinate with the carinae on the anteroposterior axis running continuously from the crown base to the apex. There is no basal constriction of the crown, but the crown-root junction is clearly visible from color and texture (Fig. 12). The length of each tooth roots is at least twice the height of each crown. In cross section, the base of the crown is sub-circular to ovoid with the labial face thicker than the lingual one. The tooth roots are ovoid in cross-section. Closer to the apex, the mediolateral compression of the teeth is increasing. Most tooth crowns are over 20 mm high, but the tooth height is not uniform. The average crown height is 21.94 mm with a standard deviation of 4.38. These measures includes all of the 15 teeth. It is to be noted that some of them have their apex broken but not enough to significantly affect the crown height. The shortest tooth crown is 13.23 mm high and the highest is 29.28 mm. The average width at the tooth base is 10 mm (standard deviation 0.9) for a length of 11.36 in average (standard deviation 0.87). The average difference between the width and the length is 1.36 mm (standard deviation 0.93), but the minimal value is 0.3 mm and the maximal value is 3.67 mm with no correlation to the crown height. The teeth are curved lingually and posteriorly. The morphology of the alveoli and the teeth still in place, especially the anterior ones, indicate that teeth were implanted slightly forward in the jaws. Crowns are heavily ornamented with long longitudinal subparallel ridges on at least the basal two thirds of the crown. Ridges are denser on the lingual face than on the labial one. On the apex, the ornamentation becomes low, short ridges forming an anastomosed pattern of drop-shaped ornaments (Figs. 12, 13). This distinctive ornamentation pattern has only been described in other Torvoneustes species (Andrade et al., 2010; Young et al., 2013b; Barrientos-Lara et al., 2016; Foffa, Young & Brusatte, 2018; Young et al., 2019) as well as in the machimosaurids Machimosaurus von Meyer, 1837 and Lemmysuchus obtusidens Andrews, 1909 (Johnson et al., 2017). As seen in To. carpenteri, To. mexicanus and isolated Torvoneustes teeth (Andrade et al., 2010; Young et al., 2013b, 2014b, 2019; Barrientos-Lara et al., 2016; Foffa, Young & Brusatte, 2018; Madzia et al., 2021), the enamel ridges on the upper part of the teeth shift and bend toward the carinae. The carina is well developed, as in other Torvoneustes specimens. The developed keel and the shift of enamel ridges toward the carinae are both autapomorphies of Torvoneustes (Andrade et al., 2010; Young et al., 2013b, 2014b, 2019; Barrientos-Lara et al., 2016; Foffa, Young & Brusatte, 2018; Madzia et al., 2021). On the basal two thirds of the crown, the carinae are serrated with no involvement of the enamel ornamentation. On the apical third of the crown, there is a clear shift in the enamel ornamentation pattern, and the enamel ridges bend toward the carinae and touch the serrated keel (Fig. 13A). The serration is present from the base of the carinae to the apex and is high, especially on the apical half of the crown. Both of these features are also found in Torvoneustes (Andrade et al., 2010; Young et al., 2013b, 2014b, 2019; Barrientos-Lara et al., 2016; Foffa, Young & Brusatte, 2018; Madzia et al., 2021). The serration is only faintly visible on macroscopic observation. The tooth apex is sharp, similar to what is described for Torvoneustes mexicanus and different from the blunter tooth tips of To. carpenteri, To. coryphaeus and Torvoneustes teeth from the UK (Wilkinson, Young & Benton, 2008; Andrade et al., 2010; Young et al., 2013b, 2019; Barrientos-Lara et al., 2016; Foffa, Young & Brusatte, 2018; Madzia et al., 2021). In an unpublished work on isolated thalattosuchian teeth from the Pal A16 collection, Schaefer (2012) already noted the resemblance of the teeth of MJSN BSY008-465 with the ones of To. carpenteri, while pointing out the greater sharpness of the former.

Observed with optic aids, the serration of the carinae is formed by a continuous row of poorly isomorphic, isolated denticles weakly affecting the keel height (poorly developed incipient denticles). This serration corresponds to the microziphodont condition as defined by Andrade et al. (2010) with the denticles all smaller than 300 µm in height and length. The denticles base length varies between 120 and 200 µm with an average of 160 µm (Fig. 13), differing from the regular and 142-µm-long (average) denticles of To. mexicanus (Barrientos-Lara et al., 2016; Table 1). The denticle density (number of denticles/5 mm, following Andrade et al., 2010) ranges from 31 to 40 depending on the area on the teeth with an average of 35, which is higher than the average measured on To. mexicanus (Barrientos-Lara et al., 2016) and G. grandis, but corresponds to the number found on Geosaurus indet. (SMNS 81834; Andrade et al., 2010; Table 1). The denticle density increases in the middle of the tooth crown and decreases at the apex, making the denticles more densely packed and narrower on the part where the carina is the highest. At the apex, the enamel ornamentation joins the carina (Figs. 13A, 13B), similar to the condition in To. carpenteri and To. mexicanus (Andrade et al., 2010; Barrientos-Lara et al., 2016) and contrary to what is observed in To. coryphaeus (Young et al., 2013b; Foffa et al., 2017). The denticles are harder to discern in this region. This false-ziphodont condition (Prasad & de Lapparent de Broin, 2002) is found in Torvoneustes but also in Machimosaurus (Andrade et al., 2010; Young et al., 2014b). However, the combination of true and false ziphodonty, as defined by Andrade et al. (2010), is only known in To. carpenteri, To. mexicanus, and Torvoneustes sp. (OUMNH J.50061 and OUMNH J.50079-J.50085; Young et al., 2013b, 2019; Barrientos-Lara et al., 2016). The base of the carina around the upper middle of the tooth shows structures resembling the “inflated base” seen in To. carpenteri (Fig. 13B; Andrade et al., 2010). MJSN BSY008-465 does not present the faceted teeth of Ieldraan and Geosaurus (Young et al., 2013a; Foffa et al., 2017), nor the macroscopic denticles of Dakosaurus (Andrade et al., 2010; Pol & Gasparini, 2009), nor the characteristic flanges on the side of the carinae seen in Plesiosuchus (Owen, 1883; Young et al., 2012b; Table 1).

Postcranial elements

From the postcranial skeleton, many vertebrae and ribs are preserved as well as a few elements of the pelvis and hindlimbs. Despite the number of postcranial elements, no osteoderms were found. The absence of osteoderms is an apomorphy of Metriorhynchidae (Fraas, 1902; Andrews, 1913; Young et al., 2010).

Vertebrae

MJSN BSY008-465 was found with 22 of its vertebrae including three cervicals, nine dorsals and 10 caudals (Fig. 14). The atlas-axis complex is missing as well as the sacral vertebrae. The vertebrae suffered different level of damage and deformation. Some show stretching, with the centrum deflected from its natural, vertical plane and the apophyses not aligned anymore, or crushing with no preferential direction of deformation. This is mainly the case in the dorsal vertebrae. All vertebrae are amphicoelous, as in all metriorhynchids (Fraas, 1902; Pierce & Benton, 2006; Cau & Fanti, 2011; Young et al., 2013a; Parrilla-Bel & Canudo, 2015). The concavity is shallow and similarly developed in the anterior and posterior articular surfaces.

Ribs

Cervical ribs

Three cervical ribs are identified. They form short, slender ribs directed posteriorly with an acute end (Figs. 15A–15C). The external face forms a ridge starting from the tuberculum and capitulum. The medial face is concave. In cross section, the cervical ribs are roughly triangular at mid-shaft. In lateral view, the ribs are “V-shaped”. This correspond to the typical condition in Metriorhynchidae (Andrews, 1913; Wilkinson, Young & Benton, 2008; Abel, Sachs & Young, 2020).

Appendicular skeleton

Ischium

The left ischium is badly preserved, lacking a major portion of its ventral part where the bone would widen the most (Fig. 16C). The proximal part is also missing, as well as the anterior process it should be bearing (Andrews, 1913; Wilkinson, Young & Benton, 2008; Herrera, Fernández & Gasparini, 2013; Young et al., 2013b). Like in other metriorhynchids, the neck of the ischium is narrow and thick, measuring 2.5 cm wide. The ischium widens and flattens distally in an overall triangular wing with a thickness of about two millimeters only. The partly preserved wing is similar to the classic metriorhynchid morphology as seen in To. carpenteri, C. araucanensis (Andrews, 1913; Wilkinson, Young & Benton, 2008; Herrera, Fernández & Gasparini, 2013; Le Mort et al., 2022). The surface of the wing is covered by striations corresponding to muscular attachment marks on both sides but better expressed on the lateral one.

Phylogenetic analysis

Following the methodological protocol of Young et al. (2020b), the eight parsimony analyses resulted in eight strict consensus topologies with lengths ranking from 2,417 steps for the unweighted analysis to 2,033 steps for the weakly downweighted topologies (k = 20 and k = 50; see Table 2). The complete eight strict consensus trees are provided in Material S3, while descriptive statistics for each of these trees are presented in Table 2. Focusing on the internal relationships of Geosaurinae, four distinct topologies are recovered (Fig. 17). They all have a similar structure, except for the position of Tyrannoneustes lythrodectikos, Ieldraan melkshamensis, Geosaurus lapparenti, “Metriorhynchus” westermanni, and “Metriorhynchus” casamiquelai.

In all of the strict consensus trees, Torvoneustes jurensis is included in a polytomous clade with all terminal taxa assigned to Torvoneustes (Fig. 17). This clade forms a polytomy with the E-clade and Purranisaurus potens. In the unweighted analysis, this group made of Torvoneustes, the E-clade and P. potens forms a polytomy with Ty. lythrodectikos and a clade consisting of Geosaurina + Plesiosuchina (Fig. 17A). These taxa form together the Geosaurini. In the strongly downweighted analyses (k = 1 and k = 3), “M.” westermanni and “M.” casamiquelai assume more basal positions outside of geosaurines (Fig. 17B). Tyrannoneustes lythrodectikos is sister group to the E-clade, which corresponds to the ‘subclade T’ of Foffa, Young & Brusatte (2018). Within Geosaurina, I. melkshamensis and G. lapparenti switch positions. The moderately downweighted analyses (k = 7 and k = 10) result in a topology overall consistent with that of the strongly downweighted analyses, except that “M.” westermanni and “M.” casamiquelai regain a basal position among geosaurines (Fig. 17C). In the moderately to weakly downweighted analyses (k = 15, k = 20 and k = 50), I. melkshamensis and G. lapparenti switch back to the positions they have in the strict consensus of the unweighted analysis (Fig. 17D). Overall, the results of our parsimony analyses are consistent with those of Young et al. (2020b) and Abel, Sachs & Young (2020), except our strict consensus for the unweighted analysis that is less well resolved (subclade T only recovered by the weighted analyses).

To improve the resolution of relationships, unstable taxa were pruned a posteriori from the consensus trees to produce a maximum agreement subtree for each of the parsimony analysis (see Material and methods). For the unweighted analysis, a total of 13 OTUs, including eight geosaurines (“Metriorhynchus” brachyrhynchus, Tyrannoneustes lythrodectikos, Geosaurus lapparenti, Purranisaurus potens, Druegendorf merged, English rostrum, Torvoneustes sp., Torvoneustes mexicanus), are pruned from the original set of 180 OTUs. In contrast, 33 OTUs, including 12 geosaurines (“Metriorhynchus” brachyrhynchus, Neptunidraco ammoniticus, Purranisaurus potens, Druegendorf merged, English rostrum, Mr. Passmore’s specimen, Chouquet cf. hastifer, Torvoneustes sp., Torvoneustes mexicanus, Geosaurus grandis, Geosaurus giganteus, and Ieldraan melkshamensis or Geosaurus lapparenti), are pruned for the weighted analyses. The four new topologies obtained for relationships within Geosaurinae are presented in Fig. 18 and the complete pruned consensus trees are available in Material S4.

All maximum agreement subtrees suggest that Torvoneustes sp. and Torvoneustes mexicanus are unstable taxa. This is probably the result of the partial nature of these terminals represented by single specimens consisting of an incomplete occipital region and a portion of rostrum, respectively. The pruning of these taxa reveals the internal relationships of the Torvoneustes clade. Torvoneustes coryphaeus is recovered as the most basal taxon in a sister group relationship with a clade consisting of cf. Torvoneustes and To. jurensis + To. carpenteri (Fig. 18). From a more general perspective, the internal relationships of the E-clade and Geosaurina are also identified as unstable.

The Bayesian analysis results in a resolved, but poorly supported tree (Fig. 19). Torvoneustes jurensis is resolved as the sister taxon of To. carpenteri. These two species form the most derived clade within the clade Torvoneustes. Torvoneustes mexicanus is found as the sister taxon of To. jurensis + To. carpenteri, whereas To. coryphaeus appears as the most basal unit. Most nodes in the Torvoneustes clade and the E-clade are weakly supported. The Bayesian topology for Geosaurini is similar to the one obtained by Young et al. (2020b), with only two exceptions: (1) the position of cf. Torvoneustes and Torvoneustes sp. are switched; (2) To. mexicanus and To. carpenteri are no longer sister taxa. In our analysis, the node supports within subclade T is slightly lower, which can be explained by the inclusion of To. jurensis as a new terminal. the complete maximum compatibility is are available in Material S5.

The different phylogenetic analyses performed as part of the present study all consistently find To. jurensis (MJSN BSY008-465) nested within a Torvoneustes clade, supporting our identification. Both the Bayesian analysis and the maximum agreement subtrees of the parsimony analyses support a close relationship between To. jurensis and To. carpenteri.

MJSN BSY008-465 assigned to Geosaurinae

The absence of mandibular fenestrae, the orbits facing laterally and overhung by the prefrontals, and the absence of osteoderms (despite the preservation of numerous postcranial remains) unambiguously place MJSN BSY008-465 among Metriorhynchidae (Fraas, 1901, 1902; Andrews, 1913; Young et al., 2010). In this section, we discuss the assignment of this specimen to Geosaurinae.

The dental characteristics of the Late Jurassic metriorhynchids allow to discriminate the Geosaurini from the Metriorhynchinae. The latter usually have smooth to faintly ornamented teeth, uncarinated or with low, non-serrated carinae, whereas Geosaurini have smooth to heavily ornamented teeth with high, serrated carinae (Table 1). The presence of prominent serrated carinae appears to be restricted to the Geosaurini tribe (Andrade et al., 2010; Young et al., 2011). Metriorhynchids genera can even be identified based on teeth only and within Geosaurini, teeth can be used for species identification, (Young & Andrade, 2009; Andrade et al., 2010; Schaefer, 2012; Young et al., 2013a, 2013b; Barrientos-Lara et al., 2016; Foffa et al., 2017; Foffa, Young & Brusatte, 2018; Schaefer et al., 2018; Madzia et al., 2021). MJSN BSY008-465 shares with Torvoneustes the presence of conspicuous apicobasal ridges on the first two-thirds of the crown shifting to an anastomosed pattern on the apex, as well as the bending of the enamel ridges toward the carinae. The teeth of some Metriorhynchinae (Cricosaurus spp., Maledictosuchus nuyivijanan) resemble those of Torvoneustes with conspicuous apicobasal ridges on the first two-thirds of the crown, but in their case the apex is smooth and the carinae are low and non-serrated (Sachs et al., 2019; Table 1).

Derived geosaurines, such as Plesiosuchus, Dakosaurus, Torvoneustes, Geosaurus, Purranisaurus and the E-clade, show an extreme reduction in interalveolar space associated with a reduction of the tooth count and a moderate enlargement of the teeth (Young et al., 2012b, 2013a, 2013b; Herrera, Gasparini & Fernández, 2015; Abel, Sachs & Young, 2020). Metriorhynchines, including those with a low tooth count such as ‘Cricosaurus’ saltillensis (see below), have large and variable interalveolar spaces (Buchy, Young & Andrade, 2013; Young et al., 2020a; Herrera, Aiglstorfer & Bronzati, 2021; Herrera, Fernández & Vennari, 2021). The only exception is Gracilineustes leedsi in which reduced interalveolar spaces are associated with a high tooth count (+30 per maxilla; Young et al., 2013b). MJSN BSY008-465 has reduced interalveolar spaces associated with moderately enlarged teeth, which corresponds to the condition in derived geosaurines. In addition, the teeth of MJSN BSY008-465 are on average larger than the typical height observed for metriorhynchine teeth, which are usually shorter than two centimeters (Wilkinson, Young & Benton, 2008; Herrera, Fernández & Vennari, 2021).

The tooth count is often used to differentiate geosaurines from metriorhynchines (Young et al., 2013b), but it should be noted that the absolute tooth count is known only in a limited number of species (Table 1). Geosaurini are usually considered to have 16 or less teeth per maxilla (Cau & Fanti, 2011). With this in mind, the estimated tooth count for MJSN BSY008-465 (17–18 or more, see above) may seem high for a Geosaurini, especially when some non-rhacheosaurin metriorhynchines, such as “C.” saltillensis and “C.” macrospondylus, present comparable tooth counts (Table 1; Buchy, Young & Andrade, 2013; Aiglstorfer, Havlik & Herrera, 2020). However, this relatively low tooth count in some rhacheosaurins seems to be linked to a pronounced shortening of the skull. On the other hand, it appears that the tooth count is poorly estimated in Torvoneustes because no complete maxilla is known, and some species have a comparable tooth count as MJSN BSY008-465. For example, To. coryphaeus is estimated to have up to 19 alveoli per maxilla (Young et al., 2013b), which also falls into the range of other geosaurines such as Chouquet’s “Metriorhynchus” cf. hastifer and its at least 20 maxillary teeth (Lepage et al., 2008). Therefore, the tooth count for Torvoneustes is maybe underestimated for the moment based on the available material. It is also possible that the tendency toward the great reduction in the number of teeth is restricted to the clade uniting Geosaurina, Dakosaurina, and Plesiosuchina. In any case, it appears that tooth count, as a tool for identification, should be handled with care, especially when based on estimations.

MJSN BSY008-465 shares with Geosaurinae the presence of an acute angle of about 60° between the medial and lateral processes of the frontal. This angle is closer to 90° in most Metriorhynchinae, to the exception of Cricosaurus and Maledictosuchus in which this angle is around 45°–50° (Wilkinson, Young & Benton, 2008; Cau & Fanti, 2011; Buchy, Young & Andrade, 2013; Parrilla-Bel et al., 2013; Foffa & Young, 2014). The new specimen described herein shares several additional cranial and mandibular features with the macrophagous predators of the Geosaurini tribe: an inflection point of the prefrontals relative to the skull midline of 70° or less; a high glenoid fossa and retroarticular process; a strongly expressed surangular-dentary groove (Young & Andrade, 2009; Young et al., 2012b, 2013b; Foffa & Young, 2014). In addition to some dental characteristics discussed above, MJSN BSY008-465 also has some cranial features that may recall the metriorhynchine Cricosaurus. MJSN BSY008-465 notably has a smooth and unornamented cranial surface, but this is also the case of D. andiniensis and P. potens (Pol & Gasparini, 2009; Herrera, Gasparini & Fernández, 2015). The frontal of MJSN BSY008-465 differs in shape from that of other geosaurines and somewhat resembles that of “C.” saltillensis (Buchy, Young & Andrade, 2013), but there is a great diversity of frontal shapes among metriorhynchids (Foffa & Young, 2014, fig. 10; Herrera, 2015). Despite these few similarities, MJSN BSY008-465 lacks some cranial characters that are typical of Cricosaurus, such as the presence of a bony septum on the premaxillary and the presence of reception pits on the maxilla (as seen in C. bambergensis and C. albersdoerferi; Sachs et al., 2019, 2021). Therefore, the craniomandibular characters, like the dental characters, indicate that MJSN BSY008-465 should be assigned to Geosaurinae and suggest that the resemblances with Cricosaurus are only superficial.

The total body length of MJSN BSY008-465 is estimated to be around four meters (De Sousa Oliveira et al., 2023), which exceeds the sizes typically measured and estimated for Rhacheosaurus (157 cm), Cricosaurus (200 cm), and Geosaurus (270 cm), but falls in the range of large-bodied geosaurines such as Suchodus durobrivensis Lydekker, 1890 (410 cm), D. andiniensis (430 cm); To. coryphaeus (370 cm) and To. carpenteri (400–470 cm) (Young et al., 2010, 2019). Because this specimen is one of the few metriorhynchids found with a significant part of its postcranium (see also Sachs et al., 2019; Le Mort et al., 2022), some remarks relative to the assignment of the specimen should be made also on this part of the skeleton.

Geosaurinae (Neptunidraco ammoniticus, “M.” brachyrhynchus, Ty. lythrodectikos, To. carpenteri, Geo. lapparenti, D. maximus) and MJSN BSY008-465 share a centrum length subequal to centrum width on cervical vertebrae, whereas in other Metriorhynchidae (Thalattosuchus superciliosus, Rhacheosaurus gracilis, C. araucanensis, C. suevicus, C. bambergensis, C. albersdoerferi) the centrum is shorter than wide (see character 423 of the phylogenetic matrix; Parrilla-Bel & Canudo, 2015). In MJSN BSY008-465, the neural length spine of the dorsal vertebrae is about half the length of the centrum and its dorsal margin is rounded. This is markedly different from C. suevicus and C. albersdoerferi in which the neural spine of the dorsal vertebrae is wide and rectangular with a flat dorsal margin and subequal in length to the length of the centrum (Sachs et al., 2021). The centrum of the dorsal vertebrae is also distinctly longer than high in C. albersdoerferi and Cretaceous metriorhynchids (Sachs, Young & Hornung, 2020; Sachs et al., 2021), whereas the centrum length is subequal to its height in MJSN BSY008-465 as in E-clade member PSHME PH1 and N. ammoniticus (Cau & Fanti, 2011; Abel, Sachs & Young, 2020).

In metriorhynchids there is a drastic reduction of the length of the tibia and fibula compared to the femur (Fraas, 1902; Andrews, 1913; Foffa et al., 2019). The tibia of MJSN BSY008-465 is not preserved, but the fibula is usually subequal or slightly longer than the tibia in metriorhynchids and can therefore be used as a proxy (Sachs et al., 2019). As preserved, knowing that each bone is missing parts of the articular heads, the fibula of MJSN BSY008-465 is about 30% of the femoral length, which corresponds to the proportions usually observed in metriorhynchids. Members of the tribe Rhacheosaurini appear to be an exception because their tibia is less than 30% of the femur length (Andrews, 1913; Wilkinson, Young & Benton, 2008; Young & Andrade, 2009; Cau & Fanti, 2011; Foffa et al., 2019). However, this character (see #518 in the phylogenetic matrix; Supplementary data in Foffa et al., 2019) can only be scored in a small number of metriorhynchids, and its repartition should be further investigated, especially in derived geosaurines such as Dakosaurus. Although not as diagnostic as the dental and cranial characters, the postcranial characters of MJSN BSY008-465 tend to suggest an affinity of this specimen with geosaurines rather than with non-metriorhynchine metriorhynchid and to exclude a relationship with rhacheosaurins such as Cricosaurus.

Taxonomic diversity in the geosaurine genus Torvoneustes

The genus Torvoneustes is a member of Geosaurinae and is currently represented by three valid species: the type species To. carpenteri from the upper Kimmeridgian, a skull heavily crushed and not described in many details, and some postcranial remains from a second specimen (Grange & Benton, 1996; Wilkinson, Young & Benton, 2008; Andrade et al., 2010); To. coryphaeus from the lower Kimmeridgian, a 3D preserved skull missing the anterior part of the rostrum (Young et al., 2013b); and To. mexicanus, likely from the Kimmeridgian, represented by a piece of a rostrum (Barrientos-Lara et al., 2016). Other specimens were referred to the genus and include several isolated teeth form the Oxfordian (BRSMG Cd5591, Cd5592; CAMSM J.13305J, 13309, J.13310, YORYM:2016.306–2016.309; OUMNH J.52428, J.47587a, J.47560; Foffa, Young & Brusatte, 2018); three specimens referred to Torvoneustes sp., MJML K1707 from the Upper Kimmeridgian; OUMNH J.50061 and OUMNH J.50079-J.50085 from the lower Tithonian (Young et al., 2019); cf. Torvoneustes (MANCH L6459) from the middle Oxfordian (Young, 2014), and Torvoneustes? (NHMW 2020/0025/0001) an isolated tooth crown from the upper Valanginian. All specimens are from England, except To. mexicanus and Torvoneustes?, which are from Mexico and Czech Republic respectively (Table 1).

Based on the three valid species, the genus Torvoneustes is defined by the following characteristics (Wilkinson, Young & Benton, 2008; Andrade et al., 2010; Young et al., 2013b; Barrientos-Lara et al., 2016): great reduction of the interalveolar space; acute angle (around 60°) between the medial and lateral processes of the frontal; inflexion point on the lateral margin of the prefrontals directed posterolaterally at an angle of ~70° from the anteroposterior axis of the skull; circular to subcircular tooth cross section; carina formed by a keel and a contiguous row of poorly defined microscopic denticles difficult to observe even under SEM observation; conspicuous enamel ornamentation consisting of subparallel apicobasal ridges on the first two thirds of the crown shifting to short, low relief tubercles on the apex. The new specimen described herein closely follows this definition, but also presents significant differences with each of the recognized species.

The frontal of MJSN BSY008-465 is shaped differently than those of To. carpenteri and To. coryphaeus. In To. carpenteri, the frontal is shorter than in To. coryphaeus and MJSN BSY008-465. Torvoneustes coryphaeus has an ornamented frontal while in To. carpenteri and MJSN BSY008-465 the frontal is smooth. Finally, MJSN BSY008-465 is characterized by a clear angle between the anterior and posterolateral processes of the frontal, at the meeting point of the frontal, nasals and prefrontals. In the aforementioned two species, there is no visible angle and the processes are aligned in an almost straight line. Variation in shape of the frontals in the genus Torvoneustes is not well known, as only two described specimens preserve this element in addition to the new material described herein. Within metriorhynchids, we can note the great interspecific variation in the shape of the frontal (Foffa & Young, 2014). For example in Cricosaurus, there is a great variation of the frontal shape, as seen in C. araucanensis (Herrera, 2015).

MJSN BSY008-465 cannot be compared with Torvoneustes sp. (MJML K1707). The latter consists of an incomplete occipital region and this part is completely lost in our specimen. However, MJSN BSY008-465 can be distinguished from all other English specimens. Torvoneustes jurensis differs from the type species To. carpenteri based on the following characters: smooth maxillae without grooves; anterior process of the frontal reaching the anterior margin of the prefrontals; posterolateral edges of the prefrontals lacking “finger-like” projections; teeth slenderer, more curved, and with a sharp apex. MJSN BSY008-465 also differs from To. coryphaeus in having: a smooth skull; rounded posterolateral edges of the prefrontals (no acute angle); slender teeth with sharp apex; tooth ornamentation touching the carina in the upper part of the crown. Isolated Torvoneustes sp. teeth from the UK (BRSMG Cd5591, Cd5592; CAMSM J.13305J, 13309, J.13310, YORYM:2016.306–2016.309; OUMNH J.52428, J.47587a, J.47560, J.50061, J.50079-J.50085) preserve tooth crowns and roots very similar to To. carpenteri, As noted above, MJSN BSY008-465 has slender teeth. Finally, MJSN BSY008-465 is different from cf. Torvoneustes MANCH L6459 by its smooth cranium.

Torvoneustes mexicanus is only known by a single specimen that consists of a fragment of snout with preserved teeth. The species diagnosis is based only on the teeth with the following characters: conical, lingually curved, bicarinate, and more slender than in other Torvoneustes species; sharp apex; microziphodont condition with well-defined isomorphic denticles; crown enamel ornamentation consisting of apicobasally aligned ridges on the basal two-thirds of the crown and shifting to short drop-shaped tubercles meeting the carina on the apex (Barrientos-Lara et al., 2016). On macroscopic observation, the teeth of MJSN BSY008-465 and To. mexicanus are very similar, but they differ on microscopic features. The denticles of To. mexicanus are well defined, regular in shape, size and distribution with a denticle basal length of about 142 µm and a denticle density (#denticles/5 mm) of 30 (Barrientos-Lara et al., 2016). The crown height seems to range between 1 and 2.5 cm. In MJSN BSY008-465, the basal length and distribution of denticles are irregular; on the tooth total length, the base of the denticles can vary from 120 to 200 µm in length with an average of 160 µm measured on four different teeth; denticles are also more densely packed in the upper middle of the tooth with a density reaching 40 while they can drop to 30 on the basal most part of the carina and the tooth apex. These observations are homogenous on the four observed teeth. It is to be noted however that denticles are sometimes hard to discern, because of their shape and size but also due to areas where they are worn or where the carina is broken. We compared these results with measurements based on formerly published SEM photographs of teeth of To. carpenteri (Young et al., 2013a). In this species, the denticle basal length varies between 120 and 220 µm, with most measured denticles having a basal length between 160 and 200 µm. Unfortunately, denticle density cannot be determined. Again based on published SEM photographs (Madzia et al., 2021), the denticles of Torvoneustes? (NHMW 2020/0025/0002) range in size from 200 to 270 µm in basal length and have a density of 19.

Previous studies showed that the denticle density is variable between geosaurines species such as Dakosaurus and Geosaurus (Andrade et al., 2010). Measurements took in the middle of the carina gives the following densities for the microziphodont specimens: 28.1 in Geosaurus indet. (NHM R.486), and 33.3 for the mesial carina and 41.7 for the distal carina in G. grandis. These results suggest possible interspecific variations in denticles density for a same tooth morphotype among geosaurines and that these variations should not be overlooked for systematic purposes (Andrade et al., 2010). In this study however, the density of denticles is only measured on one tooth. The intraspecific and the individual variation, which were documented in other studies on crocodylomorphs teeth (Prasad & de Lapparent de Broin, 2002), were not explored in this case. However, the observations on MJSN BSY008-465, in addition to previous studies on metriorhynchid teeth, support the idea that microscopic dental characteristics in metriorhynchids have a potential to be used in systematics and should be further investigated.

From the above discussion and considering that To. mexicanus is only known by a very incomplete specimen of uncertain stratigraphical origin, it seems reasonable to conclude that MJSN BSY008-465 represents a different species (Table 3), which we name Torvoneustes jurensis. Future discoveries of more complete fossil specimens of To. mexicanus will allow a better understanding of the differences between these species.

Macroevolution trends in Torvoneustes

Young et al. (2013b, 2019) discussed macroevolutionary trends in the genus Torvoneustes, we here include the new taxon To. jurensis within these putative trends and discuss how its addition strengthens or contradicts them. The first potential trend noted by Young et al. (2013b, 2019) is a reduction of the maxillary tooth count with time from “relatively high” in cf. Torvoneustes (MANCH L6459; Young et al., 2019) and 17–19 in To. coryphaeus (Young et al., 2013b), to 14 in To. carpenteri (Wilkinson, Young & Benton, 2008). However, it should be noted that no complete maxilla is known for any specimen referred to Torvoneustes, so these tooth counts represent estimations (see above). Another trend noted in Torvoneustes is a decrease in dermocranial external ornamentation, with the upper Kimmeridgian To. carpenteri having a smoother skull than the older representatives To. coryphaeus and cf. Torvoneustes MANCH L6459 (Grange & Benton, 1996; Wilkinson, Young & Benton, 2008; Young et al., 2013b; Young, 2014).

Concerning the tooth morphology, the following macroevolutionary trends were proposed: increasing enamel ornamentation; blunter crown apices; tooth crown losing the lingual curvature; and crown cross section becoming subconical (Young et al., 2019). These trends are interpreted to be linked to an increasingly durophagous diet (Young et al., 2013b, 2019; Foffa et al., 2018). Torvoneustes coryphaeus, To. carpenteri, Torvoneustes? (NHMW 2020/0025/0001) and Torvoneustes sp. (OUMNH J.50061 and OUMNH J.50079-J.50085) fit relatively well into this proposed evolutionary trend. However, that is not the case of To. mexicanus which has more slender teeth than To. coryphaeus and To. carpenteri, as well as more curved teeth than the latter. The teeth of To. mexicanus are also sharper than those of To. coryphaeus, To. carpenteri and Torvoneustes sp. OUMNH J.50061 and OUMNH J.50079-J.50085.

The acquisition of false serration (false ziphodont dentition; Prasad & de Lapparent de Broin, 2002; Young & Andrade, 2009; Andrade et al., 2010) is another macroevolutionary trend proposed for Torvoneustes (Young et al., 2019). Torvoneustes coryphaeus is the only species lacking the false ziphodont dentition and is also the oldest specimen whose teeth are known (Table 3). The other specimens preserving teeth (To. carpenteri; Torvoneustes sp., OUMNH J.50061 and OUMNH J.50079-J.50085; Torvoneustes?, NHMW 2020/0025/0001; To. mexicanus) all present the false ziphodont condition. Therefore, the authors propose the acquisition of the false ziphodont condition in all species younger than To. coryphaeus (Young et al., 2019).

The discovery of the occipital region of a Torvoneustes specimen of great size from the Tithonian of England led (Young et al., 2019) to propose an increase of body size as a possible evolutionary trend within Torvoneustes. This specimen is estimated to be around 6 m long while other Torvoneustes specimens are estimated to be between 3.70 and 4.70 m long (Young et al., 2011, 2019). And finally, Young et al. (2019) also suggested that the increase in length of the suborbital fenestrae leading to an enlarged pterygoid musculature and the ventralization of basioccipital tuberosities would be another evolutionary trend of Torvoneustes. However, because this part of the cranium is not preserved in To. jurensis, this trend will not be further discussed below.

Torvoneustes jurensis fits relatively well with some of the aforementioned macroevolutionary trends. First, To. jurensis is younger than To. coryphaeus and indeed presents the false ziphodont condition on its teeth. However, it should be noted that in the current state of knowledge the distribution of this character does not per se corresponds to an evolutionary trend, especially considering the observation of this character among Oxfordian isolated teeth (Foffa, Young & Brusatte, 2018), older than To. coryphaeus. Therefore, false ziphodont condition may be a synapomorphy uniting Torvoneustes species but lost in To. coryphaeus.

Torvoneustes jurensis, which is from the early late Kimmeridgian, has an even smoother cranium than To. carpenteri. Dermocranial ornamentation appears early during ontogeny in crocodylomorphs, usually developing in specimens with a skull longer than 200 mm (de Buffrenil, 1982; de Buffrénil et al., 2015). This indicates that the smooth cranium of To. jurensis is not linked to its ontogenic stage and fits into the evolutionary trend proposed by Young et al. (2019). Within Thalattosuchia, the trend toward smoother dermocranial bones is believed to improve hydrodynamic efficiency (Young et al., 2013b) and overall follows the idea that pelagic species have less ornamented skulls than semi-aquatic one (comparing metriorhynchids and pelagic teleosauroids to non-pelagic teleosauroids for example; Clarac et al., 2017; Foffa et al., 2019). This trend is also present in Dakosaurus with D. andiniensis, the geologically younger species, showing smoother cranial bones than D. maximus (Pol & Gasparini, 2009; Young et al., 2012a). However, the functional role of bone ornamentation remains controversial in crocodylomorphs (de Buffrénil et al., 2015). Clarac et al. (2017) presented evidence that the evolution of ornamentation in pseudosuchians is influenced by both natural selection and Brownian motion. The study shows that heavy ornamentation is present in pseudosuchians with semi-pelagic lifestyle and linked to basking for animals with low mobility, these results are backed by findings from Pochat-Cottilloux et al. (2022). Therefore, the loss of ornamentation in Thalattosuchia may rather be linked to an increasingly pelagic lifestyle than directly to hydrodynamic efficiency. The reduction of ornamentation of the cranium and osteoderms was also observed by Foffa et al. (2019) in teleosauroids and similarly linked it to adaptation to pelagic lifestyle.

The description of To. jurensis contradicts some aspects of the other proposed evolutionary trends. Torvoneustes jurensis presents slenderer and sharper teeth than the other Torvoneustes species, except To. mexicanus. In their description, Barrientos-Lara et al. (2016) raised the question of whether the slender and sharp teeth of To. mexicanus could be linked to ontogeny, but the lack of data to characterize ontogenic changes within Torvoneustes teeth led them to consider the differences between To. mexicanus and other Torvoneustes as a specific characteristic. The teeth of To. jurensis are very similar to those of To. mexicanus. However, the skull length of the holotype of To. jurensis is similar to that of the holotype of To. carpenteri. Their body length is estimated to be close to 4.0 m for To. jurensis and between 4.0 and 4.70 m for To. carpenteri (Grange & Benton, 1996; Wilkinson, Young & Benton, 2008; Young et al., 2011; De Sousa Oliveira et al., 2023). Therefore, ontogeny cannot explain the differences in tooth morphology between these species. The sharp and slender teeth of To. jurensis and To. mexicanus may instead represent a diverging tooth morphotype within the genus. It might indicate that To. mexicanus and To. jurensis are less specialized than species with more robust teeth. They might be opportunist feeders with durophagous tendencies. This would be consistent with the Kimmeridgian environment of the Jura platform: Torvoneustes jurensis was found in a carbonate platform environment where remains of teleosauroids (Sericodon jugleri von Meyer, 1845; Proexochokefalos cf. bouchardi and another durophagous genus Machimosaurus hugii Meyer, 1837) are abundant, as well as many coastal marine turtles (Thalassochelydia Anquetin, Püntener & Joyce, 2017) and fishes with hard scale (e.g., Scheenstia sp., Fig. 20).

The maxillary tooth count of To. jurensis is estimated to be at least 16 or 17, but is probably higher. Therefore, it seems like the reduction of maxillary tooth count is not a homogenous trend in the genus. However, it should be stressed once more that no definitive maxillary tooth count is known at the moment for any specimen referred to Torvoneustes. It is then possibly too early to conclude on any trend for this character.

Regarding the increase in body size, it should be noted that crocodilians continue to grow well into adulthood (Sebens, 1987; Grigg & Kirshner, 2015) and that only a handful of specimens are known for Torvoneustes. In these circumstances, any conclusion on size evolutionary trends must therefore be taken with care. As noted above, the holotype specimens of To. carpenteri and To. jurensis are roughly of comparable total length, which could agree with the proposed evolutionary trend as the two species are roughly of the same age. However, many isolated teeth showing similar characteristics as those of MJSN BSY008-465 were found in the same stratigraphical layers during the excavation on the A16 highway, along with teeth of Sericodon, Proexochokefalos, Machimosaurus and Dakosaurus (Schaefer, 2012; Schaefer et al., 2018): teeth with sub-circular to ovoid cross-section, bicarinate with micro-ziphodont condition, enamel ornamentation composed of sub parallel apicobasal ridges on the basal two thirds of the crown shifting into low relief drop-shaped ridges forming an “anastomosed pattern” on the remaining upper third of the crown. These teeth show a great variation of height and base length (Schaefer, 2012; L. Girard, 2021, personal observation). One of these teeth in particular (MJSN TCH007-91) shows a base length 31% longer than the largest tooth associated with MJSN BSY008-465. The great variation in crown size of isolated teeth indicates that specimens of various sizes (and potentially ages) visited the area, including specimens significantly larger than MJSN BSY008-465. Considering there is only a few Torvoneustes specimens known between the middle Oxfordian and the Tithonian, it seems premature to consider size increase as an evolutionary trend in Torvoneustes for the moment.