Revision of the Late Jurassic teleosaurid genus Machimosaurus (Crocodylomorpha, Thalattosuchia)

Machimosaurus was a large-bodied genus of teleosaurid crocodylomorph, considered to have been durophagous/chelonivorous, and which frequented coastal marine/estuarine ecosystems during the Late Jurassic. Here, we revise the genus based on previously described specimens and revise the species within this genus. We conclude that there were three European Machimosaurus species and another taxon in Ethiopia. This conclusion is based on numerous lines of evidence: craniomandibular, dental and postcranial morphologies; differences in estimated total body length; geological age; geographical distribution; and hypothetical lifestyle. We re-diagnose the type species Machimosaurus hugii and limit referred specimens to only those from Upper Kimmeridgian–Lower Tithonian of Switzerland, Portugal and Spain. We also re-diagnose Machimosaurus mosae, demonstrate that it is an available name and restrict the species to the uppermost Kimmeridgian–lowermost Tithonian of northeastern France. We re-diagnose and validate the species Machimosaurus nowackianus from Harrar, Ethiopia. Finally, we establish a new species, Machimosaurus buffetauti, for the Lower Kimmeridgian specimens of France and Germany (and possibly England and Poland). We hypothesize that Machimosaurus may have been analogous to the Pliocene–Holocene genus Crocodylus in having one large-bodied taxon suited to traversing marine barriers and additional, geographically limited taxa across its range.


Summary
Machimosaurus was a large-bodied genus of teleosaurid crocodylomorph, considered to have been durophagous/ chelonivorous, and which frequented coastal marine/estuarine ecosystems during the Late Jurassic. Here, we revise the genus based on previously described specimens and revise the species  Machimosaurus/'Steneosaurus' obtusidens-like tooth crowns are also found in Bathonian deposits of the Great Oolite Group from England. One such tooth (TWCMS K1239) from Maidford in Northamptonshire has a blunt apex, and numerous apicobasal enamel ridges on the lingual and labial surface. This tooth crown is similar to the posterior-most tooth crowns of Steneosaurus larteti skulls (OUMNH J.29850 and OUMNH J.29851) from the Great Oolite Group of England (figure 1). It is therefore possible that the 'Machimosaurus' tooth taxa from France are also referable to S. larteti; however, investigating that hypothesis is beyond the scope of this study.

Callovian
'Steneosaurus' obtusidens has been considered to be a subjective junior synonym of the Kimmeridgian taxon M. hugii, although most studies which drew this conclusion did note that further study was necessary on the anatomy and taxonomy of blunt-toothed teleosaurids [4,12,13]. 'Steneosaurus' obtusidens is known from the Oxford Clay Formation of central England, and a specimen from the Marnes de Dives Formation of northern France has been referred to this species [1,13,14]. Recent studies, however, have considered 'S.' obtusidens to be distinct enough to warrant its own genus [10,15], which we agree with. The holotype of this species is currently being re-described, which will help elucidate its anatomy and evolutionary relationships.
An isolated tooth crown (OUMNH J.14464) referred to M. rigauxi was found at Hanborough railway station, Oxfordshire, England (Cornbrash Formation) [16]. The description and figures of this specimen [16, p. 26-27, plate 1 fig. 5a-c] match the posterior dentition of the 'S.' obtusidens holotype, as the carinal keels are prominent and the apicobasal enamel ridges near the keels converge and contact the keel itself [1,15].
Based upon the Bathonian tooth taxa being more similar to Steneosaurus (in particular S. larteti) and 'S.' obtusidens being a distinct taxon, Machimosaurus is therefore unknown in the Middle Jurassic.

Oxfordian-Kimmeridgian
From near Harrar in Ethiopia (Oxfordian or Kimmeridgian aged deposits), an anterior region of dentary was described as a new species of pliosaurid sauropterygian, cf. Simolestes nowackianus [19] (figure 3). However, based on the dental morphology, the spatulate anterior region, arrangement of the dentary alveoli and thecodont tooth replacement, Bardet & Hua [20] demonstrated that it is in fact a large specimen of Machimosaurus. Currently, this is the only Machimosaurus specimen known from outside Europe.

Kimmeridgian
Prior to this study, two valid species of Machimosaurus were recognized in the Kimmeridgian of Europe: the type species M. hugii and M. mosae [4,21]. Based on the numerous European localities outlined below, during the Kimmeridgian Machimosaurus commonly frequented shallow marine ecosystems, with the occasional individual known from brackish and open-shelf environments, and possibly also freshwater environments.  From Portugal, Machimosaurus is known from two sites: -In 1943, the geologist Carlos Teixeira reported an isolated M. hugii tooth from Lagares (Colmeias, near Leiria), Portugal [22]. This tooth (MG 25) is from the Alcobaça Beds Formation (Upper Kimmeridgian). -The largest known specimen of Machimosaurus is known from the Guimarota site, Leiria, Portugal (Alcobaça Beds Formation) [6,7] (figures 4-10). The Guimarota site was deposited in either a lagoonal environment with some freshwater influx or a wooded swamp similar to extant mangrove forests. This locality has yielded several species of crocodylomorphs such as Lusitanisuchus mitracostatus, Goniopholis baryglyphaeus and Theriosuchus guimarotae [23,24].
From Spain: -Isolated cf. Machimosaurus teeth have been reported from the Tereñes Formation of the Asturias coast, Northern Spain [25]. This formation is considered to represent a shallow tide-less sea [26]. Teeth are also known from the Lastres Formation in Asturias [25], a fluvial-dominated deltaic system in origin [26] ( figure 11). -The Kimmeridgian ichnogenus Hatcherichnus is known from coastal and deltaic units of Asturias, Spain. It has been suggested that these track ways were made by either Machimosaurus or a large goniopholidid [27].
During the Kimmeridgian-Tithonian this region was a lagoonal environment [29].  Formation was deposited in a nearshore or shallow-shelf marine environment off the west coast of the London-Brabant Massive [30].              narrow mandible with a high tooth count; (ii) the anterior-most preserved alveoli have inter-alveolar spaces which are too long, being greater than the length of the adjacent alveoli; (iii) the splenial is very elongated and has at least 16 pairs of symphyseal alveoli adjacent; and (iv) the Meckelian groove is very deeply excavated, especially at the mandibular midline [1,4,10,14]. From England, Machimosaurus is solely known from a single incomplete isolated tooth crown discovered at Smallmouth Sands, Dorset (Lower Kimmeridge Clay Formation) [5, fig. 215A       absence of Machimosaurus in the Tithonian of Portugal is solely due to a shift in palaeoenvironment, rather than a true disappearance of the genus in the Tithonian. Sauvage [18] reported an isolated tooth from the Upper Kimmeridgian-Lower Tithonian of Santa-Cruz (Praia Azul Member, Lourinhã Formation [44]    A large vertebra previously referred to cf. Machimosaurus [47] is here no longer regarded as a crocodylomorph. The Lourinhã Formation has a rich and diverse dinosaur fauna, including eggs and tracks, but crocodylomorph remains are also common, mostly Goniopholis and Theriosuchus.

Tithonian
An isolated Machimosaurus tooth is known from marine deposits of the Higueruelas Formation at Buñol, Valencia Province, Eastern Spain [48].

Berriasian
Isolated teeth from Spain, with a possible Berriasian age, have previously been attributed to Machimosaurus. In 1916-1918, JM Catalá discovered a series of fossil vertebrates from Benagéber, Valencia province, Spain. The description of Beltrán [49] stated that the fossils were from the 'Wealdense' (i.e. Wealden facies) and that Gonophilus (sic) teeth were among the collection. Royo y Gómez [50] reviewed the Catalá collection when it was temporarily loaned to the Museo Nacional de Ciencias Naturales in Madrid. Two species of crocodylomorphs were reported as being present: Goniopholis and Machimosaurus. One year later, Royo y Gómez assigned the Benagéber crocodiles to Steneosaurus cf. obtusidens and Machimosaurus sp. nov. [51], specifying its age as 'Purbequiense' (i.e. Purbeck facies) [52]. Therefore, it is possible that the teeth were Tithonian or Berriasian in age.
Unfortunately, these teeth were never described or figured and are currently missing. It is possible that they were destroyed in the fire that ruined the Museo de Historia Natural de la Universidad de Valencia in 1932, were Beltrán was a professor [53]. As such, the presence of Machimosaurus in the earliest Cretaceous of Spain cannot currently be confirmed.

Valanginian
A partial dentary with in situ tooth crowns from the Valanginian of southern France (Département des Bouches-du-Rhône) was described as Steneosaurus sp. due to a superficial similarity in dental morphology between it and 'S.' obtusidens [54]. However, recent re-examination of the Valanginian specimen demonstrated it to be a metriorhynchid closely related to P. manselii [55].

Upper Hauterivian-Lower Barremian
Sanz et al. [56] referred isolated tooth crowns from the Lower Cretaceous of Galve (Teruel province, Spain) to cf. Machimosaurus sp. These teeth come from the sediments on the top of El Castellar Formation (uppermost Hauterivian-lowermost Barremian), which is lacustrine in origin [57]. Intriguingly, Machimosaurus was not cited in the subsequent papers on the Galve crocodylomorphs made by the same authors (e.g. [58,59]), indicating that they may have been hesitant in this taxonomic assignment.
The teeth were described as being: 'distinguished by blunt crowns with a very characteristic enamel ornamentation of anastomosed and braided ridges' [52, p. 207]. One of the teeth was figured [56, p. 208, fig. 3a,b], and based on that line drawing we do not consider it referable to Machimosaurus. This is due to the tooth's strong lingual curvature, concave lingual surface and strongly pronounced mesial and distal carinae, all characteristics not seen among Machimosaurus teeth.
Gasca et al. [60] mentioned aff. Machimosaurus teeth in the Lower Cretaceous of Allepuz (Teruel province, Spain). These teeth came from a microvertebrate site in the Camarillas Formation (Lower Barremian) that originated in an avulsion deposit, namely an ephemeral fluviatile pond, and presents a mixture of terrestrial (theropod dinosaurs and crocodylomorphs: Bernissartiidae, Atoposauridae, aff. Machimosaurus) and freshwater vertebrates (hybodontid sharks, bony fishes and amphibians). Unfortunately, these teeth have never been described or figured, thus we cannot determine the reliability of this taxonomic assessment. Therefore, there is no evidence that Machimosaurus, or any other teleosaurid, survived into the Cretaceous.

The diverse and long-lived Machimosaurus hugii
Recently, Pierce et al. [9] proposed that M. hugii was the senior subjective synonym of various Callovian teleosaurids: 'S.' obtusidens, Steneosaurus durobrivensis and Steneosaurus hulkei. No evidence for this taxonomic revision was given. However, they considered M. mosae to be distinct from this long-lived (more than 10 Ma) M. hugii species. Martin & Vincent [10, p. 194] criticized the content of their species diagnoses, as: 'most of the content of these diagnoses reveal to be either diagnostic at the genus level or to characterize all Teleosauridae'. Martin & Vincent [10, pp. 194-195] went on to show that the very high variation in maxillary and dentary tooth counts among the various Callovian teleosaurids is: 'sufficient difference to discard such an interpretation (the synonymy)'. We concur with this assessment. Below, we propose a revised diagnosis for Machimosaurus, which has numerous autapomorphies absent in these Callovian species. Moreover, multiple phylogenetic analyses falsify the synonymy of M. hugii with S. durobrivensis and/or 'S.' obtusidens [10,41,61].

The diverse Kimmeridgian Machimosaurus hugii
Recently, Martin & Vincent [10] fig. 1, it looks like the rostrum is too close to the orbital region of the skull. As such, the skull was probably mesorostrine (ratio of rostrum length to basicranial length would have been slightly more than 55%).
detail by Hua [4] and Vignaud (1995, unpublished PhD thesis), who studied teleosaurid ontogeny using the numerous European skulls available. Both of these authors considered M. mosae to be taxonomically distinct from M. hugii. We reject the hypothesis that the M. mosae neotype is a juvenile of M. hugii below, based on four fundamental flaws in this 'juvenile hypothesis': (i) the M. mosae neotype is comparable in size to the French [3] and German [10] skulls referred to M. hugii, all three of which differ in length by only 7 cm (93-100 cm; table 1); (ii) the lack of juvenile characteristics in any of the French [3,4] and German [10] skulls [58]; (iii) the M. mosae neotype has exostoses (formation of new bone on the surface of bones, usually seen in mature individuals) in the femur, right pubis and on the transverse processes of some caudal vertebrae [4]; and (iv) the M. mosae neotype is from the uppermost Kimmeridgian, whereas the two skulls referred to M. hugii are from the Lower Kimmeridgian, a temporal gap of some 3-5 million years. Moreover, there is another character that shows that M. mosae is a distinct taxon from M. hugii, the presence of the prearticular (figure 25). Martin & Vincent [10] described the first prearticular ever mentioned for a teleosaurid (although they did not highlight its significance). Prearticular bones were previously only known in Metriorhynchidae among thalattosuchians [1]. Interestingly, and most importantly for Machimosaurus systematics, the prearticulars are not found in the mandible of M. mosae [4]. Therefore, the loss of the prearticulars is a specific character for M. mosae (as the prearticulars are also present in S. larteti; figure 1a,b). (Owing to the status of the M. mosae neotype, we cannot currently test the hypothesis whether the absence of the prearticulars is a preservational artefact.) Postcranial characteristics supporting the distinction of M. hugii and M. mosae-not examined by Martin & Vincent [10], who unfortunately mainly focused on the skull-include: -Coracoids are highly variable among teleosaurid species [1]. The same is true between the new specimen of M. hugii and M. mosae, especially in the shape and size of the postglenoid and glenoid processes (figure 36a,b). In the German M. hugii skull [10], the coracoid glenoid process (process near the glenoid fossa that projects posterodorsally) is elongate, extending considerably from the head of the coracoid, and is a sub-isosceles triangle in shape when seen in lateral view; the coracoid postglenoid process anterior margin is very slightly concave and terminates approximately in the same frontal plane as the glenoid; and the postglenoid process posterior margin is strongly concave and terminates approximately in the same frontal plane as the posterior-end of the glenoid process. However in the M. mosae neotype [4], the coracoid glenoid process is very short, not extending far from the head of the coracoid, and is a rightangled triangle in shape when seen in lateral view; the coracoid postglenoid process anterior  Thanks to the new specimen described by Martin & Vincent [10] there are now numerous postcranial characteristics, along with the absence of the prearticular (assuming that it is not a preservational artefact), that allow easy identification of M. mosae and differentiate it from M. hugii. Furthermore, they described a depression on the dorsal surface of the quadrates near the hemicondyles on the German M. hugii skull. These depressions are not seen in M. mosae and comprise another feature differentiating these two species [4].  [66]. Curiously, throughout all the various competing arguments over specimen synonymies regarding M. hugii, the description and figures of von Huene [63] and Krebs [6,7] (as well as the Solothurn teeth) depict a taxon which is distinct from the M. mosae and the 'M. hugii' specimens described during the 1980s-2010s from France and Germany. This taxon is known from Swiss and Portuguese material. The distinctiveness of this taxon relative to other Machimosaurus specimens has not been clearly recognized until now. As we show below, this taxon is M. hugii.

Description of
von Huene [63]   anteroposterly; (ii) sub-globidont dentition (blunt apices, low apicobasal height to basal width ratio, but the teeth lack the pronounced 'globular'/bulbous morphology of true globidonty); (iii) apicobasally aligned enamel ridges immediately adjacent to the apical anastomosed region that are closely packed on both the labial and lingual tooth surfaces; (iv) uniform inter-alveolar spaces in the posterior-mid region of the maxillae, with the inter-alveolar spaces becoming slightly larger anteriorly but still being largely uniform in size; and (v) uniformly narrow inter-alveolar spaces in the dentaries. These characteristics are distinct from the morphologies seen in the M. mosae neotype [4,21] and the French and German 'M. hugii' specimens [2,3,10,17] (figures 12-14 and 21-27), in which: (i) the external surfaces of the snout bones are more strongly ornamented, with higher relief ridges and sub-circular/oval pits; (ii) no tooth crowns are sub-globidont; (iii) apicobasally aligned enamel ridges immediately adjacent to the apical anastomosed region are closely packed only on the lingual tooth surface, but on the labial surface these ridges are more widely spaced; (iv) the maxillary and (v) dentary inter-alveolar spaces are variable in size, some of which can be quite large proportionaly to the adjacent alveoli.
Krebs' [6,7]   dentition; (iii) apicobasally aligned enamel ridges immediately adjacent to the apical anastomosed region that are closely packed on both the labial and lingual tooth surfaces; (iv) paroccipital processes that are greatly enlarged, both elongated mediolaterally and with lateral ends that are expanded; (v) basioccipital tuberosities (basal tubera) that are very large in size and are sub-rectangular in shape when seen in occipital view; (vi) the inter-basioccipital tubera notch is a large inverse 'U'-shape when seen in occipital view; and (vii) uniformly spaced inter-alveolar spaces in the mid region of the maxillae. Moreover, the apicobasal ridge characteristic is also seen in the lectotype of M. hugii [6,7]. Once again, these characteristics are distinct from the morphologies seen in the M. mosae neotype [4,21] and the French and German 'M. hugii' specimens [2,3,10,17]. Even though these Swiss and Portuguese specimens are fragmentary, they share four autapomorphies that are not seen in any other teleosaurid: (i) the external surfaces of the snout bones are poorly ornamented with low relief ridges mostly orientated anteroposterly; (ii) sub-globidont dentition; (iii) apicobasally aligned enamel ridges immediately adjacent to the apical anastomosed region that are closely packed on both the labial and lingual tooth surfaces; and (iv) uniform inter-alveolar spaces in the maxillae that are proportionally narrow relative to alveoli. None of these characteristics are found in the French or German specimens (table 2). They indicate that the Swiss and Portuguese material belongs to the same diagnostic taxon.  p r e m a x i l l a r y a l v e o l i c o u n t   We estimate there to be 17 or 18 maxillary alveoli, and most likely a similar number of dentary alveoli. This is higher than the 16 maxillary and dentary alveoli estimate of Sauvage & Liénard [62]. c The seven pre-palatine maxillary count is based on the figure in Sauvage & Liénard [62]. However, this count assumes that the reconstruction is accurate, which it may not be. This is due to the unusual shape of the anterior process of the palatines and the breaks in the specimen.

Steneosaurus bouchardi skull of von Huene [63]
von Huene [63] also described an incomplete skull and mandible from Solothurn and referred them to Steneosaurus bouchardi, an Upper Kimmeridgian species of longirostrine teleosaurid. The skull (NMS 7049; figures 41-43) lacks most of the snout, right supratemporal arch, both quadrates and the palatal surface is poorly preserved. Its specimen labels show that 'Zangerl, Chicago' referred this skull to M. hugii in 1947. The Solothurn skull (NMS 7049) shares three autapomorphies with the braincase of the Leiria skull (MG-8730-2): (i) paroccipital processes are greatly enlarged, both elongated mediolaterally and with lateral ends that are expanded; (ii) basioccipital tuberosities (basal tubera) are very large in size and are sub-rectangular in shape when seen in occipital view; and (iii) the inter-basioccipital tubera notch is a large inverse 'U'-shape when seen in occipital view. Moreover, like the other Swiss specimens and the Portuguese specimen, the external surfaces of the skull bones are poorly ornamented. These features support the referral of NMS 7049 to the same taxon as the Swiss and Portuguese material described above. As we show below that this taxon is M. hugii, we therefore conclude that Zangerl's referral of NMS 7049 to M. hugii is correct.

Geological range
Middle Oxfordian to Lower Tithonian. (As noted above, we cannot confirm the presence of Machimosaurus in deposits younger than the Lower Tithonian.)

Geographical range
Africa (Ethiopia) and Europe (England, France, Germany, Portugal, Spain and Switzerland).

Generic diagnosis
Teleosaurid crocodylomorphs with the following unique combination of characters (autapomorphic characters among teleosaurids are indicated by an asterisk * ): large body size (basicranial length typically 90-110 cm, but can exceed 140 cm) * ; three alveoli per premaxilla (shared with Peipehsuchus teleorhinus); the first premaxillary alveoli are orientated strongly anteroventrally * ; 18-22 alveoli per maxilla * ; 19-25 alveoli per dentary * ; vertically orientated, interlocking dentition, with pronounced reception pits at the premaxillary, maxillary and dentary inter-alveolar spaces * ; conical teeth with blunt/rounded apices (shared with 'S.' obtusidens); tooth enamel ornamentation varies along the crown, in the basal region enamel ornamentation is composed of numerous apicobasally aligned ridges of high relief, which become an anastomosed pattern in the apical region (shared with 'S.' obtusidens); presence of carinae is variable, in anterior teeth they can be prominent but in shorter blunter teeth carinae are either very poorly developed or absent * ; ratio of crown apicobasal height to basal transverse width can be as low as 1.5 in the posterior teeth; rostrum is broad and mesorostrine, constituting less than 60% of basicranial length * ; antorbital fenestrae are absent (possibly shared with Steneosaurus heberti); supratemporal fossae are parallelogram in shape (shared with 'S.' obtusidens); ratio of maximum supratemporal fossa length to basicranial length is greater than 27% * ; three sacral vertebrae (possibly shared with 'S.' obtusidens) * ; medial quadrate hemicondyle is considerably smaller than the lateral hemicondyle * ; exoccipital is excluded from the occipital condyle (composed solely of the basioccipital) (shared with S. heberti); axis neural spine is tall in lateral view, terminating in a transverse plane that is notably dorsal to the plane of the pre-and postzygapophyses * ; axis neural spine posteriorly expanded when seen in lateral view, with the posterior margin terminating approximately in the same frontal plane as the posterior margin of the postzygapophyses * . There are, to our knowledge, no Machimosaurus specimens known from Austria. Therefore, Pierce et al. [9] and Martin & Vincent [10] were incorrect in referring to a holotype for this taxon. They appear to have considered the lectotype as being the holotype, as the same specimen number is given (the old 96 number which Krebs [6] used). Also, Pierce et al. [9] stated that the 'holotype' is from the Palaeontologische Sammlung im Museum der Stadt Solothurn, the former name for the Naturmuseum Solothurn (again, the old name which Krebs [6] used). Martin & Vincent [10], however, listed the 'holotype' as being from the Staatliche Naturhistorische Sammlungen Dresden, which is the former name for the Senckenberg Naturhistorische Sammlungen Dresden. No reason is stated why they believed that the 'holotype' was moved to a different museum in a different country. Regardless, we can confirm that the lectotype tooth is still in Solothurn. tooth count (three to four pairs); inter-alveolar spaces between the maxillary and dentary alveoli are very small (closely packed alveoli) * ; the premaxilla is notably wide at the level of the external nares, much wider than the width of the anterior end of the maxilla * ; orbits are sub-rectangular in shape * ; paroccipital processes are greatly enlarged, both elongated mediolaterally and with lateral ends that are expanded * ; basioccipital apophysis has a 'U-shaped' cross section (teleosaurid symplesiomorphy); basioccipital tuberosities (basal tubera) are very large in size and are a sub-rectangular shape in occipital view * ; the inter-basioccipital tubera notch is a large inverse 'U'-shape when seen in occipital view * ; dorsal osteoderm ornamentation is composed of small-to-large irregularly shaped pits arranged in a random manner, that are well separated from one another (somewhat similar to Steneosaurus leedsi).

Species diagnosis
Teleosaurid crocodylomorph within the genus Machimosaurus with the following unique combination of characters (autapomorphic characters are indicated by an asterisk * ): 21-22 alveoli per maxilla (approx. [16][17] of which are anterior to the palatines); 24/25 alveoli per dentary (19-20 of which are adjacent to the mandibular symphysis); low post-symphyseal dentary tooth count (two pairs) * ; inter-alveolar spaces between the maxillary and dentary alveoli are variable in size (thalattosuchian symplesiomorphy); orbits are sub-circular in shape (transverse and anteroposterior axes are sub-equal; the Steneosaurus brevior holotype also has circular orbits) * ; the quadrates have a single large circular depression on the dorsal surface near the hemicondyles * ; basioccipital apophysis has a 'U-shaped' cross section (teleosaurid symplesiomorphy); the inter-basioccipital tubera notch is a wide and gentle inverse semicircle when seen in occipital view (teleosaurid symplesiomorphy); basioccipital tuberosities (basal tubera) are reduced in size when seen in occipital view (apomorphy shared with M. mosae); axis neural arch dorsal margin is strongly concave when seen in lateral view * ; axis postzygapophyses terminate significantly posterior to posterior surface of the centrum view (somewhat similar to that seen in S. durobrivensis); coracoid glenoid process (process near the glenoid fossa that projects posterodorsally) is elongate, extending considerably from the head of the coracoid, and is a sub-isosceles triangle in shape when seen in lateral view * ; coracoid postglenoid process anterior margin is very slightly concave and terminates approximately in the same frontal plane as the glenoid * ; coracoid postglenoid process posterior margin is strongly concave and terminates approximately in the same frontal plane as the posterior end of the glenoid process * ; dorsal osteoderm ornamentation is composed of small-to-large irregularly shaped pits arranged in a random manner, that are well separated from one another (somewhat similar to S. leedsi).
Steneosaurus burgensis. The names S. burgensis and S. burgensis chanti have been applied to the Machimosaurus skull from Ain, France [3,[72][73][74]. These specific and sub-specific names are however nomina nuda. Both Jarrin [72] and Chanel [74] stated that the Ain skull was sent to Caen for preparation and study by Eugène Eudes-Deslongchamps, who proposed the name S. burgensis for the specimen, in consultation with the Société d'émulation de l'Ain. Neither Jarrin [72,73] nor Chanel [74] established the name under Article 12 of the International Commission on Zoological Nomenclature (ICZN) Code as: they did not describe the specimen, nor did they provide a definition of the species; they simply reported that the name was proposed by Eudes-Deslongchamps. Unfortunately, Eudes-Deslongchamps never published his description of the Ain skull [3,74]. The specimen was not formally described until 1982, and then it was referred to M. hugii [3]. The sub-specific epithet chanti was apparently established by those who did not fully understand zoological nomenclature [3], as it was added to the Ain skull's specimen plaque solely to honour the discoverer [3,74].
Our decision to establish a new taxon based on SMNS 91415, and not formally establish S. burgensis for the Ain skull, was for several reasons: (i) the Ain skull is partially reconstructed, and it is unclear how much is plaster and how much is real bone [3]; (ii) the cranium and lower jaw of the Ain skull are in articulation, meaning that the palatal and dorsal mandibular morphologies cannot be seen [3,74]; (iii) the German skull SMNS 91415 has the cranium and lower jaw disarticulated, allowing these morphologies to be observed [10] only slightly posterior to posterior surface of the centrum * ; coracoid glenoid process is very short, not extending far from the head of the coracoid and is a right-angled triangle in shape when seen in lateral view * ; coracoid postglenoid process anterior margin is strongly concave and terminates in a frontal plane anterior to the glenoid * ; coracoid postglenoid process posterior margin is strongly concave distally but shifts to being somewhat convex proximally and terminates in a frontal plane posterior to the posterior end of the glenoid process * ; dorsal osteoderm ornamentation is composed of numerous small, irregularly shaped pits arranged in an anastomosed pattern, these pits can fuse and become elongate grooves that radiate from the keel (similar to 'S.' obtusidens) * ; ventral osteoderms have a longitudinal keel * .

Validity of Machimosaurus mosae
Martin & Vincent [10, p. 193 However, this is key to why T. mosae is an available name under the ICZN Code. Immediately below those headings is a detailed species diagnosis, followed by an eight-page description of the skeleton, a long discussion on the affinities of Machimosaurus, and four plates with line drawings of the specimen. As such, along with the paper being published in a scientific journal, this description clearly fulfils the criteria set out in Articles 8, 11 and 12 of the ICZN Code. Article 11.6 of the ICZN Code, publication as synonymy, states: 'A name which when first published in an available work was treated as a junior synonym of a name then used as valid is not thereby made available'. Article 11.6.1 states: 'However, if such a name published as a junior synonym had been treated before 1961 as an available name and either adopted as the name of a taxon or treated as a senior homonym, it is made available thereby but dates from its first publication as a synonym'. Clearly, Sauvage & Liénard [62] treated T. mosae as an available name, and they adopted its specific name for their validly described taxon. Consequently, under Article 11.6.1 M. mosae is an available name.
However, the authorship is Sauvage & Liénard, 1879 not Liénard, 1876. This is due to Article 50.7 of the ICZN Code, which states: 'If a scientific name (taken, for example, from a label or manuscript) was first published in the synonymy of an available name and became available before 1961 through the provisions of Article 11.6, its author is the person who published it as a synonym, even if some other originator is cited, and is not the person who subsequently adopted it as a valid name.' Therefore, contra Martin & Vincent [10], M. mosae is not invalid, and the nominal authors of the specific name are indeed Sauvage & Liénard [62].

Tooth taxa of Sauvage [11]
Sauvage [11] listed three Machimosaurus species living in the Late Kimmeridgian-Early Tithonian of Northern France (around Boulogne-sur-Mer): M. hugii, M. interruptus and M. ferox. The latter two were established by Sauvage [11] for isolated tooth crowns from the area. Krebs [8, p. 48] stated that Sauvage invoked insignificant differences between the crowns when establishing his species and that Sauvage himself later withdrew the name M. ferox. Owing to the geological age and location of these isolated teeth, it is possible they are referable to M. mosae. If so, M. interruptus would have priority. As the holotype of M. interruptus cannot be found, and the neotype of M. mosae is currently unavailable for study, this possibility cannot be explored.

Holoype horizon
It is not clear what formation the specimen was found in [19,20]. It was described as being found in sandy clays, 8 m above a crystalline basement.

Species diagnosis
Teleosaurid crocodylomorph within the genus Machimosaurus with the following unique combination of characters (autapomorphic characters are indicated by an asterisk * ): the anterior dentary inter-alveolar spaces are reduced, being notably smaller in size than in M. buffetauti and M. mosae, in particular the D1-D2 and D2-D3 inter-alveolar spaces which are both less than half the length of the D2 alveoli * ; the interdentary distance between the D3 and D4 alveolar couplets is notably smaller than in M. buffetatuti and M. mosae, such that the transverse width of the left and right D4 alveoli are as large as, or greater than, the immediately adjacent flat dentary region * .
Note. While the two anterior dentary characteristics readily differentiate M. nowackianus from both M. buffetauti and M. mosae, no known M. hugii specimen preserves the anterior region of the dentary.

Ontogeny
The differences in tooth count, symphyseal length and length of the supratemporal fossae enable easy identification for two of the four Machimosaurus species (M. buffetauti and M. mosae; table 1). Interestingly, there are three skulls of M. buffetauti and M. mosae between 93 and 100 cm in basicranial length. All of these have a snout (=preorbital) length to basicranial length ratio of approximately 58%, even though there is variation in maxillary tooth count, dentary tooth count, dentary symphyseal tooth count and mandibular symphysis length. This shows that proportional snout length does not vary between these two species (even though these dental and mandibular characteristics do vary). This indicates that the differences between these two species are not due to ontogeny, because it would be unusual for juveniles and adults of the same taxon to have nearly identical skull lengths and nearly identically proportioned snouts. Moreover, when we look beyond tooth counts and biometric ratios and examine cranial, mandibular and postcranial morphologies, we find numerous characteristics that differentiate these two species (see diagnoses and tables 2 and 3).
This leaves the type species M. hugii as a remaining point of discussion for the European species. Unfortunately, the postcranial characteristics that differentiate M. buffetauti and M. mosae are unknown in M. hugii (table 3), as are the differences in cranial biometrics and tooth counts (tables 1 and 2). Machimosaurus hugii does have three braincase apomorphies: (i) paroccipital processes are greatly enlarged, both elongated mediolaterally and with lateral ends that are expanded; (ii) basioccipital tuberosities (basal tubera) are very large in size; and (iii) inter-basioccipital tubera notch is a large inverse 'U'-shape when seen in occipital view. Machimosaurus hugii also has two dental apomorphies:     (i) sub-globidont dentition and (ii) apicobasally aligned enamel ridges immediately adjacent to the apical anastomosed region that are closely packed on both the labial and lingual tooth surfaces. These five characteristics readily allow M. hugii to be identified, and seem unlikely to be under ontogenetic control (for example, a drastic change between non-globidont and sub-globidont dentition is not known in any other crocodylomorph and would be greatly unexpected). Therefore, it is unlikely that M. hugii is the adult version of either M. buffetauti or M. mosae. As further support of this conclusion, we note that the smaller M. buffetauti and M. mosae individuals do not show any juvenile characteristics (such as having proportionally large orbits to basicranial length or paired frontals (P. Vignaud 1995, unpublished PhD thesis)). Moreover, there is variation in orbit shape, frontal shape and premaxilla width (figure 44).

Age and locality
The four Machimosaurus species were not spatio-temporally contemporaneous. All of the specimens we refer to M. buffetauti are from the Lower Kimmeridgian of France and Germany (and possibly also England  The other M. hugii specimens are known from Portugal and Spain. As noted above, the exact age of the Leiria specimen is unknown, with it being either Early or Late Kimmeridgian in age. The Lourinhã specimens are close to the Kimmeridgian-Tithonian boundary. Isolated Spanish teeth are Kimmeridgian and Early Tithonian in age [25,48]. All the Iberian specimens have the same dental morphologies as the uppermost Kimmeridgian Swiss specimens (apicobasally aligned enamel ridges immediately adjacent to the apical anastomosed region that are closely packed on both the labial and lingual tooth surfaces [6,48]  Machimosaurus nowackianus is the only known non-European taxon, and thus was widely separated from the three European taxa. It is currently difficult to assess whether it may have been temporally separated from the European taxa as well, as the age of the Ethiopian deposits is only coarsely constrained to the Oxfordian-Kimmeridgian. In summary, all four Machimosaurus species were either separated by several million years in time or by hundreds or thousands of kilometres.

Body size
The neotype of M. mosae gives the first definitive evidence of body length in this genus. The skeleton was approximately 6 m in length and had a basicranial length of 96.5 cm [21]. This therefore allows us to make an estimate of total body length for any Machimosaurus specimen with a basicranial length (assuming    [62]. This results in a body length estimate of 8.08 m, making it substantially larger than the older M. buffetauti. The type species M. hugii was larger still. The incomplete Leira skull was estimated to have a basicranial length of 149 cm [6]. This gives a body length estimate of 9.26 m. As such, Krebs' [6,7] estimate of the Leira specimen exceeding 9 m long in length was reasonable. Therefore, there were three European species of Machimosaurus, which appear to have differed in total body length ( figure 46). The Lower Kimmeridgian species M. buffetauti is both the geologically oldest and smallest of the three Machimosaurus species, with a body length in the range of 5.8-6.2 m. The uppermost Kimmeridgian-lowermost Tithonian species M. mosae was larger, with a body length in the range of approximately 6-8 m. The Upper Kimmeridgian-Lower Tithonian species M. hugii was the largest of all these species, with a body length exceeding 9 m. Unfortunately, M. nowackianus is only known from an anterior dentary, so its body length cannot be reliably estimated. However, this hypothesis requires further testing as new specimens are discovered, because the sample size used here is small (although it includes all known complete skulls).
Therefore, at over 9 m in length, M. hugii is the largest known crocodylomorph of the Triassic and Jurassic Periods, and until the Cretaceous it was the largest crocodylomorph that had ever existed in Europe.

Hypothetical lifestyles
Two very different lifestyles have been hypothesized for this genus. Krebs [6,7] presented evidence that the Leiria M. hugii specimen was adapted for living in open seas, whereas Hua [4] put forward evidence that the M. mosae neotype was adapted for living in coastal, high-energy environments. These  [10] were correct and there is only one Machimosaurus taxon, this would present a major anomaly. However, as we have already shown, Machimosaurus taxa differ in craniomandibular, dental and postcranial morphologies. They also differed in geological age, locality and total body length. As such, there is overwhelming evidence that there were three non-sympatric Machimosaurus species in Europe during the Kimmeridgian, and there is no reason why these distinct species could not have had distinct lifestyles.
Krebs [6,7] postulated a convincing argument that the Leiria M. hugii specimen was well suited to an open sea lifestyle: based on the vertebral zygapophyseal articulations, M. hugii would have been well suited to swimming by lateral undulations of the tail, perhaps using the limbs for steering and balancing, and was a fast swimmer. Also, the cervicocranial depressor musculature would have been well developed, as their attachment sites on the skull were enlarged (the basioccipital tubera and the paroccipital processes), which would have greatly assisted Machimosaurus in diving [7]. Krebs' [6,7] hypothesis that M. hugii was well suited to an open sea lifestyle helps explain this species' initially odd geographical distribution: the northern Tethyian and eastern Proto-Atlantic margins (Switzerland, Spain and Portugal).
An additional line of evidence that M. hugii was better suited to a more aquatic/pelagic lifestyle comes from dermal bone and osteoderm ornamentation. The ornamentation of dermal bones and osteoderm pits in extant crocodylians are known to be vascularized [76][77][78]. Infrared thermal imaging of basking broad-snouted caimans (Caiman latirostris) demonstrates heat exchange between osteoderms and the environment. Moreover, at low temperatures (16 • C) the dorsal surface of the snout is one of the warmest regions of the body, whereas at higher temperatures (25 • C) it no longer is [78]. Furthermore, the distribution of dermal ornamentation on the skull of Pseudosuchia in general is consistent with the hypothesis that the most exposed (dorsal) parts of the skull are the most ornamented (M. Fau, M. Laurin and V. de Buffrénil 2014, personal communication). Therefore, vascularized dermal bone/osteoderm dorsal surfaces, heat exchange and blood flow alternation appear to have a role in thermoregulatory terrestrial basking behaviours [76][77][78]. However, as noted above, skull and osteoderm ornamentation of M. hugii was reduced (also see [7, fig. 17]) suggesting it was less vascularized than other Machimosaurus species. This supports current research on skull ornamentation in crocodylomorphs, which found an inverse relationship between regional dermal ornamentation levels and aquatic specialization [79].
This relationship certainly exists within the thalattosuchian clade Metriorhynchoidea, in which basal members have highly ornamented crania [80], while within the pelagic clade Metriorhynchidae there is repeated evolution of a 'smooth' skull [81]. Metriorhynchoids have a more extreme shift in thermoregulatory behaviour, as basal metriorhynchoids have osteoderms and retain external mandibular fenestrae (which enables the musculus intramandibularis of extant crocodylians to fix the jaws in a gaping position during mouth-gaping basking behaviour), whereas metriorhynchids lack osteoderms and the external mandibular fenestrae [80].
Hua [4] postulated a convincing lifestyle for the M. mosae neotype. The robust ribs, thick and keeled ventral osteoderms, thick gastralia and three sacral vertebrae would have helped M. mosae remain in place in a high-energy/turbulent environment [4]. The small paroccipital processes and basioccipital tubera show that this species was not well suited for diving [4,21]. Moreover, the high number of tightly packed dorsal osteoderm pits and the well-ornamented skull suggest that M. mosae had a highly vascularized dorsal surface, well suited for terrestrial basking behaviours.
No hypothetical lifestyle has been postulated for M. buffetauti. However, this taxon is similar to M. mosae (tables 1-3). However, comparing the figures of the ventral osteoderms and ribs in Hua [4] to those in Martin & Vincent [10] reveals that M. mosae was a more robust taxon than M. buffetauti. Interestingly, the dermal bone and osteoderm ornamentation of M. buffetauti is more similar to M. hugii than M. mosae, suggesting that this Lower Kimmeridgian species was perhaps intermediate between the two Upper Kimmeridgian extremes. As M. nowackianus is only known from an anterior dentary, no hypothetical lifestyle can be postulated.

Conclusion
We here review and clarify the systematics of Machimosaurus, one of the most distinctive teleosaurid crocodylomorphs from the Jurassic. We show that M. mosae is indeed an available name. We also show that the type specimen (the lectotype) of M. hugii is an isolated tooth crown from Switzerland, which is still curated at Solothurn, and that the Portuguese specimen was not considered to be the neotype of this species. We also established M. buffetauti sp. nov. for the Lower Kimmeridgian specimens from France and Germany, based on a previously described specimen. Most importantly, we demonstrate that there were three species of Machimosaurus in the Kimmeridgian (Upper Jurassic) of Europe, and another taxon in Ethiopia. This conclusion is not solely based on tooth counts and cranial biometric ratios, unlike recent revisions of the genus. Our revision uses these characteristics, but expands upon them to include comparative anatomy (craniomandibular, dental and postcranial), body size, hypothetical lifestyle, geological age and geographical range. This holistic approach readily identifies three nonsympatric European species and reveals that potentially contemporaneous taxa were adapted for very different ecosystems. What is surprising, however, is that much of the reasoning we outline here has been long known and thoroughly described, but had not been synthesized.
One new aspect that helps differentiate Machimosaurus species in this paper is postcranial morphology. Prior to description of the M. buffetauti holotype by Martin & Vincent [10], the only postcranial skeleton that was well described and figured was the M. mosae neotype [4]. We have attempted to begin to rectify the long neglected study of teleosaurid postcranial morphology here, as we show for the first time that the postcranial skeletons of M. buffetauti and M. mosae are distinct (table 3). We recommend that future studies on thalattosuchians do not solely focus on snout length, tooth counts and biometric ratios of skull measurements, but thoroughly investigate craniomandibular and postcranial morphologies.
As Machimosaurus specimens are becoming increasingly abundant, one potentially interesting hypothesis could be investigated with future discoveries. Was Machimosaurus in the Kimmeridgian-Lower Tithonian equivalent to the Pliocene-Holocene Asian-Australasian and American subclades of Crocodylus? The extant Asian-Australasian Crocodylus subclade has one large-bodied taxon well suited to traversing marine barriers (C. porosus) and five geographically limited taxa across its range (C. johnsoni, C. mindorensis, C. novaeguineae, C. palustris and C. siamensis), while the extant American Crocodylus subclade has one large-bodied taxon well suited to traversing marine barriers (C. acutus) and three geographically limited taxa across its range (C. intermedius, C. moreletii and C. rhombifer) [82]. In Machimosaurus, there is one large-bodied taxon well suited to traversing marine barriers (M. hugii), with geographically limited but temporally distinct taxa living in Europe and Ethiopia. Could there be more geographically limited Machimosaurus taxa along the margins of Tethys and/or palaeo-islands? It is a potentially interesting hypothesis that will require future study.
Recent phylogenetic analyses place Machimosaurus and 'S.' obtusidens as sister taxa, forming a clade of durophagous/generalist teleosaurids that lived during the Callovian-Tithonian. Their large size, robust craniomandibular and dental morphologies, and broad dietary range suggest 'S.' obtusidens and Machimosaurus were an important component of Jurassic shallow marine/brackish ecosystems. We anticipate that future discoveries, especially from outside of Europe, will further elucidate the evolution of this remarkable clade of marine crocodylomorphs.