Skip to Main Content

The Upper Cretaceous phosphates of Morocco preserve one of the world’s most diverse assemblages of mosasaurs, reflecting the adaptive radiation of this clade during the Maastrichtian. Herein, we describe a new mosasaur from these deposits. Although the teeth of this specimen resemble those of ‘Platecarpusptychodon, suggesting referral to this species, we re-examine and ultimately reject the fundamental validity of ‘P.ptychodon due to the non-diagnostic nature of its holotype and original diagnosis. We instead designate the new specimen as the holotype of a new genus and species, Gavialimimus almaghribensis, gen. et sp. nov. G. almaghribensis is characterized by a highly elongate snout, highly retracted nares and large supratemporal fenestrae, among other features. Phylogenetic analysis under multiple parsimony-based methods reveals novel substructure within the subfamily Plioplatecarpinae, consistently recovering a clade uniting the new species with Selmasaurus and the enigmatic Goronyosaurus. Synapomorphies of this clade include a highly constricted parietal, with Selmasaurus and Gavialimimus being further united by a broadly excavated medial surface of the quadrate suprastapedial process. The cranial morphology of G. almaghribensis also provides new insight into several aspects of mosasaur evolution and comparative anatomy, including adaptive radiation and niche partitioning in Moroccan marine palaeoecosystems.

Introduction

Determining phylogenetic relationships among extinct taxa is essential for understanding broad evolutionary processes such as adaptive radiation. Despite this importance, the evolutionary history of many major clades, such as mosasaurs, remains unresolved (Wright & Shannon 1988; LeBlanc et al. 2012; Simões et al. 2017a). Mosasaurs were a dominant group of marine reptiles during the Late Cretaceous (100.5–66.0 Ma), noted for their worldwide adaptive radiation during the latter stages of this period (Everhart 2005).

This diversification into various body forms and predatory specializations is particularly apparent in the Upper Cretaceous deposits of north-western Africa. Both the Maastrichtian phosphates of Morocco and the Iullemmeden Basin of the Dukamaje Formation in Nigeria and Niger record extensive mosasaur assemblages (Lingham-Soliar 1991, 1998; Bardet et al. 2015, 2018). The Moroccan phosphates preserve remarkably high taxonomic and morphological diversity, containing at least seven genera comprising species such as the halisaurine Halisaurus arambourgi Bardet & Pereda Suberbiola, 2005, and the mosasaurines Eremiasaurus heterodontus LeBlanc, Caldwell & Bardet, 2012 and Globidens simplex LeBlanc, Mohr & Caldwell, 2019 (Bardet et al. 2005, 2015, 2018; LeBlanc et al. 2012, 2019). The Dukamaje Formation preserves a similar level of diversity, with seven genera spanning the Plioplatecarpinae, Halisaurinae and Mosasaurinae; most notable among these taxa is the enigmatic Goronyosaurus nigeriensis (Swinton, 1930), a poorly known species of uncertain phylogenetic placement (Azzaroli et al. 1975; Soliar 1988). These deposits together exemplify the remarkable capacity for evolutionary radiation present in mosasaurs.

Our study builds on current knowledge of this diversity via the description and analysis of a new mosasaur specimen (MHNM.KHG.1231) from Morocco. The teeth of this specimen resemble those of ‘Platecarpusptychodon Arambourg, 1952, suggesting referral to this species. However, the holotype and diagnosis of ‘P.ptychodon exhibit critical flaws forcing a re-evaluation – and ultimately a rejection – of its taxonomic validity. Morphological and phylogenetic analyses of this specimen instead support referral to a new genus and species, closely related to Selmasaurus Wright & Shannon, 1988, a genus of plioplatecarpine mosasaur currently known only from the Santonian (86.3–83.6 Ma) and early Campanian (83.6–80.6 Ma) of Kansas and Alabama (Wright & Shannon 1988; Polcyn & Everhart 2008; Ogg et al. 2012). Specifically, the distribution of Selmasaurus russelli Wright & Shannon, 1988 within the Eutaw Formation (upper Santonian) in Alabama and within the unnamed lower member of the Mooreville Chalk Formation (lower Campanian) in Kansas results in an overall temporal range of 85–81.5 Ma (Wright & Shannon 1988; Kiernan 2002; Liu 2007; Konishi 2008; Ogg et al. 2012). Selmasaurus johnsoni Polcyn & Everhart, 2008 occurs in Kansas within the Smoky Hill Chalk Member of the Niobrara Formation, in biostratigraphic units corresponding to the lower Santonian (86.3–85 Ma) (Polcyn & Everhart 2008; Ogg et al. 2012).

Here, we present a detailed osteological description of MHNM.KHG.1231, followed by phylogenetic analyses of mosasauroid relationships incorporating this new specimen. We then assess the evolutionary implications of these phylogenetic results for the new specimen and associated taxa, including discussions of functional morphology and palaeoecology. Ultimately, this specimen provides insight into the striking diversity – both morphological and ecological – of the Late Cretaceous palaeoecosystems in which these mosasaurs lived, as well as the phylogenetic framework underlying this diversity.

Geological context

MHNM.KHG.1231 originates from the Maastrichtian phosphate deposits of Morocco. These deposits are a major component of the Mediterranean Tethyan Phosphogenic Province, a belt of phosphates that extends around the Mediterranean Sea, from North Africa to the Middle East (Lucas & Prévôt-Lucas 1996). These deposits span the Upper Cretaceous (Maastrichtian) to the middle Eocene (Lutetian) and outcrop in five major basins: the Oulad Abdoun, Ganntour, Meskala, Sous and Oued Eddahab (Bardet et al. 2010; LeBlanc et al. 2012; Fig. 1A).

Figure 1. Map and stratigraphic column of relevant geological features of Morocco. A, map of north-western Morocco, showing the Atlantic Ocean (light grey), Atlas Mountains (grey lines), Oulad Abdoun Basin (black), and other major phosphatic basins (dark grey). Each section of the scale bar represents 30 km. B, stratigraphy of the Oulad Abdoun Basin. Beds associated with the provenance of MHNM.KHG.1231 are italicized. Abbreviations: BB, basal limestone bonebed; LCIII, Lower Couche III; Li, limestone; Ma, marls; Ph, phosphates; UCIII, Upper Couche III. Modified from Arambourg (1952) and from Bardet et al. (2005, by permission of Oxford University Press).

Historically, establishing stratigraphical correlations within and among these phosphatic basins has been difficult for many reasons, most significantly the drastic local lateral facies changes and the low abundance of microvertebrate and invertebrate fossils typically used for biostratigraphical correlations (Bardet et al. 2010; LeBlanc et al. 2012). Arambourg (1952) was the first to perform a methodical biostratigraphical analysis for these deposits, focusing his efforts on the Oulad Abdoun and Ganntour basins (see also Bardet et al. 2010). Using unique assemblages of vertebrate remains (mainly fishes and reptiles), Arambourg (1952) established three main stratigraphic levels of phosphates in each basin, referring to them as ‘couches’, or ‘beds’ in English (see also Bardet et al. 2010). These beds span the Maastrichtian (Couche III) to the Ypresian (Couche I) (LeBlanc et al. 2012). The Maastrichtian component (Couche III) of the Oulad Abdoun basin contains three major subdivisions: a basal limestone bonebed, an intermediate grey phosphatic layer (Lower Couche III), and an upper layer of fossiliferous, yellow phosphates of upper Maastrichtian age (Upper Couche III) (LeBlanc et al. 2012; Fig. 1B). These layers are interspersed with marly and calcareous beds (Bardet et al. 2015).

MHNM.KHG.1231 was discovered in a Moroccan phosphate mine, corresponding to the aforementioned Couche III. This provenance is further supported by the sediments and microfossils associated with this specimen. Teeth from the selachian taxa Serratolamna serrata (Agassiz, 1843) and Squalicorax pristodontus (Agassiz, 1843) were present in the matrix surrounding MHNM.KHG.1231 (C.S., pers. obs.), consistent with Couche III (Maastrichtian) of the Oulad Abdoun basin (Arambourg 1952; Fig. 1). Furthermore, the soft, yellow, phosphatic matrix itself unequivocally matches the description of Upper Couche III (LeBlanc et al. 2012).

Material and methods

Specimen and figures

The holotype specimen is accessioned in the collections of the Museum of Natural History of Marrakech at Cadi Ayyad University in Morocco as MHNM.KHG.1231 and has been on loan to the University of Alberta Laboratory for Vertebrate Palaeontology (UALVP), where it was studied. This specimen consists of: an articulated skull, complete except for the chondrocranium (Figs 2–5); partial left and right mandibles (Figs 6, 7); a partial right quadrate (Fig. 8); and other disarticulated cranial (pterygoids) and postcranial (humerus and vertebrae) material (Fig. 9). Specimens were photographed using a Canon PowerShot 340HS digital camera. Images were traced and figured in Adobe Photoshop CC 2019 and 2020 and Adobe Illustrator 2020.

Figure 2. MHNM.KHG.1231 (articulated skull) in dorsal view. A, diagram; B, photograph. Light grey shading indicates the external nares and other openings in the skull. Dark grey shading indicates openings in the skull resulting from taphonomic breakage. Double-dashed lines indicate major breaks in or taphonomic distortion to the corresponding elements. Bracketed numbers indicate tooth positions. Abbreviations: f, frontal; h, humerus; mx, maxilla; oen, opening for the external naris; p, parietal; pf, pineal foramen; pmx, premaxilla; pof, postorbitofrontal; prf, prefrontal; sq, squamosal; stf, supratemporal fenestra; v, vomer. Paired bones are preceded by r (right) or l (left). Scale bar represents 5 cm.

Figure 3. MHNM.KHG.1231 (articulated skull) in ventral view. A, diagram; B, photograph. Black shading represents matrix. Light grey shading indicates the internal nares and other openings in the skull. Dark grey shading indicates openings in the skull resulting from taphonomic breakage. Solid arrows indicate interdental pits; fletched arrowheads indicate replacement teeth. Double-dashed lines indicate major breaks in the corresponding elements. Bracketed numbers indicate tooth positions. Abbreviations: f, frontal; fpf, frontal posteromedian flange; h, humerus; mx, maxilla; p, parietal; pal, palatine; pf, pineal foramen; pmx, premaxilla; pof, postorbitofrontal; prf, prefrontal; smx, septomaxilla; sq, squamosal; stf, supratemporal fenestra; v, vomer; vp, vomerine process of the premaxilla. Paired bones are preceded by r (right) or l (left). Scale bar represents 5 cm.

Figure 4. Select palatal views of MHNM.KHG.1231. A, anterior palate; B, mid-palatal view; C, posterior palate and ventral skull roof. Major anatomical features are labelled on the diagram or photographs, with sutures indicated on the diagram. Insets on line drawing correspond to panels denoted by the same letter. Arrows indicate interdental pits. Shading of line drawing as in Figures 2 and 3. Abbreviations: ch/os, location of cerebral hemisphere or of articulation of the orbitosphenoid; ls, lingual shelf of maxilla; mx, maxilla; oc, olfactory canal; p, parietal; pal, palatine; pmx, premaxilla; pmx-v, premaxilla–vomer articulation; pof, postorbitofrontal; prf, prefrontal; sq, squamosal; tb, tabular boss; v, vomer; vp, vomerine process of the premaxilla. Paired bones are preceded by r (right) or l (left). Scale bars each represent 2 cm.

Figure 5. Orbital region and skull roof of MHNM.KHG.1231. A, ventral view of skull roof; B, dorsal view of skull roof; C, lateral view of left orbit. Major anatomical features are labelled on the diagrams or photographs, with sutures indicated on the diagrams. Line drawings correspond to panels denoted by the same letter. Shading of line drawings as in Figures 2 and 3. In C, dashed line indicates approximate dorsal midline and double-dashed line indicates estimated location of suture given taxonomic distortion to specimen. Abbreviations: ch/os, location of cerebral hemisphere or of articulation of the orbitosphenoid; f, frontal; fa, frontal ala; fpf, frontal posteromedian flange; jp, jugal process of postorbitofrontal; mx, maxilla; oc, olfactory canal; p, parietal; pal, palatine; pof, postorbitofrontal; prf, prefrontal; sq, squamosal; tb, tabular boss. Paired bones are preceded by r (right) or l (left) where relevant. Scale bars each represent 5 cm. Line drawings not to scale.

Figure 6. Select views of the dentition of MHNM.KHG.1231, highlighting key aspects of dental anatomy and tooth development. A, anterior premaxillary teeth; B, right maxillary tooth 3; C, left maxillary tooth 11; D, left dentary tooth 3; E, left dentary tooth 5; F, left dentary tooth 6 and tooth position 7. Insets on overview photographs correspond to panels denoted by the same letter. Solid arrows point to interdental pits; fletched arrows point to replacement teeth. Replacement teeth in early (C), intermediate (B), and late (F) stages of development are preserved in this specimen. Several stages of post-eruption tooth development are also visible, from early stages in which the marginal teeth are unfused to their respective alveolus (A and F), to intermediate stages of partial fusion to the tooth-bearing element (B and D), to the final stage of being fully fused to the alveolus (E). Abbreviation: mp, medial parapet of the dentary. Scale bars each represent 1 cm. Overview photographs not to scale.

Figure 7. Mandibular elements of MHNM.KHG.1231. A, left dentary in lateral view; B, right dentary in medial view; C, D, right posterior mandibular unit in lateral view; E, F, right posterior mandibular unit in medial view. Solid arrowheads indicate interdental pits; fletched arrowheads indicate replacement teeth. Bracketed numbers indicate tooth positions. Abbreviations: a, angular; art, articular; gf, glenoid fossa; mg, Meckelian groove; mp, medial parapet of the dentary; part, prearticular; sa, surangular. Each scale bar represents 5 cm. Upper scale bar applies to panels A and B; lower scale bar applies to panels D and F. Panels C and E show the location of sutures and are not to scale.

Figure 8. Partial right quadrate of MHNM.KHG.1231 in A, lateral, B, medial, C, dorsal and D, ventral views. Abbreviations: mex, medial excavation; qs, quadrate shaft (top portion); sn, stapedial notch; sp, stapedial pit; ssp, suprastapedial process. Scale bar represents 2 cm.

Figure 9. Disarticulated cranial and postcranial material of MHNM.KHG.1231. A, left pterygoid in ventral view; B, left pterygoid in dorsal view; C, right pterygoid (ventral view) and two vertebrae, the left one cervical and the right one caudal; D, opposite view of C (i.e. right pterygoid in dorsal view); E, complete cervical vertebra in left lateral view and crushed vertebral centrum; F, opposite view of E (i.e. the more complete vertebra is in right lateral view); G, right humerus in extensor view (proximal is to the top). Abbreviations: bp, basisphenoid process; ecpp, ectopterygoid process of pterygoid; ect, ectepicondyle; ent, entepicondyle; fl, flange; hae, haemapophysis; hyp, hypapophysis; ns, neural spine; pozyg, postzygapophysis; pt, pterygoid; pzyg, prezygapophysis; qr, quadrate ramus; syn, synapophysis; vc, vertebral centrum; zg, zygantrum. Scale bar represents 5 cm; all bones to scale.

Dataset selection

The two datasets used for this analysis (see , S1–S4) were modified from the two character matrices of Simões et al. (2017a). Both of these source matrices include 44 mosasauroid taxa and are based on the same fundamental characters but implement different coding schemes. The source matrix applying multistate character coding contained 125 characters before modification, whereas the source matrix applying contingent character coding contained 131 characters (Simões et al. 2017a). These matrices were chosen because they comprise the most recent large-scale mosasauroid phylogeny and contain numerous species from every major mosasaur group. This thorough taxon sampling is ideal, as it tests the phylogenetic placement of MHNM.KHG.1231 relative to all mosasaurs (rather than just a select subfamily), thus eliminating ingroup choice as a potential source of bias.

Because these source datasets use both multistate and contingent coding schemes in both unweighted and implied weighting maximum parsimony analyses, they therefore allow the phylogenetic position of MHNM.KHG.1231 to be tested and compared under various conditions and methods. Contingent coding schemes are useful for comparison to other mosasauroid phylogenies because most previous phylogenies have been generated using this scheme (e.g. Palci et al. 2013; Jiménez-Huidobro et al. 2016; Simões et al. 2017a). However, multistate coding schemes are more logically sound, as they evaluate dependent characters in a single transformational series, rather than parsing them into separate characters (Simões et al. 2017a, b). Similarly, unweighted maximum parsimony analysis is the approach most commonly used in mosasauroid phylogenetics (Simões et al. 2017a), though implied weighting has been proposed as a more accurate method of phylogenetic reconstruction because it down-weights the impact of homoplasy in tree construction (Goloboff 1993; Goloboff et al. 2008a; Simões et al. 2017a). Given this diversity of phylogenetic methods, the chosen source datasets therefore enable a range of analyses and comparisons concerning the phylogenetic position of MHNM.KHG.1231 and related taxa. Performing and reporting these analyses also accounts for algorithm choice as a potential source of methodological bias or error.

Furthermore, the inclusion of multiple dolichosaur and aigialosaur species in these source matrices – with Adriosaurus suessi Seeley, 1881 being designated as the official outgroup – ensures that the outgroup is closely related to the ingroup (Mosasauridae), as opposed to the traditional usage of Varanidae as an outgroup. These datasets are therefore preferred over other mosasauroid matrices, in which the outgroup is typically an artificial operational taxonomic unit (OTU) generated using a compilation of varanid lizard character states (Simões et al. 2017a). Using a theoretical unit as an outgroup can cause bias in character polarization, as character polarity for the ingroup is often designated based on previous assumptions of character evolution rather than on the actual character states observed in a tangible outgroup taxon (Simões et al. 2017a). The use of varanids or a varanid-based artificial OTU is also problematic because varanids are more distantly related to mosasaurs than dolichosaurs or aigialosaurs are, thus increasing the possibility that independent character evolution along the outgroup and ingroup branches has changed the outgroup state from the true ancestral state (Simões et al. 2017a). Given these issues, using an actual (i.e. non-theoretical), closely related taxon such as A. suessi as the outgroup is preferred, as this approach addresses the aforementioned inaccuracies that can result from outgroup choice.

Dataset modifications

Three taxa were added to the source matrices: MHNM.KHG.1231 as a new species (see below), Goronyosaurus nigeriensis and Selmasaurus russelli. The latter two taxa were included because they each exhibit certain morphological similarities to MHNM.KHG.1231 (e.g. snout elongation and deep interdental pits for G. nigeriensis; similar medial excavation of the suprastapedial process of the quadrate for S. russelli), suggesting the potential for a close evolutionary relationship worth being tested phylogenetically. Both of these taxa were scored directly. Published descriptions were used to complement these direct observations but were not used as the sole basis for scoring any particular character; if a feature was described in the manuscript but could not be directly observed by us, it was scored as ‘?’ rather than relying on written descriptions. Although this approach may have reduced the number of characters that could be scored, it was employed in order to ensure soundness of methodology.

Certain characters within the source matrices were also modified. Character 51/55 (basioccipital tubera shape) in the multistate and contingent source matrices, respectively, was deleted from this analysis on the basis of poor character construction. This character’s states (tubera not anteroposteriorly elongate [0], or anteroposteriorly elongate with rugose ventrolateral surfaces [1]) conflate different features – elongation and rugosity – thus exemplifying problematic character type I A.6 as outlined by Simões et al. (2017b). Character 74/79 (tooth replacement mode) in the multistate and contingent source matrices, respectively, was also deleted on the basis of poor character construction, as its character states (replacement teeth form in shallow excavations [0], or in subdental crypts [1]) simply refer to different stages of tooth replacement (e.g. Caldwell 2007) and therefore carry no phylogenetic signal. A new character (character 124/130: quadrate mid-shaft lateral deflection; absent [0], or present [1]) was added to both the multistate and contingent datasets. This is a modified version of character 95 in Konishi & Caldwell (2011) and was included because the quadrate shaft deflection to which it refers is a synapomorphy of Selmasaurus.

The state for character 12/13 (frontal olfactory canal embrasure) in the modified multistate and contingent datasets was changed from ‘1’ (canal almost or completely enclosed ventrally by descending processes) to ‘0’ (canal not embraced ventrally by descending processes) for Selmasaurus johnsoni in light of re-interpretation of this character and its condition in S. johnsoni and S. russelli. The character states for multistate character 26 (prefrontal–postorbitofrontal contact) were originally mislabelled as ‘0’, ‘0’ and ‘1’ in the source character list; this typographic error was corrected to ‘0’, ‘1’ and ‘2’ in the modified dataset. For character 54/59 (dentary medial parapet), a new character state (state 3: medial parapet taller than lateral wall of bone) was added to the original states (parapet positioned at base of tooth roots [0], or parapet elevated and strap-like, enclosing about half of height of tooth attachment in shallow channel [1], or parapet equal in height to lateral wall of bone [2]). This new state accounts for additional variation present in MHNM.KHG.1231 and some individuals of Plioplatecarpus (cf. Konishi & Caldwell 2011, character 54[2]).

Phylogenetic analysis

Unweighted maximum parsimony

Unweighted maximum parsimony analysis was performed for both the multistate and contingent character matrices using TNT v. 1.5 (Goloboff et al. 2008b; Goloboff & Catalano 2016) via a heuristic ‘traditional search’ using the tree bisection and reconnection (TBR) algorithm with 1000 random-addition-sequence replicates. This method is consistent with that used by Simões et al. (2017a).

Implied weighting maximum parsimony

In these analyses, trees were generated from both the multistate and contingent matrices using an implied weighting function of k = 3.0, as per the methods of Simões et al. (2017a). Values of k = 7.0 and k = 11.0 were also used to test the effect of stringency of the concavity function on the resultant topology. All other settings and methods were retained from the unweighted maximum parsimony analysis.

Character mapping

Synapomorphies were mapped onto all most-parsimonious trees (MPTs) in TNT v. 1.5 (Goloboff et al. 2008b; Goloboff & Catalano 2016). Character history was traced in Mesquite v. 3.51, with ancestral states reconstructed using parsimony (Maddison & Maddison 2018). Trees were visualized in Mesquite v. 3.51 (Maddison & Maddison 2018) and prepared for figures using Adobe Illustrator 2020.

Institutional abbreviations

CMN, Canadian Museum of Nature, Ottawa, Canada; FHSM, Sternberg Museum of Natural History, Fort Hays State University, Hays, Kansas; IGF, Institute of Geology and Paleontology of the University of Florence, Florence, Italy; IRSNB, Institut Royal des Sciences Naturelles de Belgique, Brussels, Belgium; MHNM, Museum of Natural History of Marrakech at Cadi Ayyad University, Marrakech, Morocco; MNHN, Muséum National d’Histoire Naturelle, Paris, France; NHMUK, Natural History Museum (formerly British Museum [Natural History]), London, UK; UALVP, University of Alberta Laboratory for Vertebrate Palaeontology, Edmonton, Canada; YPM, Yale Peabody Museum, New Haven, USA.

Systematic palaeontology

Reptilia Linnaeus, 1758

Squamata Oppel, 1811

Mosasauridae Gervais, 1852

Plioplatecarpinae (Dollo, 1884) Williston, 1897a

Gavialimimus gen. nov.

Type species

Gavialimimus almaghribensis, sp. nov.

Diagnosis

As for the type and only species.

Derivation of name

Meaning ‘gharial mimic’, from the Gallicized Hindi root ‘gavial’ and the Greek root ‘mimus’, the genus name refers to morphological convergence between the holotype specimen and the extant gharial (Gavialis gangeticus), primarily regarding their distinctive longirostry and interlocking teeth.

Occurrence

As for the type and only species.

Gavialimimus almaghribensis sp. nov.

(Figs 2–9)

Holotype

MHNM.KHG.1231, an articulated skull (including the premaxilla, maxillae, prefrontals, frontal, parietal, postorbitofrontals, squamosals, vomers and palatines), isolated left and right dentaries, articulated right posterior mandibular unit, isolated pterygoids, a partial right quadrate and four isolated vertebrae (one caudal, two cervical and one indeterminate) and the right humerus.

Diagnosis

Plioplatecarpine mosasaur species bearing the following autapomorphies: snout highly elongate, with openings for the external nares small and highly retracted; supratemporal fenestrae large, nearly as long as the entire frontal including the anterior frontal processes; dentary medial parapet taller than lateral wall of dentary; Meckelian groove of the dentary terminating anteriorly at the sixth tooth position; parietal markedly constricted, resulting in a triangular parietal table with a distinct midsagittal crest posteriorly; prefrontal extending posteriorly nearly to level of frontoparietal suture. Also distinguished by a unique combination of the following features: marginal teeth not medially striated; interdental pits deep, present along most of the length of the marginal tooth rows; pineal foramen elongate and in slight contact with the frontoparietal suture; postorbitofrontal wide in dorsal view, equal to roughly half the width of the frontal, and lacking a transverse dorsal ridge; zygosphenes and zygantra absent; humerus ectepicondyle present. Differs from Selmasaurus johnsoni in the following respects: premaxillary predental rostrum absent; postorbitofrontal in contact with prefrontal; maxillary–premaxillary suture terminating posteriorly between the eighth and ninth maxillary teeth. Differs from Selmasaurus russelli in the following respects: posterior parietal shelf absent. Differs from Goronyosaurus nigeriensis in the following respects: ectopterygoid processes of pterygoid not forked; maxilla terminating just posterior to anterior border of orbit, as opposed to extending beyond posterior margin of orbit; premaxillary teeth not caniniform.

Derivation of name

The specific epithet is a romanized version of the Arabic name for Morocco, paired with the Latin suffix ‘-ensis’, thus denoting the country of origin of the holotype.

Occurrence

Upper Couche III (upper Maastrichtian) of the Oulad Abdoun Basin in northern Morocco.

Description and comparisons

General comments

MHNM.KHG.1231 consists of an articulated skull (mostly complete except for the chondrocranium: Figs 2–5), left and right dentaries (Figs 6, 7A, B), right posterior mandibular unit (Fig. 7C–F), a partial right quadrate (Fig. 8), isolated right and left pterygoids (Fig. 9A–D), four isolated vertebrae (Fig. 9C–F) and the right humerus (Fig. 9G). Most of these elements were recovered in three major blocks: one containing the skull, one containing the left dentary and partial quadrate, and one containing the right dentary. The other elements were small enough to be recovered individually from the matrix surrounding the three main blocks. Several isolated teeth (UALVP 57049) matching the morphology of those present in the jaw bones were also recovered from the matrix surrounding the other elements.

The preorbital region of the skull comprises almost two-thirds of the total length of the skull (Figs 2, 3). This striking snout elongation is uncommon among mosasaurs, with Ectenosaurus clidastoides (Merriam, 1894) and, to a lesser extent, Plotosaurus bennisoni (Camp, 1942) and Goronyosaurus nigeriensis being the only mosasaurs exhibiting a comparable condition.

Although the skull is articulated, taphonomic deformation has displaced some of the bones from their natural positions. The entire skull is slanted to its right. Most significantly, the palate and skull roof are collapsed onto each other, with the palate being offset from the skull roof such that the right internal naris is aligned with the left external naris. The right maxilla is also collapsed underneath the skull, such that it underlies the right external naris.

Skull

Premaxilla

In MHNM.KHG.1231, the premaxilla is preserved largely in natural position, though the internarial bar is broken slightly posterior to its midpoint (Fig. 2). The premaxilla is smoothly rounded anteriorly, terminating just beyond the anterior-most premaxillary tooth and thus lacking a rostrum (Figs 2, 3, 4A, 6A). The dorsal surface of the premaxilla is smooth, possessing neither a sulcate texture nor a dorsal crest (Fig. 2B). Foramina marking the exits of the ophthalmic ramus (V1 branch) of the trigeminal nerve are loosely dispersed on either side of the dorsal midline (Russell 1967; Fig. 2). Although most mosasaurs exhibit six or seven such foramina on each side of the premaxilla, MHNM.KHG.1231 possesses approximately 10; this comparatively high number is similar to the state present in Goronyosaurus nigeriensis (Lingham-Soliar 1991).

The premaxilla articulates with the maxilla via a long suture that is posteriorly displaced due to deformation of the skull. Based on our interpretation of the skull, this suture likely would have extended to between the eighth and ninth maxillary tooth positions, at which point the bones separate to surround the anterior extent of the openings for the external nares (Fig. 2). The internarial bar forms the anteromedial border of the external nares, extending to the posterior extent of the external narial openings, where it articulates on either side with the paired anterior processes of the frontal (Fig. 2). The internarial bar is mostly uniform in width, being slightly narrower than half the width of the rostrum and tapering slightly toward its posterior terminus.

The external nares themselves are unique in being highly retracted, extending posteriorly almost to the level of the orbits and being restricted to the posterior 25% of the preorbital region (Fig. 2); similar external narial retraction has been noted in Goronyosaurus (Azzaroli et al. 1972; Soliar 1988).

The ventral surface of the premaxilla – including the junction of the internarial bar and dentigerous portion of the premaxilla – is largely obscured by the vomers (Figs 3, 4A, B). However, the vomerine process of the premaxilla is visible as a median ridge between the second pair of premaxillary teeth, separated from the roots of the teeth on either side by a shallow groove (Figs 3, 4A). Posteriorly, this ridge articulates with the vomers, though the exact suture is difficult to discern; anteriorly, it terminates just rostral to the second pair of premaxillary teeth. A slight groove separates the anterior pair of premaxillary teeth (Fig. 4A).

Maxilla

The maxillae terminate posteriorly at the anterior border of the orbits (Figs 2, 3, 4C, 5). They are fairly uniform in height along their extent, though taper anteriorly following the curve of the maxillary–premaxillary suture (Fig. 2). The left maxilla is broken – either pathologically or taphonomically – at the level of the eighth maxillary tooth (Fig. 2). Several large foramina are present along the ventral border of the lateral surface of the maxilla, above roughly every other tooth position. Several smaller foramina are scattered over the rest of the lateral surface, being mostly concentrated toward the anterior of the maxilla (Fig. 2). These foramina mark the exits of the maxillary (V2) branch of the trigeminal nerve (Russell 1967).

The maxilla articulates with the prefrontal via a straight suture that extends posterolaterally from the posterolateral edge of the external naris to the anterodorsal edge of the orbit (Figs 2, 5C). Thus, the maxilla forms most of the lateral border of the external naris, with the prefrontal surrounding the posterior and posterolateral edges of the external nares (Fig. 2).

The posterior terminus of the maxilla extends beyond the posterior-most maxillary tooth for a distance roughly equivalent to one alveolus length (Figs 3, 4C). The posterior extent of the lateral border of the left maxilla is marked by several large indentations or pits (Figs 2, 5C). However, as these pits are absent on the right maxilla, they are likely the result of taphonomic deformation.

The maxillae each bear 12 tooth positions (Fig. 2); the morphology of these teeth is described below (in Description, Dentition). The maxillary tooth row is slightly inset from the smoothly rounded ventrolateral border of the maxilla (Figs 3, 4B, C). The lingual shelf of the maxilla is best visible posteriorly on the left maxilla (Fig. 4C) and anteriorly on the right maxilla (Fig. 4B). It extends along the medial border of the maxillary tooth row and is quite pronounced, especially posteriorly, causing the tooth row to effectively occupy a one-sided trough (Figs 3, 4B, C). Although the lingual shelf is about 1.8 cm wide at its widest point posteriorly and remains quite pronounced along most of its length, it does become thinner anteriorly, with its thinnest width of 0.1 cm occurring medial to the third maxillary tooth. Anterior to this point, the lingual shelf is no longer visible, though it is unclear whether this absence is biological or simply the result of being obscured by matrix.

Dentition

Two pairs of teeth are present on the premaxilla (Figs 2, 3, 4A). The anterior pair is moderately recurved, with the teeth showing slight wear on their tips (Figs 4A, 6A). The right anterior tooth is slightly more prognathous than its left counterpart (Figs 3, 4A, 6A), likely the result of taphonomic deformation to the associated alveolus (see below; Fig. 6A). This displacement may also have been facilitated by the lack of fusion between this tooth and its alveolus (see below; Fig. 6A). The second premaxillary tooth is absent on the left side of the premaxilla and taphonomically distorted on the right side; however, based on the dimensions of the alveolus and the remnants of the tooth that are present, this tooth was likely of similar size to the anterior-most premaxillary tooth (Figs 3, 4A). As such, this specimen lacks the prominent caniniform premaxillary dentition characteristic of Goronyosaurus (Lingham-Soliar 2002).

There are 12 tooth positions on the maxilla (Figs 2, 3) and at least 10 tooth positions on the dentary (Fig. 7A, B). All of the marginal teeth are strongly fluted, bicarinate and lack serrations (Fig. 6). The teeth are widely spaced, indicating strong interdigitation of the tooth rows when occluded. As in Goronyosaurus, deep interdental pits are present along the premaxillary, maxillary and dentary tooth rows to accommodate this interdigitation (Lingham-Soliar 1991; Figs 3, 4, 6). These interdental pits can be distinguished from the replacement tooth pits because they are located lateral to the tooth row, whereas the replacement teeth form posteromedial to the tooth roots (Figs 3, 6). The interdental pits are deepest anteriorly on the jaws and get progressively shallower until finally disappearing posterior to the seventh dentary (Figs 6F, 7A) and at least the fifth maxillary tooth positions (Figs 3, 4B). The interdental pits may extend beyond the fifth tooth position on the maxilla, but this location is obscured by the displaced humerus. This shallowing is related to progressive changes in the dimensions of the teeth: the premaxillary and anterior maxillary teeth are long and distinctly recurved whereas the more posterior maxillary teeth are progressively shorter, stouter and less recurved (Figs 3, 4, 6). This moderate heterodonty allows the posterior teeth to interlock without interdental pits.

The tooth bases are oval in cross-section and are posteriorly elongated toward the posterior extent of the maxillary tooth row (Figs 3, 4, 6). The teeth also exhibit progressive changes in the degree of wear. The premaxillary and anterior maxillary and dentary teeth are barely worn (Figs 4A, 6), whereas the posterior maxillary teeth exhibit much more pronounced wear (Figs 4B, C, 6C). In the 10th and 11th left maxillary teeth, this wear extends from the tip of the tooth crown along the tooth’s mesial carina.

The marginal tooth roots are broadly exposed along their respective tooth rows, constituting between one-third and one-half of the height of the overall tooth (Figs 2B, 3B, 4, 6). Taphonomic damage to the alveolus of the right anterior premaxillary tooth has caused the root of this tooth to be further exposed beyond the level of the maxillary tooth row, revealing the root to be elongate and curved underneath the interdental pit between the right premaxillary teeth (Figs 3B 4A, 6A). Similar breakage at the fourth tooth position on the left dentary reveals similar curvature of the tooth root underneath the interdental pit posterior to this tooth position, with the exposed root being at least twice the length of the tooth crown (Fig. 7A). The presence of this root curvature at both the first premaxillary and fourth dentary tooth positions therefore indicates that this curvature is present and pronounced at least anteriorly on the marginal tooth rows, though whether it continues into the posterior extent of the tooth rows remains uncertain. Similar curvature is also present in the tooth roots of Prognathodon solvayi Dollo, 1889a (C.S., pers. obs.).

Three replacement teeth are visible on the marginal jaw elements: one in the early stages of development occurs in association with the 11th tooth on the left maxilla (Figs 3A, 6C); another replacement tooth at a slightly more advanced stage, indicated by its slightly larger size, occurs in association with the third tooth on the right maxilla (Figs 3A, 6B); and a final replacement tooth at a much more advanced developmental stage – indicated by its much larger size – occurs in association with the seventh tooth position on the left dentary (Figs 6F, 7A). The two smaller replacement teeth occur in shallow pits directly posteromedial to their respective tooth positions (Fig. 6B, C), whereas the largest replacement tooth is present in a crypt underneath the functional tooth (Fig. 6F). As in the functional teeth, none of the replacement teeth are serrated. However, these replacement teeth differ from the functional teeth in having smooth faces, thus lacking the plicae present on the fully developed, functional teeth.

Different stages of post-eruption tooth development are also visible along the left dentary tooth row, reflecting the progressive fusion that occurs throughout this late-stage tooth development. The roots of the second, fourth and sixth teeth are separate from their surrounding alveoli, with each alveolus forming a rim around the associated tooth root (Figs 6F, 7A). The third tooth root is partially fused to the bone, such that the aforementioned rim is less pronounced; instead, the junction between the tooth root and the alveolus is marked by a slight groove encircling the root (Figs 6D, 7A). The fifth tooth is completely fused to its alveolus, with the root being smoothly confluent with the surrounding dentary bone (Figs 6E, 7A). Similar developmental stages are evident along the upper marginal tooth row: the right first premaxillary and first maxillary teeth are unfused to their respective tooth-bearing elements (Fig. 6A); the right second premaxillary and second, third, 10th and 11th maxillary teeth are partially fused (Fig. 6B); and the right fifth maxillary tooth is completely fused to the maxilla. Various empty alveoli are present along all marginal tooth-bearing elements (Figs 3, 4, 6, 7).

Prefrontal

The prefrontal articulates laterally with the maxilla, medially with the frontal and posteriorly with the postorbitofrontal, and is slightly curved to form the anterodorsal border of the orbit (Figs 2, 5). The suture with the maxilla is described above (Description, Maxilla).

The anterior border of the prefrontal is smoothly and tightly curved, with two processes – one medial and one lateral – that surround the posterior and posterolateral border of the external naris (Fig. 2). The lateral process is much longer than the medial one, such that the prefrontal fully surrounds the posterolateral margin of the external naris but only barely rounds the posteromedial portion before articulating with the frontal. This condition is consistent with the lack of anterolateral processes of the frontal in MHNM.KHG.1231 and in Selmasaurus, as discussed below (see Description, Frontal); i.e. rather than the frontal bearing anterolateral processes that surround the posterolateral borders of the external nares, it is the prefrontal which embays the external narial openings (Fig. 2).

The prefrontal articulates with the frontal via a straight suture that extends from the posteromedial corner of the external naris to the posterolateral corner of the frontal, another unusual feature among mosasaurs, where in dorsal aspect the prefrontal typically terminates posteriorly at the mid-length of the frontal supraorbital border (Figs 2, 5; see Russell [1967, fig. 4] and Bell [1997, fig. 6] for comparisons). The posterior terminus of the prefrontal is deeply notched, with diverging medial and lateral processes that articulate with the anteriorly projecting main body of the postorbitofrontal (Figs 2, 3, 4C, 5). The lateral process extends to the midpoint of the orbit, thus forming the anterolateral border of the orbit, with the medial process being slightly longer (Figs 2, 3, 5C). This articulation of the prefrontal with the postorbitofrontal excludes the frontal from the orbital margin. Similar to the condition in Plotosaurus (see LeBlanc et al. 2013, figs 2, 6), the prefrontal and postorbitofrontal exhibit extensive contact along the ventrolateral, lateral and dorsolateral borders of the orbit (Fig. 5). This pronounced articulation, previously considered unique to Plotosaurus among mosasaurs, contributes to reduced kinesis between the muzzle and posterior skull (LeBlanc et al. 2013).

Lacrimal

In many non-ophidian squamates, a lacrimal is present between the prefrontal and maxilla (see Konishi et al. [2016] for proposed location in mosasaurs). However, because this area of the skull is taphonomically distorted in MHNM.KHG.1231 (Figs 2, 5C), the presence of the lacrimal cannot be assessed in this specimen.

Frontal

The frontal is a triangular bone articulating anteriorly with the internarial bar of the premaxilla, laterally with the prefrontal, posterolaterally with the postorbitofrontal, and posteriorly with the parietal (Figs 2, 5). In MHNM.KHG.1231, the frontal possesses a single pair of long anterior projections, about half the length of the main body of the frontal, which surround the posterior extent of the internarial bar and thus border the external nares posteromedially (Fig. 2). These processes bear short but sharp ridges concentrated on the medial half of their respective dorsal surfaces (Fig. 2). Though the posterior-most portion of the premaxillary internarial bar is broken, the tight ‘V’-shaped notch formed by the frontal’s anterior processes indicates that the internarial bar would have invaded the frontal to a point roughly in line with the posterior extent of the openings for the external nares (Fig. 2B). Uniquely shared with Ectenosaurus and Selmasaurus among russellosaurines, MHNM.KHG.1231 lacks a pair of frontal anterolateral processes, thus precluding the frontal from forming the posterolateral corner of the naris (compare Fig. 2 with Russell [1967, figs 83, 86], Wright & Shannon [1988, fig. 1], Holmes [1996, fig. 2A], Konishi & Caldwell [2007, fig. 5A] and Polcyn & Everhart [2008, fig. 3]).

Though the dorsal surface of the frontal is quite deeply fractured, the sutures between the frontal and surrounding bones are still clear and the ventral surface is much more intact. The frontal is tightly joined to the parietal via a complex and interdigitating suture (Figs 2, 3, 4C, 5), similar to the condition described in Goronyosaurus nigeriensis (Lingham-Soliar 1999) and Selmasaurus russelli (Wright & Shannon 1988) and also present, albeit to a lesser extent, in Ectenosaurus (FHSM VP-401; C.S., pers. obs.). The junction between the frontal and prefrontal is similarly tightly articulated, though lacks clear interdigitation (Figs 2, 3, 4C, 5). The posterolateral corners of the frontal each bear an ala overlying the postorbitofrontal between its anterior and medial processes and forming the main junction between the frontal and postorbitofrontal (Figs 2, 5B, C). These alae are particularly long and narrow, resembling the condition in Goronyosaurus and Phosphorosaurus ponpetelegans Konishi, Caldwell, Nishimura, Sakurai & Tanoue, 2016, but differing from most other mosasaurs, including closely related taxa such as Selmasaurus (C.S., pers. obs.; see also Azzaroli et al. [1972], Soliar [1988], Wright & Shannon [1988] and Konishi et al. [2016] for comparisons to the aforementioned taxa).

Ventrally, the frontal is underlain posterolaterally by the postorbitofrontals (Figs 3, 4C, 5A). These overlap the frontal alae so that the posterolateral corners of the frontal’s ventral surface are laterally concave (Figs 3, 4C, 5A) rather than triangularly projecting as the alae are dorsally (Figs 2, 5B, C). Highly unusually for mosasaurs, the prefrontal extends nearly as far posteriorly as to the frontoparietal suture (Figs 2, 3, 4C, 5; see also Description, prefrontal, above). The posteromedian flanges of the frontal project posteriorly in ventral view to a level about one-third along the length of the pineal foramen (Figs 3, 5A). These flanges articulate medially with the parietal and so do not directly border the pineal foramen (Figs 3, 5A). The only contact between the frontal and the pineal foramen occurs where the anterior terminus of the pineal foramen contacts the midpoint of the posterior border of the frontal (Figs 3, 4C, 5A).

Anterior to the pineal foramen, the frontal bears a pronounced, roughly triangular boss on its ventral surface that tapers anteriorly to join the olfactory canal (Figs 3, 4C, 5A). The edges of the tabular boss form a sharp ridge, whereas the central portion of the boss is slightly concave and is at about the same level as the frontal’s main ventral surface (Figs 4C, 5A). The olfactory canal continues anteriorly along the ventral midline of the frontal, though is only visible for about 1 cm before being obscured by the displaced palatal bones (Figs 3, 4C, 5A). The posterior portion of the canal is completely surrounded by bone, such that it appears on the frontal’s ventral surface as a narrow rectangular tube extending from the anterior portion of the aforementioned boss (Figs 4C, 5A); however, as only this posterior-most extent is visible, it is uncertain whether this ventral enclosure continues along the rest of the canal or not. Lateral to the anterior half of the tabular boss and the posterior portion of the olfactory tract, narrow and smoothly concave depressions preserve the location of the cerebral hemispheres sensu Russell (1967) or, alternatively, the articulation surface for the orbitosphenoids sensu Konishi & Caldwell (2011) (Figs 3, 4C, 5A). These depressions are approximately 0.5–0.6 cm in width and 3.2 cm in length, though may extend further anteriorly under the superimposed palatal bones.

Parietal

The parietal is an elongate element that forms the medial and anteromedial borders of the supratemporal fenestrae. The anterior portion of the parietal forms a flattened table approximately 7 cm wide and almost 4 cm long, articulating with the frontal anteriorly and the postorbitofrontal laterally (Figs 2, 3, 5B). The pineal foramen barely contacts the frontoparietal suture anteriorly on this roughly triangular table (Figs 2, 3, 4C, 5A, B). This foramen is large and elongate, measuring 2.4 cm long and 0.6 cm wide. Dorsally, its edges are smoothly rounded (Figs 2B, 5B), whereas the ventral opening of this foramen is surrounded by a high, sharp, oval ridge (Figs 3, 4C, 5A) akin to the condition in Mosasaurus spp. (e.g. Konishi et al. 2014, fig. 4C) and at least one specimen of Ectenosaurus sp. (YPM 4674; T.K., pers. obs.). This ridge projects sharply from the level of the rest of the parietal, reaches its peak about 1 cm from the pineal foramen’s border, and curves down smoothly into the foramen itself (Figs 4C, 5A). Anteriorly, this ridge is confluent with the raised tabular boss on the ventral midline of the frontal (Figs 4C, 5A). Posteriorly, the sides of the ridge on either side of the pineal foramen converge to form a groove that extends the length of the parietal’s ventral midline, terminating where the parietal diverges posteriorly to form the suspensorial rami (Figs 3A, 4C, 5A). This groove is laterally offset near the anterior origin of the parietal descending processes, though this discontinuity is likely the result of taphonomy, as evidenced by the other breaks and grooves visible on the parietal.

The majority of the parietal consists of a narrow, elongate median body forming the medial borders of the supratemporal fenestrae (Figs 2, 3). This portion of the parietal is extremely constricted, a condition otherwise noted only in Goronyosaurus and Selmasaurus (Wright & Shannon 1988; Polcyn & Everhart 2008). Dorsally, an overhanging lateral shelf is present on each side of the parietal table posterolateral to the pineal foramen (Figs 2, 5B). This shelf extends diagonally from the centre of each side of the flattened anterior parietal to a point about one-fifth of the way along the lateral surface of the constricted body of the parietal (Fig. 2). This is in distinct contrast to the typical mosasaur condition: in other mosasaurs, these overhanging crests typically extend to the suspensorial rami of the parietal, causing almost the entire dorsal surface of the parietal to be distinctly flattened. In MHNM.KHG.1231, by comparison, these crests are greatly shortened, restricting the parietal table to the anterior one-third of the parietal.

Due to its marked constriction, the rest of the parietal forms a midsagittal crest. As noted above, this marked parietal constriction is also present in Selmasaurus russelli (see Wright & Shannon 1988, fig. 1) and S. johnsoni (see Polcyn & Everhart 2008, fig. 3); however, the condition in MHNM.KHG.1231 is more pronounced than in these species, resulting in a triangular, rather than sub-rectangular, parietal table and a posterior midsagittal, rather than parasagittal, crest(s) (Fig. 2). As mentioned, the parietal of Goronyosaurus shows similar constriction; however, it is too incomplete to fully assess the state of the sagittal crest or table.

Ventrally, the parietal bears two broadly crescentic descending flanges that descend at an angle from either side of the parietal’s posterior body, forming the medial wall of the supratemporal fenestrae (Figs 2, 3). These flanges extend from the anteromedial corner of the supratemporal fenestrae to a point about 4 cm anterior to the divergence of the parietal into the suspensorial rami. In life they would have articulated with the prootics, though in MHNM.KHG.1231 these bones, as well as other elements of the chondrocranium, are not preserved. These descending flanges are about 0.5–0.6 cm thick, tapering ventrally to a width of about 0.1 cm.

In mosasaurs, the posterior portion of the parietal is typically comprised of the suspensorial rami, though in MHNM.KHG.1231 these structures are broken such that only the most proximal portion of each ramus is preserved (Figs 2, 3). From what remains of the posterior parietal, it is evident that the rami diverge horizontally at about 30° to the sagittal midline. Proximally, the rami are relatively similar in height and width, though soon after their divergence become dorsoventrally depressed so as to be wider than they are tall, as is typical of non-halisaurine mosasaurs (e.g. Konishi et al. 2016).

Postorbitofrontal

The postorbitofrontal articulates anteriorly with the prefrontal, anteromedially with the frontal, medially with the parietal, and posteroventrally with the squamosal (Figs 2, 3, 4C, 5). Anteriorly, the main body of the postorbitofrontal extends to form the posterior half of the dorsal orbital border, articulating with the notched posterior terminus of the prefrontal just anterior to the midpoint of the orbit (Figs 4C, 5). The prefrontal–postorbitofrontal suture thus delineates an anteriorly facing wedge of the postorbitofrontal surrounded on either side by posteriorly projecting medial and lateral processes of the prefrontal. A similarly extensive articulation of the postorbitofrontal and prefrontal has been described in Plotosaurus as an adaptation for reduced cranial kinesis (LeBlanc et al. 2013).

The postorbitofrontal’s medial process to the parietal forms the anterolateral border of the supratemporal fenestra. Dorsally, this medial process is overlapped by the posterolaterally projecting frontal ala (Figs 2, 5B). The dorsal suture between the frontal ala and the postorbitofrontal is complex and interdigitating, in contrast to the more linear sutures connecting the postorbitofrontal to the prefrontal (Figs 2, 5B, C). Ventrally, the postorbitofrontal meets the parietal in a laterally concave suture marked by a distinct groove in the ventral skull roof (Figs 3, 4C, 5A).

At the junction between its anterior, medial and posterior processes, the postorbitofrontal bears a descending process that in life would have articulated with the jugal (Russell 1967; Figs 3, 4C, 5A). This process is anteroposteriorly compressed and tapers to a smoothly rounded ventrolateral border.

Posteriorly, the postorbitofrontal bears a long, thin process that articulates with the dorsal surface of the squamosal (Figs 2, 3, 5C). Whereas the anterior and medial processes of the postorbitofrontal are dorsoventrally depressed, this squamosal process is laterally compressed. The junction of this process with the main body of the postorbitofrontal forms the anterolateral corner of the supratemporal fenestra and occurs at approximately a right angle (Fig. 2). This corner is obtuse in another plioplatecarpine Platecarpus tympaniticus Cope, 1869 (e.g. Konishi & Caldwell 2009, fig. 5D), with Selmasaurus showing an intermediate condition (e.g. Wright & Shannon 1988, fig. 1). The squamosal process extends nearly to the posterior extent of the squamosal, becoming progressively thinner as it approaches its posterior terminus (Figs 2, 3). A notch at the ventral junction of the postorbitofrontal’s squamosal and jugal processes articulates with the pointed anterior terminus of the squamosal (Fig. 3).

Squamosal

The squamosal is a laterally compressed, rod-like bone that articulates dorsally with the squamosal process of the postorbitofrontal (Figs 2, 3, 5C). In MHNM.KHG.1231, the left squamosal is broken near its anterior extent and is twisted such that its dorsal surface now lies medially (Fig. 2). The dorsal border of the squamosal is gently inclined, such that the squamosal shaft is 2.8 cm high at its posterior terminus and tapers downward anteriorly to a height of 1.1 cm, terminating anteriorly in a wedge that inserts between the postorbitofrontal’s jugal and squamosal processes (Figs 3, 5C).

The posterior terminus of the squamosal forms a ventromedially expanded head (Fig. 3) that in life would have articulated medially with the supratemporal and parietal and ventrally with the quadrate. The dorsal surface of the squamosal head is continuous with the dorsal surface of the squamosal shaft (Fig. 2), whereas the ventral portion of the body expands below the level of the ventral shaft (Fig. 3). The internal face of the squamosal head is teardrop-shaped and medially concave, forming a facet for the posterior portion of the supratemporal (Fig. 3). The squamosal head bears a flat facet anteroventrally that would have articulated with the dorsal surface of the quadrate, possibly along the suprastapedial process (Fig. 3).

The squamosal bone and the squamosal process of the postorbitofrontal together form the lateral border of the supratemporal fenestra. The junction of the squamosal shaft and squamosal head forms the posterolateral corner of this fenestra, as the squamosal head expands ventromedially from the shaft. The dorsal articulation between the squamosal and postorbitofrontal extends nearly the full length of the squamosal, terminating at the approximate dorsal midpoint of the squamosal head (Fig. 2).

Quadrate

A partial right quadrate was recovered from the matrix surrounding the right dentary (Fig. 8). Very little is preserved, with this fragment representing the dorsal-most portion of the quadrate shaft and most of the suprastapedial process. The medial surface of the suprastapedial process bears a broad excavation, causing the suprastapedial process to be constricted dorsally (Fig. 8B, C). This condition is limited to MHNM.KHG.1231, Ectenosaurus and Selmasaurus among the Plioplatecarpinae (Konishi 2008). Dorsal constriction of the suprastapedial process has been recognized in previous studies as a trait typical of mosasaurines (e.g. Bell 1997, character 45); however, given the results of our and previous phylogenetic analyses (e.g. Bell 1997; Caldwell et al. 2008; Jiménez-Huidobro & Caldwell 2019), we discard the possibility that the condition in mosasaurines could be homologous with that observed in Gavialimimus as unparsimonious. (For further discussion of this condition, see description and scoring of characters 38 and 42 in our multistate [ S1] and contingent [ S2] character lists, respectively.)

The dorsal surface of the suprastapedial process is generally consistent in width, measuring about 1.3 cm wide along most of its length, though it does flare outward distally and taper proximally (Fig. 8C). This ventral flaring is mostly medial with minor lateral expansion, likely reflecting medial deflection of the distal terminus of the suprastapedial process combined with a minimal degree of distal expansion of this process. However, the distal-most portion of the suprastapedial process was sheared off during excavation of the specimen (see unfinished surface in Fig. 8D), so its exact morphology is unclear. The proximal tapering of this quadrate fragment represents the small preserved remnant of the quadrate shaft. This portion of the quadrate is exceptionally narrow, which, if anatomically accurate, would be quite unique among mosasaurs; however, the highly fragmentary nature of this element makes confident interpretation of the quadrate shaft morphology in its entirety impossible.

The alar concavity on the lateral surface is smooth and flat, curving tightly dorsolaterally to give rise to the quadrate ala around its rim (Fig. 8A). The quadrate ala is mostly broken off, though what is preserved is quite thin. The dorsal extent of the stapedial notch is present and is smoothly and tightly curved, measuring about 0.8 cm in width (Fig. 8A, B, D). The stapedial pit overlies the stapedial notch anteromedially at the junction between the suprastapedial process and the quadrate shaft (Fig. 8B, D). It is a relatively narrow oval, measuring approximately 1.2 cm and 0.6 cm along its long and short axes, respectively. Its long axis is oriented parallel to the anterodorsal border of the stapedial notch (Fig. 8D).

Several of the features noted above are similar to the quadrate morphology of Selmasaurus. For example, a medially excavated suprastapedial process and corresponding narrow dorsal surface are also present in the holotypes of S. russelli and S. johnsoni and in quadrates referred to Selmasaurus sp. (C.S., pers. obs.; see also Konishi 2008; Polcyn & Everhart 2008, fig. 6). In Selmasaurus, the suprastapedial and infrastapedial processes are in broad contact (Konishi 2008), with both processes being medially deflected. This medial deflection also occurs in the suprastapedial process of MHNM.KHG.1231; however, the infrastapedial process is not preserved and so this contact is impossible to assess. Furthermore, in Selmasaurus the suprastapedial process simply tapers to a blunt terminus, without the distal expansion that appears to be present in MHNM.KHG.1231. A synapomorphy of the genus Selmasaurus is the presence of a distinct lateral deflection midway along the length of the quadrate shaft, a condition also present in Taniwhasaurus antarcticus (Novas, Fernández, de Gasparini, Lirio, Nuñez & Puerta, 2002) and certain species of Plioplatecarpus (e.g. Plioplatecarpus houzeaui Dollo, 1889b) (C.S., pers. obs.; see also Konishi 2008; Fernandez & Martin 2009). Again, the incomplete nature of the quadrate of MHNM.KHG.1231 makes this feature impossible to assess in this specimen. Therefore, although what little is left of the quadrate in MHNM.KHG.1231 is broadly similar to the morphology of Selmasaurus, the degree of incompleteness of this specimen limits the extent of comparisons that can be made and synapomorphies evaluated.

Vomer

The vomers in MHNM.KHG.1231 are long, splint-like bones that extend from the midpoint of the main body of the palatine to the posterior extent of the premaxilla, lying medially to and in close contact with the lingual shelves of the maxillae for most of their length, thus forming a nearly solid bony palate (Figs 3, 4).

The vomer–palatine suture is well preserved. The posterior terminus of the vomer articulates with the anteromedial corner of the main palatine body (Figs 3A, 4C). The lateral border of the vomer then curves medially to form the rounded posteromedial corner of the internal naris (Figs 3, 4C). From this point onward, the lateral margin of the vomer extends straight anteriorly toward its contact with the premaxilla (Figs 3, 4A, B). Medially, the vomers articulate with each other along their entire length, forming a straight anteroposterior suture along the midline of the palate (Figs 3, 4). As the snout thins anteriorly, so too do the vomers, becoming narrower and less distinguishable from each other as they approach their anterior articulation with the vomerine processes of the premaxilla (Figs 3, 4A). The ventral oblique crest of the left vomer extends farther ventrally than that of the right vomer, due to displacement of the palate.

Anterior to the disarticulated humerus – which obscures part of the posterior palate – a small splint of bone is present lateral to the right vomer (Figs 3, 4B). This splint lies medial to the fourth, fifth and sixth tooth positions on the right maxilla. This bone may represent part of the right septomaxilla; alternatively, it may simply be a broken portion of the right vomer. Due to taphonomic distortion of the palate, especially at its anterior extent, it is not possible to definitively discount either of these options.

Palatine

Due to taphonomic compression of the skull in MHNM.KHG.1231, the skull roof is in direct contact with the palate. The palate is also offset laterally, such that the palatal midline is to the left of the skull roof midline and the right internal naris is in line with the left external naris (Figs 3, 4). The palatal bones themselves are quite well preserved, including preservation of the articulation between the palatines and vomers (Figs 3, 4C), a contact rarely preserved in mosasaurs (Russell 1967).

The palatine in MHNM.KHG.1231 consists of a roughly triangular main body bearing a process projecting anteriorly from the anterolateral corner of its ventral surface (Figs 3, 4C). The vomero–palatine contact creates a diagonal suture running posteromedially from the internal naris to the midpoint of the medial border of the palatine (Fig. 4C). Posterior to this suture, the medial border of the palatine runs almost straight posteriorly (Figs 3, 4C). The palatines possess only a narrow contact with one another medially (Fig. 4C). This separation may be the result of taphonomic distortion, but may alternatively represent where the pterygoids, here not preserved in situ, would have articulated medially to the palatines (e.g. as in Plesioplatecarpus planifrons [Cope, 1874], as figured in Konishi & Caldwell [2007, fig. 3]).

As in most mosasaur specimens, the palatine–pterygoid articulation is not preserved (Russell 1967; Konishi & Caldwell 2007). The posterior extent of each palatine is broken, such that the posterior border is transversally straight on the right palatine and curved on the left palatine (Figs 3, 4C).

The palatine articulates laterally with the maxilla via an unusually long suture, at least four alveoli long based on the right side, due in part to a long anterolateral process (see below). Posteriorly, the maxilla–palatine contact is directed anteromedially at about 45° in ventral view (Fig. 4C). Anterior to this, the palatine bears a process projecting directly anteriorly and articulating with a groove on the medial surface of the posterior maxilla (Figs 3, 4C). This process is about half the length of the main body of the palatine, terminating just anterior to the 11th maxillary tooth position and forming the posterolateral border of the internal naris (Fig. 4C). The anterior border of the palatine is tightly curved to form the oblong posterior border of the internal naris (Fig. 4C). The ventral surface of the palatine slopes gently dorsally as it anteriorly approaches the internal naris.

Due to the taphonomic distortion of the skull roof and palate, as well as the placement of a disarticulated humerus midway along the palate, the anterior extent of the internal nares is unclear. However, based on the left internal naris – the more intact of the internal nares in MHNM.KHG.1231 – the internal nares are teardrop-shaped, tapering anteromedially and extending to at least the ninth maxillary tooth position (Fig. 3). Posteriorly, the internal nares terminate between the 11th and 12th maxillary tooth positions (Figs 3, 4C), whereas this posterior termination typically occurs more anteriorly along the tooth row in other mosasaurs. This is particularly true in short-snouted plioplatecarpines such as Plioplatecarpus primaevus Russell, 1967 (see Holmes 1996, fig. 2) and Platecarpus tympaniticus (see Russell 1967, fig. 84). This more anterior termination also occurs in long-snouted taxa such as early-diverging mosasaurines (e.g. NHMUK PV R2946, C.S., pers. obs.). In this latter specimen, the internal nares terminate posteriorly near the 13th of its 16–18 maxillary teeth and are therefore comparatively more anteriorly placed than those of MHNM.KHG.1231. The internal nares in MHNM.KHG.1231 are thus somewhat retracted compared to other mosasaurs.

Pterygoid

Both pterygoids are partially preserved in MHNM.KHG.1231 (Fig. 9A–D). The anterior and posterior termini of both pterygoids are broken, such that the pterygoids both terminate anteriorly midway along the tooth row on the main pterygoid body and posteriorly just beyond the divergence between the basisphenoid process and quadrate ramus. The ectopterygoid process is also preserved on both pterygoids (Fig. 9A–D). Both pterygoids are noticeably fractured, with the right pterygoid also being partially obscured laterally and dorsally by two vertebrae (Fig. 9C, D).

The main body of the pterygoid tapers to a thin, flat flange that projects laterally from the pterygoid tooth row, causing the tooth row to be medially offset from the pterygoid midline (Fig. 9A, B). Dorsally, the main body extends to form a narrow, rounded ridge that sits directly above the tooth row (Fig. 9B).

The ectopterygoid process projects generally transversally, though taphonomic distortion and differences in preservation between the left (Fig. 9A, B) and right (Fig. 9C, D) pterygoids cause subtle differences in morphology and orientation of the respective ectopterygoid processes. The ectopterygoid process tapers anterolaterally, also bearing a posterodistal projection angled posteromedially toward the proximal end of the quadrate ramus of the pterygoid. The posterior border of the ectopterygoid process meets the pterygoid body posterolateral to the posterior terminus of the pterygoid tooth row (Fig. 9A, C). This condition resembles that of Goronyosaurus nigeriensis, the holotype of Platecarpus somenensis Thévenin, 1896, and Selmasaurus johnsoni (but unknown in S. russelli), in which the base of the ectopterygoid process is also unusually broad (T.K., pers. obs.; see also Soliar 1988, fig. 5; Polcyn & Everhart 2008, fig. 4). The former two species also possess a posterodistal projection of the ectopterygoid process, as described above for MHNM.KHG.1231. This projection contacts the quadrate ramus of the pterygoid in G. nigeriensis but not in P. somenensis. This process also appears to contact the quadrate ramus in the left pterygoid of MHNM.KHG.1231 (Fig. 9A, B); however, this contact is likely taphonomic, as the distal-most tip of this projection is deeply broken on the left pterygoid, giving rise to an artificial contact between these processes. The true anatomical configuration would therefore likely resemble that of P. somenensis, with the posterodistal projection of the ectopterygoid process approaching but not contacting the quadrate ramus, as seen in the right pterygoid of MHNM.KHG.1231 (Fig. 9D).

The basisphenoid process and quadrate ramus begin to diverge slightly posterior to the pterygoid tooth row. This divergence is visible dorsally, as the dorsally rounded surfaces of these processes extend posteriorly from the main pterygoid body, forming ridges that border a progressively deepening sulcus (Fig. 9B, D). About 5 cm posterior to the posterior-most pterygoid tooth – a distance equivalent to approximately three pterygoid tooth positions – this sulcus gives way to a complete divergence between these processes. The basisphenoid process projects directly posteriorly from the main pterygoid body, with the quadrate ramus projecting laterally at a 45° angle (Fig. 9B, D). The dorsal surface of the basisphenoid process is smoothly rounded, whereas the quadrate ramus bears a thin groove lateral to its dorsal midline. The ventral surface of the quadrate ramus is uniformly smooth, whereas the basisphenoid process bears a narrow, rounded ridge along the medial border of its ventral surface.

The pterygoid tooth row preserves six tooth positions (Fig. 9A, C). A separate fragment of the anterior pterygoid tooth row is present in a block containing two vertebrae (Fig. 9E, F), although it is uncertain to which pterygoid this fragment belongs. Only four pterygoid tooth crowns are preserved: two on the separated fragment and two on the right pterygoid at the fifth and sixth posterior-most tooth positions (Fig. 9C, E). These teeth exhibit narrow plicae, similar to the marginal dentition. The pterygoid teeth are much more recurved and much smaller than the marginal teeth, an expected condition for plioplatecarpines among russellosaurines. The tooth row itself is straight; the anterior-most teeth preserved on the right pterygoid project ventromedially from the tooth row, but this is clearly the result of taphonomy (Fig. 9C). On the left pterygoid, breakage of the thin flange lateral to the tooth row gives the appearance of the anterior part of the tooth row rising above the level of the ventral surface (Fig. 9A); however, the posterior extent of both the left and right pterygoid tooth rows is unbroken and shows the tooth row to be level with the ventral surface formed by the diverging processes.

Mandible

Dentary

Both the left and right dentaries are heavily fractured posteriorly, with several less severe fractures throughout (Fig. 7A, B). The posterior portion of the left dentary is also separated from the rest of the dentary by a large break (Fig. 7A), though whether this break is due to predation or to breakage during excavation of the specimen is indeterminable.

The ventral border of the dentary is horizontal along most of its length, though curves dorsally at the anterior-most tooth position to form the anterior terminus of the dentary, and curves ventrally posterior to the 10th tooth position (Fig. 7A). The ventral border of the dentary is tightly rounded and is thicker anteriorly than posteriorly. The dorsal border of the dentary is flattened to accommodate the tooth row and interdental pits (Fig. 7A). The morphology of the teeth and associated features is described above (Description, Dentition). Medial to the tooth row, the medial parapet of the dentary rises vertically so as to be taller than the lateral surface of the dentary (Figs 6D–F, 7A), a condition otherwise only known to occur in Plioplatecarpus (see Konishi & Caldwell 2011). The medial parapet increases in height posteriorly, progressing from covering the ventral one-third of the exposed portion of the dentary tooth roots anteriorly to covering about one-half of the exposed roots of the posterior dentary teeth (Fig. 7A). Replacement tooth pits occur lateral to the medial parapet and posteromedial to their respective tooth positions (Figs 6D–F, 7A).

The lateral surface of the dentary is almost entirely flat, though bulges slightly outward at its anterior-most extent under the first and second tooth positions (Fig. 7A). The height of the dentary increases along its length, such that the height of the posterior terminus is about threefold that of the anterior terminus (Fig. 7A, B). The lateral surface of the dentary bears several foramina marking the exits of the mandibular (V3) branch of the trigeminal nerve (cranial nerve V) (Russell 1967; Fig. 7A). These foramina are the same size as those on the premaxilla and maxilla (Fig. 2). They form three loose anteroposterior rows along the lateral surface of the anterior dentary, below the first four dentary tooth positions (Fig. 7A).

The medial surface of the dentary is mostly flat, though it bears a deep excavation marking the Meckelian groove (Fig. 7B). This groove runs along the ventral half of the dentary’s medial surface, with its anterior terminus under the sixth anterior-most tooth position. The Meckelian groove in MHNM.KHG.1231 thus extends markedly less anteriorly compared to other mosasaurs, in which this groove typically extends almost all the way to the anterior terminus of the dentary (e.g. Platecarpus tympaniticus [Russell 1967, fig. 29]; Selmasaurus johnsoni [Polcyn & Everhart 2008, fig. 7S]). This posteriorly retracted position of the anterior terminus of the Meckelian groove therefore represents an autapomorphy of this species. The Meckelian groove is tallest posteriorly, tapering from a height of 4.7 cm at its posterior terminus to a height of 0.8 cm at its anterior terminus (Fig. 7B). The dorsal margin of the Meckelian groove is horizontal and uniformly inset from the main plane of the medial surface of the dentary (Fig. 7B). In contrast, the ventral margin is inset along the anterior-most one-third of its length, though posterior to this, the Meckelian groove opens directly onto the ventral margin of the dentary (Fig. 7B).

Angular

The right angular is preserved in articulation with the rest of the right posterior mandibular unit (Fig. 7C–F). The anterior terminus of the angular is broken, so the morphology of the splenial-angular articulating surface is unknown. The angular is much more exposed laterally (Fig. 7C, D) than it is medially (Fig. 7E, F). Laterally, the dorsal border of the angular is parallel to its ventral border and is slightly ventrally concave (Fig. 7C, D). In contrast, in medial view the angular tapers sharply posteriorly to form a long, thin sliver of bone that underlies the articular-prearticular (Fig. 7E, F). The angular extends posteriorly much farther than is typical of mosasaurs (e.g. Platecarpus tympaniticus: Russell 1967, fig. 29), terminating below the glenoid fossa, just anterior to a posteroventral flange on the articular that likely represents a crushed and distorted retroarticular process (Fig. 7E, F).

Surangular

The surangular is quite fragmented, especially anteriorly (Fig. 7C–F). The dorsal border of the surangular is straight with a slight posterodorsal bump marking the anterodorsal portion of the glenoid fossa. In lateral view the surangular becomes taller anteriorly, consistent with the anterior dorsoventral expansion of the overall posterior mandibular unit (Fig. 7C, D). Posteriorly, the surangular meets the articular at the middle of the glenoid fossa (Fig. 7E, F). The medial surface of the surangular is slightly thickened near its posterodorsal corner, delimiting the anteromedial border of the glenoid fossa. The suture between the surangular and articular-prearticular is marked by a deep groove on the medial surface of the posterior mandibular unit (Fig. 7E, F); such a groove is lacking on the lateral surface, with the sutures between the surangular and surrounding elements being much less pronounced (Fig. 7C, D).

Articular-Prearticular

The anterior extent of the prearticular is long and narrow, underlying most of the surangular (Fig. 7E, F). The surangular–prearticular suture is strongly grooved medially, representing the adductor fossa, though taphonomically compressed (Fig. 7F). The prearticular bears a pronounced ridge ventral to the anterior half of this suture, with the surangular bearing a similar – though less pronounced – ridge on its ventral border at the same position (Fig. 7F). Posteriorly, the dorsal border of the prearticular curves posterodorsally at a 50° angle relative to its flat anterodorsal border as it gives way to the articular (Fig. 7E, F). The surangular–articular suture terminates dorsally within the glenoid fossa (Fig. 7C–F). This suture is marked by a medial thickening of the articular at the posterior border of the glenoid fossa, causing the posteroventral corner of the surangular to be inset relative to the articular (Fig. 7F). The dorsal border of the articular is relatively straight and dips at a slight posteroventral angle. The posterior extent of the articular is relatively complete except for a displaced flange preserved at its posteroventral corner, possibly representing a taphonomically deformed and displaced retroarticular process (Fig. 7C–F).

Postcranial

Vertebrae

Three complete though disarticulated and distorted vertebrae are preserved, as well as one highly crushed vertebral centrum (Fig. 9C–F). The synapophyses of each vertebra project horizontally at a slight posterior angle, though the exact angle varies among the vertebrae due to taphonomic distortion (Fig. 9C–F). The articulatory facet of each synapophysis is a depressed oval distinctly shorter than the centrum height, suggesting a position in the anterior to intermediate cervical region (Russell 1967). Two of the vertebrae possess prominent hypapophyses (Fig. 9C, vertebra on the left; Fig. 9E), consistent with this possible anterior cervical position. However, the other vertebra (Fig. 9D, vertebra on the left) instead possesses a pair of haemapophyses, one intact and the other mostly sheared off (Fig. 9D). The presence of these haemapophyses, in conjunction with the narrower, taller, and less concave and convex (i.e. flatter) vertebral centrum, suggests that this is a caudal vertebra.

The neural spine of each vertebra is slightly more than twice the height of the centrum, though all are broken at their dorsal extent (Fig. 9C–F). Each neural spine projects from the centrum at about a 20° posterodorsal angle. The pre- and postzygapophyses are particularly well preserved in one of the vertebrae (Fig. 9E, F). On this vertebra, the prezygapophysis projects at a slight anterolateral and upturned angle from the lateral border of the neural arch (Fig. 9E). The postzygapophyses are smaller than the prezygapophyses, forming circular tubers that project laterally from the base of the neural spine just dorsal to the neural arch (Fig. 9E, F). The articulatory facet of each postzygapophysis is directed ventrolaterally (Fig. 9E, F). A small depression is present under one of the postzygapophyses, potentially representing a zygantrum (Fig. 9F). However, the lack of zygosphenes on any of the vertebrae, as well as the breakage of the neural spine of this vertebra, suggest that this feature is the result of taphonomic distortion; it is more likely that this specimen lacks zygantra and zygosphenes.

Humerus

The right humerus, in extensor view, is preserved on the underside of the palate (Figs 3, 4B, 9G). Due to the fragility of the palate, the humerus was not removed. Breakage and abrasion to the humerus obscure the position of muscle attachment scars. The length and distal width of the humerus are essentially equal, with the length measuring 8.6 cm and the distal width measuring 9.0 cm (Fig. 9G). The distal end of the humerus is expanded relative to the proximal end, similar to the condition in Platecarpus (Russell 1967; Fig. 9G). Both the entepicondyle and ectepicondyle are present (Fig. 9G). Though both processes are broken, the portions that are preserved indicate that the ectepicondyle was likely larger and more laterally prominent than the entepicondyle. Both processes extend from the distal border of the humerus to about midway along the humerus shaft. The shaft itself is smoothly concave along its preaxial and postaxial borders (Fig. 9G). The extensor surface of the shaft is gently constricted along its longitudinal axis, resulting in a slight ridge extending along this surface from the proximal to distal ends of the humerus (Fig. 9G). This long ridge is not typical of plioplatecarpines, though the thin cracks bordering this ridge suggest that it may be taphonomically exaggerated in MHNM.KHG.1231 (Fig. 9G). The proximal end of the humerus is mediolaterally deeper than the comparatively compressed distal end. The proximal end exhibits a spongy texture, indicating the presence of a cartilage cap, similar to the condition in Platecarpus and Tylosaurus (Russell 1967). As such, specific features such as the postglenoid process and the glenoid condyle are not preserved.

Phylogenetic analysis

Multistate coding scheme – unweighted maximum parsimony analysis (Mu-UMP)

Unweighted maximum parsimony analysis of the dataset using a multistate coding scheme recovered 26 most parsimonious trees (MPTs) with lengths of 464 steps (consistency index CI = 0.34913793; retention index RI = 0.70304818). A strict consensus tree was generated from these optimal trees (). For improved resolution, a 50% majority rule consensus tree was also generated ().

Multistate coding scheme – implied weighting maximum parsimony analysis (Mu-IWMP)

Implied weighting maximum parsimony analysis of the dataset using a multistate coding scheme and a k-value of 3.0 recovered one MPT with a length of 468 steps (fit = 46.630811; CI = 0.34615385; RI = 0.69911504; Fig. 10A). Implied weighting analyses using k-values of 7.0 (Fig. 10B) and 11.0 (Fig. 10C) each recovered one MPT with a length of 465 steps (CI = 0.3483871; RI = 0.7020649), with fit values of 28.187371 and 20.361057, respectively. This analytical approach produced the trees with the highest resolution (i.e. a single MPT for each k-value). Each k-value also recovered the same topology within Mosasauridae, except within the Mosasaurinae.

Figure 10. Phylogenetic trees generated via implied weighting maximum parsimony analysis of the dataset utilizing multistate character coding (Mu-IWMP). A, single most parsimonious tree (MPT) (468 steps) generated when k = 3.0; B, single MPT (465 steps) generated when k = 7.0; C, single MPT (465 steps) generated when k = 11.0; D, enlarged phylogeny of the subfamily Plioplatecarpinae. Abbreviations: Ha, Halisaurinae; Mo, Mosasaurinae; Pl, Plioplatecarpinae; Te, Tethysaurinae; Ty, Tylosaurinae; Ya, Yaguarasaurinae.

Contingent coding scheme – unweighted maximum parsimony analysis (Co-UMP)

Unweighted maximum parsimony analysis of the dataset using a contingent coding scheme recovered 112 MPTs with lengths of 468 steps (CI = 0.34615385; RI = 0.70520231). Strict consensus and 50% majority rule consensus trees were generated from these optimal trees ().

Contingent coding scheme – implied weighting maximum parsimony analysis (Co-IWMP)

Implied weighting maximum parsimony analysis of the dataset using a contingent coding scheme and a k-value of 3.0 recovered two MPTs with lengths of 472 steps (fit = 48.300075; CI = 0.34322034; RI = 0.70134875). A strict consensus tree was generated from these optimal trees (). Analysis using a k-value of 7.0 () recovered one MPT with a length of 468 steps (fit = 29.016181; CI = 0.34615385; RI = 0.70520231). Analysis using a k-value of 11.0 recovered two MPTs with lengths of 468 steps (fit = 20.889202; CI = 0.34615385; RI = 0.70520231), from which a strict consensus tree was generated ().

Topology

All of the phylogenies recovered generally similar topologies, with the major differences among them resulting from variation in resolution. A detailed comparison of the various phylogenies is provided in the . We focus herein on the phylogenetic relationships within Plioplatecarpinae (Fig. 10D), as this is the subfamily in which MHNM.KHG.1231 was consistently recovered in all analyses (Fig. 10; , Figs S5–7).

Plioplatecarpinae exhibited a highly variable topology across different analyses, with polytomies greatly reducing internal resolution in the Mu-UMP consensus trees () and essentially collapsing this clade in the Co-UMP strict consensus tree (). That being said, certain aspects of its internal topology were fairly consistent across the different analytical conditions. The clade (Latoplatecarpus (Plioplatecarpus, Platecarpus tympaniticus)) was recovered in every analysis, contra Konishi & Caldwell (2011) (Fig. 10; –7). In the Mu-IWMP (Fig. 10), Co-UMP majority rule consensus () and Co-IWMP () trees – i.e. all of the trees in which the Plioplatecarpinae was monophyletic and fully resolved – Ectenosaurus was the earliest-diverging taxon, followed by Plesioplatecarpus planifrons. In the Mu-IWMP trees (Fig. 10), this was followed by (Latoplatecarpus (Plioplatecarpus, P. tympaniticus)), then Angolasaurus as the sister group to a clade consisting of (Goronyosaurus (Selmasaurus johnsoni (S. russelli, MHNM.KHG.1231))); in the contingent trees (), the positions of (Latoplatecarpus (Plioplatecarpus, P. tympaniticus)) and Angolasaurus were switched.

This GoronyosaurusSelmasaurus–MHNM.KHG.1231 clade is of particular interest for this study, as MHNM.KHG.1231 represents the species of interest, Gavialimimus almaghribensis, gen. et sp. nov. This clade is supported by the following synapomorphies: highly constricted parietal; sides of the frontal in front of the orbits relatively straight. It was recovered in 93% of the MPTs in the Co-UMP majority rule consensus tree () and in all of the IWMP trees (Fig. 10; ). Polytomies collapsed this clade in the Mu-UMP and Co-UMP strict consensus tree (). Interestingly, the Mu-UMP majority rule consensus tree () recovered G. nigeriensis as sister to (Latoplatecarpus (Plioplatecarpus, Platecarpus tympaniticus)).

Gavialimimus almaghribensis was recovered as the sister to Selmasaurus russelli in 100% of the MPTs in the Mu-UMP majority rule consensus tree () and 91% of the MPTs in the Co-UMP majority rule consensus tree (). This sister-group relationship was also recovered in the Mu-IWMP trees (Fig. 10) and Co-IWMP trees (). G. almaghribensis and Selmasaurus share several synapomorphies from the multistate and contingent datasets, comprising: parietal table elongate, triangular to sub-rectangular, and highly medially constricted, with a distinct mid- or parasagittal crest anterior to the divergence of the suspensorial rami (character 18/20 in the multistate and contingent character lists, respectively); frontal not invaded by posterior end of nares (character 9/10); and lack of a frontal midline dorsal keel (character 10/11). These taxa also share a distinct broad excavation of the medial surface of the quadrate suprastapedial process (Konishi 2008; Fig. 8). G. almaghribensis further shares the following synapomorphies with S. russelli: pineal foramen ventral opening surrounded by a rounded, elongate ridge (character 21/23); and pineal foramen large (character 19/21). Note that, although this latter character state does occur in most other plioplatecarpines, it does not occur in S. johnsoni, thus distinguishing S. russelli and G. almaghribensis from S. johnsoni.

Discussion

Rejection of ‘Platecarpusptychodon

The diagnosis for ‘Platecarpusptychodon is as follows (translated from Arambourg 1952, p. 285):

Platecarpus (?) with teeth that are relatively low and broad at the base of the crown, slightly compressed with wide edges and no serrations along the symphyseal and commissural borders; the lingual and labial faces are ornamented with numerous irregular vertical plicae extending 2/3 of the tooth’s height from the neck.

The teeth of MHNM.KHG.1231 match this diagnosis, suggesting referral of this specimen to ‘P.ptychodon. However, although the teeth of ‘P.ptychodon were considered by Arambourg (1952) – and subsequent workers (e.g. Bardet et al. 2000, 2018) – to be highly diagnostic, their putatively unique features are also present in the teeth of other mosasaur species, such as Platecarpus somenensis (Fig. 11). Given that this latter species was named in 1896, well before the creation of ‘P.ptychodon in 1952, ‘P.ptychodon is therefore not a valid species, as its diagnosis and holotype, represented by a single isolated tooth (Fig. 11A), are insufficient in distinguishing it from the already-established P. somenensis (Fig. 11G–I).

Figure 11. Comparison of various marginal teeth of A–C,Platecarpusptychodon, D–F, Gavialimimus almaghribensis, G–I, Platecarpus somenensis, J–L, Prognathodon solvayi, and M–O, Mosasaurus lemonnieri. Note how all of the teeth match the diagnosis of ‘P.ptychodon, rendering this diagnosis – and thus this species – invalid. A–C, isolated teeth as photographed by Arambourg (1952, pl. 39, fig. 2: holotype of ‘P.ptychodon, fig. 4, and fig. 6, respectively); D, E, MHNM.KHG.1231 (holotype), maxillary; F, MHNM.KHG.1231 (holotype), premaxillary; G–I, MNHN 1895-7 (holotype), maxillary; J, IRSNB R33-4672 (holotype), maxillary; K, L, IRSNB R33-4565, dentary; M, N, IRSNB 3109, maxillary and dentary; O, IRSNB 3119, maxillary. Scale bar represents 1 cm; all teeth to scale.

The teeth of ‘Platecarpusptychodon also resemble those of Prognathodon solvayi (Fig. 11J–L), Tylosaurus ivoensis (Persson, 1963) (see Lindgren & Siverson 2002, fig. 4), Taniwhasaurus (see Caldwell et al. 2008, fig. 5), and some specimens of Mosasaurus lemonnieri Dollo, 1889b (Fig. 11M–O). However, a feature distinguishing the teeth of ‘P.ptychodon from the mosasaurines Prognathodon solvayi and Mosasaurus lemonnieri is the presence of slight crenulations or serrations in the enamel along the carinae of mosasaurine teeth, best visible in histological thin-section. Importantly, though, these crenulations are largely limited to the replacement teeth in mosasaurines, being worn down very quickly in functional teeth (A. LeBlanc, pers. comm.). Therefore, even this feature is unreliable as the sole basis for distinguishing taxa such as those listed above. M. lemonnieri and Pr. solvayi thus also exemplify already-established species with which the holotype of ‘P.ptychodon may be confounded. Furthermore, no plioplatecarpine – including ‘P.ptychodon and Platecarpus somenensis – has been observed to possess these crenulations (Bell 1997; A. LeBlanc, pers. comm.). As such, ‘P.ptychodon remains indistinguishable from plioplatecarpines such as P. somenensis regarding both histological and coarser-scale morphology.

Indeed, although Mosasaurus lemonnieri, Platecarpus somenensis, Prognathodon solvayi, Tylosaurus ivoensis, Taniwhasaurus and MHNM.KHG.1231 are clearly different taxa, their teeth overlap so greatly in morphology that the ‘Platecarpusptychodon type material and similar isolated teeth often cannot be referred to a single species; this tooth morphology is ultimately entirely non-diagnostic (Fig. 11). As cautioned by various authors (e.g. Massare 1987; LeBlanc et al. 2012), tooth morphology in general is highly susceptible to convergence; as such, diagnosing a taxon based solely on such a homoplastic feature is a fundamentally unreliable taxonomic approach. ‘P.ptychodon is thus a nomen dubium, as its type material is indeterminate and lacks any autapomorphic features distinguishing it from already-named species (International Commission on Zoological Nomenclature 1999, Glossary and 75.5; see also Coombs [1990] and Modesto [1996] for similar designations of other taxa as nomina dubia based on non-diagnostic type material). As a consequence of this invalidity, all specimens that have since been referred to ‘P.ptychodon no longer bear this designation and therefore must be re-examined for correct referral to a valid species. Ultimately, because ‘P.ptychodon is not a valid species, MHNM.KHG.1231 is free to be assigned as the holotype of a new taxon.

Phylogenetic relationships of Gavialimimus almaghribensis

Before discussing the phylogenetic relationships of Gavialimimus almaghribensis, it is important to first assess the various analytical approaches presented above. From a theoretical viewpoint, we prefer implied weighting to unweighted parsimony analyses because of the ability of implied weighting analyses to reduce the impact of homoplasy on phylogenetic reconstruction (Goloboff 1993; Goloboff et al. 2008a; Simões et al. 2017a). We also prefer multistate character coding schemes, rather than contingent schemes, because of their treatment of dependent characters as a single transformational series (Simões et al. 2017a, b). This theoretical background was ultimately supported by our results: of all the analyses presented above, the Mu-IWMP analyses (Fig. 10) produced the trees with the highest resolution and among the lowest number of steps (465, second only to the Mu-UMP trees with 464 steps). For these methodological and empirical reasons, this is our preferred analysis (Fig. 10).

The recovered GoronyosaurusSelmasaurus–MHNM.KHG.1231 clade also warrants further discussion. As noted above (see Phylogenetic analysis, Topology), the latter two genera share several synapomorphies, raising the possibility that MHNM.KHG.1231 should be assigned to Selmasaurus rather than a new genus. However, because of the many autapomorphies of MHNM.KHG.1231, doing so would result in a high degree of intrageneric variation within Selmasaurus. MHNM.KHG.1231 also differs temporally from Selmasaurus, occurring in the upper Maastrichtian at least 9 million years after the upper Santonian–lower Campanian distribution of S. russelli (Wright & Shannon 1988; Kiernan 2002; Liu 2007; Konishi 2008; Ogg et al. 2012) and at least 13 million years after the lower Santonian range of S. johnsoni (Polcyn & Everhart 2008; Ogg et al. 2012). For these reasons, although MHNM.KHG.1231 is in many ways morphologically similar to Selmasaurus and was recovered herein as nested within this genus, we have concluded that it is sufficiently different in various aspects to warrant the establishment of a new genus.

Furthermore, both Goronyosaurus and Selmasaurus are known from only a few specimens, a level of incompleteness which makes assessing possible synapomorphies with MHNM.KHG.1231 difficult (see below). As more specimens of these taxa are discovered and more of their anatomy becomes known, it is quite possible that the exact phylogenetic relationships hypothesized herein for this clade may change. Specifically, although MHNM.KHG.1231 nests within Selmasaurus in the current analysis (Fig. 10), future phylogenies may recover a different topology. This is especially likely given the overall uncertainty surrounding mosasaur phylogenetics (e.g. Simões et al. 2017a); indeed, even across our analyses, different methodological approaches produced disparate phylogenies, including within Plioplatecarpinae (Fig. 10; , Figs S5–7). Therefore, although our assignment of MHNM.KHG.1231 to a new genus renders Selmasaurus paraphyletic under the current phylogeny (Fig. 10), we do not consider this a major concern, nor a valid impetus for modifying the taxonomic status of MHNM.KHG.1231 or of either existing Selmasaurus species.

Goronyosaurus was recovered as basal to Selmasaurus within this GoronyosaurusSelmasaurusGavialimimus clade; however, it does share certain features with Gavialimimus almaghribensis, including highly retracted external nares, a distinctly elongated snout, deep interdental pits, and a high number of premaxillary foramina (C.S., pers. obs.; see also Azzaroli et al. 1972; Soliar 1988; Lingham-Soliar 1991). These features are unknown in S. russelli, as the related elements are not preserved in specimens of this taxon. However, these elements are preserved in S. johnsoni, revealing the aforementioned features to be absent in this latter species. This is potentially indicative of a closer relationship between Goronyosaurus and Gavialimimus than recovered in this phylogenetic analysis.

However, this possible affiliation remains equivocal. For example, interdental pits were described by Lingham-Soliar (1991) as the defining characteristic of the genus Goronyosaurus, but have since been noted to varying extents in several mosasaurines and were listed by LeBlanc et al. (2012) as a diagnostic feature of the Mosasaurini; as such, this characteristic by itself does not inherently necessitate referral of Gavialimimus as a close relative of Goronyosaurus. Furthermore, although snout elongation and external narial retraction are diagnostic of both Goronyosaurus and Gavialimimus, they are not necessarily synapomorphic for these taxa. Both of these features result from the modification of several constituent elements, such as the premaxilla, maxillae, vomers, palatines, frontals and prefrontals. As such, although the gross morphology of these taxa may be similar, the underlying components and modifications giving rise to that morphology may not be homologous (Caldwell et al. 1995; Caldwell 2012; Simões et al. 2017b). As a hypothetical example, snout elongation could incorporate elongation of the vomers in one taxon but elongation of the vomerine processes of the palatines in the other taxon. In order to properly assess the phylogenetic significance of this similarity, these overall conditions cannot be treated in terms of gross morphology; rather, the homology of each separate component of these integrated structures must be assessed independently (Caldwell et al. 1995; Caldwell 2012; Simões et al. 2017b). However, the poor quality of preservation of the only known articulated Goronyosaurus skull (IGF 14750) severely hinders our ability to make such precise anatomical comparisons, leaving the potential homology of these conditions equivocal.

Indeed, the interrelationships among Goronyosaurus, Selmasaurus and Gavialimimus almaghribensis are further complicated by the fragmentary nature of existing Selmasaurus material. For example, because snout elongation and narial retraction are not preserved in the holotype of S. russelli, it is difficult to determine whether these features are unique to Goronyosaurus and Gavialimimus only or are more broadly characteristic of this overall clade. Of note, however, the degree of narial retraction can be assessed in the holotype of S. johnsoni, whose naris begins very anteriorly above the level of the third maxillary tooth (T.K., pers. obs.). Based on this observation, at least the overall condition of narial retraction appears unique to Goronyosaurus and Gavialimimus based on currently available specimens; importantly, though, the above arguments regarding homology assessments for complex morphologies still apply to this condition, and the condition in S. russelli remains unknown. Similarly, because the posterior portion of the parietal of Goronyosaurus is not sufficiently preserved in any currently available specimens, it is uncertain whether the parietal constriction present in Goronyosaurus matches the condition described by character 18/20 above and uniting Gavialimimus and Selmasaurus; i.e. parietal constriction in general is a synapomorphy of the GoronyosaurusSelmasaurus–Gavialimimus clade, but the specific condition of the parietal as conveyed by our dataset is restricted to the latter two genera.

A final intriguing feature involves the ectopterygoid process of the pterygoid. As noted above (see Description, Pterygoid), this process has an unusually broad base in Selmasaurus johnsoni, Goronyosaurus nigeriensis, the holotype of Platecarpus somenensis and Gavialimimus almaghribensis. Furthermore, the latter three taxa possess a posterodistal projection on this process, which contacts the quadrate ramus of the pterygoid in G. nigeriensis but not the latter two taxa. The presence of this projection again suggests a possible closer relationship between Goronyosaurus and Gavialimimus than recovered herein, with the similarity of this feature in P. somenensis bringing yet another poorly-known taxon into the mix. However, a rigorous assessment of this condition would require detailed re-analysis of both the Goronyosaurus and P. somenensis holotype material, which falls outside the scope of this study. Furthermore, the Goronyosaurus type material is quite poorly preserved as previously noted, with much of the subsequently referred material possibly being chimaeric in nature.

Unfortunately, the aforementioned uncertainties regarding the anatomy and phylogeny of these taxa can only be remedied with the discovery of more complete specimens. In all, although Goronyosaurus and Gavialimimus almaghribensis do share several intriguing cranial similarities, these similarities do not provide conclusive evidence of a closer evolutionary relationship between these taxa than that recovered in this study’s phylogenetic analyses, and at this point do not necessitate referral of MHNM.KHG.1231 to the genus Goronyosaurus. However, the presence of such similarities – including some shared with the equally enigmatic Platecarpus somenensis – underlines the need for further research into the relationships and morphologies of these taxa.

Palaeoecology and functional morphology

Massare (1987) divided Mesozoic marine reptiles into seven gradational predatory guilds based on tooth morphology, emphasizing shape of the tooth apex, type of tooth wear, presence of carinae or cutting edges, and shape and size of the tooth crown. These guilds unite taxa based on primary prey preference, regardless of phylogenetic placement. The three ‘end member’ guilds are: ‘Pierce I’, consisting of taxa with long, slender, pointed teeth used to pierce soft-bodied prey items; ‘Cut’, containing taxa with robust, pointed, carinated teeth used to rip the flesh of large marine vertebrates; and ‘Crush’, consisting of taxa with robust, blunt teeth used to crush thick-shelled invertebrates (Massare 1987; Fig. 12). The other guilds, which bridge these end-point morphotypes, are ‘Pierce II’, ‘Smash’, ‘Crunch’ and ‘General’ (Massare 1987; Fig. 12).

Figure 12. Predatory guilds of Mesozoic marine reptiles based on tooth morphology (modified from Massare [1987] and Bardet et al. [2015] with permissions from Taylor & Francis and Elsevier, respectively). These marine predators – including Moroccan mosasaurs, indicated in italicized text – can be classified into eight gradational feeding guilds based on tooth size, shape, carinae and pattern of wear. These tooth morphoguilds reflect similar prey preference, regardless of phylogenetic position. Ideal ‘end member’ tooth morphologies are depicted at each apex of the diagram. Teeth assigned to ‘Platecarpusptychodon (morphologically indistinguishable from those of Gavialimimus almaghribensis, pictured in the diagram) were originally used to classify this organism as a member of the ‘Pierce II’–‘Cut’ guilds (Bardet et al. 2015). However, cranial adaptations of G. almaghribensis suggest ecological specialization as a highly adapted piscivore, thus supporting placement as a ‘Pierce I’–‘Pierce II’ predator.

In a palaeoecological analysis of niche partitioning within the Maastrichtian phosphates of Morocco, Bardet et al. (2015) modified this system, adding a new guild, ‘Crush-Cut’ (Fig. 12). These authors included ‘Platecarpusptychodon in their analysis (Bardet et al. 2015). Although we reject this species as a nomen dubium, the teeth used by Bardet et al. (2015) in their analysis are morphologically equivalent to those of Gavialimimus almaghribensis (see above and Fig. 11). As such, these authors’ conclusions regarding the ‘P.ptychodon morphotype can be extrapolated to Gavialimimus, as they should be also to the other taxa depicted in Figure 11. Though these authors assigned ‘P.ptychodon to a combination of the ‘Pierce II’ and ‘Cut’ guilds, reflecting a diet of cephalopods and fishes (Bardet et al. 2015), new information from the G. almaghribensis holotype allows refinement of this interpretation.

The narrow, highly elongate snout and interlocking teeth of Gavialimimus almaghribensis are convergent with the condition in gharials, a distinctive taxon of longirostrine crocodilians. In gharials, this morphotype reflects predation on rapidly moving fish, with the interlocking teeth entrapping highly agile prey and the narrow snout reducing drag and displacement of water as the head swings laterally and the jaws snap shut (McHenry et al. 2006; Pierce et al. 2008; Cuff & Rayfield 2013; McCurry et al. 2017). Several aspects of the cranial morphology of Gavialimimus also converge upon that of the derived mosasaurine Plotosaurus, such as the elongate snout (though this longirostry is far more pronounced in Gavialimimus), interlocking teeth, and reduced cranial kinesis, especially along extensively articulating frontoparietal and prefrontal–postorbitofrontal sutures (LeBlanc et al. 2013). A recent functional analysis of the skull of Plotosaurus interpreted these adaptations as evidence of predatory specialization on small, rapid prey, thus contributing toward niche partitioning in the Maastrichtian of California (LeBlanc et al. 2013). Finally, although the specific function of the fluted tooth morphology present in Gavialimimus remains uncertain, similar fluting occurs convergently in several other mosasaur taxa (Fig. 11) and in several piscivorous snakes, such as Acrochordus and homalopsids (Vaeth et al. 1985). The widespread presence of tooth fluting across these taxa suggests that this feature may represent an adaptation for piscivory. Altogether, these convergences with gharials and other piscivorous squamates therefore suggest similar piscivory specialized on rapid, evasive fish.

This diet would likely have been composed primarily of teleosts, which underwent an extensive radiation during the late Mesozoic enabled in large part by their more efficient and rapid swimming compared to earlier fishes (Schaeffer 1969; Massare 1988). The radiation of mosasaurs as a whole during the Late Cretaceous has been tied to their ability to predate upon teleosts more effectively than pursuit predators such as ichthyosaurs could (Massare 1988). The unique adaptations of Gavialimimus relative to other mosasaurs therefore suggest a scenario of predation-based niche partitioning, with this morphology reflecting a particularly specialized form of piscivory in Gavialimimus under this hypothesis.

In terms of feeding guilds, this interpretation of predatory behaviour in Gavialimimus almaghribensis adds detail to the findings of Bardet et al. (2015). Although these authors assigned the morphologically equivalent teeth of ‘Platecarpusptychodon to the ‘Pierce II’–‘Cut’ guilds, Massare (1987) describes the ‘Cut’ morphoguild as reflecting predation upon large marine vertebrates, a diet that would cause significant breakage and wear to the teeth (Massare 1987). This wear would logically occur along the entire tooth row, which is not the case in G. almaghribensis. Instead, this specimen displays very minimal tooth wear anteriorly on the tooth row (Fig. 6A, B), with this wear only becoming prominent on the posterior teeth (Figs 4B, C, 6C). This minimal anterior tooth wear is more consistent with the ‘Pierce I’ or ‘Pierce II’ feeding guilds. Rather than being an indication of specific diet, the progressive increase in wear posteriorly is largely an expected consequence of adductor muscle attachment on the posterior mandible and resultant distribution of force along the jaws while biting (e.g. Callison 1967; Russell 1967). Extensive tooth replacement would also be necessary to cope with the considerable breakage described by Massare (1987) for members of the ‘Cut’ guild; in G. almaghribensis, however, tooth replacement is quite minimal, with replacement teeth visible at only three positions on the maxillae and dentaries (Figs 3, 6, 7). Finally, although the teeth of G. almaghribensis do possess carinae, which Massare (1987) uses to characterize the ‘Cut’ guild, these cutting edges could reasonably have aided in capturing and killing evasive prey and therefore are not automatically indicative of membership in the ‘Cut’ guild.

Re-interpretation of cranial features also present in Goronyosaurus further supports removal from the ‘Cut’ guild. In a biomechanical analysis of the skull of Goronyosaurus, Lingham-Soliar (1999) discussed two possible feeding methods for this taxon based on cranial morphology: predation upon small and evasive organisms, or predation upon larger prey items involving the need for processing prior to ingestion. The prognathous condition of the dentary teeth was used as evidence for the former mode, though the majority of cranial adaptations were interpreted as support for the latter. Lingham-Soliar (1999) concluded that this taxon’s deep interdental pits and lack of cranial kinesis would have strengthened the skull and enabled tight interlocking of the jaws. Combined with the elongate snout and extensive area for muscle attachment in the posterior skull, these features were interpreted as adaptations for predation on large aquatic vertebrates, comparable to the ‘Cut’ guild (Massare 1987; Lingham-Soliar 1999). These adaptations also occur in Gavialimimus almaghribensis, such as the large supratemporal fenestrae which would have accommodated massive jaw adductor musculature (Figs 2, 3), as well as the interlocking frontoparietal suture (Figs 2, 3, 5) which would have prevented or minimized movement along this mesokinetic axis (Wright & Shannon 1988; Lingham-Soliar 1999; LeBlanc et al. 2013). However, the strong convergence between G. almaghribensis and extant gharials suggests re-interpretation of these adaptations.

Instead of being used to predate upon large vertebrates, these modifications for powerful closure and interlocking of the jaws are also consistent with a predation style reliant upon rapidly closing the jaws and grasping evasive prey items (see also LeBlanc et al. 2013). As suggested by Lingham-Soliar (1999) for Goronyosaurus, the recurved and slightly prognathous anterior teeth of Gavialimimus could have contributed to this specialized piscivory by directing prey into the mouth (Massare 1987; Lingham-Soliar 1999), where the posterior marginal teeth could then grasp this prey. Finally, as noted above, a similar connection between cranial akinesis and predatory specialization upon small, evasive prey is hypothesized to occur in the long-snouted mosasaurine Plotosaurus (LeBlanc et al. 2013). These adaptations in G. almaghribensis are therefore consistent with a gharial-like specialization on smaller, highly agile aquatic prey.

Altogether, the cranial adaptations and dental features of Gavialimimus almaghribensis reflect membership in the ‘Pierce II’ guild, closer to the ‘Pierce I’ apex than the ‘Cut’ apex (Fig. 12). Although this places G. almaghribensis in a similar position to Halisaurus arambourgi, the preorbital region of the skull of G. almaghribensis represents almost two-thirds of the total skull length (Figs 2, 3), compared to only 50% in H. arambourgi (Bardet et al. 2015). As well, H. arambourgi is a halisaurine, for which Konishi et al. (2016) reported well-developed binocular vision possibly linked to nocturnal foraging upon small, bioluminescent animals such as 10-armed cephalopods. Unlike in halisaurines, the orbits of G. almaghribensis are neither forward facing nor large, therefore not conducive to binocular vision. Thus, based on its extensive snout elongation, alongside its other cranial adaptations, G. almaghribensis altogether exhibits increased specialization for piscivory, most likely during the daytime. This proposed ecological niche specialization would make G. almaghribensis unique among Moroccan mosasaurs, if not mosasaurs as a whole. Under this hypothesis, these adaptations would therefore have allowed G. almaghribensis to exploit prey items with reduced or minimal competition from sympatric mosasaur species, enabling its survival in such a diverse ecosystem (Fig. 13).

Figure 13. Life reconstruction of Gavialimimus almaghribensis, gen. et sp. nov., hunting a school of teleosts. Image credit: Tatsuya Shinmura.

Narial retraction

Pronounced retraction of the openings for the external nares is listed herein as a diagnostic feature of Gavialimimus almaghribensis. We interpret this condition as ‘true’ narial retraction, in contrast to the traditional concept of ‘posteriorly retracted external nares’ as a synapomorphy linking mosasauroids and varanoids (e.g. Rieppel et al. 2007; Conrad 2008; Conrad et al. 2008, 2011). This distinction is important, as this traditional view of ‘narial retraction’ is pervasive in the squamate literature (e.g. the global squamate dataset of Conrad 2008, and all phylogenies derived therefrom) and has long played a key role in shaping perceptions of mosasaur evolution and phylogenetics (Caldwell 2012). However, as discussed below, it is deeply flawed as a primary homology concept (see also Caldwell et al. 1995; Caldwell 2012). This clarification is particularly timely in light of recent discussions of character construction (e.g. Simões et al. 2017b) and applications of these guidelines in revising morphological phylogenetic datasets (e.g. Simões et al. 2018; Garberoglio et al. 2019). As a longirostrine organism exemplifying ‘true’ narial retraction, Gavialimimus therefore provides novel insight into and clarification of this anatomical condition.

The hypothesized homology of the openings for the external nares in varanoids with those in mosasauroids is flawed in numerous aspects. Although the overall fenestration of the snout may be superficially similar, this condition is achieved in these two groups via different modifications to the shape and topology of the surrounding bones (Caldwell et al. 1995; see Caldwell 2012 for a detailed discussion of this issue). Employing the test of similarity for primary homology (Rieppel & Kearney 2002), posterior expansion of the narial fenestrae in varanoids and mosasauroids is therefore not homologous and does not unite these taxa. Furthermore, the external nares themselves in varanoids are not in fact retracted compared to the condition in other lizards. Instead, the soft-tissue external narial openings remain at an anteriorly terminal position in varanoids (as in other lizards), with only the bony fenestrae being posteriorly extended (Caldwell 2012). Although not terminal, slit-like soft-tissue external nares also occur anteriorly inside the anterior-most, expanded section of the bony fenestrae in Platecarpus tympaniticus, a condition that again does not exemplify retraction of the nares and which could have been widespread among all mosasaurs except those few with ‘true’ narial retraction (Lindgren et al. 2010; Caldwell 2012).

In contrast to this erroneous conceptualization of the external narial openings in typical varanoids and mosasauroids, the condition in Gavialimimus almaghribensis is distinct in portraying true retraction of the openings for the external nares relative to the tip of the snout. The bony snout fenestrae terminate anteriorly around the eighth maxillary tooth position, limiting this fenestration to only the posterior-most 25% of the preorbital region of the skull (Fig. 2). The only non-plioplatecarpine taxa that potentially exhibit an analogous narial retraction would be halisaurines, in which the premaxillary–maxillary suture is longer than the length of the entire external naris, which is hence closer to the frontal than to the snout tip (e.g. Konishi et al. 2016, fig. 18).

In comparison, in other mosasaurs the bony openings for the external nares either comprise a much greater proportion of the preorbital skull region and/or terminate anteriorly much closer to the snout (e.g. Platecarpus tympaniticus [Konishi et al. 2012, fig. 3]; Prognathodon overtoni [Williston, 1897b], see Konishi et al. [2011, fig. 2]; Tylosaurus bernardi [Dollo, 1885], see Jiménez-Huidobro & Caldwell [2016, fig. 1]). Even in other mosasaurs exhibiting snout elongation, the bony openings for the external nares are distinct from the condition in Gavialimimus almaghribensis. For example, these openings extend over most of the preorbital region of the skull in Plotosaurus bennisoni (e.g. LeBlanc et al. 2013, fig. 2). In Ectenosaurus clidastoides – the mosasaur most comparable to Gavialimimus in terms of pronounced snout elongation – these openings are small, similar to the condition in Gavialimimus, but are positioned much closer to the tip of the snout (e.g. Russell 1967, fig. 86). Finally, although the external narial openings appear to be small and retracted in Goronyosaurus nigeriensis, as in Gavialimimus, the poor condition of the G. nigeriensis holotype skull somewhat obscures this comparison (C.S., pers. obs.; see also Azzaroli et al. 1972; Soliar 1988).

Therefore, even if the soft-tissue narial openings were located at the anterior termini of the bony fenestrae in Gavialimimus almaghribensis, as in varanoids, the fenestrae themselves are retracted enough that this position would still be distinctly posterior relative to the typical condition in varanoids and presumably mosasaurs, save halisaurines. As such, Gavialimimus provides a rare example of true retraction of the external nares, in contrast to the traditional but inaccurate conceptualization of external narial retraction in mosasaurs and varanoids.

Conclusions

The Maastrichtian phosphates of Morocco record an incredibly diverse assemblage of numerous mosasaur genera from many subfamilies (Bardet et al. 2015, 2018). This study builds on our knowledge of this diversity via the description and analysis of a new mosasaur specimen (MHNM.KHG.1231) from Morocco, consisting of relatively complete cranial and isolated postcranial material (Figs 2–9). Although the teeth of this specimen initially suggest referral to ‘Platecarpusptychodon, closer examination of this species and others exhibiting similar tooth morphology (Fig. 11) reveals the diagnosis of ‘P.ptychodon to be fundamentally flawed, leading to the rejection of this species as a nomen dubium.

Phylogenetic analysis under multiple parsimony-based methods consistently recovers MHNM.KHG.1231 as the sister taxon to Selmasaurus russelli, with these taxa forming a clade within the Plioplatecarpinae alongside S. johnsoni and Goronyosaurus nigeriensis (Fig. 10). Due to its unique skull – with unequivocal features such as a highly elongate snout, highly retracted nares, and large supratemporal fenestrae – we propose that MHNM.KHG.1231 represents a new mosasaur genus and species. Features uniting this specimen with Selmasaurus include the nature of the parietal constriction and the broadly excavated medial surface of the quadrate suprastapedial process, with general constriction of the parietal also present in G. nigeriensis. This new genus is distinguished from Selmasaurus by a large, oval pineal foramen that contacts the frontoparietal suture, certain features of the dentary and vertebrae, and a wide postorbitofrontal, among other features such as the aforementioned autapomorphies.

Both Selmasaurus and Goronyosaurus have only rarely been included in mosasaur or squamate phylogenies, and never in the same analysis in the literature. As such, recovery of the aforementioned GoronyosaurusSelmasaurusGavialimimus clade provides novel insight into plioplatecarpine evolution. Cranial features present in this clade – including snout elongation and akinetic cranial sutures – suggest morphological specialization into distinct ecological and predatory niches to reduce competition and enable coexistence in diverse ecosystems.

The elongate snout of Gavialimimus almaghribensis also provides insight into ‘true’ external narial retraction in mosasaurs, in contrast to how ‘narial retraction’ is typically but incorrectly interpreted in many studies of mosasaur and varanoid evolution.

Supplemental material

Supplementary Material - S4 MosaMatrix Contingent

Download (18 KB)

Supplementary Material - S3 MosaMatrix Multistate

Download (17 KB)

Supplementary Material

Download MS Word (2158 KB)

Acknowledgements

We thank: A. Murray for discussions regarding taxonomic guidelines; A. LeBlanc, I. Paparella and O. Vernygora for helpful discussions regarding phylogenetic analysis and mosasaur anatomy; A. LeBlanc for discussions regarding tooth morphology and histology; and O. Vernygora for assistance in identifying isolated shark teeth used to support the provenance of the holotype. We are especially grateful to L. A. Lindoe for his guidance in the preparation of MHNM.KHG.1231. We also thank Drs N.-E. Jalil and M. Ghamizi (Museum of Natural History of Marrakech, Cadi Ayyad University, Marrakech) for their assistance with the specimen. Finally, we thank J. Lively and H. Street for their helpful and thorough reviews, both of which greatly improved the quality of this manuscript. Funding was provided via an Alexander Graham Bell Canada Graduate Scholarship awarded by the Natural Sciences and Engineering Research Council of Canada (NSERC) to CRCS, as well as an NSERC Discovery Grant (#23458) and Chair’s Research Allowance to MWC. Figures 1 and 9 were modified with permission from the respective copyright holders. These include: figure 1 (map and stratigraphic column of Oulad Abdoun Basin) of Bardet et al. (2005), included herein by permission of Oxford University Press; figure 15 (tooth morphoguilds of Mesozoic marine predators) of Massare (1987), adapted herein by permission of the Society of Vertebrate Paleontology, , and of the publisher, Taylor & Francis; and figure 7 (tooth morphoguilds and niche partitioning in Maastrichtian Moroccan mosasaurs) of Bardet et al. (2015), adapted herein with permission from Elsevier. All authors report no conflicts of interest.

Supplementary material

Supplementary material for this article can be accessed here: .

Associate Editor: Jennifer Olori

    References

  • Agassiz, L. 1843. Recherches sur les poisson fossiles. Volume 3 . Imprimerie de Petitpierre, Neuchâtel, 332 pp. 
  • Arambourg, C. 1952. Les vertébrés fossiles des gisements de phosphates (Maroc–Algérie–Tunisie). Notes et mémoires du Service géologique du Maroc , 92 , 1372. 
  • Azzaroli, A. , de Giuli, C. , Ficcarelli, G. & Torre, D. 1972. An aberrant mosasaur from the Upper Cretaceous of north-western Nigeria. Accademia Nazionale dei Lincei, Rendiconti. Classe di scienze fisiche matematiche e naturali , 3 , 398404. 
  • Azzaroli, A. , de Giuli, C. , Ficcarelli, G. & Torre, D. 1975. Late Cretaceous mosasaurs from the Sokoto district, Nigeria. Accademia Nazionale dei Lincei, Memorie. Classe di scienze fisiche matematiche e naturali , 13 , 2134. 
  • Bardet, N. , Cappetta, H. , Pereda Superbiola, X. , Mouty, M. , Al Maleh, A. K. , Ahmad, A. M. , Khrata, O. & Gannoum, N. 2000. The marine vertebrate faunas from the Late Cretaceous phosphates of Syria. Geological Magazine , 137 (3), 269290. doi:10.1017/S0016756800003988  [Crossref], [Web of Science ®]
  • Bardet, N. , Pereda Suberbiola, X. , Iarochène, M. , Bouya, B. & Amaghzaz, M. 2005. A new species of Halisaurus from the Late Cretaceous phosphates of Morocco, and the phylogenetical relationships of the Halisaurinae (Squamata: Mosasauridae). Zoological Journal of the Linnean Society , 143 , 447472. [Crossref], [Web of Science ®]
  • Bardet, N. , Pereda Suberbiola, X. , Jouve, S. , Bourdon, E. , Vincent, P. , Houssaye, A. , Rage, J.-C. , Jalil, N.-E. , Bouya, B. & Amaghzaz, M. 2010. Reptilian assemblages from the latest Cretaceous–Palaeogene phosphates of Morocco: from Arambourg to present time. Historical Biology , 22 , 186199. doi:10.1080/08912961003754945  [Taylor & Francis Online], [Web of Science ®]
  • Bardet, N. , Houssaye, A. , Vincent, P. , Pereda Suberbiola, X. , Amaghzaz, M. , Jourani, E. & Meslouh, S. 2015. Mosasaurids (Squamata) from the Maastrichtian Phosphates of Morocco: Biodiversity, palaeobiogeography and palaeoecology based on tooth morphoguilds. Gondwana Research , 27 , 10681078. doi:10.1016/j.gr.2014.08.014  [Crossref], [Web of Science ®]
  • Bardet, N. , Gheerbrant, E. , Noubhani, A. , Cappetta, H. , Jouve, S. , Bourdon, E. , Pereda Suberbiola, X. , Jalil, N.-E. , Vincent, P. , Houssaye, A. , Sole, F. , Elhoussaini Darif, K. , Adnet, S. , Rage, J.-C. , de Lapparent de Broin, F. , Sudre, J. , Bouya, B. , Amaghzaz, M. & Meslouh, S. 2018. Les vertébrés fossiles des phosphates crétacés-paléogènes (72,1–47,8 Ma) du Maroc. Pp. 351452 in S.Zouhri (ed) Paléontologie des Vertébrés du Maroc: état des connaissances . Mémoires de la Société Géologique de France, Paris. 
  • Bell, G. L. 1997. A phylogenetic revision of North American and Adriatic Mosasauroidea. Pp. 293332 in J. M.Callaway & E. L.Nicholls (eds) Ancient marine reptiles . Academic Press, San Diego. [Crossref]
  • Caldwell, M. W. 2007. Ontogeny, anatomy and attachment of the dentition in mosasaurs (Mosasauridae: Squamata). Zoological Journal of the Linnean Society , 149 , 687700. [Crossref], [Web of Science ®]
  • Caldwell, M. W. 2012. A challenge to categories: “What, if anything, is a mosasaur?”. Bulletin de la Société Géologique de France , 183 (1), 734. [Crossref]
  • Caldwell, M. W. , Carroll, R. L. & Kaiser, H. 1995. The pectoral girdle and forelimb of Carsosaurus marchesetti (Aigialosauridae), with a preliminary phylogenetic analysis of mosasauroids and varanoids. Journal of Vertebrate Paleontology , 15 (3), 516531. doi:10.1080/02724634.1995.10011245  [Taylor & Francis Online], [Web of Science ®]
  • Caldwell, M. W. , Konishi, T. , Obata, I. & Muramoto, K. 2008. A new species of Taniwhasaurus (Mosasauridae, Tylosaurinae) from the upper Santonian–lower Campanian (Upper Cretaceous) of Hokkaido, Japan. Journal of Vertebrate Paleontology , 28 (2), 339348. doi:10.1671/0272-4634(2008)28  [Taylor & Francis Online], [Web of Science ®]
  • Callison, G. 1967. Intracranial mobility in Kansas mosasaurs. University of Kansas Paleontological Contributions , 26 , 115. 
  • Camp, C. L. 1942. California mosasaurs. Memoirs of the University of California , 13 , 168. 
  • Conrad, J. L. 2008. Phylogeny and systematics of Squamata (Reptilia) based on morphology. Bulletin of the American Museum of Natural History , 310 , 1182. doi:10.1206/310.1  [Crossref], [Web of Science ®]
  • Conrad, J. L. , Rieppel, O. & Grande, L. 2008. Re-assessment of varanid evolution based on new data from Saniwa ensidens Leidy, 1870 (Squamata, Reptilia). American Museum Novitates , 3630 , 115. doi:10.1206/596.1  [Crossref]
  • Conrad, J. L. , Ast, J. C. , Montanari, S. & Norell, M. A. 2011. A combined evidence phylogenetic analysis of Anguimorpha (Reptilia: Squamata). Cladistics , 27 , 230277. doi:10.1111/j.1096-0031.2010.00330.x  [Crossref], [Web of Science ®]
  • Coombs, W. P. 1990. Teeth and taxonomy in ankylosaurs. Pp. 269279 in K.Carpenter & P. J.Currie (eds) Dinosaur systematics: approaches and perspectives . Cambridge University Press, New York. [Crossref]
  • Cope, E. D. 1869. On the reptilian orders, Pythonomorpha and Streptosauria. Boston Society of Natural History Proceedings , 12 , 250266. 
  • Cope, E. D. 1874. Review of the Vertebrata of the Cretaceous Period found west of the Mississippi River. United States Geological Survey of the Territories Bulletin , 1 , 348. 
  • Cuff, A. R. & Rayfield, E. J. 2013. Feeding mechanics in spinosaurid theropods and extant crocodilians. PLoS ONE , 8 (5), e65295. doi:10.1371/journal.pone.0065295  [Crossref], [PubMed], [Web of Science ®]
  • Dollo, L. 1884. Le mosasaure. Revue des Questions Scientifiques , 16 , 648653. 
  • Dollo, L. 1885. Le hainosaure. Révue des Questions Scientifiques , 18 , 285289. 
  • Dollo, L. 1889a. Note sur les vertébrés fossiles récemment offerts au Musée de Bruxelles par M. Alfred Lemonnier. Bulletin de la Société belge de Géologie, de Paléontologie et d’Hydrologie , 3 , 181182. 
  • Dollo, L. 1889b. Première note sur les mosasauriens de Mesvin. Bulletin de la Société Belge de Géologie, de Paléontologie, et d’Hydrologie , 3 , 271304. 
  • Everhart, M. J. 2005. Rapid evolution, diversification and distribution of mosasaurs (Reptilia; Squamata) prior to the K-T Boundary. Pp. 1627 in Tate Museum 2005 11th annual symposium in paleontology and geology . Tate Museum, Casper, WY. 
  • Fernandez, M. & Martin, J. E. 2009. Description and phylogenetic relationships of Taniwhasaurus antarcticus (Mosasauridae, Tylosaurinae) from the upper Campanian (Cretaceous) of Antarctica. Cretaceous Research , 30 , 717726. doi:10.1016/j.cretres.2008.12.012  [Crossref], [Web of Science ®]
  • Garberoglio, F. F. , Apesteguía, S. , Simões, T. R. , Palci, A. , Gómez, R. O. , Nydam, R. L. , Larsson, H. C. E. , Lee, M. S. Y. & Caldwell, M. W. 2019. New skulls and skeletons of the Cretaceous legged snake Najash, and the evolution of the modern snake body plan. Science Advances , 5 (11), eaax5833. doi:10.1126/sciadv.aax5833  [Crossref], [PubMed], [Web of Science ®]
  • Gervais, P. 1852. Zoologie et paléontologie Françaises (animaux vertébrés) . First Edition. Libraire de la Société de Géographie, Paris, 271 pp. 
  • Goloboff, P. A. 1993. Estimating character weights during tree search. Cladistics , 9 , 8391. doi:10.1111/j.1096-0031.1993.tb00209.x  [Crossref], [Web of Science ®]
  • Goloboff, P. A. , Carpenter, J. M. , Arias, J. S. & Esquivel, D. R. M. 2008a. Weighting against homoplasy improves phylogenetic analysis of morphological data sets. Cladistics , 24 , 758773. doi:10.1111/j.1096-0031.2008.00209.x  [Crossref], [Web of Science ®]
  • Goloboff, P. A. , Farris, J. S. & Nixon, K. C. 2008b. TNT, a free program for phylogenetic analysis. Cladistics , 24 , 774786. doi:10.1111/j.1096-0031.2008.00217.x  [Crossref], [Web of Science ®]
  • Goloboff, P. A. & Catalano, S. A. 2016. TNT version 1.5, including a full implementation of phylogenetic morphometrics. Cladistics , 32 , 221238. doi:10.1111/cla.12160  [Crossref], [Web of Science ®]
  • Holmes, R. 1996. Plioplatecarpus primaevus (Mosasauridae) from the Bearpaw Formation (Campanian, Upper Cretaceous) of the North American Western Interior Seaway. Journal of Vertebrate Paleontology , 16 (4), 673687. doi:10.1080/02724634.1996.10011357  [Taylor & Francis Online], [Web of Science ®]
  • International Commission on Zoological Nomenclature 1999. International code of zoological nomenclature . Fourth Edition. The International Trust for Zoological Nomenclature, London, 70 pp. ( 75.5), 79–98 (Glossary). [Crossref]
  • Jiménez-Huidobro, P. & Caldwell, M. W. 2016. Reassessment and reassignment of the early Maastrichtian mosasaur Hainosaurus bernardi Dollo, 1885, to Tylosaurus Marsh, 1872. Journal of Vertebrate Paleontology , 36 (3), e1096275. doi:10.1080/02724634.2016.1096275  [Taylor & Francis Online], [Web of Science ®]
  • Jiménez-Huidobro, P. , Simões, T. R. & Caldwell, M. W. 2016. Re-characterization of Tylosaurus nepaeolicus (Cope, 1874) and Tylosaurus kansasensis Everhart, 2005: Ontogeny or sympatry? Cretaceous Research , 65 , 6881. doi:10.1016/j.cretres.2016.04.008  [Crossref], [Web of Science ®]
  • Jiménez-Huidobro, P. & Caldwell, M. W. 2019. A new hypothesis of the phylogenetic relationships of the Tylosaurinae (Squamata: Mosasauroidea). Frontiers in Earth Science , 7 , 47. doi:10.3389/feart.2019.00047  [Crossref], [Web of Science ®]
  • Kiernan, C. R. 2002. Stratigraphic distribution and habitat segregation of mosasaurs in the Upper Cretaceous of western and central Alabama, with an historical review of Alabama mosasaur discoveries. Journal of Vertebrate Paleontology , 22 (1), 91103. doi:10.1671/0272-4634(2002)022  [Taylor & Francis Online], [Web of Science ®]
  • Konishi, T. 2008. A new specimen of Selmasaurus sp., cf. S. russelli (Mosasauridae: Plioplatecarpini) from Greene County, western Alabama, USA. Pp. 95105 in M. J. Everhart (ed) Proceedings of the second mosasaur meeting . Hays, Kansas. 
  • Konishi, T. & Caldwell, M. W. 2007. New specimens of Platecarpus planifrons (Cope, 1874) (Squamata: Mosasauridae) and a revised taxonomy of the genus. Journal of Vertebrate Paleontology , 27 (1), 5972. doi:10.1671/0272-4634(2007)27  [Taylor & Francis Online], [Web of Science ®]
  • Konishi, T. & Caldwell, M. W. 2009. New material of the mosasaur Plioplatecarpus nichollsae Cuthbertson et al., 2007, clarifies problematic features of the holotype specimen. Journal of Vertebrate Paleontology , 29 (2), 417436. doi:10.1671/039.029.0225  [Taylor & Francis Online], [Web of Science ®]
  • Konishi, T. , Brinkman, D. , Massare, J. A. & Caldwell, M. W. 2011. New exceptional specimens of Prognathodon overtoni (Squamata, Mosasauridae) from the upper Campanian of Alberta, Canada, and the systematics and ecology of the genus. Journal of Vertebrate Paleontology , 31 (5), 10261046. doi:10.1080/02724634.2011.601714  [Taylor & Francis Online], [Web of Science ®]
  • Konishi, T. & Caldwell, M. W. 2011. Two new plioplatecarpine (Squamata, Mosasauridae) genera from the Upper Cretaceous of North America, and a global phylogenetic analysis of plioplatecarpines. Journal of Vertebrate Paleontology , 31 (4), 754783. doi:10.1080/02724634.2011.579023  [Taylor & Francis Online], [Web of Science ®]
  • Konishi, T. , Lindgren, J. , Caldwell, M. W. & Chiappe, L. 2012. Platecarpus tympaniticus (Squamata, Mosasauridae): osteology of an exceptionally preserved specimen and its insights into the acquisition of a streamlined body shape in mosasaurs. Journal of Vertebrate Paleontology , 32 (6), 13131327. doi:10.1080/02724634.2012.699811  [Taylor & Francis Online], [Web of Science ®]
  • Konishi, T. , Newbrey, M. G. & Caldwell, M. W. 2014. A small, exquisitely preserved specimen of Mosasaurus missouriensis (Squamata, Mosasauridae) from the upper Campanian of the Bearpaw Formation, western Canada, and the first stomach contents for the genus. Journal of Vertebrate Paleontology , 34 (4), 802819. doi:10.1080/02724634.2014.838573  [Taylor & Francis Online], [Web of Science ®]
  • Konishi, T. , Caldwell, M. W. , Nishimura, T. , Sakurai, K. & Tanoue, K. 2016. A new halisaurine mosasaur (Squamata: Halisaurinae) from Japan: the first record in the western Pacific realm and the first documented insights into binocular vision in mosasaurs. Journal of Systematic Palaeontology , 14 (10), 809839. doi:10.1080/14772019.2015.1113447  [Taylor & Francis Online], [Web of Science ®]
  • LeBlanc, A. R. H. , Caldwell, M. W. & Bardet, N. 2012. A new mosasaurine from the Maastrichtian (Upper Cretaceous) phosphates of Morocco and its implications for mosasaurine systematics. Journal of Vertebrate Paleontology , 32 (1), 82104. doi:10.1080/02724634.2012.624145  [Taylor & Francis Online], [Web of Science ®]
  • LeBlanc, A. R. H. , Caldwell, M. W. & Lindgren, J. 2013. Aquatic adaptation, cranial kinesis, and the skull of the mosasaurine mosasaur Plotosaurus bennisoni . Journal of Vertebrate Paleontology , 33 (2), 349362. doi:10.1080/02724634.2013.726675  [Taylor & Francis Online], [Web of Science ®]
  • LeBlanc, A. R. H. , Mohr, S. R. & Caldwell, M. W. 2019. Insights into the anatomy and functional morphology of durophagous mosasaurines (Squamata: Mosasauridae) from a new species of Globidens from Morocco. Zoological Journal of the Linnean Society , 10 , 127. doi:10.1093/zoolinnean/zlz008/540165  [Crossref]
  • Lindgren, J. & Siverson, M. 2002. Tylosaurus ivoensis: a giant mosasaur from the early Campanian of Sweden. Transactions of the Royal Society of Edinburgh: Earth Sciences , 93 , 7393. doi:10.1017/S026359330000033X  [Crossref]
  • Lindgren, J. , Caldwell, M. W. , Konishi, T. & Chiappe, L. M. 2010. Convergent evolution in aquatic tetrapods: insights from an exceptional fossil mosasaur. PLoS ONE , 5 (8), e11998. doi:10.1371/journal.pone.0011998  [Crossref], [PubMed], [Web of Science ®]
  • Lingham-Soliar, T. 1991. Mosasaurs from the Upper Cretaceous of Niger. Palaeontology , 34 , 653670. [Web of Science ®]
  • Lingham-Soliar, T. 1998. A new mosasaur Pluridens walkeri from the Upper Cretaceous, Maastrichtian of the Iullemmeden Basin, southwest Niger. Journal of Vertebrate Paleontology , 18 (4), 709717. [Taylor & Francis Online], [Web of Science ®]
  • Lingham-Soliar, T. 1999. A functional analysis of the skull of Goronyosaurus nigeriensis (Squamata: Mosasauridae) and its bearing on the predatory behaviour and evolution of this enigmatic taxon. Neues Jahrbuch für Geologie und Paläontologie, Abhandlungen , 213 (3), 355374. [Crossref], [Web of Science ®]
  • Lingham-Soliar, T. 2002. First occurrence of premaxillary caniniform teeth in the Varanoidea: presence in the extinct mosasaur Goronyosaurus (Squamata: Mosasauridae) and its functional and paleoecological implications. Lethaia , 35 , 187190. [Crossref], [Web of Science ®]
  • Linnaeus, C. 1758. Systema Naturae , Volume 1. (Systema naturae per regna tria naturae, secundum classes, ordines, genera, species, cum characteribus, differentiis, synonymis, locis. Tomus I. Editio decima, reformata) . Tenth edition . Laurentii Salvii, Holmiae, 824 pp. 
  • Liu, K. 2007. Sequence stratigraphy and orbital cyclostratigraphy of the Mooreville Chalk (Santonian–Campanian), northeastern Gulf of Mexico area, USA. Cretaceous Research , 28 , 405418. doi:10.1016/j.cretres.2006.06.005  [Crossref], [Web of Science ®]
  • Lucas, J. & Prévôt-Lucas, L. 1996. Tethyan phosphates and bioproductites. Pp. 367391 in A. E. M.Nairn, L.-E.Ricou, B.Vrielynck & J.Dercourt (eds) The ocean basins and margins: the Tethys Ocean . Springer, Boston. 
  • Maddison, W. P. & Maddison, D. R. 2018. Mesquite: a modular system for evolutionary analysis. Version 3.51.  
  • Massare, J. A. 1987. Tooth morphology and prey preference of Mesozoic marine reptiles. Journal of Vertebrate Paleontology , 7 (2), 121137. doi:10.1080/02724634.1987.10011647  [Taylor & Francis Online]
  • Massare, J. A. 1988. Swimming capabilities of Mesozoic marine reptiles: implications for method of predation. Paleobiology , 14 (2), 187205. [Crossref], [Web of Science ®]
  • McCurry, M. R ., Evans, A. R. , Fitzgerald, E. M. G. , Adams, J. W. , Clausen, P. D. & McHenry, C. R. 2017. The remarkable convergence of skull shape in crocodilians and toothed whales. Proceedings of the Royal Society B , 284 , 20162348. doi:10.1098/rspb2016.2348  [Crossref]
  • McHenry, C. R. , Clausen, P. D. , Daniel, W. J. T. , Meers, M. B. & Pendharkar, A. 2006. Biomechanics of the rostrum in crocodilians: a comparative analysis using finite-element modeling. The Anatomical Record , 288 , 827849. doi:10.1002/ar.a.20360  [Crossref], [Web of Science ®]
  • Merriam, J. C. 1894. Ueber die Pythonomorphen der Kansas Kreide. Palaeontographica , 41 , 139. 
  • Modesto, S. P. 1996. Noteosaurus africanus Broom is a nomen dubium. Journal of Vertebrate Paleontology , 16 (1), 172174. doi:10.1080/02724634.1996.10011296  [Taylor & Francis Online], [Web of Science ®]
  • Novas, F. E. , Fernández, M. , de Gasparini, Z. , Lirio, J. M. , Nuñez, H. J. & Puerta, P. 2002. Lakumasaurus antarcticus, n. gen. et sp., a new mosasaur (Reptilia, Squamata) from the Upper Cretaceous of Antarctica. Ameghiniana , 39 , 245249. [Web of Science ®]
  • Ogg, J. G. , Hinnov, L. A. & Huang, C. 2012. Cretaceous. Pp. 793853 in F. M.Gradstein, J. G.Ogg, M. B.Schmitz & G. M.Ogg (eds) The geologic timescale 2012 . Elsevier, Amsterdam. [Crossref]
  • Oppel, M. 1811. Die Ordungen Familien, und Gattungen der Reptilien als Prodrom einer Naturgeschichte derselben . J. Lindauer, München, 86 pp. [Crossref]
  • Palci, A. , Caldwell, M. W. & Papazzoni, C. A. 2013. A new genus and subfamily of mosasaurs from the Upper Cretaceous of northern Italy. Journal of Vertebrate Paleontology , 33 (3), 599612. doi:10.1080/02724634.2013.731024  [Taylor & Francis Online], [Web of Science ®]
  • Persson, P. O. 1963. Studies on Mesozoic marine reptile faunas with particular regard to the Plesiosauria. Publications from the Institutes of Mineralogy, Paleontology, and Quaternary Geology, University of Lund, Sweden , 118 , 115. 
  • Pierce, S. E. , Angielczyk, K. D. & Rayfield, E. J. 2008. Patterns of morphospace occupation and mechanical performance in extant crocodilian skulls: a combined geometric morphometric and finite element modeling approach. Journal of Morphology , 269 , 840864. doi:10.1002/jmor.10627  [Crossref], [PubMed], [Web of Science ®]
  • Polcyn, M. J. & Everhart, M. J. 2008. Description and phylogenetic analysis of a new species of Selmasaurus (Mosasauridae: Plioplatecarpinae) from the Niobrara Chalk of western Kansas. Pp. 1328 in M. J. Everhart (ed) Proceedings of the second mosasaur meeting . Hays, Kansas. 
  • Rieppel, O. & Kearney, M. 2002. Similarity. Biological Journal of the Linnean Society , 75 , 5982. [Crossref], [Web of Science ®]
  • Rieppel, O. , Conrad, J. L. & Maisano, J. A. 2007. New morphological data for Eosaniwa koehni Haubold, 1977 and a revised phylogenetic analysis. Journal of Paleontology , 81 (4), 760769. [Crossref], [Web of Science ®]
  • Russell, D. A. 1967. Systematics and morphology of American mosasaurs. Bulletin of the Peabody Museum of Natural History , 23 , 1241. 
  • Schaeffer, B. 1969. Adaptive radiation of the fishes and the fish-amphibian transition. Annals of the New York Academy of Sciences , 167 , 517. [Crossref], [Web of Science ®]
  • Seeley, H. G. 1881. On remains of a small lizard from Neocomian rocks of Comen, near Trieste, preserved in the Geological Museum of the University of Vienna. Quarterly Journal of the Geological Society of London , 37 , 5256. [Crossref]
  • Simões, T. R. , Vernygora, O. , Paparella, I. , Jimenez-Huidobro, P. & Caldwell, M. W. 2017a. Mosasauroid phylogeny under multiple phylogenetic methods provides new insights on the evolution of aquatic adaptations in the group. PLoS ONE , 12 (5), e0176773. doi:10.1371/journal.pone.0176773  [Crossref], [PubMed], [Web of Science ®]
  • Simões, T. R. , Caldwell, M. W. , Palci, A. & Nydam, R. L. 2017b. Giant taxon-character matrices: quality of character constructions remains critical regardless of size. Cladistics , 33 , 198219. doi:10.1111/cla.12163  [Crossref], [Web of Science ®]
  • Simões, T. R. , Caldwell, M. W. , Tałanda, M. , Bernardi, M. , Palci, A. , Vernygora, O. , Bernardini, F. , Mancini, L. & Nydam, R. L. 2018. The origin of squamates revealed by a Middle Triassic lizard from the Italian Alps. Nature , 557 , 706709. doi:10.1038/s41586-018-0093-3  [Crossref], [PubMed], [Web of Science ®]
  • Soliar, T. 1988. The mosasaur Goronyosaurus from the Upper Cretaceous of Sokoto state, Nigeria. Palaeontology , 31 , 747762. [Web of Science ®]
  • Swinton, W. E. 1930. On fossil Reptilia from Sokoto Province. Bulletin of the Geological Survey of Nigeria , 13 , 156. 
  • Thévenin, A. 1896. Mosasauriens de la craie grise de Vaux-Éclusier près Péronne (Somme). Bulletin de la Société Géologique de France , 24 , 900916. 
  • Vaeth, R. H. , Rossman, D. A. & Shoop, W. 1985. Observations of tooth surface morphology in snakes. Journal of Herpetology , 19 (1), 2026. [Crossref], [Web of Science ®]
  • Williston, S. W. 1897a. Range and distribution of the mosasaurs. Kansas University Quarterly , 6 , 177189. 
  • Williston, S. W. 1897b. Brachysaurus, a new genus of mosasaurs. Kansas University Quarterly , 6 , 9598. 
  • Wright, K. R. & Shannon, S. W. 1988. Selmasaurus russelli, a new plioplatecarpine mosasaur (Squamata, Mosasauridae) from Alabama. Journal of Vertebrate Paleontology , 8 (1), 102107. doi:10.1080/02724634.1988.10011686  [Taylor & Francis Online]