Neuroanatomy of the crocodylian Tomistoma dowsoni from the Miocene of North Africa provides insights into the evolutionary history of gavialoids

The interrelationships of the extant crocodylians Gavialis gangeticus and Tomistoma schlegelii have been historically disputed. Whereas molecular analyses indicate a sister taxon relationship between these two gavialoid species, morphological datasets typically place Gavialis as the outgroup to all other extant crocodylians. Recent morphological‐based phylogenetic analyses have begun to resolve this discrepancy, recovering Gavialis as the closest living relative of Tomistoma; however, several stratigraphically early fossil taxa are recovered as closer to Gavialis than Tomistoma, resulting in anomalously early divergence timings. As such, additional morphological data might be required to resolve these remaining discrepancies. ‘Tomistoma’ dowsoni is an extinct species of gavialoid from the Miocene of North Africa. Utilising CT scans of a near‐complete, referred skull, we reconstruct the neuroanatomy and neurosensory apparatus of ‘Tomistoma’ dowsoni. Based on qualitative and quantitative morphometric comparisons with other crocodyliforms, the neuroanatomy of ‘Tomistoma’ dowsoni is characterised by an intermediate morphology between the two extant gavialoids, more closely resembling Gavialis. This mirrors the results of recent studies based on the external anatomy of these three species and other fossil gavialoids. Several neuroanatomical features of these species appear to reflect ecological and/or phylogenetic signals. For example, the ‘simple’ morphology of their neurosensory apparatus is broadly similar to that of other long and narrow‐snouted (longirostrine), aquatic crocodyliforms. A dorsoventrally short, anteroposteriorly long endosseous labyrinth is also associated with longirostry. These features indicate that snout and skull morphology, which are themselves partly constrained by ecology, exert an influence on neuroanatomical morphology, as has also been recognised in birds and turtles. Conversely, the presence of a pterygoid bulla in Gavialis and several extinct gavialoids, and its absence in Tomistoma schlegelii, could be interpreted as a phylogenetic signal of crocodylians more closely related to Gavialis than to Tomistoma. Evaluation of additional fossil gavialoids will be needed to further test whether these and other neuroanatomical features primarily reflect a phylogenetic or ecological signal. By incorporating such previously inaccessible information of extinct and extant gavialoids into phylogenetic and macroecological studies, we can potentially further constrain the clade's interrelationships, as well as evaluate the timing and ecological association of the evolution of these neuroanatomical features. Finally, our study supports recent phylogenetic analyses that place ‘Tomistoma’ dowsoni as being phylogenetically closer to Gavialis gangeticus than to Tomistoma schlegelii, indicating the necessity of a taxonomic revision of this fossil species.


| INTRODUC TI ON
Crocodylia is a clade of semi-aquatic, ambush predators that inhabit both freshwater and estuarine environments (Grigg & Kirshner, 2015).
They are broadly restricted to the subtropical latitudinal belt, with over 25 extant species currently recognised, consisting of alligators, caimans, crocodiles and gavials (Grigg & Kirshner, 2015). Crocodylian interrelationships have been debated for decades, most notably due to conflicting results from phylogenetic analyses based on molecular (Densmore & Owen, 1989) versus morphological (Brochu, 1997) data. The most notable discrepancy pertains to the position of the extant species Gavialis gangeticus: whereas molecular data recover it as the closest living relative of Tomistoma schlegelii, forming the clade Gavialidae, morphological data have typically placed it as the sister taxon to all other extant crocodylians (Brochu, 1997). For the first time based solely on morphological data, Rio and Mannion (2021) robustly recovered Gavialis gangeticus as the closest living relative of Tomistoma schlegelii (see also Ristevski et al., 2022). Additionally, their analyses recovered more widespread similarities between fossil taxa traditionally referred to Tomistominae and Gavialoidea (see also Iijima et al., 2022;Iijima & Kobayashi, 2019). Despite this, a temporal incongruence is still evident in Gavialoidea, with several stratigraphically early fossil taxa recovered as closer to Gavialis than Tomistoma, resulting in inferred divergence timings for Gavialidae that greatly predate those estimated from molecular data (Rio & Mannion, 2021).
Although the incorporation of further taxa (e.g. Iijima et al., 2022), as well as methodological approaches such as total evidence analyses (Darlim et al., 2022;Lee & Yates, 2018), might help resolve this discrepancy, one potential additional resource might emanate from the internal anatomy of specimens (Gold et al., 2014), utilising previously inaccessible data increasingly made available through computed tomography.
Here, we present a reconstruction of the neuroanatomy and neurosensory apparatus of the Miocene North African fossil crocodylian species 'Tomistoma' dowsoni. In recent phylogenetic analyses, 'Tomistoma' dowsoni has been recovered as a gavialoid, more closely related to Gavialis gangeticus than to Tomistoma schlegelii (Groh et al., 2020;Rio & Mannion, 2021). Given its 'intermediate' position within Gavialoidea (sensu Iijima et al., 2022), the neuroanatomy of both extant gavialoids is also reconstructed for comparative purposes. Additionally, we quantitatively evaluate morphological variation in crocodyliform neuroanatomy, especially that of crocodylians and their close relatives, as well as test how this corresponds to the environment they inhabit. previously inaccessible information of extinct and extant gavialoids into phylogenetic and macroecological studies, we can potentially further constrain the clade's interrelationships, as well as evaluate the timing and ecological association of the evolution of these neuroanatomical features. Finally, our study supports recent phylogenetic analyses that place 'Tomistoma' dowsoni as being phylogenetically closer to Gavialis gangeticus than to Tomistoma schlegelii, indicating the necessity of a taxonomic revision of this fossil species.

| Specimens and CT scan reconstructions
The neuroanatomy of 'Tomistoma' dowsoni was interpreted from NHMUK PV R 4769. This is a referred specimen purchased by the NHMUK in 1920 from Lady Moon, and it was collected from near the Siwa Oasis, in the Western Desert of Egypt (Hamilton, 1973). Its precise stratigraphic provenance is uncertain, but it is likely to be from the lower Miocene Moghra (=Moghara) Formation (Hamilton, 1973).
NHMUK PV R 4769 has been used as the basis for the 'Tomistoma' dowsoni operational taxonomic unit in recent phylogenetic analyses (Groh et al., 2020;Rio & Mannion, 2021), in which it has been recovered as more closely related to Gavialis gangeticus than to Tomistoma schlegelii. It is represented by a near-complete and well-preserved skull, missing the quadratojugals, pterygoids and ectopterygoids.
The specimen is supported by a longitudinal metal rod which affects the reconstruction of some neuroanatomical features. Similarly, an incomplete left squamosal and postorbital also impacted reconstruction of the paratympanic region.
NHMUK PV R 4769 was characterised at the NHMUK with Xray micro-computed tomography using a Nikon Metrology XTH 225 ST system (Nikon Metrology,Leuven,Belgium). Acquisition of the full skull was implemented in five parts, with a voltage of 215 kV and a current of 698 μA, resulting in a reconstructed isotropic voxel size of 75.999 μm 3 , and 4476 projections with an average of four frames, with an exposure time of 0.708 seconds per frame. Three acquisitions were carried out for the posterior portion of the skull, followed by the skull being turned upsidedown and the last two acquisitions captured the anterior portion of the skull. Datasets were merged into a single volume using Avizo v. 9.7 (FEI Visualization Science Group; https://www.therm ofish er.com ), using the protocol described in (Butler et al., 2022).
The neuroanatomy of NHMUK PV R 4769 was subsequently segmented in Avizo v. 9.7, smoothed in Blender (Stichting Blender Foundation, Amsterdam) and rendered in Inkscape (Inkscape Project, 2020).
The neuroanatomy of the extant gavialoids Gavialis gangeticus (FLMNH UF 118998) and Tomistoma schlegelii (TMM M6342) was reconstructed based on data available in MorphoSource (https:// www.morph osour ce.org/). As the paratympanic region is not preserved in NHMUK PV R 4769, these features were not segmented in Gavialis gangeticus and Tomistoma schlegelii. Both specimens used are adults, for accurate comparison to NHMUK PV R 4769, as brain volume varies throughout ontogeny (Jirak & Janacek, 2017;Watanabe et al., 2019). These two extant taxa, as well as published neuroanatomical reconstructions of the extinct gavialoid Gryposuchus neogaeus from the Miocene of Argentina (Bona et al., 2015), the non-crocodylian allodaposuchid eusuchian Agaresuchus fontisensis from the Late Cretaceous of Spain (Serrano-Martínez et al., 2021), and the thalattosuchian Pelagosaurus typus from the Early Jurassic of the UK (Pierce et al., 2017), were used as a comparative framework during segmentation.

| Reptile encephalisation quotient
The reptile encephalisation quotient (REQ) was developed by Hurlburt (1996) from the encephalisation Quotient of Jerison (1973), based on extant reptile species. The REQ is a commonly used metric to measure relative brain size of extinct species (Paulina-Carabajal & Currie, 2017), and has been previously applied to eusuchian neuroanatomy to infer cognitive capabilities (Serrano-Martínez et al., 2021).
Measuring REQ requires an estimation of body and brain mass. Body mass was calculated for NHMUK PV R 4769 using the regression equation Ln (Total Length of Skull) = 0.32Ln (Body Mass) + 2.05 (Platt et al., 2009), which was subsequently rearranged to interpret body mass: Body mass = (Total Length of Skull x e −2.05 ) 1/0.32 . Brain mass was estimated using the endocast volume, applying a density of 1 g/ cm 3 (Franzosa, 2004). As the endocast volume would not necessarily be the same as the brain volume, given that the endocast represents the brain and its associated tissues (Hopson & Gans, 1979;Jirak & Janacek, 2017;Watanabe et al., 2019), the relative brain volume was

| Olfactory capability and visual acuity calculations
The olfactory capabilities of 'Tomistoma' dowsoni were calculated using the methodology of Zelenitsky et al. (2011). Olfaction acuity is dependent on the size of mitral cells, as well as odour receptors, which can be estimated from the relative size of the olfactory bulb (Lautenschlager et al., 2012;Serrano-Martínez et al., 2019;Zelenitsky et al., 2009). The greatest diameter of the olfactory bulb of each of NHMUK PV R 4769, Gavialis gangeticus, and Tomistoma schlegelii was compared to the greatest diameter of their respective cerebrum hemispheres, which was subsequently normalised via a log transformation (Serrano-Martínez et al., 2021).
Visual acuity is usually estimated from the size of the eyeball, which can be inferred from the sclerotic ring (Lautenschlager et al., 2012). As eusuchians lack sclerotic rings, Serrano-Martínez et al. (2021) estimated the relative size of the optic region using the optic lobes, which can be inferred from the rhombencephalon region of the endocast (Jirak & Janacek, 2017). The relative volume of the optic region was calculated with the Arithmetic function in Avizo v. 9.7, by comparing the volume of the optic lobe to the volume of the whole endocast.

| Morphometric data
Morphometric data were collated from the endocasts and endosseous labyrinths of NHMUK PV R 4769, Gavialis gangeticus and Tomistoma schlegelii, using the 'Measurement' tool in Avizo v. 9.7 (Table 1). Selected dimensions followed Pierce et al. (2017). Our dataset was augmented by measurements from specimens of taxa presented in the published literature (Erb & Turner, 2021;Pierce et al., 2017;Ristevski, 2022) Bona et al., 2015), Gyprosuchus neogaeus (Bona et al., 2015), the Miocene South American caimanine Mourasuchus grendsi (Bona et al., 2013), and the allodaposuchids Agaresuchus fontisensis (Serrano-Martínez et al., 2021) and Arenysuchus gascabadiolorum (Puértolas-Pascual et al., 2022). Following Pierce et al. (2017), we converted the raw morphometric data into ratios, in order to interpret the relative proportions of the olfactory tract, cerebrum, pituitary fossa and the endosseous labyrinth (Table 2). Variation in the shape of the endosseous labyrinth was measured using 82 curved, semi-landmarks plotted around the inner ear, as well as around each of the semi-circular canals. We added 'Tomistoma' dowsoni to the dataset collated by Ristevski (2022), which consists of the morphologies of endosseous labyrinths across 20 crocodylomorphs, including terrestrial, semi-aquatic and pelagic species. Due to most species included in this study having a semi-aquatic ecology, species were also classified based on their skull morphology. Adapting the classification system of Busbey (1995), the ratio of the rostrum length compared to the skull length is less than 0.55 in short-snouted/'brevirostrine' taxa, 0.55 to 0.7 in 'mesorostrine' taxa, and greater than 0.7 in long-snouted/'longirostrine' taxa. Skull width was also measured at the premaxilla and at the orbits, in order to determine the snout thickness (see Table 3). In this study, we use the term 'longirostrine' to describe crocodyliform skulls that have a long and narrow snout compared to the skull table and 'brevirostrine' to describe skulls that have a snout with a relatively similar width to the skull table (Table 3). These glands occupy small concavities located on the ventral surface of the nasals, on the nasomaxillary suture (Cowgill et al., 2022;Witmer, 1995). Whereas they are restricted to the posterior half of the rostrum in Tomistoma schlegelii (Cowgill et al., 2022), these concavities extend further anteriorly in both NHMUK PV R 4769 and sinus (Fernández & Herrera, 2009;Witmer, 1995). Despite the sinus being closed off in extant crocodylians, this feature is still well developed in those taxa (Pierce et al., 2017). The olfactory bulb connects to the olfactory region in NHMUK PV R 4769, resulting in a sharp contact, as is also the case in Tomistoma schlegelii. By contrast, the expansion of the olfactory region is more gradual in Gavialis gangeticus ( Figure 3). In NHMUK PV R 4769, the olfactory region expands ventrolaterally at the contact between the nasal cavity and the nasopharyngeal duct to form the paranasal sinus ( The nasal cavity of all three species possesses two channels that extend from the olfactory region to the most anterior part of the nasal cavity. These channels have been referred to as neurovascular canals, and, more specifically, dorsal alveolar canals or ducts for the trigeminal nerve and maxillary veins and arteries (Pierce et al., 2017;Serrano-Martínez et al., 2019. In both extant gharials, these channels run anteroposteriorly on the lateral surfaces of the nasal cavity, with the channels meeting posterior to the external naris, on the dorsal surface of the nasal cavity (Figures 3 and 4). These two channels also run on the lateral surfaces of the nasal cavity in NHMUK PV R 4769; however, they do not meet on the dorsal surface ( Figure 2).

The paranasal system of Gavialis gangeticus, Tomistoma schlegelii
and NHMUK PV R 4769 is characterised by a relatively simple morphology, which is representative of longirostrine crocodyliforms (Witmer, 1997;. The morphology of the

| Endocranium
Crocodylian encephalic endocasts tend to be relatively straight in outline, with little curvature (Edinger, 1938;Hopson & Gans, 1979), including that of Gavialis gangeticus. Tomistoma schlegelii, however, appears to be the exception, showing greater curvature, as reflected in acute cephalic and pontine flexure angles (Table 1; Figure 4). The encephalic endocast of NHMUK PV R 4769 also shows a greater degree of curvature and more acute cephalic and pontine flexure angles than other crocodylians (Figures 2 and 3), although this is not to the same extent as that of Tomistoma schlegelii (Table 1) (Table 2). When the cerebrum width is compared to the encephalic endocast length, NHMUK PV R 4769 has a similar value to that of Gavialis gangeticus, with both lower than that of Tomistoma schlegelii ( Table 2). The cerebrum of Tomistoma schlegelii has a near-symmetrical expansion in dorsal view, whereas the greatest expansion in NHMUK PV R 4769 and Gavialis gangeticus occurs at the posterior end of the cerebrum (Figures 5-7). This posterior expansion has also been noted in other crocodylomorphs (Colbert et al., 1946;Edinger, 1938;Hopson & Gans, 1979;Kley et al., 2010;Pierce et al., 2017). Posteroventral to the cerebrum, and anterior to the optic lobes, is the pituitary (Figures 5-7). In NHMUK PV R 4769, the pituitary is much more laterally expansive than that of Gavialis gangeticus or Tomistoma schlegelii; however, it has a similar length to these two species (Table 2). The pituitary has two large channels that extend posterolaterally in all three species (Figures 5-7). These channels curve dorsolaterally at the posterior part in all three species and house the cerebral carotid artery (Dufeau & Witmer, 2015;Hopson & Gans, 1979;Pierce et al., 2017;. The optic lobes of the encephalic endocast are difficult to segment in crocodile-line archosaurs as a result of the thick dural envelope (Hopson & Gans, 1979;Pierce et al., 2017), but they can be deduced from the mesencephalon region of the brain (Serrano-Martínez et al., 2021). These lobes are more prominent in early ontogenetic stages, becoming less distinct as individuals mature (Hu et al., 2021;Jirak & Janacek, 2017;Ristevski, 2022). Similarly, segmentation of the cranial nerves is difficult in fossil taxa due to both preservation and quality of the scan (Pierce et al., 2017). This is particularly evident in NHMUK PV R 4769 due to the poor preservation in this area (Figure 1a). As a result, this region, as well as the paratympanic sinuses, were not segmented in Tomistoma schlegelii and Gavialis gangeticus, as there would not have been a direct comparison with NHMUK PV R 4769.

| Endosseous labyrinth
Poor preservation of the region of the skull of NHMUK PV R 4769 in which the endosseous labyrinth would have been housed made segmentation difficult; however, the overall shape could be reconstructed ( Figure 8). The endosseous labyrinths of Gavialis gangeticus and Tomistoma schlegelii have a similar morphology to one another, with the anterior semi-circular canal being larger than the posterior semi-circular canal, as is the case in most archosaurs (Brusatte et al., 2016;Witmer et al., 2003), including other extant and extinct crocodylians (Bona et al., 2013(Bona et al., , 2015Dufeau & Witmer, 2015;Georgi & Sipla, 2008;. In dorsal view, the anterior and posterior semi-circular canals appear more equidimensional in NHMUK PV R 4769 ( Figure 8); however, when the area of each canal is quantified, the anterior semi-circular canal is more than double that of its posterior counterpart (Table 2), resulting in a similar value to that of Gavialis gangeticus (Pierce et al., 2017). The cochlear duct, responsible for auditory capabilities in the inner ear, is of comparable size across the three species in this study (Tables 1 and 2   The relative size of the optic region, estimated as the ratio of the volume of the optic lobes to that of the encephalic endocast, averages between 10% and 15% in most extant crocodylians, with that of Tomistoma schlegelii and NHMUK PV R 4769 estimated to be 11% and 12% respectively (Table 5). By contrast, the relative volume for Gavialis gangeticus is estimated to be 18%

| Landmark-based morphometrics
The endosseous labyrinth varies mostly in its width and height across crocodyliforms. The first three principal components (PCs) equate to approximately 66.5% of endosseous labyrinth shape variation, with only the first five principal components explaining more than 5% of variation ( Figure 9). When broadly classified into their environmental habitats, there is a large overlap in semi-aquatic, terrestrial and pelagic taxa (Table 3), although this is most likely because of the higher number of the sampled species assigned to a semi-aquatic ecology in this study. When classified based on their skull shape (Table 3), distinct clusters of labyrinth shapes can be observed, with crocodyliforms categorised as longirostrine forming one grouping, and taxa with brevirostrine and intermediate skull morphologies forming a second grouping (Figure 9a,b).

| Ecological versus phylogenetic signal
Features of the neuroanatomy and neurosensory apparatus of gavialoids appear to show an ecological and/or phylogenetic signal.
Below we discuss several of these features, including their potential implications for reconstructing the phylogenetic relationships and macroecology of gavialoids and other crocodyliforms.
The thickness of the semi-circular canals of the endosseous labyrinths for example, is thought to be dependent on the ecology of crocodylomorph species (Schwab et al., 2020). Whereas terrestrial species have a dorsoventrally tall labyrinth, with thin semi-circular canals, and pelagic species have a compact labyrinth, semi-aquatic crocodyliforms have an intermediate labyrinth morphology (Schwab et al., 2020), which characterises 'Tomistoma' dowsoni, Gavialis gangeticus and Tomistoma schlegelii (Figure 8). The endosseous labyrinth appears to be anteroposteriorly wider and dorsoventrally shorter in longirostrine crocodyliforms, but narrower and taller in brevirostrine taxa (Figure 9b). More species, particularly fossil crocodylomorphs, need to be included in analyses to test this more robustly; however, this preliminary finding potentially indicates that snout and skull morphology, which are themselves partly constrained by ecology in crocodylomorphs, exert an influence on the shape of the endosseous labyrinth in this group. Recent studies have also recognised the influence of skull shape on the endosseous labyrinth in turtles (Evers et al., 2022) and on braincase shape in birds (Chiappe et al., 2022), suggesting that this might be a more widespread pattern.
The relatively large size of the cerebrum in birds and mammals has been associated with refined sensory inputs in these groups, as a larger cerebral region implies a greater neuronal area to execute complex behaviours (Pierce et al., 2017;Rogers, 1999). When comparing cerebrum width to skull width, Tomistoma schlegelii has a higher value than 'Tomistoma' dowsoni and Gavialis gangeticus (Table 1), which could suggest that Tomistoma schlegelii has greater behavioural complexity, as also reflected in the higher olfactory acuity estimation for this species (Table 4). Ecological studies have suggested that Tomistoma schlegelii shows complex behavioural patterns during courtship, for example, employing visual, tactile and auditory cues (Staniewicz et al., 2022). Analysis of the vocalisations produced by the two living gavialoids has revealed differences between their call structures (Bonke et al., 2015), with sounds produced by Tomistoma schlegelii having different acoustic properties and context, with a greater reliance on visual or olfactory cues, particularly in underwater environments (Staniewicz et al., 2022). These differences in underwater signals have been suggested to result from morphological differences (Dinets, 2013;Staniewicz et al., 2022), with a possible role for the larger cerebrum in Tomistoma schlegelii. Gavialis gangeticus and 'Tomistoma' dowsoni share a similar cerebrum morphology, in which its greatest expansion occurs posteriorly, whereas that of Tomistoma schlegelii has a symmetrical expansion (Figures 5-7). Although both extant gharials occupy aquatic habitats, Gavialis gangeticus is observed in streams and rivers with sandy, grassy or rocky shores, whereas Tomistoma schlegelii is observed in densely vegetated swamps and lowland forest rivers (Staniewicz et al., 2022). The palaeoenvironment of the Moghra Formation, which 'Tomistoma' dowsoni inhabited, is thought to be a tide-dominated estuary (Georgalis et al., 2020), closer to the environments inhabited by Gavialis gan-  . In both Gavialis gangeticus and Tomistoma schlegelii, the nasolacrimal ducts are located on the dorsal surface of the olfactory region, which is not seen in early marine longirostrine taxa (Pierce et al., 2017), however, their morphology differs. Similarly, the external naris, which has a dorsal inflection, is not seen in early longirostrine crocodylomorphs, but characterises eusuchians (Pierce et al., 2017;Serrano-Martínez et al., 2021). The morphology of this dorsal inflection also differs between longirostrine and brevirostrine eusuchians (Figures 2-4 dowsoni is also reflected in these taxa. Furthermore, there is evidence for a pterygoid bulla in Hanyusuchus sinensis (Iijima et al., 2022), a feature that characterises Gavialis gangeticus (Martin & Bellairs, 1977), but not Tomistoma schlegelii, and that has also been identified in extinct species of Gavialis (Gavialis lewisi and Gavialis bengawanicus), as well as the extinct gavialoid Eogavialis africanum from the late Eocene of Egypt (Hecht & Malone, 1972;Iijima et al., 2022;Lull, 1944;Martin et al., 2012). It might also be present in several South American gryposuchine gavialoids, including Dadagavialis gunai and Gryposuchus (Riff & Aguilera, 2008;Salas-Gismondi et al., 2016. In Rio and Mannion's (2021) phylogenetic analysis, 'Tomistoma' dowsoni is recovered as the sister taxon to a clade that includes Eogavialis africanum, Gavialis and gryposuchines. Given that Eogavialis africanum is potentially a 'problematic' taxon in terms of its temporal incongruence with the reconstructed divergence date of Gavialis and Tomistoma, it will therefore be informative to determine if a bulla is truly synapomorphic of this clade, or is more widespread among gavialoids, with either more than one independent origin of the bulla, or its apomorphic loss in Tomistoma schlegelii.

| Systematics of 'Tomistoma' dowsoni and contemporaneous gavialoids
Coupled with the results from recent phylogenetic analyses (Groh et al., 2020;Rio & Mannion, 2021), the neuroanatomy of 'Tomistoma' dowsoni further suggests this species is more closely related to Gavialis gangeticus than to Tomistoma schlegelii, and thus casts additional doubt as to its current generic attribution.
Similarly, revision of contemporaneous Miocene species from the Mediterranean region that have previously been referred to Tomistoma indicates that none of them share close affinities with the extant species either (Nicholl et al., 2020). An anatomical and taxonomic revision of 'Tomistoma' dowsoni, including the type material (Fourtau, 1920), is currently in preparation, along with ongoing systematic work on the Miocene gavialoids of Europe and North Africa (Burke et al., 2022).

AUTH O R CO NTR I B UTI O N S
PMJB segmented NHMUK PV R 4769, analysed the data, produced the figures and wrote the manuscript. PDM conceived the project idea and contributed to writing and revising the manuscript.

ACK N O WLE D G E M ENTS
We thank Vincent Fernandez for CT scanning NHMUK PV R 4769 and Jessie Maisano and David Blackburn for facilitating access to the CT scans of Tomistoma schlegelii and Gavialis gangeticus on MorphoSource. We also thank Cecily Nicholl and Nils Chabrol for discussion. We are grateful for comments from Francisco Barrios and Jorgo Ristevski that improved this manuscript, and we also thank the latter for facilitating access to his dataset.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data will be made openly available in a public repository after article publication.