Isolated tooth reveals hidden spinosaurid dinosaur diversity in the British Wealden Supergroup (Lower Cretaceous)

Geological context and provenance of HASMG G369a

The collection of teeth labelled as HASMG G369 consists of 10 specimens, and is associated with a note, which states:

“If no specific locality is mentioned, these specimens are from Netherfield (Purbeck)”

No specific locality is mentioned for any of the specimens, and it is unclear when or by whom this note was written. Importantly, the note is inconsistent with the accession record for HASMG G369, which details a “collection of local Wealden fossils” gifted by the Reverend Pierre Tielhard de Chardin (1881–1955); the provenance and contents of this “collection” are unknown. Tielhard is known to have collected from the Ashdown and Wadhurst Clay formations around Hastings, and donated many specimens (including some vertebrate remains) to Hastings Museum (Brooks, 2008). Thus, within the Weald sub-basin, HASMG G369a was either found from the Purbeck Group near Netherfield or the overlying Wealden Supergroup strata surrounding Hastings (Fig. 1A).

Three fault-bounded inliers result in surface exposures of the Purbeck Group within the Weald sub-basin, located north and northwest of Battle in East Sussex, and are surrounded by the overlying Hastings Group (most of which comprise deposits of the Ashdown Formation) (Howitt, 1964; Milner, 1922; Radley & Allen, 2012a). These are the oldest exposed rocks in the region, with the inliers located north of Brightling, between Hollingrove and Netherfield, and near Archer Wood (Lake & Shepard-Thorn, 1987); the foremost pair have been respectively referred to as the Rounden Wood/Brightling-Heathfield and Limekiln Wood/Mountfield inliers (Howitt, 1964; White, 1928). The Purbeck Group in the area was previously quarried and mined, with data also provided from boreholes; however, surface exposures are poor and are mainly visible following valley denudation; those exposed in stream valleys have often been disturbed by valley-bulging, landslips and slope cambers (Lake & Shepard-Thorn, 1987; Topley, 1875). Nevertheless, exposures of the Purbeck Group in the region are represented by both of its constituent Lulworth and Durlston Formations (Fig. 1B), which are principally Berriasian in age (Cope, 2007; Hopson, Wilkinson & Woods, 2008; Howitt, 1964; Lake & Shepard-Thorn, 1987). As mentioned above, Purbeck theropods are very rare, and documented specimens from Sussex outcrops include material referred to “Megalosaurus sp.” (Benton & Spencer, 1995; Topley, 1875; White, 1928).

The Hastings Group, itself the basal unit of the Wealden Supergroup within the Weald sub-basin (Batten, 2011), dominates the area surrounding Hasting and is comprised of the older (late Berriasian–early Valanginian) Ashdown Formation, followed by the Wadhurst Clay Formation (Valanginian) and Tunbridge Wells Formation (late Valanginian; Fig. 1B), several of which are well exposed along coastal sections (Hopson, Wilkinson & Woods, 2008; Lake & Shepard-Thorn, 1987; Radley & Allen, 2012a). Only a small outcrop of the overlying Weald Clay Group is known near Cooden (Lake & Shepard-Thorn, 1987). Vertebrate fossils from the coastal exposures around Hastings in particular have been collected for over a century (Benton & Spencer, 1995). Documented theropod finds from the Hastings area include an allosauroid tibia (HASMG G378) (Naish, 2003) and material referred to “Megalosaurus dunkeri” (e.g., NHMUK PV R19154) and “M. oweni” (Benton & Spencer, 1995; White, 1928). Allosauroid and spinosaurid teeth are also known from the Wadhurst Clay around Bexhill (Charig & Milner, 1997; Turmine-Juhel et al., 2019), as are the remains of a tiny maniraptoran (Naish & Sweetman, 2011). The enigmatic theropod Altispinax (NHMUK PV R1828) is also known from the Hastings Group of Battle (Maisch, 2016; Naish, 2011; Von Huene, 1923), located between Netherfield and Hastings.

We were unable to clarify the conflicting accession information surrounding HASMG G369a or ascertain its provenance. Given the rarity of Purbeck Group theropods, limited exposure of the succession around Netherfield, and accession history, we consider it highly unlikely this tooth originates from the Purbeck Group. Further, in the overlying Hastings Group, vertebrate fossils (bar fish detritus) are also extremely rare in the Ashdown Formation around Hastings and the exposures of the Weald Clay Formation are highly limited (Lake & Shepard-Thorn, 1987). Taken together, the upper units of the Hastings Group succession are thus the more likely candidates regarding HASMG G369a’s provenance, and we thus provisionally consider the specimen to be Valanginian in age.

Orientation and terminology

Dental nomenclature and protocols for crown and denticle morphometry follow the recommendations of Hendrickx, Mateus & Araújo (2015b) and references therein.

Measurements

The specimen was examined via a DinoLite (AM4113TL) digital microscope. Measurements were taken using a 150 mm digital calliper (accuracy 0.01 mm), as well as the measurement tools in DinoXcope (v2.0.4) software. A full list of measurements is provided in the Supplementary Information.

As HASMG G369a is missing its apex, several ordinary least-squares regression analyses were conducted where the specimen’s crown height (CH) was compared against crown base length (CBL) and crown base width (CBW) for other spinosaurid teeth. Measurements were collected from the dataset of Hendrickx, Tschopp & Ezcurra (2020). Variables were log-transformed to fit a normal distribution and the analyses were conducted using the Bivariate regression function (Model > Linear) in Past4 (v.4.11) (Hammer, Harper & Ryan, 2001). Of the different spinosaurid samples analysed (see Supplementary Information), logCBW from Baryonyx walkeri lateral teeth provided the most favourable regression coefficient ( r² = 0.86), the slope and intercept of which was then used to estimate crown height in HASMG G369a. Other measurements or descriptions derived from CH (e.g., mid-crown length and width, number of denticles at mid-crown etc.) were based on the estimation detailed above.

Crown angle (CA) was estimated using the Angle tool in FIJI (Schindelin et al., 2012) via the creation of a vertex delimited by the CBL and a line trending through the midpoint of the preserved apex as the specimen was observed in lateral view. The landmarks used to delineate CBL follows Hendrickx, Mateus & Araújo (2015b). Hendrickx, Mateus & Araújo (2015b) described a method to calculate CA using the law of cosines and several morphometric landmarks, but photographs and FIJI has also been employed for isolated theropod crowns (Hendrickx, Tschopp & Ezcurra, 2020).

Cladistic analysis

We examined the phylogenetic affinities of HASMG G369a by including it in an updated version of the Hendrickx & Mateus (2014) data matrix designed to test the affinities of non-avian theropod teeth (Hendrickx, Tschopp & Ezcurra, 2020). This updated matrix was used to assess the affinities of an isolated theropod tooth associated with the Aerosteon riocoloradensis holotype: the latter operational taxonomic unit (OTU) was replaced by HASMG G369a, and the final matrix was composed of 146 characters (Ch.) scored across 106 theropod OTUs (the “whole dentition” dataset). The mesial and lateral dentitions of spinosaurids are difficult to distinguish (Hendrickx, Mateus & Araújo, 2015b). However, as early spinosaurids possessed supernumerary lateral teeth (e.g., Baryonyx NHMUK PV R9951), it is more likely that HASMG G369a originated from the more distal maxillary or dentary dentition. HASMG G369a was thus scored as a lateral tooth.

We performed the cladistic analysis in TNT 1.5 (Goloboff & Catalano, 2016) following the methods outlined in Young et al. (2019) and Hendrickx, Tschopp & Ezcurra (2020), based on a backbone tree topology and the positive constraint command (force +), setting HASMG G369a as a floating terminal. The references used to create the backbone tree can be found in Hendrickx, Tschopp & Ezcurra (2020). A pair of additional cladistic analyses was also performed using the whole dentition dataset without constraints, and a reduced matrix consisting only of crown-based characters (see Young et al., 2019; Hendrickx, Tschopp & Ezcurra, 2020: 11). The latter included 91 characters (Ch. 38–122 and 141–146) scored for 101 OTUs, with all edentulous taxa removed.

The tree searching strategy involved a combination of algorithms: Wagner trees, TBR branch swapping, sectorial searches, Ratchet (perturbation phase stopped after 20 substitutions) and Tree Fusing (five rounds) were used until 100 hits of the same minimum tree length were reached. The recovered trees were subsequently subjected to an additional round of TBR branch swapping. In the unconstrained analyses, wildcard OTUs were identified using the iterPCR function (Goloboff & Szumik, 2015; Pol & Escapa, 2009), and Bremer support values were calculated as a measure of nodal support in the resulting reduced consensus.

Hendrickx & Mateus (2014) use hypodigms for their spinosaurid OTUs, given the type specimens for several do not preserve dental elements (e.g., Suchomimus) or have been lost entirely (e.g., Spinosaurus). We note that their Baryonyx OTU includes the B. walkeri holotype NHMUK PV R9951 and the Iberian specimen ML 1190, and that the latter was recently considered the type specimen of a distinct taxon, Iberospinus natarioi (Mateus & Estraviz-López, 2022). Mateus & Estraviz-López (2022) combined the dental character matrix of Hendrickx et al. (2020)—itself a version of the matrix used in the present work—with the modified pan-skeletal matrix of Arden et al. (2019) in their phylogenetic analysis of ML 1190. The latter specimen was coded for 36 observable dental characters, however it would appear that Mateus & Estraviz-López (2022) did not realise that the Baryonyx OTU employed in their analysis is a hypodigm and already contained ML 1190 (Hendrickx & Mateus, 2014). Nevertheless, the spinosaurid OTUs used in our analysis of the Hendrickx, Tschopp & Ezcurra (2020) matrix were not modified given the fact that the dental material of I. natarioi is limited, positionally overlaps with that of B. walkeri, and possesses the same (observable) character scores as the Baryonyx OTU.

Elsewhere, the OTU of Irritator also includes the type specimen of Angaturama, following previous authors who consider the latter congeneric with the former (Buffetaut & Ouaja, 2002; Charig & Milner, 1997; Dal Sasso et al., 2005; Sereno et al., 1998; Sues et al., 2002). Specimens used for the cf. Suchomimus and cf. Spinosaurus hypodigm OTUs can be found in Hendrickx & Mateus (2014: Table 1).

Regarding character scores, those of Ch. 82 (concerning the basalmost position of the mesial serration in lateral teeth) were scored by a process of elimination: although the basalmost mesial serration is not preserved in HASMG G369a, it likely possessed state 1 given the preserved extent of the mesial denticles and the probable inapplicability of states 0 and 2. Meanwhile, Ch. 90 (denticle number in lateral teeth respectively) were extrapolated from the observable data due to the incomplete nature of the carinae and preservation of denticles.

Discriminant function analyses

Cluster analysis

Cluster analyses were also performed in Past4 on the different pan-theropodan datasets mentioned above. Hierarchical clustering with a Paired group algorithm and Neighbour joining clustering were used, rooting the tree with the final branch, whilst selecting Euclidean distances as the similarity index.

Systematic palaeontology

Description

Crown

HASMG G369a is a conidont crown with a lenticular cross section at the cervix and at mid-crown (Figs. 2A–2B); as such, the crown is weakly labiolingually compressed (CBR: 0.86). The crown is not particularly large (preserved CH: 13.2 mm; reconstructed CH: 17.2 mm) and only moderately elongated (preserved CHR: 1.68; reconstructed CHR: 2.1).

The mesial and distal carinae are both denticulated (Figs. 2C–2D, 3), lacking adjacent concave surfaces. The former is straight, undivided, and not notably developed, and is positioned largely centrally on the mesial profile. Whilst the basalmost portion has been chipped off (see above), what remains suggests the mesial carina almost certainly reached the cervix. The distal carina is slightly diagonally oriented, and as mentioned above, trending towards the labial side basally. It too is not markedly developed and lacks any twisting of splitting. It extends basally past the cervix a short distance. The apical extent of either carina cannot be determined for this specimen.

The crown displays weak distal recurvature in labiolingual views (Fig. 2A). Its mesial profile is weakly convex, whilst the distal profile is almost straight for the majority of its preserved length. The preserved apex is almost centrally positioned. When viewed distally (Fig. 2D), the crown also possesses minor lingual curvature, with the apex closer to the lingual side. Both the labial and lingual crown surfaces are convex.

The cervix assumes a parabolic morphology on the better-preserved labial side of the crown, such that the basalmost extent of the enamel occurs roughly centrally (Fig. 2E). The equivalent features, or relative extent of the enamel on the lingual side, cannot be reliably ascertained due to preservation. However, the extent of the enamel on the mesial and distal surfaces appears largely similar.

Cladistic analysis

The results of the various cladistic analyses, detailed below, are summarised in Table 2. Full versions of the recovered trees are available in the Supplementary Information.

Whole dentition dataset

Two MPTs of 1318 steps were recovered following the constrained search on the whole dentition dataset (CI = 0.204097, RI = 0.451360). HASMG G369a either assumed a position outside the baryonychine + spinosaurine clade or at the base of Spinosaurinae; the latter position was supported by a single synapomorphy: a slightly convex mesial margin (Ch. 73:1). Accordingly, the strict consensus recovered HASMG G369a in a polytomous Spinosauridae alongside Baryonychinae and Spinosaurinae (Fig. 4A), with the clade supported by numerous synapomorphies. Of these, HASMG G69a shared: (1) weak labiolingual compression of the crown with a CBR exceeding 0.75 (Ch. 70:2), (2) subcircular basal cross-section of the crown (Ch. 76:0), (3) over 30 distocentral denticles per five mm (Ch. 89:0), (4) fluted enamel surfaces present on both labiolingual surfaces (Ch. 111:2) and (5) veined enamel texture (Ch. 121:3).

The unconstrained search on the whole dentition dataset initially returned 248 MPTs of 1074 steps (CI = 0.250466, RI = 0.578975). This increased to 87576 MPTs following the round of TBR. The strict consensus is largely unresolved and predominantly formed by two large polytomies containing well over 25 OTUs each. Few traditional clades can be recognised but those present include Spinosauridae and Abelisauridae. The strict consensus nevertheless recovered HASMG G369a within a polytomous Spinosauridae alongside Baryonychinae and Spinosaurinae.

A reduced consensus was achieved following the pruning of 23 wildcard OTUs (Limusaurus (juvenile), Masiakasaurus, Indosuchus, Chilesaurus, Piatnitzkysaurus, Sciuruminus, Eustreptospondylus, Afrovenator, Dubreuillosaurus, Duriavenator, Sinraptor, Allosaurus, Orkoraptor, Acrocanthosaurus, Aorun, Guanlong, Eotyrannus, Raptorex, Gorgosaurus, Alioramus, Daspletosaurus, Tyrannosaurus and Ornitholestes) identified via the iterPCR function (Fig. 4B). As above, HASMG G369a is again recovered in a polytomous Spinosauridae alongside Spinosaurinae and Baryonychinae, which is supported by several synapomorphies; those present in HASMG G369a are: (1) the basalmost denticle on the mesial carina of lateral teeth extending to the base of the crown or slightly above the cervix (Ch. 82; see comment in the “Cladistic analysis” methodology section above), (2) basalmost serration on the distal carina situated below the cervix (Ch. 85), and (3) flutes present on both labial and lingual surfaces (Ch. 111).

Crown-based dataset

The unconstrained search on the crown-based dataset initially recovered 244 MPTs of 648 steps (CI = 0.251543, RI = 0.62139). The additional round of TBR returned over 99999 trees found (overflow). The strict consensus produced a huge polytomy incorporating the vast majority of OTUs including HASMG G369a (see Supplementary Information for the full result). HASMG G369a was one of 74 OTUs acting as wildcard taxa (the others include: Daemonosaurus, Eodromaeus, Eoraptor, Dracovenator, Coelophysis, Liliensternus, Dilophosaurus, Ceratosaurus, Genyodectes, Berberosaurus, Masiakasaurus, Kryptops, Rugops, Abelisaurus, Aucasaurus, Arcovenator, Chenanisaurus, Indosuchus, Majungasaurus, Skorpiovenator, Piatnitzkysaurus, Marshosaurus, Monolophosaurus, Sciuriminus, Eustreptospondylus, Afrovenator, Dubreuillosaurus, Duriavenator, Megalosaurus, Torvosaurus, Baryonyx, Suchomimus, Irritator, Spinosaurus, Erectopus, Sinraptor, Allosaurus, Neovenator, Fukuiraptor, Australovenator, Megaraptor, Orkoraptor, Acrocanthosaurus, Eocarcharia, Carcharodontosaurus, Giganotosaurus, Mapusaurus, Bicentenaria, Aorun, Zuolong, Proceratosaurus, Guanlong, Dilong, Compsognathus, Ornitholestes, Haplocheirus, Eshanosaurus, Falcarius, Jianchangosaurus, Segnosaurus, Erlikosaurus, Incisivosaurus, Halszkaraptor, Sinornithosaurus, Graciliraptor, Dromaeosaurus, Bambiraptor, Tsaagan, Velociraptor, Sinusonasus, Zanabazar, Troodon and Archaeopteryx).

Discriminant function analysis

Pan-theropodan datasets

The analyses conducted on the whole dataset (Fig. 5), regardless of whether the absence of denticles was considered inapplicable or not, consistently classified HASMG G369a as a spinosaurid (clade-level analyses) or referred the tooth to the baryonychine spinosaurid Suchomimus (genus-level analyses) (Table 3). Reclassification rates (RR) are, however, generally low, ranging between 59.37–62.07%. Similarly, the reduced datasets based on single-author measurements classified HASMG G369a as a spinosaurid and as Suchomimus in the respective analyses (again, with low RR between 59.19–63.74%).

Spinosaurid-only datasets

The DFA results for the spinosaurid-only morphometric datasets (Table 4) consistently classified HASMG G369a as a non-Baryonyx spinosaurid. Reclassification rates are very high (98.18–100%), especially in comparison to the pan-theropodan datasets used above, with HASMG G369a classified as cf. Suchomimus in the majority of analyses (PC1 63.73–84.32%, PC2 14.84–26.12%). Interestingly, the results from the dataset including all spinosaurids and all variables classified HASMG G369a as “Suchosaurus” (PC1 72.53%, PC2 20.03%), which is also known from the Hastings Group.

Visualisation of the DFA plots also shows that spinosaurid teeth are readily differentiable based on the data from Hendrickx, Tschopp & Ezcurra (2020) (Figs. 6 and 7): spinosaurine and baryonychine taxa occupy different morphospace areas, whilst Baryonyx and cf. Suchomimus do not overlap in any iteration of the analyses. This suggests that Baryonyx and cf. Suchomimus teeth are morphologically distinct. Whether this impacts discussions regarding the congeneric status of the two taxa remains to be seen, especially given the non-cranial nature of the Suchomimus holotype skeleton (Carrano, Benson & Sampson, 2012; Sereno et al., 1998). Also of note is the tendency for “Suchosaurus” to cluster closely with the cf. Suchomimus morphospace in the analyses containing all spinosaurid specimens, whilst “Sinopliosaurus” plotted close to the morphospace occupied by spinosaurine teeth.

As an aside, the isolated specimen XMDFEC V10010 from the Santonian (Late Cretaceous) Majiacun Formation of China, referred to Baryonychinae by Hone, Xu & Wang (2010), does not cluster closely or share morphospace with any spinosaurid taxon in the DFA analyses of the spinosaurid sample. To explore this further, we tested the specimen using discriminant function and cluster analyses on the “whole”, “personal” and “large crown” pan-theropodan datasets from Hendrickx, Tschopp & Ezcurra (2020), treating XMDFEC V10010 as an unknown taxon. These results are presented in full in the Supplementary Information and are briefly discussed below.

Cluster analysis

The cluster analyses based on the pan-theropodan dataset (Table 5, Supplementary Information), regardless of the method employed (i.e., hierarchical vs. neighbour joining), unanimously support spinosaurid affinities of HASMG G369a. Almost all results recover the crown as a sister taxon to Suchomimus, except for the Neighbour joining analysis performed on the whole dataset (no denticles = “?”), where it is recovered as sister to a clade containing Irritator + Suchomimus.

Affinities of HASMG G369a and the diversity of British spinosaurids

The results from the cladistic, discriminant and cluster analyses clearly support the spinosaurid affinities of HASMG G369a. HASMG G369a shares multiple dental characters in common with spinosaurids, including a sub-circular outline, fluted enamel ornamentation and veined enamel texture, extension of the mesial carina to the cervix and a centrally positioned distal carina (Hendrickx et al., 2019).

Of particular note is the finding that HASMG G369a (its wildcard status within the crown-only phylogenetic analyses excepting) failed to associate with Baryonyx in any data run. This further supports previous arguments that the Wealden Supergroup contains multiple spinosaurid lineages (Barker et al., 2021; Buffetaut, 2010; Naish, 2011; Naish & Martill, 2007). These results also suggest that the spinosaurid diversity within the Wealden Supergroup reflects the situation of coeval Iberian localities, which appear to have contained a more diverse spinosaurid fauna than previously assumed (Isasmendi et al., 2020; Malafaia et al., 2020; Mateus & Estraviz-López, 2022).

The dentition of Ceratosuchops and Riparovenator were not scored for this analysis due to poor preservation; however, future work should aim to use cladistic and discriminant methods on spinosaurid crowns found in known strata within the Wealden Supergroup in order to further assess the diversity of its spinosaurids. It would be of particular interest to examine isolated spinosaurid teeth from the Upper Weald Clay Formation, in order to test whether these can be confidently referred to Baryonyx. Revisiting coeval Lower Cretaceous localities from Iberia may also be useful given the widespread presence of spinosaurids in these deposits (Malafaia et al., 2020); several morphometric-based (PCA and DFA) analyses have already been undertaken on Iberian spinosaurid crowns (the results of which also hint at high spinosaurid diversity) (Alonso & Canudo, 2016; Alonso et al., 2018; Isasmendi et al., 2020). However, cladistic analyses are recommended (if not preferred) for the identification of isolated theropod teeth (Hendrickx, Tschopp & Ezcurra, 2020), although some alternative machine learning techniques (e.g., decision trees) may be attractive tools with which to assess morphometric data from isolated theropod teeth (Wills, Underwood & Barrett, 2021). It should be noted that performing cladistic analyses on single teeth can be time consuming: each individual tooth in a batch of “unknown” specimens has to be tested separately, or appropriately grouped into morphotypes (Hendrickx, Tschopp & Ezcurra, 2020). This is further exacerbated by the difficulty distinguishing the position of isolated spinosaurid teeth (Hendrickx, Mateus & Araújo, 2015b); whilst we believe a lateral position for HASMG G369a is a more likely origin (see above), spinosaurid samples could alternatively be tested in both positions. Another potential technique for investigating spinosaurid diversity in the Wealden Supergroup might be to conduct specimen-level phylogenetic analyses using Bayesian methods and incorporating stratigraphical information, a method inspired by Cau (2017).

Comparative anatomy

The large number of minute denticles recalls the condition present in baryonychine spinosaurids (Hendrickx et al., 2019). The presence of minute denticles on both carinae most recalls the situation of other British spinosaurid crowns, including those of Baryonyx (Charig & Milner, 1997), Riparovenator (Barker et al., 2021), and BEXHM 1995.485 (C. Barker, pers. obs., 2022; Charig & Milner (1997) misreported the accession number of this specimen as “BEXHM 1993.485”); the carinae of Ceratosuchops are poorly preserved and its dentition will be revisited elsewhere, but denticles are present on some distal carinae at least. The denticles of the “Suchosaurus cultridens” type specimen (NHMUK PV R36536) are difficult to discern but this is probably due to wear (Buffetaut, 2010). Nevertheless, HASMG G369a differs from some Iberian spinosaurid teeth where a baryonychine dental morphotype lacking mesial denticles has been reported (Isasmendi et al., 2020).

Sporadic variation in denticle size is noted in baryonychines and is particularly developed in Baryonyx and Iberospinus (Hendrickx et al., 2019; Mateus & Estraviz-López, 2022). In contrast, those of cf. Suchomimus change more gradually and sporadic variation in denticle size is mainly observed on the basal portions of the teeth (Hendrickx et al., 2019). Those of the preserved mesial dentition of Riparovenator are similarly regular (C. Barker, pers. obs., 2022), as are baryonychine teeth from the Barremian–lower Aptian Cameros Basin of Spain (Isasmendi et al., 2020). HASMG G369a mirrors the latter specimens in this regard, with the more complete distal carina possessing a largely gradual change of denticle size.

Although damaged in its basal portion, the mesial carina likely reaches or terminates very near the cervix in HASMG G369a, as is common for spinosaurids generally (Hendrickx et al., 2019). However, a few spinosaurid crowns, notably from Lower Cretaceous Iberian deposits, do display shorter carinae that extend over only half or two-thirds of the crown height (Canudo et al., 2008; Hendrickx et al., 2019; Isasmendi et al., 2020). A similar feature is also seen in Iberospinus (Mateus & Estraviz-López, 2022). Charig & Milner (1997) described the carinae of BEXHM 1995.485 as failing to reach the cervix, however it would appear that the carinae have been chipped in places, and what remains basally seems to extend past the cervix.

Fluted enamel is typical of spinosaurid crowns (Hendrickx et al., 2019), and some have noted that these tend to be more numerous and better developed on the lingual surface (Buffetaut, 2012), further corroborating the orientation of the specimen described above. Those present on HASMG G369a, whilst generally weakly developed, are nevertheless in the range of several other spinosaurids: Baryonyx and cf. Suchomimus average around 6–7 flutes (range 4–8 and 2–10 respectively), whilst an average of 7–8 flutes are observed in Irritator (range 5–10) (Hendrickx et al., 2019). A similar range (3–9 flutes) has been observed in spinosaurid crowns from Lower Cretaceous Iberian localities (Ruiz-Omeñaca et al., 2005). However, the number of flutes in HASMG G369a differs from “Suchosaurus” (10–12 flutes) and several spinosaurines (17–20 flutes) (Hendrickx et al., 2019). The presence of flutes on both sides of the tooth also makes HASMG G369a different from Baryonyx walkeri (where the flutes are almost entirely lingually located), and is instead similar to the condition present in Ceratosuchops, Riparovenator, “Suchosaurus” and cf. Suchomimus.

Other forms of enamel ornamentation, such as the transverse undulations observed in some Baryonyx (NHMUK PV R9951), Iberospinus (ML1190) and cf. Suchomimus crowns (e.g., MNN G67-1), or the marginal undulations present in Baryonyx, Irritator (SMNS 58022), cf. Suchomimus (e.g., MNN G35-9) and indeterminate Brazilian spinosaurines (Hendrickx et al., 2019; Hendrickx, Tschopp & Ezcurra, 2020; Medeiros, 2006), are absent in HASMG G369a. Elsewhere, HASMG G369a shares with spinosaurids a lack of interdenticular sulci (Hendrickx et al., 2019).

The enamel texture of HASMG G369a is unusual in that two morphotypes are present: a veined textured basally and a more irregular texture apically. The former is common in spinosaurids and synapomorphic for the clade: it is present in Baryonyx (NHMUK PV R9951), Iberospinus (ML 1190) and various cf. Suchomimus crowns (e.g., MNN G35-9) (Hendrickx et al., 2019). Veined enamel texture is also present in Ceratosuchops inferodios (IWCMS 2014.95.5) and Riparovenator milnerae (IWCMS 2014.95.6) (C. Barker, pers. obs., 2022). Indeed, HASMG 639a also possesses the strong basal curvature of the veined texture towards the adjacent carinae, which is characteristic of the clade (Hendrickx et al., 2019; Mateus et al., 2011). However, an irregular enamel texture has so far only been reported for some Irritator crowns among spinosaurids (Hendrickx et al., 2019).

Differences in dental characters have been used to discuss the taxonomy of isolated spinosaurid teeth (Fanti et al., 2014; Richter, Mudroch & Buckley, 2013), though the utility of several traits has been questioned (Hendrickx, Mateus & Buffetaut, 2016). Tooth-bearing spinosaurid bones often lack erupted in-situ teeth, rendering variation between teeth within a complete tooth row poorly understood. Where teeth can be assigned to a single individual, as in the Baryonyx walkeri holotype NHMUK PV R9951, variation in ornamentation is documented (Hendrickx, Mateus & Buffetaut, 2016). Theropod dentition is also known to vary ontogenetically (Hendrickx et al., 2019) and it remains possible that differences in spinosaurid crown ornamentation may reflect ontogeny or tooth position more than phylogenetic position (Hendrickx, Mateus & Buffetaut, 2016).

Spinosaurid teeth are sometimes confused for those of crocodyliforms (Bertin, 2010; Buffetaut, 2010; Hone, Xu & Wang, 2010; Sánchez-Hernández, Benton & Naish, 2007), and the latter are well represented and taxonomically diverse in the Purbeck Group and Wealden Supergroup of southern England (Benton & Spencer, 1995; Salisbury, 2002; Salisbury & Naish, 2011). The crocodyliform fauna recovered from the Hastings Group is dominated by goniopholidids but also includes atoposaurids, bernissartiids and indeterminate mesoeucrocodylians and eusuchians (Salisbury & Naish, 2011). However, we can confidently dismiss a crocodyliform origin for HASMG G396a based on several lines of evidence.

Numerous “ridges” (i.e., flutes) ornament the enamel of goniopholidid and pholidosaurid crowns; in Goniopholis and Pholidosaurus for instance, these are well defined and closely packed (Allain et al., 2022; De Andrade et al., 2011; Martin, Raslan-Loubatié & Mazin, 2016; Owen, 1840–1845; Owen, 1878; Owen, 1879), whereas those of HASMG G369a are fewer and poorly defined. Interestingly, Owen (1840–1845) drew attention to the differences present between enamel ornamentation of “Suchosaurus cultridens” relative to that of Goniopholis. Smooth carinae are observed in goniopholidids generally, although false-ziphodont serrations are present in some taxa (e.g., G. kiplingi) (De Andrade et al., 2011; Puértolas-Pascual, Canudo & Rabal-Garcés, 2015; Salisbury et al., 1999). The latter are clearly distinguishable from the true denticles of HASMG G369a. Similarly, the mesial and distal carinae of pholidosaurids such as Pholidosaurus lack denticles, and can barely be differentiated from the flutes on the enamel surface (Martin, Raslan-Loubatié & Mazin, 2016). HASMG G396a is evidently not referable to atoposaurids, due both to the small size (<1m) of representative taxa (e.g., Theriosuchus) (Schwarz & Salisbury, 2005) and their distinctive distal dentition (Salisbury & Naish, 2011; Young et al., 2016). Fluted, conical teeth are present in the mesial dentition of bernissartiids, but these are also represented by small (<1m) taxa (Martin et al., 2020; Sweetman, Pedreira-Segade & Vidovic, 2015). In addition, their mesial teeth lack serrations and possess incipient cervical constriction (Martin et al., 2020; Norell & Clark, 1990). The short, rounded posterior crowns of bernissartiids are also obviously incompatible with the conidont morphology of HASMG G369a (Martin et al., 2020; Norell & Clark, 1990; Sweetman, Pedreira-Segade & Vidovic, 2015). In conclusion, we can reject with confidence the possibility that HASMG G369a might be considered referable to Crocodyliformes.

The British spinosaurid record and biogeography of early spinosaurids

Most British spinosaurid skeletal (i.e., non-dental) material has been recovered from the Barremian strata of Surrey (Upper Weald Clay Formation) and the Isle of Wight (Wessex Formation and base of the Vectis Formation) (Barker et al., 2021; Barker et al., 2022; Charig & Milner, 1986; Charig & Milner, 1997; Martill & Hutt, 1996; Milner, 2003). However, spinosaurid teeth are relatively common throughout the Wealden Supergroup (Fowler, 2007; Turmine-Juhel et al., 2019). While this is well known, the extent of the British spinosaurid record, and how it compares to that of other localities globally, has yet to be rigorously analysed.

The spinosaurid crown BEXHM 1995.485 is briefly described by Charig & Milner (1997) as originating from the “Ashdown Sand (Hauterivian)” near Bexhill in East Sussex, which Milner (2003) considered to be the earliest record of Spinosauridae. The term “Ashdown Sands” is now defunct (Hopson, Wilkinson & Woods, 2008), having been introduced by Drew (1861) before being formalised to Ashdown Formation by Rawson (1992). The latter is now considered late Berriasian to early Valanginian in age (Hopson, Wilkinson & Woods, 2008). More recently, Turmine-Juhel et al. (2019) described and figured two poorly preserved crowns (BEXHM 2019.49.251 and BEXHM 2019.49.253) which they referred to Baryonyx sp. All three teeth were found at the same site—the Pevensey Pit at Ashdown Brickworks (Turkey Road, Bexhill-on-Sea; J Porter, D Brockhurst, pers. comm., 2022)—where the only exposures are of the Valanginian Wadhurst Clay Formation (Turmine-Juhel et al., 2019). BEXHM 1995.485 therefore cannot be Hauterivian or from the Ashdown Formation, contra Charig & Milner (1997) and Milner (2003).

Modern interest in spinosaurids has resulted in the discovery of several Wealden Supergroup teeth in collections of crocodylomorph material housed in various institutions (Buffetaut, 2007; Buffetaut, 2010; Fowler, 2007; Milner, 2003). However, the historic nature of many of these specimens impacts our ability to identify their precise stratigraphic position. Fowler (2007) described a pair of spinosaurid crowns within a collection of goniopholidid teeth (NHMUK PV R1901) from the “Wealden” of Hastings, a provenance which would make them Valanginian or possibly Berriasian. Elsewhere, Bertin (2010), following Lydekker (1888), listed a “Suchosaurus cultridens” crown (NHMUK PV R635) as originating from the Berriasian-Valanginian “Hastings Sands” of Sandown. Older works suggested that the “Hastings Sands” were represented on the Isle of Wight (White, 1921). However, the oldest exposed Wealden Supergroup strata on the Isle of Wight are from the entirely Barremian upper portion of the Wessex Formation (Radley & Allen, 2012b; Sweetman, 2011) and this specimen is probably Barremian in age.

It would thus appear that the oldest British spinosaurid material is definitively Valanginian in age, with Berriasian occurrences remaining a possibility for some specimens of undetermined provenance. In comparison, the oldest specimens from Iberia—the other European hotspot for spinosaurid remains—are late Hauterivian in age (Malafaia et al., 2020). Fowler (2007) described and figured a “saurian” tooth (DCM-G95a) potentially recovered from the Purbeck Group of Swanage (Dorset, UK), which possesses several spinosaurid characters such as fluted enamel ornamentation. However, it is not dissimilar to plesiosaur tooth crowns (Fowler, 2007) and is indeed most likely from a marine reptile (D Fowler, pers. comm., 2022).

Alleged Jurassic spinosaurid teeth have been reported from Tanzania (Buffetaut, 2012) and Niger (Serrano-Martínez et al., 2015; Serrano-Martínez et al., 2016). However, similarities with other theropod clades (notably ceratosaurs and megalosaurids) have been noted and doubts have been cast on the identification of these specimens (Hendrickx et al., 2019; Soto, Toriño & Perea, 2020). An additional putative spinosaurid tooth—initially compared with the above mentioned Tanzanian material—has been described from the Jurassic of France (Vullo et al., 2014). Insufficient data exists to regard this identity as secured and, like the above Tanzanian “spinosaurid” specimens, it is probable that this tooth is also non-spinosaurid. Thus, whilst Spinosauridae likely evolved during the Jurassic (Barker et al., 2021; Carrano, Benson & Sampson, 2012), definitive Jurassic material pertaining to the group remains elusive. Moreover, associated discussion regarding the early evolution of spinosaurid teeth, with a proposed gradual acquisition of adaptations towards piscivory (Buffetaut, 2012; Serrano-Martínez et al., 2015; Serrano-Martínez et al., 2016), are best considered speculative pending further data (Hendrickx et al., 2019; Soto, Toriño & Perea, 2020).

A small, conidont crown (LPUFS 5737) from the Berriasian–Valanginian of Brazil (Sales et al., 2017) may represent one of the oldest spinosaurid occurrences globally. Additional spinosaurine teeth, as well as specimens referred to Baryonychinae (e.g., LPUFS 5870) or regarded as indeterminate spinosaurids (e.g., LPUFS 5871), have also been recently recovered from the locality (Aragão, 2021; Lacerda et al., 2023). We note that the identification of these specimens is based on (sometimes limited) qualitative data and would benefit from additional support generated using cladistic, discriminant and cluster analyses, as advocated for isolated theropod teeth in general (Hendrickx, Tschopp & Ezcurra, 2020). Nevertheless, evidence for spinosaurids in deposits of Berriasian–Valanginian age could complicate the biogeographic scenario proposed for the clade by Barker et al. (2021), as independently suggested by Lacerda et al. (2023). Barker et al. (2021) regarded Europe as the ancestral region but did not include specimens known from isolated teeth in their analyses. As a result, alternative biogeographical scenarios include earlier instances of dispersal from the proposed European ancestral area, or a different ancestral area altogether.

Spinosaurid persistence in the Late Cretaceous and status of specimen XMDFEC V10010

The results of the discriminant function analyses (Supplementary Information) show that XMDFEC V10010 does not associate with Spinosauridae when classified at either the clade or genus level. At the clade-level, the specimen was consistently classified as an allosauroid (Metriacanthosauridae or Allosauridae; reclassification rates = 54.46–62.12%; PC1 37.97–57.88%, PC2 19.11–31.01%), regardless of the dataset or whether serrations were considered inapplicable. At the genus-level, the allosauroid signal was retained, with the tooth most commonly referred to Early Cretaceous Erectopus, a tetanuran previously referred to Allosauroidea (and possibly Metriacanthosauridae) (Carrano, Benson & Sampson, 2012). XMDFEC V10010 was also referred to the megalosauroid Condorraptor and the abelisaurid Skorpiovenator in some genus-level DFAs. Reclassification rates in the genus level analyses were generally similar to those at the clade level analyses, and ranged between 57.4–63.68%.

The cluster analyses using the hierarchical clustering option consistently recovered XMDFEC V10010 as the sister taxon to an indeterminate abelisaurid. Similarly, the neighbour-joining option also commonly recovered the tooth as sister to an indeterminate abelisaurid, with several analyses of the whole dataset also recovering XMDFEC V10010 as a sister taxon to Abelisauridae indet. + Fukuiraptor.

The conflicting signals produced by the above quantitative analyses on XMDFEC V10010 are perhaps expected given that the dentition of Metriacanthosauridae and Allosauridae are considered the closest to that of Abelisauridae (Hendrickx, Tschopp & Ezcurra, 2020), although these allosauroid clades are not known from the Late Cretaceous (Carrano, Benson & Sampson, 2012). In comparison, abelisaurids were successful and diverse during the Late Cretaceous but are poorly represented in Asian deposits (outside of India) (Carrano & Sampson, 2008; Delcourt, 2018). Their teeth are nevertheless relatively diagnostic; however, the dental characters that unite Abelisauridae involve the shape of the premaxillary and maxillary alveoli (which are unknown for XMDFEC V10010) or relate to the morphology of the denticles (which are somewhat worn in XMDFEC V10010; Hone, Xu & Wang, 2010) (Hendrickx et al., 2019; Hendrickx, Tschopp & Ezcurra, 2020). Cladistic analyses of XMDFEC V10010 based on first hand examination of the specimen would be beneficial, and we refrain from referring the tooth to a theropod clade without this additional line of evidence. However the quantitative evidence presented herein corroborates previous suggestions that XMDFEC V10010 cannot be referred to Spinosauridae (Buffetaut et al., 2019; Katsuhiro & Yoshikazu, 2017; Soto, Toriño & Perea, 2020). With the Patagonian late Cenomanian-early Turonian tooth referred to Spinosauridae in Salgado et al. (2009) also likely from a different theropod lineage (Soto, Toriño & Perea, 2020), the youngest definitive spinosaurid remains appear to come from Cenomanian deposits of Africa (Benyoucef et al., 2022; Ibrahim et al., 2020b; Sereno et al., 2022).

Assuming the reinterpretation of the above-mentioned Chinese and Patagonian specimens is correct, the potential extinction of Spinosauridae around the Cenomanian–Turonian boundary (CTB) remains poorly understood (Candeiro, Brusatte & De Souza, 2017). This time interval coincides with the peak Cretaceous greenhouse climate and a major marine transgression, and a marine extinction event has been documented (Kerr, 2014; Sepkoski, 1986). However, studies of the faunal changes in terrestrial, freshwater and brackish water environments during this transition are rare, and available data from North America suggests these faunas were not (a few taxa excepting) overly affected (Benson et al., 2013; Eaton et al., 1997). Spinosaurids are not definitively known from the Mesozoic of North America, however, and it may be that results inferred from these deposits may not be applicable elsewhere. Moreover, as theropods that have been positively associated with costal palaeoenvironments (Sales et al., 2016), it is interesting to speculate upon the impact of the CTB marine transgression on available spinosaurid habitat, and certainly warrants further consideration as a potential driver of their apparent extinction.