An isolated theropod tooth was found in the Hauterivian–Barremian Itsuki Formation of the Tetori Group in the Kuzuryu district, Ono City, Fukui Prefecture, central Japan. The present specimen, OMFJ V-1, shows a thick lanceolate basal cross-section and small mesial and distal denticles. A cladistic analysis based on the dental characters suggested that OMFJ V-1 be classified as belonging to Allosauroidea or Tyrannosauroidea. Principal component and linear discriminant analyses also suggested that OMFJ V-1 belongs to either of these two theropod clades. The posterior probabilities obtained in the linear discriminant analyses indicated that the confidence of the classification as Allosauroidea is slightly higher than that for Tyrannosauridae. However, because these analyses also supported possibilities of OMFJ V-1 belonging to other theropod clades to lesser extents, its taxonomic referral remains ambiguous. If OMFJ V-1 belongs to Tyrannosauroidea, it would indicate that a medium-sized tyrannosauroid already appeared in central Japan during the Hauterivian–Barremian age. On the other hand, if OMFJ V-1 belongs to Allosauroidea, it would indicate that at least two medium-to-large-sized theropods, allosaurids and tyrannosaurids, lived almost coevally in this region. The third possibility is that OMFJ V-1 belongs to Megaraptora. If such affinities are established, it would represent the oldest record of this clade of theropods.
Introduction
Isolated teeth are among the most frequently discovered vertebrate fossils, sometimes being found even in regions that rarely yield vertebrate bony elements, and thus provide important information on taxonomic compositions of paleoecological systems in fossil-poor regions or strata.
Although several recent studies have reported nearly complete dinosaur fossils from several regions in Japan (e.g. Azuma et al., 2016; Imai et al., 2019; Kobayashi et al., 2019), most Mesozoic terrestrial vertebrate fossils from Japan are fragmentary or isolated elements (e.g. Tsuihiji et al., 2013; Tanoue and Okazaki, 2014; Hirayama et al., 2020; Sakai et al., 2020a). To fully understand environments, ecosystems, and organismal diversity in the Mesozoic of Japan, it is important to analyze such fragmentary elements. In particular, isolated teeth, especially those of dinosaurs, have been shown to provide important information on the diversity and biogeography of vertebrates in Mesozoic localities worldwide (e.g. Barrett et al., 2002; Molnar et al., 2009; Larson and Currie, 2013).
Theropod teeth have relatively simple and conservative shapes, showing a high level of similarity and/or homoplasy among taxa (e.g. Smith et al., 2005; Hendrickx and Mateus, 2014; Hendrickx et al., 2015b, c, 2019). Although it is difficult to determine species-level identification based only on isolated teeth, many studies have attempted to assign isolated teeth to higher taxa, ranging from “superfamily” to genus levels, based upon the dental morphology and through qualitative and quantitative analyses (e.g. Currie et al., 1990; Fiorillo and Currie, 1994; Rauhut and Werner, 1995; Larson, 2008; Molnar et al., 2009; Tsuihiji et al., 2013; Hendrickx and Mateus, 2014; Hendrickx et al., 2015c, 2020; Gerke and Wings, 2016; Averianov et al., 2019; Young et al., 2019). In such attempts, two kinds of analytical methods, morphometric and cladistic analyses, have been mainly employed. Although dental morphology varies even within the dentition of one jaw (e.g. Buckley et al., 2010; Reichel, 2012), principal component analysis (PCA) and discriminant function analysis (DFA) using tooth morphological data have demonstrated some, if not complete, discrimination among different theropod groups (e.g. Currie et al., 1990; Smith et al., 2005: Gerke and Wings, 2016; Hendrickx et al., 2019, 2020; Young et al., 2019). In addition, cladistic analyses conducted in recent studies identified isolated theropod teeth to certain taxonomic levels (e.g. Hendrickx and Mateus, 2014; Hendrickx et al., 2019, 2020; Young et al., 2019). Most recently, Wills et al. (2020) showed the usefulness of machine learning techniques as a third method for taxonomic identification of isolated theropod teeth although this method is yet to be in wide use.
Here, we describe an isolated theropod tooth found in the Kuzuryu district, Ono City, Fukui Prefecture, central Japan, and attempt to identify its taxonomic affinities using morphometric and cladistic analyses. Implications of the resulting taxonomic identification for the terrestrial faunal and paleoecological changes in the northern part of central Japan are also discussed.
Geological setting
The Kuzuryu district surrounding the upper reaches of the Kuzuryu River occupies the southeastern part of Fukui Prefecture in central Japan. The Mesozoic sequence in the Kuzuryu district is composed of the Middle to Upper Jurassic Kuzuryu Group (Maeda, 1952; re-defined by Yamada and Sano, 2018) and the overlying Lower Cretaceous Tetori Group (Figure 1). The Tetori Group (Oishi, 1933; re-defined by Yamada and Sano, 2018) consists of terrestrial and shallow marine sediments and is widely distributed in the Inner Zone of southwest Japan. The Tetori Group in the type area along the Itoshiro River (Yamada and Sano, 2018) is divided into the Yambara, Ashidani, Obuchi, Itsuki and Nochino formations in ascending order (Sakai et al., 2020b; Figure 2) and is overlain by Hayashidani Andesite or Upper Cretaceous acid igneous rocks.
The Itsuki Formation consists mainly of alternating beds of sandstone and mudstone and yields terrestrial plant and animal fossils. The fossil vertebrate fauna includes dinosaurians (including bird tracks), testudines, squamates, choristoderans, therapsids (including tritylodontids) and actinopterygian fish (Manabe, 1999; Azuma et al., 2002; Hirayama et al., 2020; Sakai et al., 2020a). Goto (2007) reported the occurrence of the late Hauterivian ammonoid Pseudothurmannia sp. from the Itsuki Formation in the northern part of the Kuzuryu district. The youngest zircon grain from sandstone of the upper part of the Itsuki Formation in the type area has a concordant age of 127.2 ± 2.5 Ma (Kawagoe et al., 2012). The zircon grains from the tuff of the Akaiwa Formation of the Tetori Group in the Shiramine area, which is correlated with the Nochino Formation in the Kuzuryu district, has an age of 121.1 ± 1.1 Ma (Sakai et al., 2019). The K-Ar dating of the lower part of Hayashidani Andesite overlying the Tetori Group indicates 99.4 ± 5.0 Ma (Tanase et al., 1994). Based on these data, the age of the Itsuki Formation is inferred to be Hauterivian-Barremian (Hirayama et al., 2020; Sakai et al., 2020a, b).
Material and methods
Material
The specimen described herein is cataloged as OMFJ V-1 at the Izumi Local History Museum, Ono City, Fukui Prefecture (Figure 3). It is an isolated theropod tooth found in the lower part of the Itsuki Formation in the Shimoyama region north of Kuzuryu Lake in Ono City, Fukui Prefecture (35°55.56' N, 136°39.46' E; Figure 1). At the locality, the alternating beds of coarse-grained sandstone, fine-grained sandstone, and dark gray mudstone are exposed. Occasional thin layers of coal and in situ rootlets are also present within the fine-grained layers. The freshwater molluscs, oogonia of charophytes, and ostracods commonly occur in dark gray mudstones. The molluscan assemblage is represented by the abundant occurrence of unionid bivalves and viviparid gastropods. The unionid bivalves are commonly articulated and not accumulated in layers. These fossils indicate that subaqueous deposition of sediment occurred in a lake or shallow pond on the floodplain.
OMFJ V-1 was collected in a medium to coarse-grained, lenticular sandstone layer (50 cm in maximum thickness) intercalated within the alternations. The specimen was associated with fragmentary turtle carapaces, fish scales, viviparid gastropods, numerous plant fragments, and small mud clasts.
Anatomical nomenclature and examined variables
Anatomical nomenclature of the tooth in the present description follows Hendrickx et al. (2015b). As for quantitative variables (e.g. length) used for morphometric analyses, definitions or measuring schemes have varied among past studies of theropod dental morphology, making comparison or compilation of morphometric data from such studies for the purpose of producing large data sets very difficult (Gerke and Wings, 2016; Hendrickx et al., 2020). Several studies measured theropod teeth in situ on jaw bones (e.g. Smith et al., 2005; Buckley et al., 2010). However, as Hendrickx et al. (2020) mentioned, methods used for such in situ teeth are difficult to apply to isolated tooth crowns. Therefore, the measuring scheme for OMFJ V-1 in the present study generally followed Hendrickx et al. (2015b, 2020), which dealt with isolated teeth.
Figure 1.
Geological map of the Itoshirogawa area, Ono City, Fukui Prefecture (modified from Sakai et al., 2020b). The star indicates the locality yielding the isolated theropod tooth, OMFJ V-1.
The following nine variables were taken on OMFJ V-1: CH, crown height; CBL, crown basal length; CBW, crown basal width; AL, apical length; MCL, mid-crown length; MCW, mid-crown width; MCA, mesioapical-central denticle density per 5 mm; DC, distocentral denticle density per 5 mm; DB, distobasal denticle density per 5 mm (Table 2).
MA (mesioapical denticle density per 5 mm) and DA (distoapical denticle density per 5 mm) of Hendrickx et al. (2015b) were not determined for OMFJ V-1 and thus were not used in the present analyses, because the enamel layer on the apical surface of this specimen is completely worn out. Because of such missing denticles, the average mesial and distal densities (MAVG and DAVG) also had to be estimated based on the values of preserved denticles only, whereas the denticle size density index (DSDI: the ratio of MAVG to DAVG) was obtained using the formula of Smith et al. (2005) and Hendrickx et al. (2015b). In addition, CA (crown angle), CBR (crown base ratio) and CHR (crown height ratio) were not directly measured on OMFJ V-1; they were instead calculated based on the formula proposed by Smith et al. (2005) and Hendrickx et al. (2015b).
Figure 2.
Stratigraphic column of the Tetori Group in the Itoshirogawa and Shiramine-Takinamigawa areas.
The base length of each denticle, i.e., the mesial denticle length (MDL) and distal denticle length (DDL), were measured from digital photographs taken with a Canon Eos Kiss X6i digital camera mounted on a Leica M125 stereo microscope to calculate the mesial and distal denticle densities. Photographs were taken of the carinae in labial and lingual views, and the base length of each denticle was measured using the free software Image J 1.50i (Schneider et al., 2012) in millimeter scale. After the average value of the base lengths of denticles was calculated, the inverse of each average was multiplied by five to produce the denticle density on each carina.
Although MC (mesiocentral denticle density per 5 mm) has been used as the parameter for the mesial denticle density in many studies (e.g. Smith et al., 2005; Hendrickx and Mateus, 2014; Hendrickx et al., 2015c, 2020; Gerke and Wings, 2016), denticles of OMFJ V-1 on the mesial margin of the middle-crown are worn out or absent. Therefore, for the following analyses, the mesioapical-central denticle length was used instead of the MDL and its denticle density (MCA, average number of denticles per five millimeters on the mesial carina at two-thirds crown height; Hendrickx and Mateus, 2014) was used instead of MC in the present study.
Four measurements of the length (CBL, CBW, CH and AL) were taken with calipers to the closest one hundredth mm. All values were measured three times and their averages were used for the analyses. Because the cervix is indistinct on the distal margin in OMFJ V-1 and the measuring methods of theropod teeth differ slightly between authors as mentioned above, CBL, CBW, CH and AL were measured in the following three ways. In the first way (OMFJ V-1 Pattern A), the preserved basalmost point of the distal carina is regarded as indicating the position of the cervix (dc in Figure 4) although the distal margin of OMFJ V-1 is broken at around the level of the cervix. In the second way (OMFJ V-1 Pattern B), to accommodate uncertainty caused by the broken distal margin at the level of the cervix, the basalmost point of the crown on the distal margin was inferred based upon partially preserved enamel on the lingual and labial surfaces (dc' in Figure 4). In the third way (OMFJ V-1 Pattern C), the line perpendicular to the ap-dc line (the crown height) was regarded as the crown length (dc-mc' line in Figure 4). This is the same method adopted by Gerke and Wings (2016) and White et al. (2015). It is noteworthy that the mesialmost point on the mesial margin in Pattern C is not on the cervix (mc' in Figure 4). All values based on these three methods of measurement were used for the present analyses to accommodate uncertainty of the anatomical landmarks of OMFJ V-1.
Figure 3.
Isolated theropod tooth found from the Itoshirogawa area, Ono City, Fukui Prefecture, OMFJ V-1, in mesial (A), labial (B), lingual (C), distal (D), and apical (E) views. The image in apical view (E) was produced using the image stacking software Helicon Focus 6.7.2 (Helicon Soft Ltd., Kharkov, Ukraine). Close up of distal denticles at the apical one-third of the crown height (F), distal denticles at the basal half of the crown height in labial view (G), mesial denticles at the apical one-third of the crown height in lingual view (H), and further close up of abnormal, mesial denticles (I). In I, arrowheads indicate two or more denticles of which cross-section circles of the dentine were merged, suggesting abnormally wave-shaped primary dentine. Scale bars equal 1 cm (A–E) and 1 mm (F–H).
Phylogenetic definitions
The phylogenetic definition of theropod taxa follows the review by Hendrickx et al. (2015a) and its corrigendum (Hendrickx and Carrano, 2016) with the following exceptions (Table 1). The phylogenetic position of megaraptoran theropods such as Megaraptor, Murusraptor, Orkoraptor, Australovenator and Fukuiraptor within Theropoda has been controversial as will be discussed below. In the present morphometric analyses, we regard Megaraptora as a subclade of Neovenatroidae, which in turn is considered as belonging to Allosauroidea. The phylogenetic position of Monolophosaurus inferred by cladistic analyses is highly unstable, ranging from the sister taxon of Avetherapoda (e.g. Smith et al., 2007; Zhao et al., 2010), a basal species of Tetanurae (e.g. Carrano et al., 2012; Hendrickx et al., 2015a; Rauhut and Pol, 2019), to somewhere within Megalosauroidea (e.g. Benson, 2010; Rauhut et al., 2016; Dai et al., 2020). In this study, Monolophosaurus is treated as a basal member of Tetanurae following Hendrickx et al. (2015a).
CT scanning
CT scan data were obtained using the CT scanner facility at the National Museum of Nature and Science, Tokyo (TESCO Microfocus CT TXS320-ACTIS) to investigate the internal structures of the tooth. Raw TIFF data files were converted to DICOM files using VGStudio Max 2.0.5 (Volume Graphics GmBH, Heidelberg, Germany). Digital segmentation was conducted by Amira 5.5 (FEI, Hillsboro, Oregon, U.S.A.).
Cladistic analysis
We scored OMFJ V-1 in the data matrix of tooth crown characters of Hendrickx et al. (2020; Appendix 1) excluding theropods having highly specialized teeth for herbivory or omnivory (i.e., alvarezsaurids, ornithomimosaurians, therizinosaurids, oviraptorosaurians and juvenile Limusaurus inextricabilis; Zanno and Makovicky, 2011; Wang et al., 2017). Aerosteon riocoloradensis was also removed from the matrix because it is questionable whether its tooth, which was regarded as a part of the holotype specimen in the original description, indeed belongs to this taxon (Hendrickx et al., 2020). The character matrix consists of 91 dental characters and 86 genus-level theropod taxa. The analysis was conducted with equally weighted parsimony using TNT 1.1 (Goloboff et al., 2008) using the “Traditional search” option. Herrerasaurus ischigualastensis was used as the outgroup taxon following Young et al. (2019) and Hendrickx et al. (2020). One thousand replicates of tree bisection reconnection (TBR) branch swapping were run, holding 10 trees per replicate with all zero-length branches collapsed. Because some replicates produced more than 10 most parsimonious trees, further branch-swapping starting from the most parsimonious trees initially obtained and stored in memory was conducted. Bootstrap values were obtained with 1000 replications. The consistency index (CI) and retention index (RI), as well as the Bremer supports, were also calculated for assessing the branch support.
Figure 4.
Line drawings of OMFJ V-1 in A, mesial; B, labial; C, lingual; and D, distal views. E, graph showing the basal length of the mesial denticle length (MDL); F, graph showing the basal length of the distal denticle length (DDL). Abbreviations: mc, the most mesial point of the cervix and used as the landmark in measuring Patterns A and B; dc, the most basal point of the preserved distal carina and the landmark in Patterns A and C; mc', the most mesial point on the basal cross-section at the level of dc and the landmark in pattern C; dc', the most basal point on the preserved tooth surface on the distal carina and the landmark in Pattern B; ap, the apex of the crown.
Principal component analysis (PCA) and linear discriminant functional analysis (LDA)
Principal component analysis (PCA) and linear discriminant analysis (LDA), the latter of which is a kind of discriminant function analysis (DFA), were performed using the software R ver. 3.4.2 (R Core Team, 2017) with the open R GUI software R AnalyticFlow (ef-prime, inc., Nihonbashi-Kayabacho, Japan).
The morphometric dataset of Hendrickx et al. (2020) consists of 1335 teeth belonging to 83 taxa. However, this dataset has missing data for several reasons (e.g. preservation condition, insufficient descriptions). All teeth containing any missing data were removed from the dataset, resulting in a revised dataset consisting of 721 teeth belonging to 70 theropod taxa, to which OMFJ V-1 was added. In addition, data of Fukuiraptor kitadaniensis, of which remains have been found from the Kitadani Quarry, near the locality of OMFJ V-1, and its close relative Australovenator wintonensis (e.g. Porfiri et al., 2014; Aranciaga Rolando et al., 2019) were added to the data set based on published photographs. Namely, three measurements (CBL, CH, AL) were taken from the photographs of F. kitadaniensis and A. wintonensis published in Currie and Azuma (2006, fig. 2) and White et al. (2015, figs. 7 and 8), respectively, using Image J 1.50i (Schneider et al., 2012). This addition resulted in a final dataset consisting of 738 teeth belonging to 72 theropod taxa and OMFJ V-1. Based on this dataset, the principal components were calculated using the pcomp function in R after all the measurements were standardized using the scale function.
Table 1.
Grouping of theropod genera used in the linear discriminant analysis in the present study.
It is noteworthy that Molnar et al. (2009) described the isolated tooth (NDC-P0001; Kanna Town Dinosaur Center, Kanna Town, Gunma Prefecture) from the Barremian Sebayashi Formation in the Sanchu area, Japan, and identified it as Fukuiraptor aff. F. kitadaniensis. This specimen would potentially be important in discussing the taxonomic affinity of OMFJ V-1 and paleobiogeography of megaraptorans. However, because NDC-P0001 lacks both mesial and distal denticles due to weathering, it was not included in the present dataset.
For LDA, the number of taxa was further reduced for the following reasons. First, taxa whose dental morphology was clearly different from that of OMFJ V-1 based on previous descriptions (i.e., Ornitholestes, oviraptorosaurians, troodontids and spinosaurids) were excluded. In addition, teeth possessing non-serrated mesial and/or distal carinae (i.e., MC = 0 and/or DC = 0) were removed from the dataset because such teeth with extreme MC and/or DC values greatly affect plot shapes in the morphospace and make the classification of OMFJ V-1 ambiguous. Furthermore, the tooth of Aerosteon was also excluded because of its doubtful referral to the holotype specimen (Hendrickx et al., 2020) as was the case with the cladistic analysis mentioned above. The resulting dataset for the linear discriminant analysis consists of 609 teeth belonging to 57 theropod genera. For the group-level LDA, these genera were categorized into the following higher-level clades or groups: non- and basal theropods, non-averostran neotherapods, non-abelisauroid ceratosaurians, Noasauridae, Abelisauridae, basal tetanurans, non-megalosaurian megalosauroids, Megalosauridae, Metriacanthosauridae, Allosauridae, Neovenatoridae, Carcharodontosauridae, non-tyrannosaurid tyrannosauroids, Tyrannosauridae, Therizinosauria and Dromaeosauridae (Table 1). Since the detailed phylogenetic affinities of Erectopus are uncertain, this taxon was excluded from the group-level LDA.
Table 2.
Measurements of OMFJ V-1. CBL, CBW, CH, AL, MCL, MCW, MDL and DDL in millimeters, MCA, DC and CB in numbers per 5 mm, and CA in degrees. Abbreviations: AL, apical length; CA, crown angle; CBL, crown base length; CBR, crown base ratio; CBW, crown base width; CH, crown height; CHR, crown height ratio; DAVG, average distal denticle density; DB, distobasal denticle density; DC, distocentral denticle density; DDL, distal denticle length; DSDI, denticle size density index; MAVG, average mesial denticle density; MCA, mesioapical-central denticle density per 5 mm (average number of denticles per five mm on mesial carina at two-thirds of the crown); MCL, mid-crown length; MCW, mid-crown width; and MDL, mesial denticle length.
As mentioned above, the dataset of Hendrickx et al. (2020) is a composite of those used in several previous studies and thus includes measurements made in various ways. Moreover, some measurement schemes shown in figures in previous studies such as White et al. (2015) and Gerke and Wings (2016) are not consistent with those defined by Smith et al. (2005) or Hendrickx et al. (2015b), even where they have been stated to be consistent. In fact, to consider potentially different results caused by non-uniform measuring methods, Hendrickx et al. (2020) recommended that the dataset of Smith et al. (2005) and Gerke and Wings (2016), and those for which measurements followed Hendrickx and his colleagues' scheme (e.g. data newly acquired by Young et al., 2019), should be analyzed separately. For these reasons, we constructed a reduced dataset by removing data that had been obtained using methods other than Hendrickx and his colleagues' scheme for a separate LDA. After adding OMFJ V-1 data measured according to Patterns A or B, this reduced dataset consisted of 412 teeth belonging to 66 theropod taxa.
LDA was conducted based on the datasets (without OMFJ V-1) with values standardized using lda function in MASS package (Venables and Ripley, 2002). The prior probabilities were set equal. The posterior probability of the re-classification score of each specimen and OMFJ V-1 was obtained by the predict function in R. Both analyses were performed using six length values (CBL, CBW, CH, AL, MDL and DDL). Four lengths (CBL, CBW, CH and AL) were log-transformed with base 10 to produce an approximately normal distribution for each variable (e.g. Smith et al., 2005; Gerke and Wings, 2016; Young et al., 2019; Hendrickx et al., 2020). Instead of the mesial and distal denticle densities, two denticle lengths (MDL and DDL) were used because the use of metric-based and continuous variables is preferred (Hendrickx et al., 2020). MDL and DDL of unserrated mesial and/or distal carinae were set to zero. These two denticle variables were transformed by the log10(x + 1) function to account for 0 values (Gerke and Wings, 2016).
Systematic paleontology
DIAPSIDA Osborn, 1903
DINOSAURIA Owen, 1842
THEROPODA Marsh, 1881
TETANURAE Gauthier, 1986
Horizon and locality
Lower part of the Itsuki Formation, Taniyamadani (Taniyama Valley), Ono City, Fukui Prefecture, Japan.
Description
OMFJ V-1 is a blade-like (ziphodont) theropod tooth (Figure 3). The crown is almost completely preserved with no substantial postmortem deformation although a few cracks are present on both labial and lingual surfaces, dividing the crown into several parts. The crown is moderately recurved. The most apical part is considerably abraded, with the enamel layer being completely worn. The enamel layer is preserved more basally on the crown. The basal cross-section is characteristically thick and lanceolate (Figure 6; CBR = 0.573; see details below). The mesial margin of the crown is strongly recurved distally whereas it is almost straight in the basal half. On the other hand, the distal margin of the crown is almost straight, with the apical end being slightly curved distally. Basal, straight portions of the distal and mesial margins are almost parallel to each other in labial and lingual views and are perpendicular to the transverse plane of the crown. The mesial and distal carinae are well-developed. Because the crown is labio-lingually compressed and blade-shaped with both carinae lying on almost the same sagittal plane, OMFJ V-1 is inferred to be a lateral theropod tooth (i.e., maxillary or distal dentary tooth).
In mesial view, the mesial carina appears almost straight from the apex to the half-height of the crown although it is slightly deflected to the left side. On the other hand, the distal carina is slightly deflected to the left side in distal view. In apical view, therefore, the mesial and distal carinae appear twisted with respect to each other. In general, theropod lateral teeth tend to possess a lingually twisting mesial carina and a labially deflected distal carina, in addition to a centrally positioned depression on the lingual side of the root (lingual depression; Figure 5) that may extend onto the basal part of the crown (Hendrickx and Mateus, 2014). Based on such tendency, the side toward which the distal carina is deflected in OMFJ V-1 can be inferred as the labial side and this orientation is assumed for the rest of the description.
The 3D-segmented pulp cavity reconstructed from the CT dataset is straight, thin, and conical (Appendix 2), demonstrating that it is preserved almost completely. Cross-sectional images of the basal part of the crown show that the distal carina is offset slightly from the mesiodistal axis of the pulp cavity, reflecting the labial deflection of the distal carina (Figure 6). The mesial carina is too small to be detected with the present resolution of CT slice images.
Some sediment is attached to the lingual surface of the most basal part of the crown. CT scan images of the basal part of the crown show this surface is cracked and sediment has invaded the pulp cavity (Figure 5) although crack damage is limited to the basolingual part of the crown. This mass of sediment contains many tiny tooth fragments. The remaining uncrushed part of the lingual surface, as well as removal of the attached sediment in the 3D-rendered model of the specimen based on the CT data, shows the lingual surface is slightly depressed in this area. This depression is most likely the lingual depression. This identification is consistent with the identification of the lingual and labial sides based the directions of deflections of the mesial and distal carinae described above.
On the labial surface, there are two weak transverse undulations. The marginal undulation is absent. The basal cross-sections of OMFJ V-1 show that the mesial margin is convex, whereas the distal margin is more pointed, with both labial and lingual sides being flat, making it distinctly lanceolate (Figure 6). Because these flattened labial and lingual sides are almost parallel to each other, the basal cross-section appears markedly thick and sub-rectangular, more similar to the lateral teeth of Tyrannosauridae than to the lanceolate or lenticular teeth of other theropods such as Carcharodontosaurus and Fukuiraptor (e.g. Stromer, 1931; Azuma and Currie, 2000; Molnar et al., 2009).
The root is almost completely lost except around the mesial margin, where the cervix is preserved. Apparently, the basal end of the preserved distal carina is positioned exactly at the same level as the cervix. However, it is hard to assess whether the apparent basal end of enamel on the distal margin indeed represents the cervix or it is merely an artefact of postmortem loss of enamel.
The mesial denticles persist for one-third of the crown from the apex, with the more basal part of the mesial carina being unserrated, although the exact boundary between the serrated and unserrated portions is worn and unrecognizable. Near the apical end of the crown, both mesial and distal denticles are almost completely damaged. More basally, however, they are fairly well preserved. On the mesial carina, 20 mesial denticles are preserved; 15 of these denticles retain nearly completely preserved basal parts that were used for measurement of the base length. One mesial denticle is abnormally shaped, as if two denticles had been fused together (Figure 3I). All preserved mesial denticles are missing their tips (Figure 3H). However, it is possible to infer that each mesial denticle is symmetric and parabolic in labial or lingual view based on the shape of the preserved basal parts. The distal carina, on the other hand, extends up to the preserved basal end of the distal margin and is serrated entirely. Fifty three denticles are preserved, with 40 of them retaining basal parts that were measured for the base length. Although most distal denticles are partially worn and missing their tips, nearly completely preserved ones show that they are rectangular in labial or lingual view and are almost perpendicularly oriented to the distal margin of the crown.
The base lengths of the denticles are almost constant over the mesial carina. On the other hand, those on the distal carina become slightly smaller basally (i.e., the denticle density becomes slightly larger toward the more basal part of the crown; Figure 4). MC and DC of OMFJ V-1 are 17.52 and 15.81, respectively. It follows that OMFJ V-1 has slightly higher denticle densities than teeth of medium to large-bodied theropod taxa (Appendix 3; e.g. Currie et al., 1990; Smith et al., 2005; Hendrickx and Mateus, 2014). DSDI is 1.139, indicating that the basal length of each mesial denticle is slightly smaller than that of the distal denticles typical of theropod teeth.
The interdenticular sulci are present but are rather short and poorly defined on the distal carina whereas they are absent on the mesial carina.
Results
Cladistic analysis
Cladistic analysis resulted in 10920 most parsimonious trees (CI = 0.271; RI = 0.598). The strict consensus of these trees distinguished Tsaagan, Spinosauridae, and a large clade consisting of all remaining taxa. The large clade, in turn, was an unresolved polytomy of four species, three small clades consisting of 5 to 12 species and two specious ones (Figure 7). All the nodes recovered in the strict consensus tree had low bootstrap and Bremer support values except for the spinosaurid clade. OMFJ V-1 was placed in one of the two specious clades and, more specifically, in a small clade consisting of tyrannosauroids, allosauroids (Allosaurus, Neovenator and Sinraptor) and Piatniztkysaurus (hereafter called the TAP clade). This clade was sister to a clade consisting of megaraptorans (Fukuiraptor, Orkoraptor and Australoventator) and Dromaeosaurus (hereafter called the MD clade). Successively more distant outgroups for these clades combined were Eustreptospondylus and a clade consisting of carcharodontosaurids (Giganotosaurus, Carcharodontosaurus, Mapusaurus, Acrocanthosaurus and Eocarcharia) and megalosauroids (Dubreuillosaurus, Duriavenator, Torvosaurus, Megalosaurus and Afrovenator). It is noteworthy that no ceratosaurans were included in this paraphyletic assemblage.
Table 3.
Proportions of the variance and loading of each principal component.
Figure 8.
Plots of PC 1 versus PC 2 scores obtained as the result of the principal component analysis. The proportions of variance of PC 1 and PC 2 are 85.68% and 10.11%, respectively. Convex hulls show the variation ranges of groups. Each of the six arrows indicates relative values of coefficients for each variable on PC 1 and PC 2.
Figure 9.
Plot of LD 1 versus LD 2 scores obtained as the result of the linear discriminant analysis performed at the genus-level. The proportion of trace of LD 1 and LD 2 are 75.16% and 13.39%, respectively. Convex hulls show the variation ranges of groups. Six white arrows indicate the relative values of the coefficients of each variable on LD 1 and LD 2.
To examine synapomorphic characters supporting each clade recovered in the strict consensus tree, character states were mapped using TNT 1.1 (Goloboff et al., 2008) and MESQUITE ver. 3.61 (Maddison and Maddison, 2019). The TAP clade consisting of OMFJ V-1, tyrannosauroids, allosauroids and Piatnitzkysaurus was supported by characters related to mesial teeth (Characters 41, 42 and 50 of Hendrickx et al., 2020), although all these characters could not be coded for OMFJ V-1. OMFJ V-1 and the other genera in the TAP clade except Eotyrannus shared a roughly flattened labial surface (State 1 in Character 74 of Hendrickx et al., 2020), which was also shared with some of the ceratosaurians (Ceratosaurus, Genyodectes and Berberosaurus), Dilophosaurus and Erectopus, although the derived tyrannosaurids (Tyrannosaurus, Daspletosaurus and Gorgosaurus) and Allosaurus shared a convex labial surface of the crown (State 0 in Character 74 of Hendrickx et al., 2020). The MD and TAP clades were united only by characters of mesial teeth (Character 50 and 53 of Hendrickx et al., 2020). The more inclusive clade originating from the node subtending these clades and Eustreptospondylus was supported by the presence of denticles over the tip or very close to the apex on mesial teeth and the CBR value of lateral teeth (Characters 62 and 70 of Hendrickx et al., 2020) although the former character could not be coded for OMFJ V-1. The latter character, moderately compressed lateral teeth (0.5 < CBR < 0.75), was widely shared with almost all theropod taxa in this analysis. These results suggest that OMFJ V-1 is likely either an allosauroid, tyrannosauroid or a close relative of Piatnitzkysaurus.
Morphometric analysis
Bivariate plots among six morphometric variables for PCA and LDA showed that all variables were more or less correlated with each other (Figure 8; Smith et al., 2005; Buckley et al., 2010). Although such a relationship among these morphometric variables suggests a potentially problematic lack of independence among them (i.e., multicolinearity), we proceeded to conduct PCA and LDA as previous studies had done. The result of the principal component analysis showed a general trend of theropod dental morphology (Figure 8). The proportions of variance accounted for by PC 1 through 4 were 85.7%, 10.1%, 2.35% and 1.28%, respectively (Table 3). Those explained by PC 5 and 6 were less than 1% each. The loadings of all variables on PC 1 were positive and had almost identical absolute values, consistent with the general notion that PC 1 reflects size variation. On PC 2, in contrast, the loadings for variables related to tooth size (CBL, CBW, CH and AL) were all negative whereas those for variables related to the denticle size (MDL and DDL) had large positive values, indicating that the PC 2 reflected the contrast between denticle and tooth sizes. PC 3 mainly reflected the contrast in denticle size between mesial and distal carinae whereas PC 4 reflected the contrast of the crown width versus other dimensions of the crown. The plot of PC 1 against PC 2 (Figure 8) showed weak separation between small (basal and non-theropods, non-averostran neotherapods, noasaurids, non-tyrannosaurid tyrannosauroids and dromaeosaurids) and large taxa (non-abelisauroid ceratosaurians, Abelidauridae, non-megalosaurian megalosauroids, Megalosauridae, Metriacanthosauridae, Allosauridae, Neovenatoridae, Carcharodontosauridae and Tyrannosauridae). In this plot, areas of distribution of Therizinosauria and Troodontidae, both of which are characterized by considerably large denticles, were similarly placed and approximately half of the area of each clade had no overlap with the area of any small theropod taxa. Spinosauridae, which has fairly small denticles or completely lacks them, occupied a fairly distinct area although it partially overlapped other taxa, especially Abelisauridae, due to expansion of the distribution envelopes by what appear to be outliers (cf. Baryonychinae [Xixia Museum of Dinosaur Fossil Eggs of China, XMDFEC V0010]). Although most areas of large theropods overlapped, those of some groups (tyrannosaurids [mainly Tyrannosaurus], Megalosauridae [mainly Torvosaurus] and Carcharodontosauridae [mainly Carcharodontosaurus]) were partly diverged from those of other large theropods mainly due to their large sizes of teeth and denticles. All three plots of OMFJ V-1 based on different measuring schemes (Patterns A through C) were placed inside of areas of distribution of seven large-bodied theropod groups, i.e., Abelisauridae, basal Tetanurae, non-megalosaurian megalosauroids, Megalosauridae, Allosauridae, Neoventoridae and Tyrannosauridae. The effect of using different measuring schemes on the distribution pattern was not large enough to change the classification dramatically, with the plots of OMFJ V-1 measured in Patterns A and C located at almost identical positions. On the other hand, the OMFJ V-1 plot measured in Pattern B was located at a distance from those measured in Patterns A and C, indicating that damage to the specimen had a larger effect on the classification than measurement methods.
Table 4.
Proportion of the trace and coefficient of each linear discriminants performed on the genus-level using the published dataset after some modifications (see details in the text).
Table 5.
Proportion of the trace and coefficient of each linear discriminants performed on the group-level using the published dataset after some modifications (see details in the text).
Table 6.
Proportion of the trace and coefficient of each linear discriminant performed on the genus-level using the published teeth dataset consisting of measurements taken by using only the methods of Hendrickx et al. (2015b, 2020).
Table 7.
Classification of OMFJ V-1 based on linear discriminant analyses at the genus- and group-levels using a reduced-dataset based on measurements taken only by Hendrickx's methods. Numbers in parentheses indicate the posterior probabilities.
The result of the linear discriminant analysis (Figure 9, Appendix 4) provided the possibilities for classification of OMFJ V-1. In the genus-level linear discriminant analysis, OMFJ V-1 was classified into tyrannosaurids (Alioramus and Gorgosaurus), allosauroids (Allosaurus and Erectopus), Berbrosaurus, Piatnitzkysaurus, Monolophosaurus and an unnamed dromaeosaurid (Table 7) although all probabilities were below 20%. The plot of LD 1 (proportion of trace: 75.16%) against LD 2 (proportion of trace: 13.39%) showed a slight separation between small and large theropods, similar to what appeared in the PCA plot (Figure 9, Appendix 4, Table 4). OMFJ V-1 was located at the margin of the distribution envelope of large theropod teeth mainly because of the small denticle size of OMFJ V-1 compared to other, similar-sized teeth. Group-level linear discriminant analysis classified OMFJ V-1 as belonging to Allosauridae (including Allosaurus only in this analysis) with the highest posterior probabilities regardless which measurement pattern was followed (Tables 5 and 7). OMFJ V-1 was also classified into non-abelisauroid ceratosaurians, non-megalosauran megalosauroids, Tyrannosauridae, and Neovenatoridae. The plots of LD 1 against LD 2 showed that the points of OMFJ V-1 measured in three patterns were all placed close to these theropod groups in the tooth morphospace (not shown in the figure). Although the non-abelisauroid ceratosaurians Ceratosaurus and Genyodectes clustered around the center of the large theropod assembly in the plot of LD 1 and LD 2, those of another non-abelisauroid ceratosaurian Berberosaurus were placed near the OMFJ V-1 points, leading to a high posterior probability of the latter species classified to this group. Non-megalosauran megalosauroids included Piatnitzkysaurus and Marshosaurus in the dataset of Hendrickx et al. (2020). The envelope of these theropods encompassed the points of OMFJ V-1 measured in Patterns A and C but was located distantly from the point of OMFJ V-1 measured in Pattern B. The posterior probability of OMFJ V-1 being classified as Tyrannosauridae was under 10% in the group-level LDA although it was classified as possibly belonging to two tyrannosaurids (Alioramus and Gorgosaurus) with relatively higher posterior probabilities than others in the genus-level LDA. This is because Tyrannosaurus has highly derived large teeth and its plots kept most of the tyrannosaurid envelop away from OMFJ V-1. Results of the linear discriminant analysis using only the data taken by the measuring scheme of Hendrickx et al. (2015, 2015b) were almost the same as that of the genus-level LDA (Tables 6 and 7). The exception was that OMFJ V-1 was classified to cf. Baryonychinae with a relatively high posterior probability in the LDA based on the data obtained by Hendrickx's methods only. However, this tooth was an outlier data point in Spinosauridae (Figure 8). In addition, the teeth of spinosaurids are fairly specialized among theropods with, for example, fluted labial and lingual surfaces and very fine serrations on the mesial and distal carinae (e.g. Charig and Milner, 1986), which are characteristics that are clearly different from those observed in OMFJ V-1.
It is noteworthy that the all linear discriminant analyses showed low posterior probability for Neovenatoridae including Megaraptora (< 12.0%) even though the points of Neovenator and Fukuiraptor were placed closer to OMFJ V-1 than those of other theropods in the plot of LD 1 against LD 2 (Figure 9, Appendix 4), especially in the group-level LDA. Even when megaraptorans were regarded as a distinct group separated from the Neovenatoridae, the posterior probabilities of OMFJ V-1 classified as belonging to Neovenatroridae or Megaraptora were still low (under 5.5%). The high posterior probabilities of OMFJ V-1 classified to Allosaurus, Neovenator or Erectopus in two genus-level LDA and Allosauridae in the group-level LDA are consistent with the classification of OMFJ V-1 as an allosauroid. An unnamed dromaeosaurid also showed high posterior probabilities in both genus-level linear discriminant analyses (Tables 4, 6 and 7).
In summary, the results of the present morphometric analyses suggest that OMFJ V-1 is possibly an allosauroid, tyrannosaurid, dromaeosaurid or close relative of Monolophosaurus, Piatnitzkysaurus or Berberosaurus. OMFJ V-1 has characteristically small mesial and distal denticles. Previous studies mentioned that the mesial and distal denticle sizes in some theropods become smaller (i.e., the denticle densities increase) through their ontogeny (Currie et al., 1990; Buckley et al., 2010). In contrast, Farlow et al. (1991) observed that the mesial and distal denticle densities in some theropods show negative allometric relationships to the fore-aft basal length of teeth. This observation is consistent with an apparent ontogenetic trend in tyrannosaurids. That is, Tsuihiji et al. (2011) reported that the denticle densities of lateral teeth ranged from 13.3 to 24.7 per 5 mm in a juvenile individual of Tarbosaurus bataar. Although such values in adult T. bataar are not available in the literature, those of adult individuals of very closely related Tyrannosaurus rex were reported ranging from 4 to 19.5 per 5 mm by Smith et al. (2005). Because these values are slightly higher than those in the juvenile T. bataar, the denticle densities of lateral teeth in tyrannosaurids may show negative allometry against the tooth size. If this is the case, therefore, it is possible that the small denticle size of OMFJ V-1 reflects a young ontogenetic stage of the specimen. Further studies on ontogenetic changes in the theropod denticle size are necessary to support or refute this hypothesis.
Discussion
Classification of OMFJ V-1
The present cladistic analysis suggests that OMFJ V-1 belongs to a clade including tyrannosauroids (Eotyrannus, Alioramus, Raptorex, Gorgosaurus, Tyrannosaurus and Daspletosaurus), non-neovenatorid allosauroids (Allosaurus and Sinraptor), a neovenatorid (Neovenator) and a megalosaurid (Piatnitzkysaurus) although support values for this clade were rather low. This clade in turn is the sister taxon of a clade including neovenatorids/megaraptorans (Fukuiraptor, Orkoraptor and Australovenator) and a dromaeosaurid (Dromaeosaurus). PCA placed OMFJ V-1 inside the distribution envelopes of Abelisauridae, basal Tetanurae, non-megalosaurian megalosauroids, Megalosauridae, Neovenatoridae and Tyrannosauridae, as can be seen in the plot of PC 1 against PC 2 (Figure 8). Genus-level LDA classified OMFJ V-1 as either a tyrannosaurid (Alioramus or Gorgosaurus), an allosauroid (Allosaurus or Erectopus), Berberosaurus, Piatnitzkysaurus, Monolophosaurus, or an unnamed dromaeosaurid. On the other hand, the group-level LDA classified OMFJ V-1 into Allosauridae (consisting of Allosaurus only in this analysis) with the highest posterior probabilities, followed by non-abelisaurid Ceratosauria and Neovenatoridae, non-megalosaurian Megalosauroidea, and Tyrannosauridae. These results are herein discussed in an attempt to infer the taxonomic affinity of OMFJ V-1.
The results of both cladistic and morphometric analyses suggested the possibility of OMFJ V-1 being a close relative of Piatnitzkysaurus. Teeth of Piatnitzkysauridae are different from those of Megalosauridae in having mesial denticles slightly smaller than distal denticles (i.e., DIDS > 1.2; Hendrickx et al., 2015c, 2019). The small sizes of mesial denticles probably made the envelope of Piatnitzkysauridae close to OMFJ V-1 in the resulting plots of morphometric analyses. However, OMFJ V-1 does not have this diagnostic character of Piatnitzkysauridae. Therefore, the analytical results that suggested OMFJ V-1 belongs to this clade are not convincing.
Although classification to Monolophosaurus had a relatively high posterior probability in the discriminant analyses, the result of the cladistic analysis did not support a close relationship between this taxon and OMFJV-1. Moreover, OMFJ V-1 has interdenticular sulci. The presence of these structures is typical for large-bodied theropods, i.e., large-bodied ceratosaurians (Ceratosaurus, Chenanisaurus, Kryptops and Majungasaurus), megalosauroids (Afrovenator, Megalosaurus and Torvosaurus), non-neovenatorid allosauroids (Allosaurus, Sinraptor, Carcharodontosaurus, Giganotosaurus, Mapusaurus and Eocarcharia; but not Acrocanthosaurus), neovenatorids/ megaraptorans (Neovenator and Fukuiraptor), and tyrannosaurids (Tyrannosaurus, Daspletosaurus and Gorgosaurus), as revealed by the cladistic analysis of Hendrickx et al. (2020) based on dental characteristics. However, these structures are absent in teeth of Monolophosaurus. Therefore, even though OMFJ V-1 has the denticle sizes and tooth crowns similar to those of Monolophosaurus, it is unlikely that the former belongs to the latter or its closely related taxa.
In the results of the present morphometric analyses, the points of the non-abelisauroid ceratosaurian Berberosaurus were located near those of OMFJ V-1, with the posterior probability for this theropod being relatively higher than those for other non-abelisauroid ceratosaurians, Ceratosaurus and Genyodectes. In the group-level LDA, the presence of Berberosaurus apparently caused the posterior probability of non-abelisaurid ceratosaurians to be artificially high. However, similar to the case of Monolophosaurus, the cladistic analysis did not support the affinity of OMFJ V-1with Berberosaurus, especially in that Berberosaurus does not possess the interdenticular sulci.
Results of both cladistic and morphometric analyses suggested that OMFJ V-1 is possibly classified into Allosauroidea or Tyrannosauroidea. Based on the posterior probabilities of the linear discriminant analyses, OMFJ V-1 is more likely classified into Allosauroidea than Tyrannosauridae although the tyrannosaurids Alioramus and Gorgosaurus also had relatively high posterior probabilities according to the genus-level analysis. Furthermore, although the results of the morphometric analyses showed that OMFJ V-1 was located far away from the envelope of non-tyrannosaurid tyrannosauroids, the analyses in this study did not include large-bodied non-tyrannosaurid tyrannosauroids such as Yutyrannus huali (Xu et al., 2012) whereas the results of the cladistic analysis did suggest that dental characteristics of OMFJ V-1 are close to those of a non-tyrannosaurid tyrannosauroid (i.e., Raptorex). To summarize, the tyrannosauroid or even tyrannosaurid affinities of OMFJ V-1 cannot be excluded based on the present analyses.
Another possibility suggested by the morphometric analyses as the taxonomic affinity of OMFJ V-1 is an unnamed, large dromaeosaurid, teeth of which are housed at the University of Chicago Paleontological Collection and were measured by Hendrickx and Mateus (2014), although this specimen has not been formally described. Based on the measurements provided by Hendrickx and Mateus (2014), this unnamed dromaeosaurid has remarkedly large teeth comparable to those of Dakotaraptor steini (DePlama et al., 2015), one of the largest known dromaeosaurids. In general, dromaeosaurids, including Velociraptor, Deinonychus, and Dakotaraptor, have a diagnostic feature of mesial denticles smaller than distal denticles (e.g. Currie et al., 1990; Norell and Makovicky, 2004; DePalma et al., 2015). Although OMFJ V-1 is different from typical dromaeosaurid teeth in this respect, a few other dromaeosaurids such as Dromaeosaurus and Utahraptor have teeth bearing mesial and distal denticles of almost equal size (Currie et al., 1990; Kirkland et al., 1993). Accordingly, the dromaeosaurid affinities of OMFJ V-1 cannot be completely ruled out. However, teeth of the unnamed dromaeosaurid housed at the University of Chicago have not been illustrated or described, making comparison to OMFJ V-1 impossible at this moment. A detailed descriptive study of those teeth is needed to assess the validity of classification of OMFJ V-1 into Dromaeosauridae.
To summarize, we conclude that OMFJ V-1 belongs to a tetanuran theropod, possibly Allosauroidea (including Neovenatoridae/Megaraptora) or Tyrannosauroidea. Between these alternatives, the allosauroid affinities appear to be stronger considering the result of the cladistic and linear discriminant analyses.
The significance of OMFJ V-1 in theropod evolution
Lateral teeth of Fukuiraptor kitadaniensis found from the Kitadani Formation in the Tetori Group are narrow and blade-like and have mesial and distal denticles of sub-equal sizes (Azuma and Currie, 2000; Currie and Azuma, 2006). Although such denticle morphology is also present in OMFJ V-1, F. kitadaniensis is different from OMFJ V-1 in possessing the following characteristics as described in previous studies (Azuma and Currie, 2000; Currie and Azuma, 2006): the mesial carina and its serrations extending to the level of the cervix or slightly distal to it; a highly labiolingually compressed crown; the cross-section at the cervix narrow and mesiodistally-pointed, lenticular-shaped; asymmetric or apically hooked denticles and well-developed interdenticular sulci. However, isolated teeth referred to F. kitadaniensis show that these characteristics apparently vary in the specimens, with some specimens having a relatively thick crown, a lanceolate basal cross-section and the mesial carina terminating approximately at the half height of the crown (H.U. pers. obs.). The presence of these characteristics in OMFJV-1 at least hinted at the possibilities that it may still belong to this species. Such possibilities are also supported, albeit indirectly, by the present cladistic and morphometric analyses suggesting the megaraptoran affinities of the specimen although only low posterior probabilities were obtained in the latter analysis. Because the known jaw bones of F. kitadaniensis are fragmentary and most teeth referred to F. kitadaniensis are isolated specimens, the amount of the variation in the dental morphology within the jaw is unknown. Therefore, future research on F. kitadaniensis including the dental morphology will be important in assessing the taxonomy of OMFJ V-1.
The classification of OMFJ V-1 bears implications for the geographic distribution of carnivorous theropods in the northern part of central Japan. The crown height of OMFJ V-1 is 33.3 mm, which is slightly smaller than the maximum size of the lateral teeth belonging to juvenile to subadult Alioramus altai (CH = 39.84 mm; Hendrickx et al., 2015c). Brusatte et al. (2009) reconstructed the holotype of A. altai at a mass of 396 kg. Therefore, it is possible that OMFJ V-1 belongs to a medium-sized theropod. If OMFJ V-1 indeed belongs to a tyrannosauroid, it would imply that a medium-sized tyrannosauroid already appeared in the northern part of central Japan in the Hauterivian-Barremian age. Furthermore, Manabe (1999) identified an isolated premaxillary tooth (Izumi Village Board of Education [IBEF] VP 1), which was found from almost the same horizon of the Itsuki Formation as that of OMFJ V-1, as belonging to Tyrannosauridae based on a symmetric D-shaped cross section. Manabe's (1999) “Tyrannosauridae” corresponds to current usage of the term Tyrannosauroidea including Jurassic and Early Cretaceous forms that have been described since then and share such characteristics of the premaxillary tooth with traditionally-known, large-bodied tyrannosaurids (e.g. Hutt et al., 2001; Holtz, 2004; Xu et al., 2004). Possible tyrannosauroid affinities of OMFJ V-1 suggested by the present cladistic and morphometric analyses are thus consistent with the finding by Manabe (1999). On the other hand, if OMFJ V-1 belongs to Allosauroidea, it follows that at least two clades of theropods, Allosauroidea and Tyrannosauroidea, coexisted as the top predators of the ecosystem in the northern central part of Japan during the Hauterivian-Barremian.
It is noteworthy, however, that Porfiri et al. (2014) described that premaxillary teeth of the Argentine megaraptoran Megaraptor namunhuaiquii as having a D-shaped cross section and identified such morphology as one of the synapomorphies uniting megaraptorans and previously described tyrannosauroids. Although D-shaped premaxillary teeth were not found in other megaraptorans, it is consistent with the fact that premaxillary teeth of Fukuiraptor kitadaniensis have a J-shaped cross section (Currie and Azuma, 2006) representing a transitional form from a compressed, blade-like cross section of lateral teeth to a D-shaped one in the premaxillary tooth row observed in other theropods (Fanti and Therrien, 2007; Reichel, 2012; Hendrickx et al., 2015b). Accordingly, it is possible that IBEF VP 1 described by Manabe (1999) might belong to a megaraptoran and, thus, that IBEF VP 1 and OMFJ V-1 may belong to closely related species or even the same taxon of Megaraptora.
With the phylogenetic position of Megaraptora within Theropoda still being debated (it has generally been considered a subclade of Allosauroidea; e.g. Benson et al., 2010; Carrano et al., 2012) and detailed dental morphology in Megaraptora including the amount of the variation within the jaw remaining mostly unexplored, taxonomic inference of OMFJ V-1 and IBEF VP 1 is conjectural at present. However, if the megaraptoran affinities of either of them are established, it represents one of the oldest records of megaraptorans globally, and accordingly suggests possibilities that the theropod remains from the Tetori group including the Itsuki and Kitadani formations may offer important information for the early phase of evolution of this clade of theropods.
The significance of the fossil locality in the Kuzuryu district
The Itsuki Formation in Taniyamadani where OMFJ V-1 was found has yielded various vertebrate fossils (Sakai et al., 2020a). Although most of them are fragmentary, some fossils have provided important implications for the paleoecosystem and paleobiogeography in the Kuzuryu district (Hirayama et al., 2020; Sakai et al., 2020b). A rich vertebrate fossil locality of the Tetori Group, Kitadani Quarry, is located at the Takinamigawa area, Katsuyama City, Fukui Prefecture adjacent to the Kuzuryu district. The Kitadani Formation exposed in the Kitadani Quarry consists of Aptian–Albian non-marine sediments and has yielded numerous vertebrates including dinosaurs (e.g. Azuma and Currie, 2000; Azuma et al., 2016; Sano, 2017; Sano and Yabe, 2017; Shibata et al., 2017). Terrestrial faunal changes corresponding to climate changes to warmer and dryer conditions at the Barremian–Aptian boundary were recorded in the sedimentary sequence of the Tetori Group (e.g. Sano and Yabe, 2017). Compared to the Kitadani fauna, that of the Itsuki Formation is older in age and, accordingly, is expected to reflect colder and more moderate climate conditions. Therefore, further studies of the vertebrate fauna including dinosaurs from the Kuzuryu district will facilitate our understanding of the faunal transition in the terrestrial ecosystem in the northern part of central Japan during the Early Cretaceous.
Conclusions
A theropod tooth (OMFJ V-1) found from the Itsuki Formation of the Tetori Group, cropping out in the Kuzuryu district, Ono City, Fukui Prefecture, was described. A cladistic analysis based only on the dental characters and morphometric analyses using six dental variables indicated that OMFJ V-1 belongs to a tetanuran theropod. In particular, these analyses suggested that OMFJ V-1 is most likely classified into Allosauroidea although the tyrannosauroid affinities cannot be completely excluded.
If OMFJ V-1 belongs to Tyrannosauroidea, its size indicates that a medium-sized form already lived in the northern part of central Japan by the Hauterivian-Barremian age. On the other hand, if OMFJ V-1 belongs to Allosauroidea, it indicates that an allosauroid and a tyrannosauroid, the latter of which have previously been reported based on an isolated premaxillary tooth (IBEF VP 1), coexisted. However, it is also possible that both OMFJ V-1 and IBEF VP 1 belong to Megaraptora as does Fukuiraptor kitadaniensis. If this is the case, these isolated teeth may represent the oldest records of this clade of theropods. Because of its age, the Itsuki Formation in the Kuzuryu district is important in revealing the dynamic transition of the paleoecosystem in the northern part of central Japan in the Early Cretaceous.
Acknowledgments
We thank Kazuyoshi Endo, Takenori Sasaki, Michinari Sunamura, Meg Wakui and other Evolutionary Paleobiologicaly Seminar members (the University of Tokyo) for their help with the cladistic and morphometric analyses and providing many useful comments. H.U. thanks Ayano Mizukami (the University of Tokyo) and Chihiro Hirata (Kanagawa Prefecture) for their support during his field work. S.I. and M.O were supported by JSPS KAKENHI Grant Number JP19654077. Reviewers K. Brink and T. Kubo provided constructive suggestions that greatly improved the quality of the manuscript.
References
Appendices
Author contribution
H. U. led the study and performed the statistical analysis. Y. S., S. I. and M. O. conducted field work. S. I. and M. O. collected the specimen and S. I. prepared it. T. T. CT-scanned the specimen. H. U., T. T. and M. M. contributed to discussion. H. U., Y. S., I. S. and T. T. wrote the manuscript, which was reviewed by all authors.
Electronic supplementary 1 (06-0002Electrical_Supplyment.xlsx). Morphometric data of OMFJ V-1 and teeth of Fukuiraptor and Australovenator added to the data matrix used in this study.
Electronic supplementary 2 (06PR-A-21-0002supple.pdf). Scatter plots of all pairs of six dental variables. All variables are more or less correlated with one another.
Electronic supplementary 3 (06PR-A-21-0002supple.pdf). Plots of LD 1 versus LD 2 scores obtained as the result of the linear discriminant analysis, the same plot as Figure 9 but showing with 95% confidence ellipses instead of convex hulls.
Appendix 1.
Character state values of OMFJ V-1 for the 91 characters coded for the data matrix of Hendrickx et al. (2020).