Tohoraonepu nihokaiwaiu sp. nov.

Ontogeny

In fossil archaeocetes and Neoceti, the development of cranial sutures is commonly used to estimate age (Uhen 2004; Walsh and Berta 2011). The incomplete basisphenoid prevents determining the development of the basioccipital-basisphenoid synchondrosis. However, the continuous bony bridge from the basioccipital crest to the alisphenoid is consistent with the synchondrosis at least being closed laterally.

The vertebral epiphyses are missing from most preserved vertebrae, including the anterior epiphysis from a post-axis cervical vertebra and both epiphyses from all preserved thoracic and lumbar vertebrae, excluding the anterior epiphysis from a single lumbar vertebra. Galatius and Kinze (2003) and Moran et al. (2015) show that vertebral epiphyseal fusion in some extant odontocetes and mysticetes begins at the anterior and posterior ends and is directed toward the middle of the vertebral column.

The inference of deciduous teeth (excluding provisionally identified molars) further suggests that OU 22394 was in an early stage of development. The preservation of a provisional m1 and M1 suggest the molars of OU 22394 erupted early in the dental eruption sequence, similar to Zygorhiza kochii and Dorudon atrox (Uhen 2000, 2004). The early eruption of molars in Tohoraonepu nihokaiwaiu is seemingly confirmed by the presence of abrasional and attritional enamel wear facets on M1 and m1 in addition to abrasional wear in an inferred dp4, indicating OU 22394 was feeding on solid food and was weaned. In Z. kochii and D. atrox, the fourth premolar is the first permanent premolar to erupt (Uhen 2000, 2004); a juvenile individual of D. atrox (UM 93220) preserves a skull and mandibles with a developing p4 while the P4 was not yet developing (Uhen 2000). Provisionally identified dp1 and dP1 indicate also replaced the first premolar, as in Z. kochii and D. atrox, which are the last premolars to erupt, occurring around the same time as the canine towards the end of the dental eruption sequence (Uhen 2000, 2004).

The level of sutural development and fusion of vertebral epiphyses with provisionally identified permanent and deciduous teeth likely identify OU 22394 as juvenile.

Body length

The skull size of OU 22394 is estimated using the occipital condyle breadth (ca. 40 mm) and bizygomatic width (ca. 228 mm) to infer condylobasal length (CBL), with reference to Kekenodon onamata and the unnamed kekenodontids OU 22294 and OU 22023 (), resulting in an estimated CBL of ca. 695 mm. The body size of OU 22394 is estimated by incorporating bizygomatic width into the regression equations of Lambert et al. (2010) and Pyenson and Sponberg (2011) to reconstruct total body size in fossil cetaceans. The equations for both stem Odontoceti and stem Mysticeti from both Lambert et al. (2010; supplementary Figures 6 and 9, respectively) and Pyenson and Sponberg (2011; supplementary equations i and vi, respectively) are used given the absence of equations for archaeocetes, resulting in values of 2.5 and 2.2 m, respectively. Previously, Boessenecker and Fordyce (2015) noted the estimated body lengths of Pyenson and Sponberg (2011) were consistently lower than Lambert et al. (2010). Pyenson and Sponberg (2011) note an 18% underestimation in the kentriodontid Atocetus and a 47% overestimation in the balaenopterid Balaenoptera siberi. Marx et al. (2015) suggests these discrepancies may be the result of constraints related to feeding ecology or differences in allometric scaling. Applying the 18% correction, as in Atocetus, results in a total body length estimation of 2.6 m. The holotype of T. nihokaiwaiu (OU 22394) represents the smallest kekenodontid, being markedly smaller than Kekenodon onamata (ca. 8 m; Corrie and Fordyce 2022) and more similar to, albeit smaller than, the unnamed kekenodontids OU 22294 and OU 22023. When fully mature, OU 22394 may had a total body length more similar to the ontogenetically older OU 22294 at ca. 4 m ().

Description. The anatomical description of Tohoraonepu nihokaiwaiu (OU 22394) is based on the preserved dentition. A comprehensive description of all additional cranial and postcranial material is openly available in figshare at https://doi.org/10.6084/m9.figshare.24311248.

Dentition

OU 22394 preserves 10 isolated, heterodont teeth ( and ; ). Six teeth are single-rooted, with sub-conical and caniniform crowns (). Of these, four teeth are single-rooted and single-cusped and represent inferred incisors, canines, and first premolars. All single-rooted, caniniform crowns are lingually and distally curved in anterior and posterior views, with evident sub-parallel raised ornament on the lingual faces. Abrasional enamel wear, characterised by rounded surfaces with polished margins, is absent from all inferred single-rooted teeth, but occurs on the apices of some primary denticles in the cheek teeth. Attritional enamel wear (e.g. occlusal wear) is largely absent except when noted.

Four teeth are double-rooted with denticulate crowns that are more anteroposteriorly elongate and transversely compressed than all single-rooted teeth (). The rostrum and mandibles are not preserved in OU 22394, making the identification of individual isolated teeth provisional. Root number, crown dimensions, and the presence and number of anterior and posterior accessory denticles are used to identify tooth position and number, with reference to other known and putative kekenodontid and basilosaurid archaeocetes.

The preserved teeth of OU 22394 (excluding the molars) have structures characteristic of deciduous teeth in archaeocetes (Gingerich and Russell 1990; Uhen and Gingerich 2001; Uhen 2004):

1. the pulp cavities of all teeth are unmineralized;

2. denticulate and double-rooted premolars have crowns with a triangular profile in labial and lingual views, that are more low-lying relative to homologous permanent premolars of the other putative kekenodontids OU 22294 and OU 22023 (Fordyce 2004; Corrie and Fordyce 2014a, 2014b), similar to basilosaurids (Kellogg 1936; Uhen 2004); and

3. the anterior and posterior roots of the double-rooted teeth (excluding the inferred molars which are identified as permanent teeth) are comparatively more divergent than the near vertically oriented roots of Kekenodon onamata (Corrie and Fordyce 2022) and OU 22294 (Fordyce 2004; Corrie and Fordyce 2014a,b).

Light coloured enamel and a thin enamel layer have also been identified as characteristics for deciduous teeth in fossil Cetacea (Gingerich and Russell 1990). The enamel on the preserved teeth of T. nihokaiwaiu ranges from being light to dark brown in colour. However, enamel colour is likely influenced by diagenetic processes, and this is used here to explain the broad range of enamel colour in T. nihokaiwaiu, rather than an indicator of tooth development. Loch et al. (2015) reported a relatively thin enamel thickness of a molariform tooth in the putative, unnamed kekenodontid OU 22023 at 180–210 µm. Thus, it is likely a thin to moderately thick enamel layer is characteristic of kekenodontid dentition, and its use in distinguishing deciduous and permanent teeth in kekenodontids is inconclusive. Although the lower-lying posterior cheek teeth of OU 22394 may be due to interspecific variation, the crown profile and open pulp cavities relative to inferred permanent teeth in K. onamata and the provisional kekenodontids OU 22294 and OU 22023 (Fordyce 2004; Corrie and Fordyce 2014a,b, 2022), are suggestive of deciduous teeth. The absence of a preserved rostrum or mandible showing the emergence of permanent teeth precludes definitive identification of deciduous teeth. However, based on the structural evidence, we provisionally identify all teeth (excluding molars; see below) as being deciduous and are represented by the letter ‘d’ in front of the tooth position.

The incisors and canines have a single associated root with a sub-rounded cross-section and a diameter that expands toward the crown (). The crowns of two presumed incisors are structurally and dimensionally similar with crown size varying only slightly. Both incisors lack roots with a predominately straight profile that are characteristic of the procumbent first incisors in OU 22294 (Corrie and Fordyce 2014a,b), suggesting either the first incisors of OU 22394 were not preserved or were not procumbent. The lack of additional teeth and/or jaws with alveoli prevents further allocation of these teeth to tooth position. The crown of a canine is anteroposteriorly longer and more transversely compressed than the inferred incisors, with a subequal crown height that is similar to the pattern in basilosaurids (Uhen 2004). This is similar to the pat Unlike the incisors, the root of the canine is compressed and does not show inflation toward the enamelo-cementum junction. The root for the canine is more ovoid in cross-section relative to the sub-rounded roots of the incisors. Both incisors have sections of enamel that are missing from the labial and lingual surfaces of the crown, which exposes the underlying dentin. However, this probably reflects damage during burial or during extraction and preparation.

The crowns of three single-rooted teeth differ from the incisors and canines in being more anteroposteriorly elongate, transversely compressed, and less subconical (A-C). The roots are more compressed than the incisors and canines, and have a distinct longitudinal, median sulcus located on both the labial and lingual surfaces of the tooth. The sulcus may represent a transition between single- and double-rooted teeth.

Two teeth provisionally identified as the left (E) and right dp1 (F) are structurally similar and have a minute anterior accessory denticle located above the enamelo-cementum junction. The left dp1 has moderately developed nodules located on the anterior surface of the crown below the accessory denticle. Because ento- and ectocingula are not otherwise present, the nodules may represent incipient denticles. Serrations on the posterior carina are more distinct relative to the right dp1 and include several large, nodule-like structures located above the accessory denticle.

The remaining single-rooted tooth, a presumed right dP1 (D), does not have a minute anterior accessory denticle, as in the previous two teeth. Instead, the base of the anterior margin of the crown bears two nodule-like crenulae. The crown is taller and shorter than both dp1, in addition to being more compressed than the right dp1, but less so than the left dp1. Although structurally disparate from p1, the level of compression of the crown and root suggests it is not a more anteriorly located incisor or canine, hence the provisional identification of right dP1.

On the left dp1, the apical 4 mm of the main denticle has been glued slightly out of position. The apex has a ca. 1.25 mm diameter oblique apical wear facet descending the lingual surface. The margins of the facet are sharp and unpolished, likely resulting from abrasion. In contrast, attritional wear is characterised by smooth and polished margins around the facet (Marx et al. 2023).

In the double-rooted cheek teeth, the triangular profile of the crown increasingly lowers moving posteriorly (). The accessory denticles increase in size toward the primary denticle. The posterior accessory denticles are larger than the anterior accessory denticles, whereas the primary denticle is larger than all accessory denticles. The anterior and posterior roots of all double-rooted cheek teeth are compressed, with an ovoid cross-section, and are fused at the apical third by an isthmus, below which the anterior and posterior roots diverge. The separation between the anterior and posterior roots increases away from the crown, resulting in an inverted U-shape, that transversely expands towards the base of the roots. Enamel ornament is mostly similar on the lingual and labial surfaces of all crowns. Faint longitudinal ridges cover the cervical region of the lingual surface (excluding?m1 and?M1), with predominantly smooth enamel ornamenting the apical half. Ornament on the labial surface of the crown consists entirely of smooth enamel, with a faint wrinkled ornament. All denticles have distinct anterior and posterior carinae which, on the denticulate posterior cheek teeth, are more developed on the primary denticle. There is no evidence of ento- or ectocingula on all preserved crowns. However, some teeth have cusp-like cingular nodules on the anterior and posterior extremities of the crown above the enamelo-cementum junction, in addition to instances of isolated nodules on the cervical region of the lingual surface of the crown.

The crown of the right dp3 (A) is more elongate, compressed, and shorter relative to the more anterior teeth. The crown has a tall, triangular profile in labial and lingual views. There are three anterior accessory denticles. Of these, the most apical is well-developed, with the more basal two being minute and nodule-like and located just above the enamelo-cementum junction. The posterior margin of the crown has three well-developed accessory denticles. A section of missing enamel on the lingual surface of the crown, located near the enamelo-cementum junction, exposes the internal surface of the crown; the structure is consistent with damage during excavation rather than a wear facet.

The crown of the right dp4 (B) has a lower triangular profile in labial and lingual views and is more elongate and compressed relative to dp3. The crown is more denticulate with three or four (damaged) anterior accessory denticles, and five posterior accessory denticles. The basalmost posterior denticle, located just above the enamelo-cementum junction, is minute and nodule-like. An apical wear facet ca. 1.25 mm wide is present on the primary denticle, forming a sub-horizontal plane that passes into an ovoid wear facet extending < 2 mm down the lingual surface These facets are interpreted as abrasive because of rounded surfaces and lack of polish or sharp margins. A large, triangular section of enamel is missing from the lingual surface of the secondmost apical posterior accessory denticle. Additionally, multiple sub-rounded to rectangular sections of missing enamel (occurring during collection and preparation) are scattered throughout the labial surface of the crown, exposing the underlying dentin.

The crown of the left m1 (C) is lower, more elongate, and more compressed than dp4. Two anterior accessory denticles are preserved with evidence of at least two others, indicating at least four anterior accessory denticles were present. There are five posterior accessory denticles. The apex of the primary denticle is truncated, with rounded enamel margins; this is consistent with a pre-mortem event and is considered to be abrasional wear. Two sections of missing enamel from the lingual surface of the crown are probably artifacts of preservation or preparation (one rectangular and located anterolingually; the other on the median of the lingual surface along the enamelo-cementum junction). An obliquely-oriented, attritional wear facet is located above the enamelo-cementum junction on the anterolabial surface of the crown. This narrow ovoid facet is ca. 6.5 mm high and ca. 1.5 mm wide and shows polished planar sections of enamel surrounding a window of excavated dentin.

A denticulate and double-rooted tooth is missing most of the anterior and posterior roots. The crown has a low-lying triangular profile (in labial and lingual views), similar to m1, but is less elongate and tall and more compressed. There are four anterior and posterior accessory denticles that are more widely spaced than in dp4 and m1. The anterior face of the crown is steeper than the posterior face, similar to dp4 and m1. The smaller size of the crown, in concert with the equal number of anterior and posterior accessory denticles and orientation of the crown, provisionally identify the tooth as the right M1 (D). A large section of missing enamel from the posterolabial surface of the crown exposes the underlying dentin. Two nodules are located well above the enamelo-cementum junction on the posterolingual surface of the crown. This may be a characteristic of the species but is inconclusive due to the juvenile status of OU 22394 and unknown ontogenetic trajectory of kekenodontids. However, both interpretations are provisional and inconclusive given the lack of kekenodontid specimens similar to T. nihokaiwaiu spanning an ontogenetic series. Several minute, sub-rounded to ovoid attritional wear facets are found on the lingual surface of the crown. Wear facets similar in profile are also found on the labial surface of the crown in addition to two vertically oriented and narrow (scratch-like) wear facets located on the two apicalmost anterior accessory denticles. Although suggestive of increased lateral motion of the mandibles, it is more likely that enamel wear facets on the labial surface of M1 are abrasional rather than attritional (Loch and Simões-Lopes 2013).

Both m1 and M1 have aforementioned open pulp cavities that are filled by matrix, indicating both molars were in an early stage of development. In mammals, molars do not have an antecedent form (Hillson 2005; Ungar 2010). Thus, m1 and M1 are not identified as deciduous teeth, as in the incisors, canines, and premolars.

Phylogenetic analysis

The cladistic analysis under EW recovered eight equally most-parsimonious trees (consistency index [CI]: 0.428; retention index [RI]: 0.662; tree length 802 steps; , top) while the IW analysis recovered two equally most parsimonious trees (CI: 0.427; RI: 0.660; tree length 73 steps; , bottom). Both analyses recover Tohoraonepu nihokaiwaiu within a poorly supported monophyletic Kekenodontidae that also includes Kekenodon onamata and the unnamed OU 22294 and OU 22023. The kekenodontids are crownward to the Basilosauridae, with strong support (91% and 94%, respectively), and basal to a poorly supported paraphyletic group of previously recognised toothed mysticetes that are excluded from Neoceti. This group includes the Eocene mysticetes Mystacodon selenesis and Llanocetus denticrenatus, a Coronodonidae clade (Coronodon havensteini, Coronodon newtonorum, Coronodon planifrons, and ChM PV 5720), and a Mammalodontidae clade (Mammalodon colliveri and Janjucetus hunderi). A single synapomorphy supports the kekenodontid clade: the absence of elongation of the pars cochlearis towards the cranial cavity dorsally and medially (184:0). Within the Kekenodontidae, a poorly supported T. nihokaiwaiu and OU 22023 clade is recovered as sister-group to OU 22294 under EW and K. onamata under IW; K. onamata is the earliest to diverge under EW whereas OU 22294 is the earliest to diverge under IW. Two synapomorphies support the T. nihokaiwaiu and OU 22023 clade: tooth enamel bears reticulating striae (34: 0); apex of the anterior process of the periotic blunt or pointed in ventral or dorsal view (159: 0). The results support previous analyses that recover kekenodontids crownward to basilosaurids (Corrie and Fordyce 2022; Boessenecker et al. 2023), further indicating that kekenodontids are Late Oligocene archaeocetes that were the latest to diverge.

Tooth displacement

Diphyodonty represents the primitive stage of tooth ontogeny in Cetacea and is characteristic of Eocene stem cetaceans (Uhen 2008, 2010). Currently, Late Eocene Basilosauridae represent the latest fossil record of diphyodonty in Cetacea, with fossil crania of the basilosaurids Zygorhiza kochii and Dorudon atrox preserving permanent teeth in the process of dental eruption (Kellogg 1936; Uhen 2000, 2004). The basilosaurid Chrysocetus healyorum may represent the earliest fossil evidence of monophyodonty in Cetacea (Uhen and Gingerich 2001). The single, holotype specimen of C. healyorum is a young juvenile, that preserves teeth with structural characteristics inferred for permanent dentition in archaeocetes, including: caniniform and molariform teeth have fully formed roots with no evidence of resorption in the rostrum or mandible; all incisors have moderately developed longitudinal striations on the lingual face of the crown (also seen in the deciduous p2 or P2 of an undescribed cf. Dorudon from the upper Bortonian of North Otago, South Island, New Zealand); the premolars have tall crowns with numerous, well-developed denticles; and the anterior and posterior roots of the premolars angle toward each other distally (Uhen and Gingerich 2001). In comparison, specimens of D. atrox and Z. kochii showing a similar degree of skeletal fusion (i.e. similar ontogenetic age) to C. healyorum were still in the process of dental eruption or had nearly completed dental eruption (Uhen 2000; Uhen and Gingerich 2001). From these observations, Uhen and Gingerich (2001) formed two hypotheses: (i) the loss of deciduous teeth with the accompanying eruption of permanent teeth occurred early in ontogeny; or (ii) C. healyorum represents the earliest record of monophyodonty in Cetacea, provisionally identifying the origin of monophyodonty within the Basilosauridae.

In contrast to archaeocetes, extant Mysticeti develop tooth buds in the embryonic stages of development, that are resorbed before breaking the gingival line in utero (Karlsen 1962; Ishikawa et al. 1999). Alternatively, extant Odontoceti are monophyodont and homodont, in addition to most being polydont (Uhen 2008). Archaic fossil odontocetes and toothed mysticetes both preserve specimens that are heterodont and seemingly monophyodont (Barnes et al. 1995; Fordyce 2002b; Fitzgerald 2006, 2010; Deméré and Berta 2008; Uhen 2008). Previously, Fordyce (1982) hypothesised monophyodonty for all extinct and extant Odontoceti. Evidence of dental eruption is absent from the Neoceti fossil record, thus obscuring inferences of diphyodonty or monophyodonty in stem Neoceti. Permanent dentition is inferred for Kekenodon onamata (Corrie and Fordyce 2022) and the unnamed OU 22294 and OU 22023 based on structures characteristic of permanent teeth in basilosaurids and more basal archaeocetes, including: the mineralisation of the pulp cavity, the tall triangular profile of the premolariform teeth, and double-rooted teeth having anterior and posterior roots that are vertically-oriented and do not diverge distally. Slightly divergent roots are found in some double-rooted teeth of OU 22023, for which a sub-adult to adult ontogenetic status is inferred based on the degree of closure of preserved cranial sutures, suggesting root orientation may not be an accurate proxy for distinguishing deciduous and permanent teeth in kekenodontids. Moreover, extensive attritional macrowear on the tooth enamel of K. onamata (Corrie and Fordyce 2022) confirms the identification of permanent teeth. The absence of attritional enamel wear with the aforementioned features OU 22294 is inferred to be the result of a suction-assisted feeding strategy and possibly a diet composed primarily of soft-bodied marine organisms (e.g. small fish and cephalopods) in contrast to the putative macrophagous diet of K. onamata (Corrie and Fordyce 2022), that likely included relatively large marine vertebrates (e.g. sharks, penguins, and other Cetacea) similar to the inferred diet of Basilosaurus isis (Fahlke et al. 2013). In contrast, attritional wear facets are present on multiple teeth of OU 22023 either indicating an ontogenetically more mature OU 22294 or disparate dietary preferences for different genera of cohabiting kekenodontids (Fordyce 2004; Corrie and Fordyce 2014a,b). The presence of abrasional and attritional enamel wear indicates the teeth were functional, whether or not OU 22394 was suckling. The absence of macrowear compared K. onamata (Corrie and Fordyce 2022) suggest a diet that likely consisted of relatively soft-bodied small schooling fish and cephalopods, consumed at least partially through raptorial macrophagy. The tentative identification of the preserved teeth of Tohoraonepu nihokaiwaiu as being deciduous (excluding inferred first molars; Figure 10C–D), based primarily on unmineralized pulp cavities, provisionally infers the retention of diphyodonty in at least one species of Kekenodontidae and would be the first evidence of cetacean diphyodonty outside of the Eocene.