Kekenodon onamata Beds (McKay 1882

KEKENODON CF. K. ONAMATA

Referred specimen: AUGD 2469, partial denticulate tooth crown, from Oruawharo Road, c. 5 km east of Port Albert, North Island, New Zealand, collected by M. Dowson in July 1969. New Zealand Map Series 1 N28:985316, equivalent to 36.2637ºS, 174.46725ºE (NZGD49). Fossil record number Q09/f7902 (original number N28/f0902; New Zealand fossil record file, Geosciences Society of New Zealand).

Stratigraphy and age: AUGD 2469 was collected near, and is presumably from, the Mahurangi Limestone, underlying the Miocene Waitemata Group. Matrix from the Mahurangi Limestone consists of whitish grey, fine-grained, sparsely macrofossiliferous and heavily bioturbated rock. Matrix collected from the limestone did not yield the lower Whaingaroan planktonic foraminifera Globigerina (Subbotina) angiporoides Hornibrook, 1965. The presence of Globigerina euapertura Jenkins, 1960 indicates an upper Whaingaroan- Waitakian (c. 28.1 –21.7 Mya; Carter, 1969), contemporaneous with the Kekenodon Beds that produced NMNZ Ma 306.

Description

Dentition: A single, denticulate tooth crown is preserved (Fig. 10). The apex of the primary denticle is missing and crushing and cracks are prolific on all surfaces of the crown, indicating post-mortem deformation. The crown has a tall, triangular profile that is vertically oriented. A prominent anterior carina is present on the primary denticle, which is markedly larger than all preserved accessory denticles. The lower-half of the anterior surface of the crown is eroded, although the profile of the tooth suggests at least one accessory denticle was present just above the enamelodentin junction.Two large and robust posterior accessory denticles are preserved with possible additional denticles (now lost) located more basally. There is no indication of an ecto- or entocingulum. The enamel on one surface of the crown displays a slight wrinkled ornament, which appears natural, possibly indicating the lingual surface of the crown. In contrast, enamel on the inferred labial surface is smooth. When

compared with basilosaurids and NMNZ Ma 306, the presence of accessory denticles combined with the tall, triangular profile of the crown is suggestive of a premolar. Moreover, the arrangement of accessory denticles is characteristic of a posterior premolar or anterior molar. There is no evidence of a re-entrant groove on the anterior surface of the crown to suggest the tooth is a lower molar, as in basilosaurids. However, re-entrant grooves are largely absent from the molars of heterodont cetaceans in the Oligocene, making its use here in tooth identification dubious. The crown lacks the lingual curvature of the upper denticulate teeth of NMNZ Ma 306, indicating a lower tooth. Despite roots being absent, the tooth was almost certainly double-rooted; there is no indication of a third lingual root. Given the preserved characteristics, the tooth is provisionally identified as a left p2.

Remarks: AUGD2469resembles preserved denticulate teeth belonging to the holotype of Kekenodon onamata (NMNZ Ma 306). The size of the crown is larger than all preserved teeth of NMNZ Ma 306, but the profile and spacing of the accessory denticles is similar to K. onamata. Alternatively, the tooth may belong to a toothed mysticete similar to what is known from the Oligocene of South Carolina, such as Coronodon havensteini and the unpublished specimens ChM PV 5720 and ChM PV 4745 (Geisler et al., 2017). However, there is a current lack of fossil evidence to indicate that large body-sized (> 5 m), toothed mysticetes with relatively large teeth were present in Oligocene New Zealand marine environments. Affinities with C. havensteini are further unlikely given the complete absence of enamel ornament on its cheek teeth. In contrast, small body-sized toothed mysticetes (e.g. Mammalodon hakataramea Fordyce & Marx, 2016) are known from Late Oligocene New Zealand (Fordyce & Marx, 2016). Affinities with Phococetus vasconum and cf. Phococetus sp., in addition to Inticetus vertizi and Inticetus -grade odontocetes, are unlikely for structural disparities previously discussed with K. onamata and the absence of large body-sized odontocetes from the fossil record of Late Oligocene New Zealand. Thus far, K. onamata possesses the largest teeth, and in turn one of the largest body sizes, in the cetacean fossil record from the Oligocene of New Zealand. AUGD 2469 is conservatively identified as Kekenodon cf. K. onamata based on the structural similarities of the tooth and its contemporaneous stratigraphic position with the holotype of Kekenodon onamata.

PHYLOGENY

The cladistic analysis under equal weighting recovered a single most-parsimonious tree (consistency index [CI]: 0.435; retention index [RI]: 0.654); tree length 786 steps; Fig. 11A) and three equally most-parsimonious trees under implied weighting (CI: 0.435; RI: 0.653; tree length 72 steps; Fig. 11B). The affinities of Eocene archaeocetes and within the Neoceti (excluding taxa previously identified as non-aetiocetid toothed mysticetes; see below) are well resolved, probably owing to the comprehensive nature of the taxa and morphological characters included in the cladistic data matrix to prevent bias toward a particular group or clade in determining the taxonomic affinities of Kekenodon onamata.

There are few topological differences between the equal and implied weighting analyses, namely occurring in Eocene archaeocetes and non-aetiocetid toothed mysticetes (Fig. 11). The affinities of Eocene archaeocetes are better resolved under equal weighting with a paraphyletic Protocetidae and Basilosauridae, whereas dorudontine basilosaurids form a polytomy under implied weighting. A Basilosaurus clade composed of Basilosaurus isis and Basilosaurus cetoides Owen, 1839 is formed under both equal and implied analyses with strong support (branch support = 85% and 87%, respectively) in addition to a poorly supported Cynthiacetus peruvianus Uhen, 2005 and Dorudon atrox clade. Neoceti monophyly is recovered under both analyses with identical topology, but with poor support. Odontocete monophyly is recovered under both equal and implied weighting with moderate support (branch support = 56% and 82%, respectively). A Xenorophidae clade is recovered with poor support under equal weighting and moderate support under implied weighting (branch support = 67%). Within Xenorophidae, Albertocetus meffordorum Uhen, 2008 and Xenorophus sloanii Kellogg, 1923 form a clade to the exclusion of stemward-positioned Archaeodelphis patrius Allen, 1921 with moderate support under e q u a l w e i g h t i n g (b r a n c h s u p p o r t = 8 4%) a n d strong support under implied weighting (branch support = 91%). A second clade consists of the remaining odontocete taxa from this analysis, which includes Patriocetus kazakhstanicus Dubrovo & Sanders 2000, Agorophius pygmaeus (J.P.Muller, 1849), Simocetus rayi Fordyce, 2002, Waipatia maerewhenua Fordyce, 1994, Squalodon calvertensis Kellogg, 1923, Mesoplodon europaeus (Gervais, 1855) and Physeter macrocephalus with moderate support under equal and implied weighting (branch support = 53% and 78%, respectively). A separate clade composed of P. kazakhstanicus and A. pygmaeus is recovered, but with poor support. Crownward, the two extant odontocete taxa included in this analysis (M. europaeus and P. macrocephalus Linnaeus, 1758) form a clade with strong support (branch support = 99% under both analyses). Mysticete monophyly (excluding previously recognized non-aetiocetid toothed mysticetes) is recovered with poor support under equal weighting and moderate support under implied weighting (branch support = 60%). Aetiocetidae monophyly is recovered, but with poor support. Within the Aetiocetidae, a clade composed of members of the genus Aetiocetus (Aetiocetus cotylalveus Emlong, 1966, Aetiocetus polydentatus and Aetiocetus weltoni Barnes et al., 1995) is recovered to the exclusion of Fucaia goedertorum and Chonocetus sookensis Russell, 1968 with moderate support under equal weighting (branch support = 54%) and poor support under implied weighting. A clade including A. weltoni and A. polydentatus is recovered in the Aetiocetus clade with moderate support under both equal and implied weighting (branch support = 55% and 64%, respectively). Fucaia goedertorum and Chonocetus sookensis form a clade separate to Aetiocetus with poor support under equal weighting and moderate support under implied weighting (branch support = 68%). A Chaeomysticeti clade is recovered with moderate support under equal and implied weighting (branch support = 78% and 50%, respectively), which also includes a clade of crown Mysticeti with strong support (branch support = 99% under both analyses). A complete list of synapomorphies for all clades is included in the Supporting Information (Appendix S1).

A novel position for taxa previously recognized as non-aetiocetid toothed mysticetes as sister to the Neoceti is recovered under both equal and implied weighting with poor support. Among these taxa, a clade composed of Llanocetus denticrenatus, Mystacodon selenesis, Coronodon havensteini, ChM PV 4745 and ChM PV 5720 is recovered under equal weighting, but with poor support. Under implied weighting, these taxa are paraphyletic with a L. denticrenatus and M. selenesis clade representing the earliest diverging lineage, but with poor support. Identifying either clade as the Llanocetidae (Mitchell, 1989) is speculative given the differences in topology and poor support across both analyses. A clade comprised of C. havensteini, ChM PV 4745 and ChM PV 5720 is recovered under both analyses with moderate support under implied weighting (branch support = 79%). A single synapomorphy common to both analyses supports this clade: tooth enamel smooth (character 34: 1). Specimens ChM PV 4745 and ChM PV 5720 form a clade under both equal and implied weighting with moderate support (branch support = 86% and 83%, respectively). Six synapomorphies common to both analyses supports this clade: upper-posterior cheek teeth are double-rooted with roots closely appressed for the entire length (character 23: 1); number of upper molars equalling three (character 26: 1); anteriormost point of the supraoccipital in line with the gap between the anterior edge of the zygomatic process of squamosal and the anteriormost point along the posterior edge of the supraorbital process of frontal (character 123: 2); apex of the anterior process of the periotic bears a tubercle (character 159: 1); anteroexternal sulcus absent (character 164: 1); and involucrum with groove or grooves absent (character 216: 0). A Mammalodontidae clade (Janjucetus hunderi and Mammalodon colliveri) positioned as the most crownward is recovered with moderate support under equal weighting (branch support = 58%) and strong support under implied weighing (branch support = 88%). Six synapomorphies common to both analyses support this clade: percentage of skull length relative to bicondylar breadth less than 700% (character 1: 0); preorbital process of frontal has a rounded off, anteriorly convex linguiform outline (character 56: 1); posteromedial edge of the ascending process of the maxilla is separated from the posterolateral edge of the nasal or premaxilla by a triangular wedge of frontal (character 84: 1); suture between right and left nasals and right and left frontals shifted toward the left side (character 98: 1); frontal shield breadth taken at mid-orbit relative to condylar breadth less than two (character 110: 0); and frontoparietal suture with skull in dorsal view, the suture is strongly V- or U-shaped with the vertex of the V or U pointed posteriorly such that the posteriormost edge of the frontal is posterior to the anteriormost edge of the parietal (character 119: 1).

These results are interesting given the recent surge in studies on toothed mysticete phylogeny brought on by the discovery of new morphologically and taxonomically informative fossils (e.g. Fitzgerald, 2006; Marx et al., 2015, 2016; Geisler et al., 2017; Lambert et al., 2017; Fordyce & Marx, 2018; Muizon et al., 2019), which, while having some topological differences, posit these taxa within Mysticeti. However, the position of these taxa in previous studies is seemingly vulnerable to change due to low branch support, namely when more archaeocete and odontocete taxa are included in phylogenetic analyses. Removing Kekenodon onamata from the analyses yielded no differences in topology under equal weighting, while the L. denticrenatus and M. selenesis clade moved crownward to the C. havensteini, ChM PV 4745 and ChM PV 5720 clade under implied weighting. Critically, the position of these taxa as sister to Neoceti remained under equal and implied weighting with strong support (branch support = 98% and 99%, respectively). The results of analyses with K. onamata removed can be seen in the Supporting Information (Appendix S1). Ten synapomorphies common to both analyses support mysticete monophyly: anterior-half of maxilla forms a highly acute angle with a flattened maxilla (character 4: 1); suture between maxilla and premaxilla on rostrum is unfused and marked by a deep groove (character 9: 3); nutrient foramina and sulci on palatal surface of maxilla present (character 19: 1); mandibular symphysis not sutured (character 47: 1); symphyseal surface smooth (character 48: 1); symphyseal groove present (character 49: 1); alveolar margin on the mandible is parallel to the ventral margin for its entire length (character 52: 1); lateral edges of nasals anterior to the level of the anterior edge of preorbital process of frontal diverge anteriorly (character 101: 2); inferior lamina of pterygoid present and floors most of sinus cavity (character 139: 1); and median furrow of tympanic bulla forms notch on posterior edge of bulla between medial and lateral prominences (character 212: 1). Given that nearly all of these characters relate to the feeding apparatus, the relatively more primitive rostral and mandibular structures of previously recognized non-aetiocetid toothed mysticetes may at least partially explain why these taxa are being posited in a more basal position outside of Neoceti.

It is possible that excluding additional archaeocete and odontocete taxa may cause previously recognized non-aetiocetid toothed mysticetes to be posited in Mysticeti, which would suggest evidence of long-branch attraction. In most studies focused on the affinities of fossil Mysticeti, few archaeocete and odontocete taxa are included and are primarily relegated to the outgroup. Including additional archaeocete and odontocete taxa in previous studies on mysticete phylogeny may limit taxonomic bias and more accurately reflect the affinities of basal toothed mysticetes. The novel positioning of these taxa as sister to Neoceti requires future study to evaluate whether these results more accurately reflect the affinities of these archaic toothed cetaceans, including possible expansion of Llanocetidae, or if it is the result of long-branch attraction spurred on by increased inclusion of archaeocete and odontocete taxa in the phylogenetic analyses.

The primary concern of the cladistic analysis was to discern the phylogenetic affinities of Kekenodon onamata within Cetacea. The results under both equal and implied weighting posit Kekenodon onamata crownward to Basilosauridae as sister-taxon to Neoceti with moderate support (branch support = 64% and 81%, respectively), identifying K. onamata as the latest-diverging archaeocete and expanding the temporal range of archaeocetes into the Late Oligocene. The inclusion of undescribed putative kekenodontids from the OU collections, including a wellpreserved specimen with a nearly complete skull (OU 22294), will further elucidate interfamilial relationships within the Kekenodontidae (e.g. monophyletic vs. paraphyletic) and evaluate the phylogenetic position of kekenodontids as sister to Neoceti.