Introduction

In March 1999, the second author of this contribution discovered a molar tooth and femur of Palaeoloxodon naumanni (Makiyama, 1924), an extinct elephant, in an outcrop of the Middle Pleistocene Kiyokawa Formation, Shimosa Group, at Yoshinoda, Sodegaura City, Chiba Prefecture (Fig. 1; Kaneko et al., 2000). During subsequent excavations that extended for 65 days, more than 500 specimens of terrestrial vertebrate fossils were collected from this locality (Okazaki et al., 2004a; Kaneko and Okazaki, in press). Among fossils, chelonian materials such as Mauremys yabei (Shikama, 1949), an extinct species of geoemy-did, as well as the remains of Cervus nippon (Cervidae; Artiodactylla; Hirayama et al., 2002, 2004a, b, in press; Kaneko and Okazaki, in press; Takakuwa, in press), are the most abundant. In this article, we will describe a new species of herbivorous aquatic turtle, based on a nearly complete skeleton, and discuss its phylogenetic and paleoclimatic significance.

Figure 1.

Locality map of Ocadia nipponica sp. nov. (Holotype; CBM-PV 686), and geology of the study site at Yoshinoda, Sodegaura, Chiba Prefecture (modified from Okazaki et al., 2004).

Geological setting

The Kiyokawa Formation represents the middle part of the Shimosa Group (Middle to Late Pleistocene) and is distributed in the northern part of the Boso Peninsula, central Japan (Fig. 1; Okazaki et al., 2004a, b). A tephra marker bed from the lowermost part of the Kiyokawa Formation, Ky2, is correlated with TB-7, another tephra bed from the Oiso district of Kanagawa Prefecture, which is estimated as 225 ±24 Ka by the fission track dating method (Nakazato et al., 2004; Okazaki et al., 2004a). The Kiyokawa Formation rests unconformably upon the Kamiizumi Formation at Yoshinoda, being composed of fluvial and shoreface deposits in ascending order at the fossil-vertebrates-bearing locality. The fluvial deposits were formed during an early transgressive stage of relative sea level during an interglacial period.

The flood plain deposit of the Kiyokawa Formation yielding fossil vertebrates is divided into three depositional units by lithology and fossil assemblages; units A, B and C (Fig. 2; Okazaki et al., 2004a, b). The lowermost unit A consists of massive muds with plant remains that are interpreted as wash load by flooding. Unit B is composed of ill-sorted muddy sands or sandy muds and contains numerous wood debris and isolated vertebrate remains, including isolated shells of Mauremys yabei, freshwater molluscan remains (e.g., Anodonta sp.), and mudstone gravels.

Figure 2.

Columnar section and depositional facies of the locality of Ocadia nipponica sp. nov. (Holotype; CBM-PV 686) from the Kiyokawa Formation at Yoshinoda, Sodegaura, Chiba Prefecture. Modified from Okazaki et al. (2004).

It may have accumulated during a mudflow accompanied with flooding. These fossil assemblages indicate that these sediments come from a larger area than the stream channel. The uppermost unit C consists of massive muds and alternating layers of silts and fine-grained sands. Unit C contains in-situ freshwater bivalves and a nearly perfect skeleton of a turtle, the holotype of Ocadia nipponica, sp. nov. These fossils indicate that unit C was deposited in still water, such as a pond or shallow lake.

Material and methods

Comparative materials of living turtles, including 51 species of Geoemydidae among which are 75 specimens of Ocadia sinensis from China, are deposited in the private collection of Ren Hirayama.

The fossil specimen, a nearly complete skeleton of one individual, including associated skull and shell, was prepared using mechanical methods and often reinforced with super glue, because the bones were sometimes extremely damaged and fragile. Only its mold remained in some places. The skull and lower jaw was found isolated about 15 cm from the main part of the postcranial materials (Hirayama et al., in press). The posterior portion of the carapace was disarticulated and scattered around the main part of the shell as well (Hirayama et al., in press).

Institutional abbreviations for fossil materials. BMNH) Natural History Museum, London, UK; CBM-PV) Natural History Museum & Institute, Chiba, Japan; MNHN) Muséum National d'Histoire Naturelle, Paris, France; NSM) National Science Museum, Tokyo, Japan.

Systematic paleontology

Order Testudines Batsch, 1788

Suborder Cryptodira Cope, 1868

Superfamily Testudinoidea Batsch, 1788

Family Geoemydidae Theobald, 1868

Genus Ocadia Gray, 1870

Type Species.—Ocadia sinensis (Gray, 1834)

Known Distribution.—Early Miocene? to Recent of Eastern Asia.

Emended Diagnosis.—Medium to large geoemydid with enlarged upper and lower triturating surfaces, decorated by lingual ridges; foramen orbito-nasale small; plastral buttresses moderately strong, developed along half way of the costal plates; gular scute shortened, often excluded from entoplastron; humero-pectoral sulcus intersecting anterior half of entoplastron.

Ocadia nipponica sp. nov.

Holotype.—CBM-PV 686; a nearly complete skeleton, including skull, lower jaw, much of the appendicular skeletons and shell (Figs. 3–4, 5A, 6A and C, 7–9, and 10A).

Figure 3.

Skeleton of Ocadia nipponica sp. nov. (Holotype; CBM-PV 686). A: skull, B: lower jaw, C: cervical vertebrae, D: pectoral girdle, E: pelvic girdle, F: left humerus, G: left hind limb, H: carapace, I: plastron.

Figure 4.

Skull and lower jaw of Ocadia nipponica sp. nov. (Holotype; CBM-PV 686). A, B: dorsal view of skull; C, D: ventral view of skull; E, F: left lateral view of skull; G, H: left lateral view of lower jaw; I, J: dorsal view of lower jaw; K, L: ventral view of lower jaw. Gray shaded areas denote matrix that could not be removed. Abbreviations: ang, angular; art, articular; bo, basioccipital; bs, basisphenoid; cor, coronoid; de, dentary; ex, exoccipital; f. o. n, foramen orbito-nasale; f.p.c.c.i, foramen posterius canalis carotici interni; f.p.p, foramen palatinum posterius; fr, frontal; f.s.t, foramen stapedio-temporale; ju, jugal; mx, maxilla; op, opisthotic; pa, parietal; pal, palatine; pf, pre-frontal; pm, premaxilla; po, postorbital; pr, prootic; pra, prearticular; pt, pterygoid; qj, quadratojugal; qu, quadrate; sq, squamosal; s.t, sella turcica; sur, surangular; vo, vomer.

Figure 5.

Skulls of East Asian Geoemydidae (ventral views). Shaded areas show the upper triturating surfaces. Not to scale. A: Ocadia nipponica sp. nov., reconstruction based on the holotype (CBM-PV 686; CBL 54 mm), B: O. sinensis (RH 401; condylobasal length (= CBL) 45 mm), C: Chinemys reevesii (RH 100; CBL 52 mm), D: Cuora flavomarginata (RH 65; CBL 42 mm), E: Mauremys japonica (RH 67; CBL 33 mm), F: Geoemyda japonica (collection of National Science Museum, Tokyo; CBL 30 mm).

Figure 6.

Comparison of upper and lower triturating surfaces of Ocadia nipponica sp. nov. and O. sinensis. Shaded areas show the triturating surfaces. Scale bar is 2 cm. A, C: reconstruction of O. nipponica sp. nov., based on the holotype (CBM-PV 686). B, D: O. sinensis; RH 401. A, B: ventral views of skull. C, D: dorsal views of lower jaw.

Figure 7.

Carapace of Ocadia nipponica sp. nov. (Holotype; CBM-PV 686). A: dorsal view, B: ventral view (posterior part of peripherals removed).

Type Locality.—Yoshinoda (about 1.5 km southeast from the Kisarazukita exit of the Tateyama Highway), Takinokuchi, Sodegaura City, Chiba Prefecture, Japan.

Type Horizon.—lower part of the Kiyokawa Formation, Shimosa Group (Middle Pleistocene, about 0.22 m.y.a.).

Known Distribution.—Known only from the type locality and horizon.

Collector.—Naoki Kohno, Naotomo Kaneko, Hajime Taru, Yuji Takakuwa, Shinji Isaji, and Ren Hirayama, 29 October, 2001 to 3 July, 2002.

Etymology.—From Nippon, Japanese name of the Japanese nation,

Diagnosis.—Triturating surfaces of skull and lower jaw much broader than O. sinensis, as extensive as Batagur baska; shell length up to 33 cm long, about 20% larger than the largest individual of O. sinensis; external surface of carapace and plastron smooth except for scute sulci, without distinct ridges and annual growth sulci; second and third vertebrals nearly rectangular, as long as broad, much narrower than first vertebral.

Hypodigm.—holotype only.

Description

Shell (Figs. 7–9, 10A)

Most of the carapacial elements are preserved in CBM-PV 686 (Figs. 7A, 9A). The total length of the carapace is estimated as 33 cm long when reconstructed. The shell surface is very smooth, except for the scute sulci, and lacks any distinct longitudinal ridge. In contrast, O. sinensis has three rows of interrupted knobs on the neurals and costals and well developed annual growth rings on each scute area (Fig. 7A).

Figure 8.

Plastron of Ocadia nipponica sp. nov. (Holotype; CBM-PV 686). A: ventral view, B: dorsal view.

Figure 9.

Shell reconstruction of Ocadia nipponica sp. nov., based on the holotype (CBM-PV 686). A: carapace, dorsal view. B: plastron, ventral view. Bold line shows scute sulci on shell surface. Shaded area shows missing portion.

Figure 10.

Shell comparison of Ocadia nipponica sp. nov. and O. sinensis in dorsal view. A: O. nipponica sp. nov. based on the holotype (CBM-PV 686), B: O. sinensis; RH 401. Bold line shows scute sulci on shell surface. Scale bar is 5 cm.

The carapace bones consist of one nuchal, eight neurals, two suprapygals, eight pairs of costals, eleven pairs of peripherals, and one pygal as a plesiomorphic condition of geoemydids. Only a few posterior parts, including the sixth neural and first suprapygal, were missing. The nuchal and second suprapygal are irregularly hexagonal. The first neural is barrel-shaped, the second to eighth neurals are hexagonal and short sided anteriorly, and the first suprapygal is trapezoidal, as in many geoemydids, including O. sinensis (Fig. 10). The first costal has a sutural facet for the axillary buttress of the hyoplastron, which runs about halfway along the first costal (Fig. 6B). A sutural facet for the inguinal buttress of the hypoplastron is developed between the fifth and sixth costals, running about halfway along the costals as well (Fig. 7B). Axillary and inguinal musk duct foramina are found in the third and seventh peripherals as a synapomorphic character of geoemydids (except for Morenia; Hirayama, 1985).

Scutes overlying the carapace consist of one cervical, five vertebrals, four costals, and twelve pairs of marginals as in generalized testudinoids. The cervical scute is narrow and much longer than broad. The first vertebral is rather broad, but contacts the second marginal along a corner only. Measurements of the first to third vertebrals are the following (in mm):

• First vertebral: maximum width 82, maximum length 61.

• Second vertebral: maximum width 71, maximum length 66.

• Third vertebral: maximum width 69, maximum length 72.

Thus, in O. nipponica the second and third vertebrals are nearly rectangular, nearly as long as broad, much narrower than the first, whereas in O. sinensis the second and third vertebrals are a more rectangular shape, broader than long, much broader than the first (Fig. 10). There are no unusual scute contacts in CBM-PV 686 as in some Geoemydidae and Testudinidae (e.g., the sixth marginal—third costal contact in Mauremys japonica).

The plastron is almost completely preserved in CBM-PV 686 (Figs. 8, 9B). It consists of one pair of epiplastra, one entoplastron, one pair of hyoplastra, hypoplastra, and xiphiplastra. The dorsal portion of the epiplastra, which are underlain by the gulars, is moderately large and thickened as in O. sinensis (Fig. 7B). The lateral margin of the posterior lobe, consisting of hypoplastron and xiphiplastron, expands convexly toward lateral. This suggests that this individual was a female with a relatively broader plastral lobe as seen in the population of O. sinensis (Ernst and Barbour, 1989; Zhang et al., 1998; Hirayama, pers. obs.).

The scutes overlying the plastron consist of one pair of gulars, humerals, pectorals, abdominals, femorals, and anals as in most testudinoid turtles. It is likely that the axillary and inguinal scutes were originally present, but it is hard to discern their sulci on the plastron.

In living O. sinensis, the position of the gular scutes is rather highly variable relative to the entoplastron. Among 32 prepared specimens of this species in the RH collection, which were collected in Taiwan, five individuals show gulars that are completely excluded from the entoplastron, and in a further eight individuals the gular barely overlaps the anterior portion of the entoplastron. In the remaining individuals, the gular scutes are “normally” overlying on the entoplastron as is the case in most turtles with shell scutes retained. Thus, about 40% (thirteen individuals) of this population has distinctly short gular scutes such as were figured by Bourrett (1941). In CBM-PV 686, the gular is distinctly short, completely restricted to the epiplastron (Fig. 9B). The gular of most turtles is usually much longer, partially overlapping the entoplastron, with the exception of a few geoemydids (two species of Geoemyda and O. sinensis) and testudinids (e.g., Geochelone and Kinixys) as shown in the Appendix. Thus, short gulars that are excluded from the entoplastron are considered synapomorphic for O. nipponica and O. sinensis.

The humero-pectoral sulcus intersects the anterior half of the entoplastron in O. nipponica and O. sinensis, but intersects the posterior portion of the entoplastron, or is located posterior to the entoplastron in other geoemydids.

Discussion

The presence of anterior and posterior musk duct foramina that are enclosed in the third and seventh peripherals demonstrates that CBM-PV 686 is a member of the family Geoemydidae (Hirayama, 1985; Joyce and Bell, 2004).

Derived characters, such as enlarged upper and lower triturating surfaces decorated by lingual ridges, and short gular scutes excluded from the entoplastron, are shared by both CBM-PV 686 and Ocadia sinensis. Thus, it is reasonable to hypothesize that this specimen is a close relative of O. sinensis.

CBM-PV 686, however, has much broader triturating surfaces than O. sinensis. Its degree of development seems more similar to Batagur baska (Boulenger, 1889; Gaffney, 1979). The external shell surface is quite smooth, lacking distinct ridges, whereas O. sinensis has distinct annual growth sulci and three rows of knobs on the carapace. The carapace of CBM-PV 686 is 33 cm long, whereas the maximum carapace length of O. sinensis is only reported to be 27 cm (Chen and Lue, 1998; Zhang et al., 1998; Hirayama, pers. obs.). Finally CBM-PV 686 has narrower second and third vertebral scutes than O. sinensis. All these differences are considered enough grounds to distinguish a fossil taxon from the living species. Thus, we propose a new species of the genus Ocadia, O. nipponica.

Phylogenetic hypotheses of geoemydids have been proposed based on both morphology (e.g., Hirayama, 1985; Yasukawa et al., 2001) and, more recently, molecular data (e.g., Honda, 2002a, b; Spinks et al., 2004), showing a large discrepancy among the results. Feldman and Parham (2004) provided a new view on phylogenetic relationships among nine species of Eurasian extant geoemydids such as Mauremys, Chinemys, and O. sinensis using sequence data of the mitochondrial DNA. Phylogenetic trees proposed by Feldman and Parham (2004) show that both Chinemys and O. sinensis are situated within the genus Mauremys and that species combined from a monophyletic clade within Geoemydidae. They placed both species of Chinemys (C. reevesii and C. nigricans) and O. sinensis within Mauremys, to avoid use of paraphyletic genus names and the proliferation of new names. Other molecular phylogenetic studies also support the close phylogenetic relationships of Chinemys and O. sinensis with M. japonica and/or other Mauremys species (e.g. Wu et al., 1998; McCord et al., 2000; Honda et al., 2002b; Barth et al., 2004; Sasaki et al., 2004; Spinks et al., 2004). Nonetheless, we have no good morphological characters that would corroborate this molecular hypothesis.

Even if the new phylogenetic tree was accepted, we suggest that generic names Chinemys and Ocadia should be preserved, because they exhibit extremely different morphology, especially in cranial characters, such as an extensive secondary palate (Figs. 4A–C, 5). On the other hand, species of Mauremys (sensu McDowell, 1964) retain a primitive primary palate (Fig. 4E). The presence of an extensive secondary palate with lingual ridges in O. nipponica and O. sinensis should be sufficient for the retention of their own generic name. As a consequence, members of the M. mutica group should have their own generic name such as Cathaiemys (Lindholm, 1931), and a new genus for M. japonica (and possibly M. yabei, an extinct species, as discussed later) should be proposed to ensure that all generic names refer to monophyletic groups. We suppose it should more properly reflect the morphological diversifications of this group such as large variation of the triturating surfaces.

A certain number of fossil geoemydids from Eurasia have been placed in Ocadia based on shell morphology alone (Lydekker, 1889; Kuhn, 1964). Unfortunately, all shell characters (e.g., moderately developed plastral buttresses) used in the literature for the definition of the genus Ocadia are plesiomorphic for Geoemydidae, and this is not useful for modern systematics (Hirayama, 1985). Nonetheless, at least one shell character, short gular excluded from the entoplastron, might be useful in defining this genus as a clade. Such a short gular can be found in some Paleogene geoemydids, such as Ocadia crassa (BMNH 32349) from the Late Eocene of the UK, and Palaeochelys (= Cuvierichelys) parisiensis from the Late Eocene of France (Lydekker, 1889; de Broin, 1977; Hervet, 2004; Hirayama, pers. obs.). Interestingly, the syntype of P. parisiensis (MNHN 8236 GY7), a newly prepared partial skull, includes a small part of the upper triturating surfaces of the maxilla with lingual ridge (Hirayama, pers. obs.). Short gulars are also present in large geoemydids (carapace is up to 36 cm long) from the early Miocene of Niimi, Okayama Prefecture of western Japan (Hirayama et al., 1982et al., 1983). A large geoemydid was reported as Ocadia from the Early Miocene of Fukushima Prefecture, northeastern Japan, although it lacks most of the plastron (Hasegawa et al., 2002). Thus, these geoemydids might be actual fossil taxa of the genus Ocadia, although the lack of good cranial material makes this inconclusive.

Fossil material referred to the living species, O. sinensis, was reported from the Pliocene of China and Japan (Yeh, 1963, 1994; Hirayama, 2001). Some more fragmentary shell materials reported from the Early Pleistocene of western Japan by Otsuka (1969) and Hirayama and Inoue (1981) are possibly fossil Ocadia mainly because of its large size (shell length estimated as more than 30 cm) and smoother shell surface than O. sinensis.

Therefore, CMB-PV 686 is the first example of a good shell and skull association of a fossil Ocadia. O. nipponica demonstrates that this genus might have been well diversified in East Asia in the past, possibly as far back as the Miocene.

O. sinensis is an aquatic plant-eating turtle, well adapted to coastal environments in subtropical to warm temperate zones (Bourret, 1941; Iverson, 1992; Chen and Lue, 1998, 1999). Its northern limit appears to be controlled by a mean annual temperature of about 16°C. If O. nipponica required physiological conditions similar to those living species, then its presence could be a good paleontological index for the middle Pleistocene of Japan. However, a mean annual temperature of 16°C appears to be too warm and is contrary to estimations based on floral evidences (Momohara et al., in press; Okuda et al., in press).

O. nipponica cooccurs with some postcranial remains of Mauremys yabei, another extinct species of aquatic turtle, from the Kiyokawa Formation of Yoshinoda locality (Hirayama et al., 2004b, in press). M. yabei was originally reported as Clemmys yabei on the basis of some shell material from Pleistocene cave fissure deposits of Kuzuu, Tochigi Prefecture, central Japan (Shikama, 1949). M. yabei is estimated to have had a maximum carapace length of only 20 cm long, much smaller than O. nipponica. Because M. yabei has both anterior and posterior musk duct foramina, this species appears to belong to the family Geoemydidae as well, although the genus Clemmys is a member of the family Emydidae in its more recent circumscription (McDowell, 1964; Gaffney and Meylan, 1988; Yasukawa and Hirayama, 2001). The extremely broad plastron suggests that this plesion is the sister taxon of M. japonica, a living endemic species of the Japanese main islands (Kyushu, Shikoku, and Honshu Islands; Hirayama et al., in press). M. yabei, however, is easily distinguished from the living species in the possession of a narrower, almost triangular cervical, and posterior plastral buttress that reaches the sixth costal, and by lacking apomorphic features of M. japonica, such as serrated peripherals, and a very broad first vertebral that reaches the second marginals (Hirayama et al., in press). Cuora miyatai is a third extinct Pleistocene geoemydid from the Japanese mainlands (Shikama, 1949; Hasegawa, 1981). This geoemydid is known from several localities of the cave fissure deposits of Tochigi, Yamaguchi, and Oita Prefectures, and is extremely similar in both shell and skull morphology to living Cuora flavomarginata, although a few plastral features such as much narrower entoplastron suggest that C. miyatai should be considered an extinct species of this genus with hinged plastron (Yasukawa and Hirayama, 2001). Fossil material of M. japonica is restricted to a few remains from Neolithic sites of the Jomon period of western Japan (Yabe, T., pers. comm.). M. yabei seems too different and specialized to be considered the direct ancestor of the living species.

Thus, at least three fossil endemic species of geoemydids lived on the Japanese main islands during the Pleistocene. Together with the trionychid Pelodiscus sinensis, M. japonica is the sole remaining turtle to naturally inhabit the Japanese main islands (Hirayama et al., 2004b). Based on the lack of any fossil record, Chinemys reevesii, a geoemydid widely distributed in Japan today, may be considered a recent immigrant, possibly from the Korean Peninsula (Yasukawa, Y., pers. comm.). The fossil record suggests that geoemydid turtles were diverse in Japan during the Pleistocene. Their endemic speciation seems deeply linked with the geographical isolation of Japan from the continental region since the beginning of the Pleistocene. The extinction of these endemic terrestrial turtles may have taken place during the Late Pleistocene, possibly caused by the extremely low temperatures during the latest glacial period, or by predatory pressure from humans who were immigrating from the continent as well as the Ryukyu Islands (Takahashi et al., 2003, 2004).

Conclusion

A new species, Ocadia nipponica, is described on the basis of a nearly complete skeleton (CBM-PV 686) from the Middle Pleistocene nonmarine deposits of the Kiyokawa Formation at Yoshinoda, Sodegaura City, Chiba Prefecture, central Japan. O. nipponica is distinguished from the living species, O. sinensis, by the possession of a more extensive secondary palate, smoother shell surface, narrower second and third vertebral scutes, and its larger size. Two extinct species of turtles, O. nipponica and Mauremys yabei, from the Kiyokawa Formation are considered as endemic to the Japanese main islands as well, suggesting a much larger taxonomic diversity of geoemydid turtles in the Japanese Islands during the Pleistocene than hitherto considered. This also implies warmer paleoclimatic conditions in the Japanese Islands during the Pleistocene than hitherto estimated on the basis of floral evidence alone.

Continued.

Continued.