INTRODUCTION

The Senkaku mole,Nesoscaptor uchidai, is the most recently discovered non-volant Japanese mammal. Abe et al. (1991) described the unique holotype from near the western coast of Uotsurijima, in the Senkaku Group of the Ryukyu Archipelago (25°44′ N, 123°28′ E: Fig. 1), as a new genus and species. Based on comparisons with other Eurasian talpine genera, including Euroscaptor, Mogera, Parascaptor, Scaptochirus, and Talpa, Nesoscaptor was diagnosed by a combination of 19 morphological characters. Of these, the dental arrangement was thought to be the most important and to differentiate the genus Nesoscaptor from the other talpine genera (Abe et al., 1991).

The genera Talpa in Europe and Euroscaptor in Asia have primitive mammalian dentition, which consists of 44 teeth, including three pairs of incisors, one pair of canines, four pairs of premolars, and three pairs of molars in both the upper and lower jaws (Abe et al., 1991). The other genera have reduced dentition. Mogera has 42 teeth due to the loss of one lower incisor pair; Parascaptor has 42 due to the loss of one upper premolar pair; and Scaptochirus has 40 due to the loss of one upper and one lower premolar pair (Abe et al., 1991; Corbet and Hill, 1992). Currently, these five genera are validated primarily by such differences in dental arrangement (e.g., Hutterer, 1993), although Corbet (1978) and Corbet and Hill (1992) questioned the taxonomic validity of dental formula in mole genera. The genus Nesoscaptor has no more than 38 teeth, since it lacks one pair of lower incisors and two pairs of premolars, one in the upper jaw and one in the lower, compared to the putative primitive dentition of 44 teeth (Abe et al., 1991: see above).

Uotsurijima is located approximately 160 km ENE of Taiwan, and is the largest island in the Senkaku Group. It is about 3.3 km long, 1.3 km wide, 4.3 km² in area, and 362.0 m in maximum elevation (Abe et al., 1991; Ota et al., 1993). Abe et al. (1991) postulated that the ancestor of N. uchidai migrated to Uotsurijima from the mainland in the Miocene, when the Senkaku Group was part of the mainland, and was subsequently isolated on Uotsurijima. Subsequent authors, such as Hutterer (1993) and Abe (1994), followed the taxonomic view of Abe et al. (1991), and regarded both the genus Nesoscaptor and the species N. uchidai as valid. Although tentatively supporting the validity of Nesoscaptor and N. uchidai, Nowak (1998) suggested that the genus Nesoscaptor was synonymous with the genus Mogera.

The genus Mogera currently consists of four species (Abe, 1995; Motokawa and Abe, 1996): M. tokudae, M. imaizumii, M. wogura, and M. insularis. Of these, the first two species are endemic to the main islands of Japan, and the third occurs in the Korean Peninsula, northeastern China, the Russian Far East, and the western part of the main islands of Japan (Abe, 1995). The fourth species is known from Taiwan, the southern part of continental China, and Hainan Island (Abe, 1995). Abe et al. (1991) compared N. uchidai with M. tokudae, M. imaizumii, and M. wogura, and found several major morphological differences between it and the latter species.

Recently, Motokawa (2000) pointed out the similarity in some morphometric characters between N. uchidai and the Taiwanese M. insularis using the reports of Abe et al. (1991) and Abe (1995). He surmised that this similarity reflected a close relationship, which corroborates the recent paleogeo-graphical hypothesis of a land connection joining the Senkaku Group, including Uotsurijima, with Taiwan and the continent in the late Pleistocene (Ota et al., 1993; Ota, 1998). To test Motokawa's (2000) prediction, we examined the holotype of N. uchidai and specimens of M. insularis from Taiwan, and analyzed their morphological variation in detail.

MATERIALS AND METHODS

We examined the holotype ofNesoscaptor uchidai deposited in the Zoological Laboratory, Faculty of Agriculture, Kyushu University, Fukuoka. For comparison, we also examined 26 specimens of Mogera insularis from nine localities in Taiwan (Fig. 1). These specimens are deposited in the National Museum of Natural Science, Taichung (NMNS), the Taiwan Endemic Species Research Institute, Chichi (TESRI), the National Pingtung Science and Technology University, Neipu (NPSTU), the National Science Museum, Tokyo (NSMT), the Osaka City University Medical School, Osaka (OCUMS), and the Zoological Collection of Kyoto University, Kyoto (KUZ) (Appendix).

Fig. 1

Map showing the Ryukyu Archipelago (left), the type locality (Uotsurijima) ofNesoscaptor uchidai, and the sampling localities of Mogera insularis (right). A, Yangmingshan, Taipei City; B, Tsejen Village, Fuhsing County, Taoyuan Prefecture; C, Kuanwu, Taian County, Miaoli Prefecture; D, Houlung Town, Miaoli Prefecture; E, Miaoli Sanyi Houyenshan Nature Reserve, Miaoli Prefecture; F, Chushan Town, Nantou Prefecture, and Chichi Town, Nantou Prefecture, and Yuchih County, Nantou Prefecture, and Shuili County, Nantou Prefecture; G, Tatachia, Hsinyi County, Nantou Prefecture; H, Neipu County, Pingtung Prefecture; I, Mt. Nanjenshan, Manchou County, Pingtung Prefecture.

The following external measurements (in mm) taken by the collectors were used for this study: head and body length (HB), tail length (T), forefoot length without claw (FFsu), forefoot length with claw (FFcu), forefoot width (FFW), hind foot length without claw (HFsu), and hind foot length with claw (HFcu). The senior author took 21 cranial measurements to the nearest 0.01 mm with digital calipers (Mitsutoyo Co., Ltd.) (Fig. 2). These included the greatest length of the skull (GLS), condylobasal length (CBL), palatal length from the anterior tip of first incisor to the posterior lip of the palate (PL), rostral length from the anterior tip of the first incisor to the anterior-most portion of the orbit in the infraorbital foramen (RL), postpalatal length from the posterior edge of the postpalatal lip to the anterior-most point on the foramen magnum (PPL), maximum length of the auditory bulla (ABL), postpalatal depth measured from a point just posterior to the posterior lip of the palate to the closest point on the cranial surface (PPD), braincase depth (BD), rostral breadth at the canines (RB), greatest interorbital breadth (IOB), zygomatic breadth (ZB), braincase breadth (BB), breadth across upper molars (BAM), distance from the first upper incisor to the third upper molar (UTR), distance from the canine to the third upper molar (CM3), distance from the fourth upper premolar to the third upper molar (P4M3), breadth across the canines (CC), mandibular length (ML), mandibular height at the coronoid process (MH), distance from the first lower incisor to the third lower molar (LTR), and distance from the first lower premolar to the third lower molar (P1M3). Braincase depth (BD) differs from that of Abe et al. (1991) in excluding the auditory bulla.

Fig. 2

Cranial and mandibular measurements used in this study. The abbreviations used for the measurements are described in the text.

Principal component analysis (PCA) was carried out with the PRINCOMP procedure of SAS Version 6 (SAS Institute Inc., 1990) based on the correlation matrix of cranial measurements. Condylobasal length (CBL) was not included in the PCA, because it did not differ much from GLS (Fig. 2). PPD and ZB were also excluded from the PCA, because relevant portions of these measurements were not available in some specimens. All measurements were log-transformed before PCA. Males and females were combined in each analysis, because the sexual differences in the Japanese moles were slight (Abe, 1967) and the number of available specimens was limited. All specimens have the dentition worn slightly or moderately, and thus considered to be adult.

RESULTS

The external and cranial measurements are presented in Tables 1 and 2, respectively. In most characters, values for the holotype of N. uchidai fell within the range for M. insularis. The values for N. uchidai were slightly larger than the maximum values for M. insularis in one cranial measurement (P4M3).

Table 1

External measurements ofNesoscaptor uchidai and Mogera insularis (mm). Values are given as the means ± SD, followed by the ranges, and sample sizes in M. insularis. See text for abbreviations used for the measurements.

Table 2

Cranial and mandibular measurements forNesoscaptor uchidai and Mogera insularis (mm). Values are given as the means ± SD, followed by the ranges, and sample sizes in M. insularis. See text for abbreviations used for the measurements.

The body was small in both N. uchidai (42.7 g, 129.9 mm HB) and M. insularis (40.8–72.5 g [n=4], 99.5–151.5 mm HB [n=13]). The manus was relatively small in both N. uchidai (14.0 mm FFsu, 19.2 mm FFcu, 15.4 mm FFW) and M. insularis (12.3–15.6 mm FFsu [n=11], 16.7–19.5 mm FFcu [n=10], 12.3–15.6 mm FFW [n=11]). The hind foot (HFsu) was 16.0 mm in N. uchidai and 14.0–16.0 mm (n=12) in M. insularis. The tail is relatively short in both species, with T being 12.0 mm in N. uchidai, and 7.0–17.0 mm (n=13) in M. insularis, showing extreme individual variation. In both species, the nostril was directed outward, and the naked portion on the upper side of the muzzle was rectangular in outline.

The crania and mandibles of N. uchidai and M. insularis are shown in Fig. 3. There was substantial local variation in M. insularis (see below), and two individuals of M. insularis from different localities are shown in Fig. 3. The skull was small and the rostrum was short in both N. uchidai and M. insularis. The anterior portion of the palate of N. uchidai was shorter and narrower than that of M. insularis. The interorbital and palatal portions were broad in both species. There was distinct local variation in the relative breadth of the rostrum and palate in M. insularis. A specimen with a broad rostrum and palate and another with distinctly narrower ones are shown in Figs. 3B and 3C, respectively. The zygomatic arch curved in N. uchidai, but was straight in M. insularis (see lateral views in Fig. 3). The anterior margin of the zygomatic arch reached to the level of the second upper molar (see the ventral view in Fig. 3A) in N. uchidai. In M. insularis, it also reached to a similar level in some individuals (Fig. 3B), but only to the level of the third upper molar in the remaining specimens (Fig. 3C).

Fig. 3

Dorsal, ventral, and left lateral views of the cranium, and lateral and lingual views of the mandible (from top to bottom) ofNesoscaptor uchidai (A, holotype) and Mogera insularis (B, KUZ M2535; and C, KUZ M3362). The bar indicates 10 mm.

The upper incisor row projected forward in both species. It clearly projected in N. uchidai (see the ventral view in Fig. 3A) and the specimens of M. insularis with a narrow rostrum (Fig. 3C), whereas in specimens of M. insularis with a broad rostrum, the degree of projection was lower (Fig. 3B). The first upper incisor was well developed in both species, but its size relative to the second upper incisor was variable within M. insularis. Most specimens of M. insularis had four upper premolar pairs, while N. uchidai had three. Of 23 specimens of M. insularis examined, 22 specimens had 42 or fewer teeth due to complete or partial loss of some after eruption. In the latter case, the roots or other remaining portions of teeth were located in their normal positions. Excluding these variants, we recognized numerical variation in dentition in only one specimen; a specimen from Mt. Nanjenshan (KUZ M3362, Fig. 3C) lacked two teeth, most likely corresponding to the second upper premolars on both sides, judging from the position of the existing teeth. In this specimen, the dental space existed between the first and third upper premolars, but the roots or other remaining portions of teeth were not found. The upper molars were very large in N. uchidai and M. insularis. The length of their row clearly exceeded the distance from the canine to the last premolar in N. uchidai, and it was equal to or slightly smaller than the latter distance in M. insularis. The degree of lingual development of the last upper premolar (= the fourth upper premolar) in N. uchidai was different on the right and left sides: it was more developed on the left side than on the right. In N. uchidai, the tympanic bulla was large, roundish, and flat, with its posterior tip extending backwards beyond the line connecting the mastoid processes. On the other hand, the tympanic bulla of M. insularis was similar to that of N. uchidai in the ventral view, but it was not flattened or well developed ventrally (see the lateral view of Fig. 3).

The mandibles of N. uchidai and M. insularis were slender, and their general appearances were similar. There were three large lower incisor-like teeth in both species. In N. uchidai, there were three lower premolars: the first was caniniform, the second was the smallest, and the third was relatively large and bore two roots. On the other hand, M. insularis had four lower premolars: the first was caniniform, the second and third were small, and the fourth was relatively large, unicuspid, and had two roots. Size comparison suggested that the missing teeth of N. uchidai correspond to the second lower premolars of M. insularis. There was some morphological variation between the right and left fourth lower premolars in N. uchidai; the right was characterized by two clearly developed main cusps, whereas the left only had one main cusp (Fig. 3). The lower molars were large, and the length of their row exceeded the distance between the first lower incisor and the fourth lower premolar in both N. uchidai and M. insularis. The third lower molar was more developed in N. uchidai than in M. insularis. The shape of the coronoid process was quite different between the two species; it was narrow and slender in N. uchidai, whereas it was broad and robust in M. insularis. The size and shape of the humerus were similar in N. uchidai and M. insularis.

Eighteen cranial morphometric characters of N. uchidai and M. insularis were subjected to PCA. The first and second principal component axes explained 62.3 and 10.3% of the total variation, respectively (Table 3). In the first axis, all variables except for PPL showed positive loadings. In the second axis, RL, MH, and PPL had relatively large positive loadings. Individual scores for the first and second principal component variables (PC1 and PC2) are plotted in Fig. 4.

Table 3

Eigenvectors of the principal component analysis (PCA) based on 18 morphometric characters ofNesoscaptor uchidai and Mogera insularis. See text for abbreviations used for the measurements.

Fig. 4

Scatter plots of the scores on the first and second principal component axes based on 18 morphometric characters ofNesoscaptor uchidai (U) and Mogera insularis (A-I, see Fig. 1 for localities).

In this ordination, the plot of the holotype of N. uchidai fell within the range of M. insularis (Fig. 4). Four specimens from the southern part of Taiwan (Neipu and Mt. Nanjenshan; H and I) were distinguished from the other samples by small PC1 and large PC2 scores. The plots of specimens from the northern (Yangmingshan, Houlung Town, and Miaoli Sanyi Houyenshan Nature Reserve; A, D, and E) and central parts of Taiwan (Chushan Town, Chichi Town, and Yuchih County; F) largely overlapped. A specimen from Kuanwu (C) had the smallest PC1 score (−5.55). Kuanwu is approximately 2,000 m above sea level, whereas the elevations of the localities of the other specimens were less than 1,000 m. Six specimens from central Taiwan (F) were collected from localities within 40 km of each other. PC1 scores of these specimens was variable. In descending order, it was largest in a specimen from Teshanli (about 200 m in altitude) (5.12), followed by one from Chushan (about 200 m), two from the Taiwan Endemic Species Research Institute at Chichi (about 300 m), one from Neiliao (about 500 m), and one from Lienhuachi (about 800 m) (−2.00).

DISCUSSION

Taxonomic status ofNesoscaptor uchidai

The present results indicate that N. uchidai substantially resembles M. insularis morphologically. On the other hand, N. uchidai is diverged from the other Mogera species (M. wogura, M. imaizumii, and M. tokudae) in several important morphological characters as shown by previous authors (Abe et al., 1991; Motokawa, 2000). These results suggest that N. uchidai is more closely related to M. insularis than to the remaining Mogera species.

Most of the diagnostic characters for the genus Nesoscaptor and the species N. uchidai (Abe et al., 1991) were actually shared by M. insularis. Moreover, both univariate and multivariate analyses also indicate that most external and cranial measurements in the holotype of N. uchidai fell within the range of variation of the corresponding measurements in M. insularis. The major difference between the two species is the number of premolars. The holotype of N. uchidai lacked four teeth (one premolar pair in both the upper and lower jaws) compared to M. insularis. This numerical reduction of premolars in N. uchidai may have been caused by the decrease in palate size, especially in the anterior portion. The difference between UTR and P4M3, representing the distance between the first upper incisor and the fourth upper premolar (the size of the anterior portion of the palate), is 5.13 in the holotype of N. uchidai. This value is distinctly smaller than the corresponding values in M. insularis (5.17–6.24). In the lateral view, the space for unicuspid premolars between the upper canine and the fourth upper premolar is much smaller in the holotype of N. uchidai than in M. insularis, as shown in Fig. 3. There are three unicuspid premolars in M. insularis, but only two (the first and the third) in the holotype of N. uchidai. Since intraspecific variation in tooth number is sometimes observed in other Mogera species (2.8% in the three species from Japan and the Korean Peninsula [Abe et al., 1991], 4.3% in M. insularis [this study]) and the available material of N. uchidai is limited to the holotype, it is uncertain whether the reduced tooth condition (total 38) observed in this specimen is a reliable diagnostic feature for defining the genus Nesoscaptor. Thus, we do not consider the tooth number as the valid diagnostic character of the genus Nesoscaptor.

Nesoscaptor uchidai and M. insularis are also distinguished by some morphological differences, such as in the zygomatic arch, auditory bulla, and coronoid process (Table 4). These characters may be variable among species of same genus in the talpine moles. For example, Euroscaptor mizura has the curved zygomatic arch, and narrow and slender coronoid process (Abe et al., 1991). On the other hand, the con-generic E. klossi has the straight zygomatic arch, and broad and robust coronoid process (Yoshiyuki, 1988).

Table 4

Morphological comparison betweenNesoscaptor uchidai and Mogera insularis.

We could not find the vertebrata of the holotype specimen of N. uchidai.Nesoscaptor uchidai is known to be different in the number of sacral vertebrata (7 bones) from the genera Euroscaptor (5) and Mogera (6) (Abe et al., 1991). We have examined this condition in five specimens of M. insularis, and found the individual variation. Four have six sacral bones, but the one (NMNS 1942) has seven sacral bones, which is similar to N. uchidai. Therefore, the validity of this character in mole taxonomy needs verification.

We could not examine one of the diagnostic characters for the genus Nesoscaptor and the species N. uchidai pro-posed by Abe et al. (1991). Abe et al. (1991) said that the ear bones of N. uchidai were similar to those of the genera Euroscaptor and Mogera.Nesoscaptor uchidai has malleus with small apophysis orbicularis as in M. tokudae and E. mizura, different from M. wogura, M. imaizumii, and M. insularis having malleus with well developped apophysis orbicularis (Abe et al., 1991; Abe, 1995). This character is different between N. uchidai and M. insularis, but it does not help distinguish Nesoscaptor from Mogera.

The Senkaku Group is located on the southeastern margin of the East China Continental Shelf northwest of the Okinawa Trough. Therefore, the geohistory, fauna, and flora of the Senkaku Group are assumed or known to be distinct from those of the other Ryukyu islands (Ota et al., 1993). Abe et al. (1991) discussed how N. uchidai migrated to Uotsurijima and has been isolated there since the Miocene. However, that scenario is discordant with the current paleogeographical view and the high level of morphological similarity between N. uchidai and M. insularis. During the late Pleistocene, the global land area is thought to have been much larger than that at present because of sea level lowering as a result of continental glaciation (Fairbanks, 1989). During this period, several islands of the Senkaku Group, including Uotsurijima, are thought to have been connected to the continent and to Taiwan (Ota et al., 1993). Therefore, these islands have not been isolated from those land masses for more than 15,000 years, since the post glacier rise in sea levels (Ota et al., 1993).

The ancestor of N. uchidai may have migrated to Uotsurijima in the late Pleistocene via the expanded portion of the southeastern continent, which included Taiwan, and then been isolated on Uotsurijima since then. Based on the current paleogeographical view and the distribution of extant terrestrial vertebrates (Ota, 1998, 2000; Motokawa, 2000), the central Ryukyus (the Amami and Okinawa Groups) are thought to have been isolated from the continent and Taiwan since the Pliocene, and the southern Ryukyus (the Miyako and Yaeyama Groups) since the middle Pleistocene. Therefore, migration of the ancestor of N. uchidai to the central and southern Ryukyus in the late Pleistocene may have been prevented by collapse of the preceding land bridge.

Another mammalian species from Uotsurijima, the striped field mouse, Apodemus agrarius, (Shiraishi and Arai, 1980), also occurs on the continent and in Taiwan. The Uotsurijima population shows slight morphological and karyological differences from populations in the latter two areas (Shiraishi and Arai, 1980; Motokawa, 2000). Abe et al. (1991) suggested that A. agrarius colonized Uotsurijima in the late Pleistocene. The population size and effective habitat area of these two small mammals native to Uotsurijima must be much smaller than those in Taiwan, because Uotsurijima is a much smaller island (4.3 km²). The small population size may have accelerated differentiation between N. uchidai and M. insularis, and between the populations of A. agrarius from Uotsurijima and Taiwan over a short period of time (<15,000 years).

Based on molecular data, Tsuchiya et al. (2000) argued that the four Mogera species are monophyletic, and that the M. insularis lineage diverged from the remaining species in the Tertiary. The present results may show that the M. insularis lineage was split into two species, M. insularis and N. uchidai, in the late Pleistocene. Thus, the phylogenetic position of N. uchidai suggests that the genus Nesoscaptor should be included within the genus Mogera as a junior synonym.

Mogera uchidai (Abe, Shiraishi et Arai, 1991)Nesoscaptor uchidai Abe, Shiraishi et Arai, 1991: 53; Hutterer, 1993: 126; Abe, 1994: 33; Nowak, 1999: 236.

Holotype: Young adult female deposited in the Zoological Laboratory, Faculty of Agriculture, Kyushu University, Fukuoka; collected from a grassy field at the western coast of Uotsurijima, in the Senkaku Group of the Ryukyu Archipelago, by Satoshi Shiraishi and Shusei Arai on 2 June 1979 (Abe et al., 1991).

Diagnosis: A small species of Mogera; tail relatively short; nostrils directed outward; rostrum of skull short and narrow; interorbital portion broad and long; palatal portion very broad; upper incisor rows projected forward; upper and lower molars large; lower incisor-like teeth three and large; the coronoid process slender and narrow.

Distribution: Uotsurijima, in the Senkaku Group of the Ryukyu Archipelago. This species is represented only by the holotype.