2015 Volume 123 Issue 3 Pages 177-184
The Penghu channel, along the western shore of Taiwan, is well known for yielding numerous Middle to Late Pleistocene mammalian fossils, including an archaic human mandible. In this work, we examine a fossil mandible of the raccoon dog, Nyctereutes procyonoides, which was recovered from the bed of the Penghu channel, and compare this with extant raccoon dog subspecies in East Asia. Two-dimensional geometric morphometric analysis of the lower first molar of the Penghu specimen demonstrates that continental and Japanese populations were clearly separated. This indicates that amongst continental populations, the Penghu specimen has a shape intermediate between extant Chinese and Korean populations. Results indicate that the Penghu raccoon dog is phylogenetically closer to the extant continental raccoon dog, rather than those of Japan, and that the Chinese and Korean populations were not greatly separated at that time. When considering the fact that recent phylogenetic and population genetic studies report that the Korean raccoon dog population separated from other continental populations at the last glacial maximum, we can state that the Penghu fauna is at least 20000 years old.
The bed of the Penghu channel has yielded numerous mammalian fossils including extinct elephants (Palaeoloxodon), water buffalo (Bubalus teilhardi), Père David’s deer (Elaphurus davidianus), deer (Cervus), and hyenas (Crocuta crocuta) (Hu and Tao, 1993; You et al., 1995; Tseng and Chang, 2007; Ho et al., 2008). Recently a mandible of archaic Homo was also reported (Chang et al., 2015). On the basis of the mammalian fauna, in particular hyena fossils, the Penghu fauna is dated to the Middle to Late Pleistocene, i.e. after 190000 BP (Ho et al., 1997, 2008; Chang et al., 2015). Although the Penghu fauna is very similar to faunal assemblages in China (Huang and Yu, 2003), extensive phylogeographic investigations of this fauna are as yet lacking. In this study, we investigate a fragmentary mandibular fossil of the raccoon dog, Nyctereutes procyonoides (Canidae, Carnivora) (Figure 1), which is the smallest taxon amongst the recovered Penghu fauna (Chang et al., 2015).
The right mandible fossil specimen of Nyctereutes procyonoides from Penghu channel NMNS006391, F051710 preserving P4 and M1: (A) mandible from lingual view, (B) buccal view, (C) occlusal view, (D) lower first and second molars 3-D reconstructed from occlusal, (E) labial, (F) buccal view, and (G) drawing from lingual view with reconstruction of broken parts shown by dotted lines. Scales are 10 mm.
Living N. procyonoides is widely distributed in East and Southeast Asia (Figure 2: Wozencraft, 2008; Saeki, 2009; Sillero-Zubiri, 2010), and is classified into six geographically separated subspecies: N. p. procyonoides (west and southwest China and north Indochina), N. p. orestes (Gansu, Guizhou, Shaanxi, Sichuan, and Yunnan provinces in China), N. p. koreensis (Korea), N. p. ussuriensis (southeast Russia, northeast China, and east Mongolia), N. p. viverrinus (Japan except for Hokkaido), and N. p. albus (Hokkaido, Japan) (Wozencraft, 2008; Sillero-Zubiri, 2010) (Figure 2).
Distribution range of extant subspecies of N. procyonoides and locality of fossil specimens cited and used in this study. The grey area indicates the current distribution of N. procyonoides reconstructed from Saeki (2009), Wozencraft (2008) and Sillero-Zubiri (2010). The black circle indicates the fossil localities from which we cited measurements. The white circle indicates the fossil localities discussed. Locality: 1, Qingshantou; 2, Zhoukoudian; 3, Anyang; 4, Tingtsun; 5, Jiangzhai; 6, Aoshima; 7, Kuzuu; 8, Shenxiandong; 9, Renzidong; 10, Jiangxi province; 11, Hang Hum. The continental shelf 100 m below sea level around the Korean peninsula to Indochina is shown by a dotted line.
On the other hand, two extinct species of Nyctereutes have been described in East Asia comprising Nyctereutes tingi (in China during the latest Miocene to the Middle Pliocene) and N. sinensis (in China during the Middle Pliocene to the Middle Pleistocene) (Kurtén, 1968; Tedford and Zhanxiang, 1991; Wang and Tedford, 2008). In addition, one species, N. megamastoides, a close relative of N. sinensis, is reported from Pliocene Europe (Tedford and Zhanxiang, 1991; Kauhala and Saeki, 2004; Dong et al., 2013). These extinct species are larger than the extant raccoon dogs and became extinct before the late Middle Pleistocene (Kauhala and Saeki, 2004). Therefore, Kurtén (1968) hypothesized that the extant species evolved from N. sinensis, with a gradual decrease in body size.
Recent phylogenetic studies reported that extant continental and Japanese subspecies are clearly separated into different clades (Kim, 2011; Kurose et al., 2012; Kim et al., 2013). Kim et al. (2013) also reported the partial separation of the Korean population from other continental populations, and a rapid population expansion in Korea after the last glacial period. It was concluded that there were more than two refugia in the continent, including the Korean population and other populations during the last glacial period. However, as fossil specimens have not been well studied, the evolutionary history of their decreasing size and distribution dynamics during glacial periods are still unclear. In this paper, we investigate the phylogeographic position of the Penghu raccoon dog amongst East Asian subspecies based on a geometric morphometric analysis of the lower first molar.
We used a mandible specimen of the fossil raccoon dog from the Penghu channel (NMNS006391, F051711 housed in the National Museum of Natural Science, Taiwan; Ho et al., 1997) (the specimen is hereinafter referred to simply as F051711) and 62 specimens of extant subspecies deposited in several museums, namely the National Museum of Nature and Science, Tokyo; Botanic Garden, Hokkaido University; the American Museum of Natural History; the United States National Museum of Natural History; the Museum of Vertebrate Zoology at Berkeley; and the College of Veterinary Medicine in Seoul National University.
The specimen F051711 from the Penghu channel is a fragmentary right mandible, whose anterior part from the P1 alveolus, the coronoid process above the condyloid process, and the distal top of the angular process are broken and lost. Its carnassial shape is clearly that of canidae and its developed subangular lobe (Figure 1A, B) indicates that the specimen apparently belongs to the genus Nyctereutes, in particular similar to N. procyonoides and N. sinensis, rather than to N. tingi, whose angular lobe is less developed than the extant species (Tedford and Zhanxiang, 1991). P4 and M1 are preserved but the main cusp of P4 and metaconid of M1 are broken. The mandible preserves the posterior part of the alveolus of the canine root, and alveoli of P1 (which has one root), P2 (two), P3 (two containing broken tooth roots), M2 (two containing broken tooth roots), and M3 (one). The number of teeth roots is the same as that of the extant raccoon dog. Two mental foramina exist around the distal root of P3 and mesial root of P2, as in the extant raccoon dog. Three-dimensional images of the fossil specimen were obtained using a One-shot 3D Measurement Macroscope (VR-3050, Keyence Corp., Osaka, Japan) (Figure 1E–G).
We measured the size of the first lower molar (mesiodistal length and buccolingual width) of the specimen in the 2-D images of the occlusal view, using ImageJ software (NIH, Bethesda) (Figure 3A). We cite the length and width of the lower first molar from publications on living and fossil N. procyonoides and its extinct relatives, and compare this with F051711 (listed in Table 1 and Figure 2). Although these measurements cited from publications are based on caliper measurements, our measurements from 2-D images are similar (Table 1). Therefore, the effect of methodological differences in measurement is considered to be negligible.
Analysis of the M1 shape: (A) landmarks and linear measurements used in the analysis on M1, figure is modified from Asahara (2013); (B) shape deformation along the RW1 and RW2 axes, average-to −0.1, circles and arrows indicate landmarks and moving of them, respectively; (C) bivariate plots of RW1 and RW2 on M1, the fossil specimen from Penghu channel (indicated by an allow) positioned about the center of the continental subspecies/populations.
| Species or subspecies | N | Mean | Range | SD | Status | Age | Reference |
|---|---|---|---|---|---|---|---|
| Penghu specimen NMNS006391, F051711 | 12.76 | Penghu channel | |||||
| N. sinensis | 7 | 14.20–16.50 | Extinct (China) | Middle Pleistocene | Pei (1934) | ||
| 16.40 | Middle Pleistocene | Jin et al. (1984) | |||||
| 15.10 | 14.80–15.40 | Early Pleistocene | Zhang (2001) | ||||
| N. megamastoides | 12 | 15.60 | 14.40–16.80 | Extinct (Europa) | Late Pliocene | Argant (2004) | |
| N. tingi | 9 | 18.20 | 17.50–19.30 | 0.71 | Extinct (China) | Early to Middle Pliocene | Tedford and Zhanxiang (1991) |
| N. procyonoides (fossils from China) | 3 | 11.77 | 11.00–12.30 | Fossil (China) | Late Pleistocene | Jin et al. (1984) | |
| 3 | 12.50 | 12.30–12.60 | Pei (1934) | ||||
| 14.00 | Pei et al. (1958) | ||||||
| 12.10 | Qi (1988) | ||||||
| 4 | 12.50–13.10 | Li and Lei (1980) | |||||
| N. procyonoides (fossils from Japan) | 13.00–14.50 | Fossil (Japan) | Late Pleistocene to Holocene | Shikama (1949) | |||
| N. procyonoides procyonoides | 3 | 11.33 | 10.78–12.04 | 0.65 | China | ||
| 6 | 12.38 | 11.89–13.38 | 0.56 | Vietnam | |||
| 53 | 12.20 | 0.59 | Kim (2011) | ||||
| N. procyonoides orestes | 6 | 12.14 | 11.80–12.61 | 0.32 | China | ||
| N. procyonoides koreensis | 18 | 12.74 | 12.02–13.26 | 0.37 | Korea | ||
| 41 | 12.47 | 0.48 | Kim (2011) | ||||
| 37 | 12.63 | Kim et al. (2012) (male) | |||||
| 26 | 12.47 | Kim et al. (2012) (female) | |||||
| N. procyonoides ussuriensis | 70 | 12.67 | 0.55 | Russia | Kim (2011) | ||
| 12.62 | 0.51 | Finland (introduced) | Kauhala et al. (1998) | ||||
| N. procyonoides viverrinus | 26 | 11.31 | 10.07–12.13 | 0.54 | Japan | ||
| 11.61 | 0.59 | Kauhala et al. (1998) | |||||
| 70 | 11.82 | 0.49 | Haba et al. (2008) | ||||
| 70 | 12.02 | 0.52 | Kim (2011) | ||||
| 53 | 11.68 | 0.45 | Kim (2011) | ||||
| 55 | 11.50 | 0.57 | Kim (2011) | ||||
| N. procyonoides albus | 9 | 10.97 | 10.64–11.27 | 0.22 | Japan | ||
| 81 | 11.41 | 0.49 | Haba et al. (2008) | ||||
| 67 | 11.55 | 0.37 | Kim (2011) |
We also compared the shape of the lower first molar using 2-D geometric morphometrics (Zelditch et al., 2004). We considered the number of specimens of easily available subspecies (i.e. Japanese subspecies), in order to avoid bias in our results arising from the large sample size of any particular subspecies (Table 1). We took 2-D images of the mandible specimens from occlusal view by a camera and then digitized eight landmarks on images of M1 in the occlusal view (Figure 3A; Asahara, 2013) using tpsDig (Rohlf, 2010a), and conducted a generalized Procrustes analysis to align the landmark coordinates and divide the shape and size data. We performed principal component analysis on the multivariate shape data (partial warp scores) using tpsRelw (Rohlf, 2010b), termed relative warp (RW) analysis (Zelditch et al., 2004). We drew shape deformations along the RW axis using tpsRelw (Rohlf, 2010b). To quantify similarity of M1 shapes, we computed the Procrustes distances between all individuals using tpsSmall (Rohlf, 2003). Procrustes distances are calculated as a square root of the sum of squared distances between landmarks of each configuration after Procrustes superimposition, indicating differences between the overall shapes of individuals (Zelditch et al., 2004). We compared Procrustes distances amongst all individuals and compared the average scores between subspecies/populations and F051711 to assess their shape similarity.
M1 lengths and widths of the specimens examined and cited are listed in Table 1. F051711 is smaller than the range of other extinct species of Nyctereutes examined in this study. It has much smaller M1 length (12.76 mm) and width than N. sinensis (Figure 4), whereas this is larger than the averages of any other living subspecies of N. procyonoides (ranging from 10.89 to 12.74 mm) (Table 1). Its size is comparable to large individuals of the northern subspecies, N. p. koreensis (Korea) and N. p. ussuriensis (Russia) (Table 1, Figure 4). M1 length increases in order from the extant Chinese and Japanese subspecies, extant Vietnam subspecies, Korean and Russian subspecies, to the fossil Chinese N. procyonoides, fossil Japanese N. procyonoides, and to the extinct relative N. sinensis (Figure 4).
M1 morphological difference among subspecies and species using linear measurements (length versus width). Measurements of extant populations cited from various publications were an average of the particular populations; therefore variation of fossil populations was emphasized.
In the results of the relative warp analysis, the RW1 axis explained 30.61% and the RW2 axis explained 16.08% of the total shape variation. Within each subspecies, RW1 and RW2 scores did not correlate to M1 length (P > 0.05). Shape deformation along the RW1 and RW2 axis are shown in Figure 3B. As the RW1 score increases, landmark 1 moves posteriorly and landmark 5 moves anteriorly, indicating that the total M1 and carnassial blade are shortened in proportion to tooth size (Figure 3B). As the RW2 score increases, the trigonid becomes smaller and the talonid basin and entoconid becomes larger in proportion to tooth size (Figure 3B). Bivariate plots of RW1 versus RW2 are shown in Figure 3C. Continental subspecies occupy the left side of the plots and Japanese subspecies the right side (Figure 3C). Therefore, the RW1 axis seems to extract shape differences between continental and Japanese subspecies. Chinese populations exhibited relatively smaller RW2 scores but Korean and Vietnamese populations displayed relatively larger RW2 scores; hence, the RW2 axis seems to be affected by inter-population differences amongst continental populations (Figure 3C). Amongst these plots, F051711 was plotted around the center of the continental populations (Figure 3C). For the average Procrustes distances from F051711 to several extant subspecies/populations, we note that the distance to N. p. procyonoides in China (0.055) is the smallest and that to N. p. koreensis in Korea (0.057) follows next (Table 2, Figure 5). Procrustes distances within the three continental subspecies or two Japanese subspecies tend to have fewer differences than those between continental and Japanese subspecies, indicating that continental and Japanese subspecies are separated by M1 shape (Procrustes distances: 0.053– 0.069 within continental, 0.059 within Japanese, and 0.064– 0.078 between continental and Japanese) (Table 2).
| Locality | Penghu | China | China (procyonoides) | China (orestes) | Vietnam | Korea | Japan (viverrinus) | |
|---|---|---|---|---|---|---|---|---|
| Taiwan | Penghu | |||||||
| China | 0.062 | |||||||
| China | procyonoides | 0.055 | ||||||
| China | orestes | 0.065 | 0.065 | |||||
| Vietnam | procyonoides | 0.064 | 0.064 | 0.069 | 0.061 | |||
| Korea | koreensis | 0.053 | 0.062 | 0.066 | 0.060 | 0.059 | ||
| Japan | viverrinus | 0.076 | 0.073 | 0.078 | 0.071 | 0.064 | 0.071 | |
| Japan | albus | 0.077 | 0.068 | 0.069 | 0.067 | 0.067 | 0.075 | 0.059 |
Procrustes distances between individuals of various population/subspecies and the Penghu specimen F051711 for M1 shape. The central lateral bar indicates the average, the top and bottom of boxes indicate quartiles, and the vertical bars indicate the range of the specimen with outliers indicated by asterisks.
The Penghu specimen, F051711, is apparently smaller than any other East Asian extinct raccoon dogs (N. tingi and N. sinensis) and European forms (N. megamastoides) (Table 1). Based on size, Ho et al. (1997) identified that the specimen belongs to a smaller extant species, N. procyonoides. This specimen is larger than the living Chinese subspecies and comparable to larger individuals of the northern subspecies in Korea and Russia (Table 1).
As regards M1 shape, F051711 clearly belongs to the continental groups (Figure 3C). Our results of the average Procrustes distances indicate that the closest subspecies/population to F051711 in overall shape is N. p. procyonoides in China, and the second is the Korean N. p. koreensis (Table 2, Figure 5).
Taken together, the specimen F051711 is similar to Korean or Russian subspecies in M1 size (Table 1, Figure 4A), and similar to Chinese and Korean subspecies in M1 shape (Table 2, Figure 5). Extant raccoon dogs are separated into two kinds of karyotypes: continental and Japanese subspecies (Kauhala and Saeki, 2004; Saeki, 2009). Genetic studies also support the fact that continental and Japanese subspecies are separated (Kurose et al., 2012; Kim, 2011; Kim et al., 2013). The differences in shape that we found (i.e. Procrustes distances) correspond to these phylogenetic relationships between continental and Japanese subspecies (Table 2). Accordingly, we consider that the similarity of F051711 to the Chinese population of N. p. procyonoides and to the Korean population of N. p. koreensis in M1 shape suggests their close phylogenetic relationship.
Conceptual diagram of the evolutionary history of the raccoon dog and the possible phylogenetic position and age of the Penghu fossil. The Penghu fossil is estimated to date after the separation of the continental and Japanese populations but before that of the Chinese and Korean populations.
The Penghu channel is geographically close to the distribution range of extant Chinese N. p. procyonoides (Figure 2). However, with the low sea levels that occurred during the ice age, the Penghu strait dried up, and the geographic range of the raccoon dog is likely to have moved southward. For example, genetic studies suggest that the Korean subspecies lived in a refugium during the last glacial period (Kim et al., 2013). In addition, the continental shelf around the Yellow Sea and East China Sea, south of the Korean peninsula, would have been exposed when the Penghu channel dried up (Figure 2). Therefore, our results are likely to reflect the phylogenetic affinity between F051711 from the Penghu channel and the two living populations of China and Korea. This result not only supports the affinity between the Penghu fauna and that of contemporary fauna on the opposite shore in the continent, but also suggests important implications for the evolutionary history of the continental raccoon dog. Thus, contemporary with F051711, extant Chinese and Korean raccoon dog subspecies might have not been separated greatly, while continental and Japanese subspecies may have been clearly separated from each other.
Recent population genetic studies reveal that the Korean population was separated from other populations, as a refugium during last glacial period, with an estimated population expansion occurring around 20000 years ago (ranging from 0–40000 BP) (Kim et al., 2013). Another study based on mitochondrial genetics estimated subclades within the continental population (using the Russian raccoon dog N. p. ussuriensis, which was introduced to Europe), suggesting a division around 457800 years ago (ranging from 223300 to 773900 BP) (Pitra et al., 2010). These researchers also estimated the diversification between the Russian and Japanese raccoon dogs (N. p. viverrinus) at around 870000 years ago (ranging from 480000 to 1370000 BP) (Pitra et al., 2010). Together, these studies, along with our results based on morphological examination, suggest that the Penghu fauna is older than the last glacial maximum (20000 BP) and is younger than 870000 BP, which corresponds to the Middle to Late Pleistocene (190000 BP, estimated by the appearance of Crocuta crucuta ultima in East Asia) (Ho et al., 1997; Chang et al., 2015), although molecular and morphological divergence could have occurred at different times, and age estimation via ‘molecular clocks’ is not entirely accurate. Moreover, based on estimations of sea levels (Waelbroeck et al., 2002), Chang et al. (2015) suggests that members of the Penghu fauna, including archaic humans, may be dated to between the two phases of low sea levels around 10000–70000 or 130000–190000 years ago. Our results on the raccoon dog are consistent with this estimation, but suggest that the age could be pushed back to 20000 years ago.
At present, Chinese and Japanese raccoon dogs have a smaller M1 (average length: 10.89–12.20 mm) as compared to the Korean and Russian raccoon dogs (average length: 12.47–12.74 mm) which may reflect Bergman’s rule (when we consider tooth size as a measure of the total body size; Table 1). It is hypothesized that, in the period postdating the age of specimen F051711, a size decrease occurred in Chinese raccoon dogs, reflecting climate change, while this may not be visible in the Korean population as their range moved to north, i.e. to the colder Korean peninsula.
Chang et al. (2015) argues that the archaic Homo specimen from the Penghu channel is closely related to the Hexian specimen from Anhui province, China, which is located slightly north of this channel. When considering raccoon dog phylogeography and the possible range dynamics of this species as affected by climate change, it is suggested that the mammalian fauna in the Anhui province during the interglacial period could resemble the Penghu fauna.
Fossil N. procyonoides have been reported across China, from the north to the southern regions such as Jiangxi (the site was not notified: Handbook of Chinese Vertebrate Fossils Editorial Group, 1979), Jiangsu (Shenxiandong: Li and Lei, 1980), and Anhui provinces (Renzidong: Jin and Liu, 2009) (Figure 2). The Penghu specimen, which is clearly classified as an extant species, indicates that N. procyonoides was distributed in the more southern areas of China prior to the Late Pleistocene.
Kurtén (1968) indicates an evolutionary trend of size reduction in raccoon dogs, which is consistent with our results based on data from China and Taiwan. That is, M1 sizes have decreased from the larger N. sinensis, intermediate fossil N. procyonoides from the Penghu channel and several localities of China, to the smaller living N. procyonoides in China (Table 1). In Japan, Late Pleistocene to Holocene fossils of N. procyonoides (from Kuzuu and Aoshima: Figure 2), are obviously related to the living Japanese raccoon dog, and are larger than living ones in terms of M1 length (Shikama, 1949) (Table 1, Figure 3A). Shikama (1949) argues that size reduction occurred independently both in the continent and in Japan. Our results of the affinity between smaller living Chinese raccoon dogs and the larger fossil raccoon dog from the Penghu channel, which is clearly different from Japanese raccoon dogs, thus supports this argument. While adaptation to the colder climate of the ice age can be an explanation for the large size of fossils, it is important to note that these fossil raccoon dogs were larger than the present northern-most raccoon dog N. p. ussuriensis (Table 1). Therefore, size reduction of the raccoon dog may be caused not only by the climate change, but by other factors as well. This result could also suggest that the Penghu specimen is Late Pleistocene in age because it is intermediate in size among Late Pleistocene fossils in China (Table 1) which show a tendency of decreasing size. While the present study was based on the shape of only one M1, further discoveries of appropriate fossils of N. procyonoides would help to reconstruct a more robust evolutionary history for raccoon dogs.
We thank S. Kawada (National Museum of Nature and Science, Tokyo), F. Takaya (Botanic Garden, Hokkaido University), E. Westwig (American Museum of Natural History), E. Langan (US National Museum of Natural History, Smithsonian Institution), C. Conroy (Museum of Vertebrate Zoology at Berkeley), and A.C.C. Lau (College of Veterinary Medicine, Seoul National University) for arranging specimens of the extant species. This study was financially supported by Kyoto University Foundation and Grant-in-Aid from Japan Society for the Promotion of Science (11J01149) (to M. Asahara).
