A Review of the Processes of Mammalian Faunal Assembly in Japan: Insights from Molecular Phylogenetics

Abstract

To elucidate the origins of the Japanese mammalian fauna from the perspectives of biogeography and community ecology, I reviewed molecular phylogenetic and phylogeographic studies for all non-volant terrestrial mammals indigenous to the Japanese Archipelago (63 species), with a particular focus on obtaining reliable chronological data. The results of this review demonstrate that geological vicariance events in the Tsugaru and Korea (Tsushima) Straits can explain the distribution of many Japanese mammals, in particular the Hokkaido-endemic species with Late Pleistocene origins and the Honshu–Shikoku–Kyushu-endemic species with Middle Pleistocene or earlier origins. Phylogenetic relatedness also contributed to the observed patterns of distribution through the processes of competitive exclusion and species assortment, and abiotic environmental filtering was another important factor. Later colonists of Honshu–Shikoku–Kyushu, from northern Hokkaido or the southern Tsushima Islands, were mostly excluded owing to the competitive dominance of earlier residents or environmental filtering. On the other hand, the fragmented distributions of some species with more ancient origins in both Hokkaido and Honshu–Shikoku–Kyushu may be a result of the competitive dominance of later migrants. Ecological coexistence can be achieved by phylogenetically dispersed species, supporting the principle of species assortment. Because almost all aspects of the mammalian faunal assembly in Japan can be explained by geological events or community ecological processes, the Japanese Archipelago may be an ideal model island system in which to study the mechanisms of faunal assembly.

Appendix

Here, I have reviewed molecular phylogenetic and phylogeographic studies of Japanese mammals to obtain reliable estimates of the divergence time between the Japanese and the continental species (or lineage) and the time to the most recent common ancestor (MRCA) of the Japanese mammals. Where no chronological data were provided in the previous study, even though some DNA sequences were determined and available in the DNA database, I briefly calculated these times by using the lineage-specific evolutionary rate. It should be noted here that the most commonly used evolutionary rates proposed by Brown et al. (1979) and Irwin et al. (1991) were not considered effective because the former was based on only the restriction fragment-length polymorphisms (RFLP), which provide less sufficient variations than the sequence data (McKay 2012 used this rate), and both rates were estimated from a limited mammalian group (primates in the former and ungulates in the latter; McKay 2012 applied this rate to all the mammalian lineage divergence), so that the applications of these rates to other groups would be difficult because of the differences in the evolutionary rates among mammalian groups. For example, a corollary of the application of the evolutionary rate obtained from a slowly evolving lineage (e.g., ungulates) into a rapidly evolving lineage (e.g., rodents) is an overestimation of the divergence times on the rapidly evolving lineage.

1.1 Eulipotyphla

1.1.1 Soricidae

Soricidae is the most speciose insectivorous family in the order Eulipotyphla and includes 26 genera and 376 species distributed worldwide (Hutterer 2005). These species could be classified into two subfamilies, Soricinae and Crocidurinae (Dubey et al. 2007; but Myosoricinae was proposed as the third subfamily by Hutterer 2005). In a molecular phylogenetic study with two mitochondrial (Cytb and 16S rRNA) and two nuclear (Apob and Brca1) genes, these subfamilies were demonstrated to have diverged in the late Early Miocene [ca. 20 MYA (million years ago) (Fumagalli et al. 1999; Dubey et al. 2007)]. In the Japanese Archipelago, seven soricine (six Sorex and one Chimarrogale) and five crocidurine (four Crocidura and one Suncus) species exist (Ohdachi et al. 2015). As regarding Sorex species in the subfamily Soricinae, four of six species (S. minutissimus, S. caecutiens, S. unguiculatus, and S. gracillimus) have a similar distribution across the northern Eurasian continent and eastern marginal islands including Hokkaido, Sakhalin, and adjacent small islands [not present in the Honshu–Shikoku–Kyushu (hereafter designated as HSK) and Ryukyu regions], representing BP2 (Table 3.1). The other two Sorex species (S. hosonoi and S. shinto) are both endemic to the HSK region (BP6; Table 3.1) and are phylogenetically close to S. minutissimus and S. caecutiens, respectively (Ohdachi et al. 1997, 2001, 2006). This phylogenetic structure suggests that the generations of S. hosonoi and S. shinto lienages were earlier than those of the Hokkaido lineage of each sister species, S. minutissimus and S. caecutiens, respecitively. It could therefore be assumed that S. hosonoi and S. shinto were earlier colonizers of the Japanese archipelago than S. minutissimus and S. caecutiens, respectively. Ohdachi et al. (2001) examined intraspecific genetic variations based on the Cytb gene for all the six Japanese Sorex species and found that, despite the similarity in the distribution patterns of the four Palaearctic Sorex species (S. caecutiens, S. minutissimus, S. unguiculatus, and S. gracillimus), their phylogeographic patterns are different from each other; where the Hokkaido lineage of S. caecutiens is well differentiated from the continental one, no genetic differentiations are observed between Hokkaido and the continental populations in S. minutissimus and S. unguiculatus, and there are three distinct haplogroups specific to the Hokkaido population of S. gracillimus. These observations suggest that the four Sorex species in Hokkaido established their lineages by different histories in spite of the similar distribution patterns. However, in Ohdachi et al. (2001), the time scale for the lineage differentiations was not assessed based on their molecular data. Hope et al. (2010) estimated that the mutation rate/lineage/myr for the Cytb gene in the S. minutissimus lineage was 5.5 % (therefore 11 % divergence rate/myr), assuming 1 MYA for the divergence between S. hosonoi and S. minutissimus on the basis of the geological evidence, and obtained the result that the time to the most recent common ancestor (MRCA) of lineages in S. minutissimus including the Hokkaido linage was in the period of the Late Pleistocene (<0.13 MYA). This time estimate is in agreement with the distribution pattern of S. minutissimus in Japan, that is, absence from HSK and Ryukyu regions that would have been caused by the Tsugaru Strait established in the Late Pleistocene, preventing the dispersals from Hokkaido to the southern major islands of Japan (Ohshima 1991).

Here, I adopt the evolutionary rate of Hope et al. (2010) for the data of Ohdachi et al. (2001) to introduce the time scale into the Japanese Sorex species evolution. However, because the evolutionary rate of Hope et al. (2010) was based on the divergence between S. hosonoi and S. minutissimus resulting from the geological features of the Japanese Archipelago, the application of this rate to test the influence of the straits around the Japanese Archipelago on the genetic diversity of Japanese Sorex species might have fallen into circular argument. Nevertheless, a similar evolutionary rate was suggested to be fit to the late Quaternary phylogeography of various taxa from rodents to carnivorans around the Bering Straits (Hope et al. 2014). Therefore, I used the 11 %/myr (million years) divergence rate in this tentative time estimations. The average pairwise genetic distances between species endemic to the HSK region (S. hosonoi and S. shinto) and each sister species (S. minutissimus and S. caecutiens, respectively) were calculated to be 8.3 % and 8.8 %, respectively, thus suggesting that these lineages diverged 0.75 and 0.8 MYA in the Early to Middle Pleistocene, respectively. On the other hand, Hokkaido lineages of S. minutissimus and S. caecutiens were estimated to have diverged 0.07 and 0.56 MYA, respectively. It is obvious to obtain the age estimate consistent with Hope et al. (2010) for S. minutissimus because of the same assumption about the evolutionary rate. However, the Middle Pleistocene origin of the Hokkaido lineage of S. caecutiens provides a novel insight (Ohdachi et al. 2001 assumed the early Late Pleistocene origin), although the most divergent lineages in Hokkaido in S. caecutiens was inferred to have split 0.09 MYA, implying that the lineage diversifications in Hokkaido might have occurred in the Late Pleistocene. Meanwhile, it was estimated that S. unguiculatus diverged from the most closely related species, S. isodon, 0.64 MYA. There is no genetic differentiation between the Japanese and continental populations in S. unguiculatus, where the average pairwise distance between Khabarovsk and Hokkaido haplotypes and that among the Hokkaido haplotypes both show 1.6 % corresponding to 0.15 MYA based on the 11 %/myr divergence rate, suggesting that the Hokkaido lineage diverged from the continental one in the final part of the Middle Pleistocene. Regarding S. gracillimus, it was calculated that the divergence between the Hokkaido and the continental lineages, that among three haplogroups in Hokkaido, and that within each of the three haplogroups, occurred 0.07, 0.05, and 0.02 MYA, respectively. All the events took place in the Late Pleistocene. The foregoing calculation suggests that the absence of S. minutissimus and S. gracillimus in Honshu and more southern islands can be explained by the barrier of the Tsugaru Strait in the Late Pleistocene (Ohshima 1991). It is not clear why S. caecutiens and S. unguiculatus, whose origin in Hokkaido is in the Middle Pleistocene, did not cross the Tsugaru Strait, when the land bridge was formed. Probably, the lineage expansion would have occurred in the Late Pleistocene, as suggested in S. caecutiens, or some community ecological mechanisms might have been applied (see Sect. 3.4.2 in the text).

The other soricine species, the Japanese water shrew Chimarrogale platycephalus, is a semiaquatic species endemic to the HSK region (BP6; Table 3.1) (the scientific name was used following Ohdachi et al. 2015). He et al. (2010) intensively examined the interspecific phylogenetic relationships among shrew species representing all the genera within the tribe Nectogalini, where Chimarrogale is included, based on eight mitochondrial genes (Cytb, Co1, Atp6, Nd2, Nd4, Nd5, and 12S and 16S rRNA) and three nuclear protein-coding genes (Apob, Brca1, and Rag2). They showed that two semi-aquatic Chimarrogale species, C. platycephalus and C. himalayica, were closely related to each other, and estimated the divergence time between them on the basis of the two nuclear protein-coding genes (Apob and Brca1) to be 1.25 MYA with several assumptions of fossil calibrations (the oldest remain of the Soricinae–Crocidurinae ancestor at 20 MYA, the oldest Blarinellini and Blarinini at 15 MYA, and the oldest known fossil species of the subgenus Sorex at 3.5 MYA). However, special caution must be given to the interpretation of their results. First, C. platycephalus in their study is designated to be of Taiwan origin (see their Table 3.1), which is not consistent with the current taxonomy because C. platycephalus is endemic to Japan. Second, because C. himalayica in their study is designated to be of Yunnan origin, the foregoing time estimate should be viewed as the divergence time between Taiwanese and continental lineages of C. himalayica. Ohdachi et al. (2006) and Yuan et al. (2013) showed that C. platycephalus is more closely related to C. himalayica in Taiwan than is the continental C. himalayica, suggesting that C. himalayica is paraphyletic to C. platycephalus. Yuan et al. (2013), using the Cytb gene, estimated the branching time of C. platycephalus to be 3.03 MYA. However, this is possibly overestimated because of the use of the Cytb gene whose substitutions would be saturated in the time scale of their study, where the calibration point was set to approximately 20 MYA, that is, at a level subjected to the saturation problem (Steppan et al. 2005). Although He et al. (2010) adopted an incorrect taxonomic framework for C. platycephalus, I suppose that their time estimates could be more realistic because they estimated divergence times by less-saturated nuclear protein-coding genes and obtained dates largely concordant with the fossil records (He et al. 2010). Taking into account that Yuan et al. (2013) estimated the divergence time between Taiwanese and continental lineages of C. himalayica to be 4.58 MYA and He et al. (2010) estimated 1.25 MYA, as already explained, 3.03 MYA estimated for the split between C. platycephalus and C. himalayica in Taiwan in Yuan et al. (2013) could correspond to age about 0.83 MYA by a very brief proportion calculation. Furthermore, Yuan et al. (2013) indicated that the genetic distance of the Cytb gene between C. platycephalus and C. himalayica in Taiwan is 10.5 %. If I use the 11 %/myr of Hope et al. (2010), the time estimate for this divergence would be 0.95 MYA, which is close to 0.83 MYA, although slightly larger.

Based on the mitochondrial Cytb variations of C. platycephalus, Iwasa and Abe (2006) clarified that there are four geographically separated phylogroups in the Japanese archipelago, one in Kyushu and the other three in Honshu. The most southern Kyushu lineage first diverged, then the Chugoku lineage in the southwestern Honshu, and in the last Kinki lineage in the east of the Chugoku region and the lineage in East and Central Honshu were separated. The time to the MRCA of the four phylogroups was inferred to be 0.39 MYA, and divergence in each phylogroup was estimated to have occurred within 0.059 MYA, on the basis of the 1.36 % transversional susbtitutions at the third codon positions per million years calculated from setting the divergence time between Soricinae and Crocidurinae to be 20 MYA, according to Fumagali et al. (1999), in which the transversional susbtitutions at the third codon positions was checked to be not saturated. In addition, by using the 11 %/myr of Hope et al. (2010) for the data of Iwasa and Abe (2006) where average pairwise genetic distances between haplotypes in Kyushu and East-Central clades is 3.0 %, the divergence time was calculated to be 0.27 MYA, which is not inconsistent with the Middle Pleistocene origin of C. platycephalus as estimated earlier (0.39 MYA). All these inferences suggest that C. platycephalus originated in the southern part of the Japanese Archipelago (except for the Ryukyu region) in the Middle Pleistocene and expanded northward from four different refugia in a post-glacial period in the Late Pleistocene (Iwasa and Abe 2006). As already described, because the Tsugaru Strait was not considered to have had a land bridge in the Late Pleistocene, the absence of this species from Hokkaido can be explained by the Late Pleistocene expansion of the Honshu lineages from each refugium. The reason for the absence of this species from the Shikoku islands is not clear.

Of all the five Crocidurinae shrew species in the Japanese archipelago, only one species, the Japanese white-toothed shrew Crocidura dsinezumi, is endemic to the HSK region (BP6), whereas the others are restricted to small islands around the Japanese archipelago, especially in the Ryukyu region: the Orii’s shrew C. orii (Amami Isl., BP9; Table 3.1), the Asian lesser white-toothed shrew C. shantungensis (Tsushima Isl., BP10; Table 3.1), the Watase’s shrew C. watasei (Amami and Okinawa Isl., BP9; Table 3.1), and the musk shrew Suncus murinus (Amami, Okinawa, Miyako, and Yaeyama Isl., BP10; Table 3.1). Dubey et al. (2008) examined the phylogenetic relationships among Crocidura and Suncus species with two mitochondrial (Cytb and 16S rRNA) and two nuclear (Apob and Brca1) genes, and estimated that the Crocidurinae shrew species were inferred to have diversified on mainly Palaearctic and Oriental regions since the Late Miocene. One of major clades derived from the diversifications is the Asian clade, where all the four Crocidura species in Japan were included (Dubey et al. 2008). Using the calibration point of 20 MYA for the divergence between Soricinae and Crocidurinae, Dubey et al. (2008) estimated that C. dsinezumi diverged from the clade that comprised C. lasiura and C. kurodai 1.42 MYA, C. watasei branched from the most closely related species C. horsfieldi 1.11 MYA, and C. orii, whose phylogenetic affinity was not unambiguously inferred, was generated 5.42 MYA during the earliest diversification among the Asian clade. C. shantungensis is also a member of the Asian clade and has been demonstrated to be the most closely related to the C. suaveolens group (Ohdachi et al. 2004; Dubey et al. 2006, 2008). Although the divergence of C. shantungensis in Tsushima Islands has not been dated in any of these previous studies, data used by Motokawa et al. (2000) and Ohdachi et al. (2004) showed that pairwise difference of the Cytb gene between the Tsushima lineage and the most closely related Cheju lineage was 1.2 %, suggesting that the divergence time is 0.1 MYA on the basis of the 11 %/myr divergence rate of Hope et al. (2010). Unfortunately, the intraspecific genetic variations of S. murinus including the Japanese lineage have never been investigated. Thus, the time scale cannot be evaluated here. Concerning the other intraspecific genetic diversity of these Crocidurinae species, it was suggested that there is a clear phylogeographic demarcation between eastern and western Japanese lineages in Crocidura dsinezumi (Han et al. 2002; Ohdachi et al. 2004). Their average pairwise genetic distance is 3.7 %, indicating that evolutionary split between eastern and western linages occurred 0.34 MYA based on the 11 %/myr divergence rate. It has also been known that individuals of C. watasei in Amami-Oshima and Okinawajima Islands have the same Cytb haplotype (Motokawa et al. 2000).

1.2 Talpidae

The family Talpidae comprises 39 species and shows a variety of ecological lifestyles, including fossorial, semi-fossorial, terrestrial, and semi-aquatic insectivorous mammals (Hutterer 2005). Eight talpid species are present in Japan, of which six species are fossorial moles (five Mogera and one Euroscaptor species) and two are semi-fossorial shrew moles (one Dymecodon and one Urotrichus species). Seven of these eight talpid species are endemic to the Japanese archipelago and only M. wogura is also found in the eastern Eurasian continent [Hutterer 2005; this species is sometimes considered endemic to Japan because the Eurasian lineage is classified as M. robusta (Zemlemrova et al. 2013)]. No talpid species are observed in Hokkaido. The biogeographic patterns for the talpid species are BP6 [the lesser Japanese shrew mole Dymecodon pilirostris, the greater Japanese shrew mole Urotrichus talpoides, the Japanese mountain mole Euroscaptor mizura, the lesser Japanese mole Mogera imaizumii, and the Echigo mole Mogera etigo (the last one is considered the same species as the Sado mole Mogera tokudae); Table 3.1], BP7 (the large Japanese mole Mogera wogura; Table 3.1), and BP9 (the Senkaku mole Mogera uchidai; Table 3.1). Their precise distributions are not considered to overlap except for the narrow geographic boundaries among them. In fact, the distribution of E. mizura is only observed in the restricted and fragmented montane regions in Japan; M. tokudae and M. etigo are confined to Sado Islands and nearby Echigo Plain in Honshu, respectively, M. imaizumii is mainly observed in the eastern part of Japan and remains in very limited regions in some western parts of Japan, and M. wogura predominantly occupies the western part of Japan and possesses the same species as in the Eurasian continent. These distribution patterns suggest the different evolutionary history among these mole and shrew mole species. Previous molecular phylogenetic analyses have clarified that the Asian fossorial moles formed a monophyletic lineage (Euroscaptor and Mogera are included), in which, in the Japanese moles, E. mizura constitutes monotypic lineage that diverged in the most basal part within the Asian mole clade, and M. imaizumii and M. wogura (including Eurasian and Japanese lineages) are sister species to each other with the M. tokudae–M. etigo clade closely related to the M. imaizumii–M. wogura clade (Tsuchiya et al. 2000; Shinohara et al. 2003, 2004, 2005, 2014; Zemlemrova et al. 2013; He et al. 2014; as no molecular studies have been conducted for M. uchidai, I could not discuss the origin of this species hereafter). Namely, among the Japanese mole species, E. mizura branched off first, M. tokudae–M. etigo second, M. imaizumii third, and the Japanese lineage of M. wogura at the last. Combined with the distributions of these moles in the Japanese Archipelago, these phylogenetic relationships suggest that the Japanese mole species migrated from the Eurasian continent following the branching order in the phylogeny, implying that the earlier the migration is, the more restricted the distributions are to montane regions, islands and small areas, and the northeastern region in Honshu. On the basis of the fossil record (16.4–20.5 MYA for the earliest Talpini fossil in East Asia and 4.0–4.3 MYA for the oldest Scaptochirus fossil), the divergence time of the E. mizura lineage was estimated to be 17.1 MYA in the analyses using the mitochondrial Cytb and 12S rRNA and nuclear Rag1 genes (Shinohara et al. 2014). Shinohara et al. (2014) also inferred that the lineage of M. tokudae (M. etigo should also be placed here) was generated 8.6 MYA and that M. imaizumii diverged from M. wogura (both Eurasian and Japanese lineages are included) 5.7 MYA. On the other hand, using the similar calibration standard based on the fossil records, He et al. (2014) estimated by the 2 mitochondrial genes (Cytb and 12S rRNA) and 12 nuclear genes (Adora3, Atp7a, App, Bche, Bdnf, Bmi1, Brca1, Crem, Plcb4, Rag1, Rag2, and Ttn) that the E. mizura lineage diverged 12.32 to 13.45 MYA and the M. imaizumii/M. wogura split occurred 3.42 to 4.5 MYA, thus indicating estimates younger than those in Shinohara et al. (2014). This difference in the time estimates would be caused by the stronger effect of the mitochondrial genes in Shinohara et al. (2014). Tsuchiya et al. (2000) also indicated similar time estimates by using the transversional substitutions of the Cytb gene. The results show that E. mizura, M. tokudae, and M. imaizumii branched off in this order 12.6–17.7, 6.6–9.2, and 2.5–3.4 MYA, respectively (Tsuchiya et al. 2000). Kirihara et al. (2013) assumed that the divergence between Talpa europaea and Talpa caucasica occurred 4.75 MYA and estimated with the Cytb gene that E. mizura, M. tokudae, M. imaizumii, and the Japanese lineage of M. wogura branched off in this order 5.21, 3.64, 2.31, and 1.20 MYA, respectively. These time estimates are much younger than those in other studies. I suppose that the divergence times inferred by Shinohara et al. (2014) and He et al. (2014) might be overestimates because they depended on the mitochondrial genes in some degree for the time scale, where the mitochondrial genes would not work well because of the saturation problem, and also adopted the calibration point based on the earliest record of Talpini in East Asia (16.4–20.5 MYA), which might also be too old for the mitochondrial gene variations to be calibrated. Although Tsuchiya et al. (2000) focused on less-saturated transversional substitutions, the depth of the calibration point seems too old (25–35 MYA for the divergence between Urotrichini and Talpini), so that the estimated time scale may be overestimated. The estimates of Kirihara et al. (2013) are in agreement with the geological and eustatic records and the migration ages of the proboscidean species into Japan. Kirihara et al. (2013) also suggested that the genetic distance between M. tokudae in Sado Islands and M. etigo in Echigo Plain is only 2 %, indicating that, on the basis of their divergence rate of the Cytb gene (4–6 %/myr), these two “species” diverged 0.3 to 0.5 MYA; this is a level of intraspecific variation (Kirihara et al. 2013).

With respect to two shrew mole species D. pilirostris and U. talpoides endemic to the Japanese Archipelago, their origins have not been clearly understood. Previous molecular phylogenetic analyses demonstrated that these two shrew mole species formed a clade (Shinohara et al. 2003, 2004). However, no time scale has been estimated yet. Furthermore, as no talpid species is closely related to one of these species (that is, the clade consists of these two species only), it is not clear which species migrated into the Japanese Archipelago first. Taking into account that D. pilirostris has a scattered distribution in the montane region as in the case of E. mizura, and U. talpoides occupies most lowlands of the Japanese Archipelago similar to Mogera species in Japan, it might be likely that D. pilirostris established its lineage in Japan earlier than U. talpoides. A simple calculation with the 4–6 %/myr Cytb divergence rate for East Asian mole species (Kirihara et al. 2013) led to 1.9–2.85 MYA for the divergence time between D. pilirostris and U. talpoides (average pairwise genetic distance was 11.4 %; data were derived from Shinohara et al. 2003).

It should also be noted here that E. mizura, D. pilirostris, and U. talpoides show a very high extent of intraspecific variations corresponding to the interspecific difference in other mole and shrew mole species (Shinohara et al. 2003, 2014). The average pairwise genetic distances between the most divergent lineages in E. mizura, D. pilirostris, and U. talpoides are 7 %, 10.6 %, and 6.9 %, respectively, on the basis of the data of Shinohara et al. (2003, 2014), which could be calculated to be 1.17–1.75 MYA, 1.77–2.65 MYA, and 1.15–1.73 MYA, respectively, following a 4–6 %/myr Cytb divergence rate (Kirihara et al. 2013). These time estimates may correspond to the interspecific levels of diversifications.

1.3 Primates

1.3.1 Cercopithecidae

Cercopithecidae is the most speciose primate family, composed of 132 species of Old World monkeys (Groves 2005). Among them, only 1 species, the Japanese macaque Macaca fuscata endemic to Japan, can be found in the HSK region (BP6; Table 3.1). Previous molecular phylogenetic analyses with the mitochondrial or nuclear gene sequences demonstrated that M. fuscata is closely related to the rhesus macaque Macaca mulatta and the Taiwan macaque Macaca cyclopis (Hayasaka et al. 1996; Tosi et al. 2003; Chu et al. 2007). Assuming that the divergence between Asian and Barbary macaques occurred 3.0 MYA, based on the fossil record and using a portion of the mitochondrial DNA sequences (Nd4, Nd5, and some tRNAs), Hayasaka et al. (1996) estimated that M. fuscata diverged from the clade of M. mulatta and M. cyclopis 0.65 to 0.73 MYA and the differentiation among three Japanese lineages took place 0.19 to 0.37 MYA. On the other hand, Tosi et al. (2003) utilized the mitochondrial (12S rRNA, tRNA-VAL, and 16S rRNA) and Y-chromosomal (Sry and Tspy) genes for the divergence time estimations postulating that the initial divergence of macaques occurred 5.5 MYA, caused by the Messinian Salinity Crisis. In this study, the time for the branching of M. fuscata from the other closely related species above was estimated to be 1.0 MYA in the Y-chromosomal gene tree. Although not estimated clearly, the mitochondrial gene tree suggested that M. fuscata diverged in a time younger than 1.2 MYA, which is consistent with the Y-chromosomal gene estimation (Tosi et al. 2003) and also is not contradicted by the estimates of Hayasaka et al. (1996; 0.65–0.73 MYA). Afterward, in a study with more extensive within-species sampling for M. fuscata, M. cyclopis, and M. mulatta, and on the basis of the divergence rate of 28 %/myr of the mitochondrial control region, the divergence of M. fuscata from the continental lineages of M. mulatta was estimated to be 0.38 to 0.42 MYA (Chu et al. 2007). Each of the species M. fuscata and M. cyclopis shows close affinity to one of the different lineages of M. mulatta (for example, M. fuscata is closely related to a M. mulatta lineage from southeastern China and Vietnam). Thus, M. mulatta is not monophyletic (Chu et al. 2007), and this further suggests that the lineages of M. fuscata and M. cyclopis are consequences of the dispersal of M. mulatta into the eastern marginal islands of the Eurasian continent. Although the exact time of the migration has not been fixed yet, the last estimate by Chu et al. (2007) is not inconsistent with the palaeontological record that the oldest fossil of M. fuscata was discovered in Japan around 0.43–0.63 MYA (Aimi 2002).

Marmi et al. (2004) conducted intraspecific phylogeographic and demographic analyses for M. fuscata by using a portion of the control region and obtained a consistent estimate for the colonization age of the ancestor of M. fuscata (0.31–0.88 MYA). They also showed that the extent of the genetic difference between M. fuscata and the eastern lineage of M. mulatta is less than that between eastern and western lineages of M. mulatta and proposed that the M. fuscata be included in the same species as the eastern M. mulatta (Marmi et al. 2004). Treating both species as a single species, they further suggested that the ancestral population of M. fuscata and the eastern M. mulatta experienced a rapid expansion 0.16 to 1.00 MYA (Marmi et al. 2004). Kawamoto et al. (2007) similarly examined intraspecific variations of M. fuscata with more individual samples from the northernmost Shimokita Peninsula (located in the northern part of Aomori Prefecture in northernmost Honshu) to the southernmost Yakushima Islands (located in the south of the Kyushu Islands) by the control region sequences, demonstrated that there are major eastern and western lineages in the Japanese Archipelago, and suggested that the eastern lineage experienced the recent northward demographic expansion from a refugium in the post-glacial periods in the Late Quaternary.

1.4 Rodentia

1.4.1 Gliridae

There are 28 species in the family Gliridae (Holden 2005), in which almost all the species are observed in Europe and Africa. The Japanese dormouse Glirulus japonicus is endemic to Japan and inhabits the HSK region (BP6; Table 3.1). Because G. japonicus is the only species within Gliridae that inhabits East Asia, it is extremely difficult to infer the origin of this species in Japan by only investigating the extant species. Nevertheless, Nunome et al. (2007) clarified with three nuclear gene exon sequences (Apob, Irbp, and Rag1) that the closest lineage of G. japonicus is the edible dormouse Glis glis that is distributed in Europe, and also estimated on the basis of the oldest fossil record of Glis (25 MYA) that their divergence occurred 24 to 30 MYA under the background of major glirid diversifications during a relatively warmer climate period from the Early Oligocene (34 MYA) to the Mid-Miocene Climatic Optimum (15–17 MYA). These time estimates suggest that the origin of the Japanese dormouse may have predated the formation of the Japanese Archipelago (about 15 MYA; Neall and Trewick 2008). Additionally, it could be argued that the past 15 million years have seen declines of East Asian dormice because of the climate change into colder and drier environments, which the dormice do not favor (Zachos et al. 2001).

On the other hand, Yasuda et al. (2007, 2012) examined intraspecific variations of G. japonicus with mitochondrial Cytb and Y-chromosomal Sry genes and clarified that G. japonicus possesses nine geographic lineages that diversified 3 to 5 MYA. Yasuda et al. (2007) concluded that the divergences into these local populations occurred within the Japanese Archipelago. Furthermore, they showed that the extent of the genetic divergence among the geographically separated populations corresponds to that of interspecific variations in other mammalian groups. Thus, they also suggested that each local population can be regarded as a cryptic species.

1.4.2 Sciuridae

The family Sciuridae is one of the species-rich mammalian groups, including 278 squirrel species distributed across the worldwide continental regions except for Australia and Antarctica (Thorington and Hoffmann 2005). There are three types of ecological lifestyles among squirrels [arboreal (tree squirrels), terrestrial (ground squirrels), and gliding (flying squirrels) locomotors]. Recent molecular phylogenetic analyses of the multiple nuclear and mitochondrial genetic loci have determined five major lineages corresponding to subfamilial level and suggested that the arboreal tree squirrel is the ancestral type that would have had a key role in the major squirrel diversifications and subsequent derivation of the terrestrial and gliding squirrel lineages (Mercer and Roth 2003; Steppan et al. 2004b; Herron et al. 2004). Six squirrels exist in the Japanese Archipelago (Ohdachi et al. 2015), which are assigned to two of these major clades, a clade of arboreal and gliding squirrels (Sciurinae in Steppan et al. 2004b; Sciurus, Pteromys, and Petaurista) and a clade of mostly ground squirrels (Xerinae in Steppan et al. 2004b; Tamias). All three locomotor styles can be observed in the squirrels in Japan: the Eurasian red squirrel Sciurus vulgaris (arboreal), the Japanese squirrel Sciurus lis (arboreal), the Siberian flying squirrel Pteromys volans (gliding), the Japanese flying squirrel Pteromys momonga (gliding), the Japanese giant flying squirrel Petaurista leucogenys (gliding), and the Siberian chipmunk Tamias sibiricus (terrestrial, but, in fact, intermediate between terrestrial and arboreal) (Steppan et al. 2004b).

The Japanese sciurid species show clear distributional trends. Namely, congeneric species do not share the same distribution although species in the different genera do coexist in the same region (S. vulgaris, P. volans, and T. sibiricus in Hokkaido and S. lis, P. momonga, and P. leucogenys in the HSK region). Hence, it follows that there are only two biogeographic patterns for the squirrels on the Japanese Archipelago, BP2 (Table 3.1; S. vulgaris, P. volans, and T. sibiricus) and BP6 (Table 3.1; S. lis, P. momonga, and P. leucogenys). Sciurus lis and P. momonga in the HSK region has been demonstrated to be each sister lineage to S. vulgaris and P. volans, respectively (Herron et al. 2004). Because S. vulgaris and P.volans are also distributed in the Eurasian continent, the Hokkaido lineage of each species would have been generated later than lineages of S. lis and P. momonga, respectively. It is therefore likely that S. lis and P. momonga were earlier colonizers in the Japanese islands than S. vulgaris and P. volans, respectively; this is the same phylogenetic structure observed in the Sorex species in the family Soricidae. Although some phylogenetic and phylogeographic studies have been known for Sciurus species (Oshida et al. 1996, 2009a; Oshida and Masuda 2000; Grill et al. 2009; Liu et al. 2014), few of these examined time scales for the lineage differentiations or demographic events. Oshida and Masuda (2000) estimated the divergence time between S. vulgaris and S. lis to be 4.0 to 5.2 MYA based on the 0.5 % transversional substitutions/myr of the Cytb gene (Irwin et al. 1991). This value would be an overestimate, probably because of the use of the underestimated substitution rate (from the application of the slowly evolving ungulate rate into the rapidly evolving rodent time scale). Later, Oshida et al. (2005) adopted a 5–10 %/myr divergence rate for the Cytb gene considering that the divergence rate was suggested to be variable among rodents within the range of 5–10 %/myr [see Brunhoff et al. (2003) for the rationale]. If I adopt the 5–10 %/myr of divergence rate for the case of Sciurus species in Oshida et al. (2000a) where the total difference in Cytb between S. vulgaris and S. lis was inferred to be 4.9–6.8 %, the divergence time between these species is 0.49–1.36 MYA (4.9–6.8 % divided by 5–10 %), which is more consistent with the palaeontological evidence suggesting that the lineage of S. lis migrated into Honshu in the Middle Pleistocene period (Kawamura 1988). Although specifying whether the ancestral S. lis lineage migrated into the Japanese archipelago via northern or southern routes is not easy, the origin from the Korean Peninsula, the southern route, may be plausible because their absence in Hokkaido can more easily be explained. However, even if it is the case, it is difficult to understand why the ancestral lineage did not reach Hokkaido despite the multiple chances of migration via the land bridges formed in the Tsugaru Strait since the Early to Middle Pleistocene period. Probably, the population expansion might have occurred in the Late Pleistocene as in Chimarrogale platycephalus. More extensive phylogeographic and demographic analyses would be needed to grasp the time scale of the population expansion of S. lis in the HSK region.

The origin of the Hokkaido lineage of S. vulgaris is not clear because there have been no molecular chronological studies comparing the continental and Hokkaido lineages. If I calculate their divergence time on the basis of the Cytb sequence data deposited in the DNA database by Oshida and Masuda (2000; Hokkaido, Korean, and Transbaikalian individuals), it was inferred that the Hokkaido lineage diverged from the monophyletic Korean and Transbaikalian lineage 0.18 to 0.36 MYA in the Middle Pleistocene, where the average between genetic distances of the Hokkaido/Korea (1.3 %) and the Hokkaido/Transbaikalia (2.3 %) was divided by the 5–10 %/myr divergence rate (Oshida et al. 2005). Liu et al. (2014) indicated that there are no geographic structures among populations of S. vulgaris from Europe to East Asia and suggested the recent rapid population expansion, although no time frame was provided. My tentative hypothesis is the Middle Pleistocene origin for the Hokkaido lineage of S. vulgaris. However, more rigorous phylogeographic studies are needed for future precise time estimations.

For the Eurasian flying squirrel P. volans, the Hokkaido population was clarified as a monophyletic lineage (Oshida et al. 2005; Lee et al. 2008), and the divergence from the Eurasian lineages was estimated to have taken place 0.2 to 0.4 MYA based on the 5–10 %/myr divergence rate for the Cytb gene (Oshida et al. 2005). For the Japanese flying squirrel P. momonga, no studies have been conducted on the intraspecific genetic diversity. Comparison of the Cytb gene sequences from the DNA database for both P. volans and P. momonga (Oshida et al. 2005) enabled me to estimate that the average sequence difference between them was 12.3 %, implying that they diverged 1.2 to 2.5 MYA on the basis of the 5–10 %/myr divergence rate as above (Oshida et al. 2005), which is earlier than the possible origin of the Hokkaido lineage of P. volans (0.2–0.4 MYA). As in the case of S. lis, the migration route of S. momonga is not clear, although the southern route may be more conceivable.

Petaurista constitutes one of the genera encompassing gliding squirrels and is the most closely related to Pteromys (Oshida et al. 2000b). In contrast to the genera Sciurus and Pteromys, species within the genus Petaurista are mainly distributed in the southern parts of the Eurasian continent rather than the northern Eurasian regions. Among them, P. leucogenys, endemic to Japan, is the most divergent lineage within the genus (Oshida et al. 2000b; Li et al. 2013a, b). Based on the evolutionary rate of Irwin et al. (1991), Oshida et al. (2000a) estimated that the generation of the P. leucogenys lineage took place around the Late Miocene period (about 8–10 MYA). Li et al. (2013) also examined the Cytb gene and obtained a much older time estimate (12.5 MYA) for the origin of the P. leucogenys lineage on the basis of the oldest sciurid fossil record (Douglassciurus jeffersoni; 33.9–37.2 MYA) and the earliest fossil of Petaurista petaurista (0.6–1.3 MYA) as the calibration points. However, these time values might be results of overestimations: first, because of using the underestimated evolutionary rate of Irwin et al. (1991) as already explained, and second, the time scale of more than 30 MYA for setting the calibration point would be too old for the mitochondrial Cytb gene to estimate reliable divergence times because of the severe saturation effect as has been noted. If I apply the 5–10 %/myr divergence rate of Cytb to the data of Oshida et al. (2000a), the divergence between P. leucogenys and the other congeneric species was estimated to have occurred 1.39 to 3.06 MYA, which is more similar to or slightly more ancient than the date of the origin of similarly the HSK-endemic sciurid species, S. lis (0.49–1.36 MYA) and P. momonga (1.2–2.5 MYA), and is also more concordant with the palaeontological evidence (the Middle Pleistocene origin) (Kawamura 1988; Kawamura et al. 1989). The phylogeographic analyses of the mitochondrial D-loop region by Oshida et al. (2001) elucidated that there are three divergent lineages in P. leucogenys in the Japanese Archipelago and suggested that the Kyushu is the ancestral region for the diversification of this species because the Kyushu population has the most divergent or ancestral haplotypes and two of three major lineages. Combined with the presence of the closely related Pataurista species in the southern part of Eurasia, these observations support the southern route immigration hypothesis, which is similar to the situation of Chimarrogale platycephalus. The divergence among the three major lineages was estimated to have occurred 0.4 to 1.0 MYA based on the evolutionary rate of the human D-loop sequences (8.6 %/million years; Vigilant et al. 1989). Later, Oshida et al. (2009b) examined the Cytb gene and found that there are five distinct phylogroups (Kyushu, Southwest, Southeast, Central, and North), including the three major lineages of Oshida et al. (2001). Using 5–10 %/myr for the divergence rate, they estimated that diversifications among five major lineages occurred 0.09 to 0.24 MYA. These estimates are in agreement with the divergence time between P. leucogenys and the closest species (1.39–3.06 MYA). Additionally, Oshida et al. (2009b) detected the trend of the sudden population expansion for only the northern phylogroup that took place 18,200 to 36,500 years ago, suggesting a possibility that the absence of this species in Hokkaido might have been caused by the formation of the Tsugaru Strait in the Late Pleistocene (Ohshima 1991).

Tamias sibiricus is the only Palaearctic species within the genus Tamias, in which the other species are all found in the American continent. There have been few phylogeographic studies including the Hokkaido lineage of T. sibiricus. Only Obolenskaya et al. (2009) examined one individual from Hokkaido together with many samples from Russia, China, and Korea. They showed that there are at least three divergent lineages in this species, northern Eurasian, Korean, and Central Chinese lineages (but the validity of the last one is still open to question because this is based on pet shop samples with unknown origin). The Japanese chipmunk is included in the northern Eurasian lineage. Although their approach for the divergence time estimation is arbitrary because they forced the divergence among these major lineages to have occurred in the onset of the Middle Pleistocene period without clear explanations for the calibration procedure, the differentiations among lineages within the northern Eurasian clade, including the lineage of Hokkaido, was calculated to have transpired 0.11 to 0.47 MYA in the Middle to Late Pleistocene period. Here again, if I apply the 5–10 %/myr rule to the data of Obolenskaya et al. (2009), the divergence time among the northern Eurasian lineages was estimated to be 0.10 to 0.19 MYA (average genetic difference among 90 individuals within the northern Eurasian clade, 0.95 %, divided by 5–10 %), which is congruent with the estimate of Obolenskaya et al. (2009).

1.4.3 Muridae

The family Muridae is the most speciose family in the order Rodentia and even in the class Mammalia, comprising 150 genera and 730 species (Musser and Carleton 2005). There are six extant species of this family in the Japanese Archipelago, except for the species commensal to our life such as rats (Rattus spp.) and house mice (Mus musculus) (Ohdachi et al. 2015). These six species are classified into three genera, Apodemus, Micromys, and Mus, all belonging to the subfamily Murinae. The Murinae is the most diversified subfamily within the Muridae, and the interrelationships among the major lineages (genera) within this subfamily have remained to be elucidated because of the rapid diversifications during a relatively short evolutionary time span (Sato and Suzuki 2004; Steppan et al. 2005).

Species in the genus Apodemus are mainly adapted to the deciduous broad-leaved forest in northern Eurasia across Europe to East Asia, including the Japanese Archipelago (Musser and Carleton 2005). Molecular phylogenetic studies have clarified that there are four major lineages among 20 species in Apodemus (Serizawa et al. 2000; Suzuki et al. 2003, 2008): the argenteus lineage endemic to Japan, the gurkha lineage endemic to Nepal, the lineage mainly occupying the East Asia, and the lineage mainly observed in Europe. Four species are found in Japan, the striped field mouse A. agrarius, the small Japanese field mouse A. argenteus, the East Asian field mouse A. peninsulae, and the large Japanese field mouse A. speciosus, showing three biogeographic patterns (BP2, BP5, and BP10, respectively; Table 3.1). Based on the nuclear gene variations (I7, Irbp, Rag1, and vwf), Suzuki et al. (2008) estimated that lineages of A. argenteus and A. speciosus occurred 7.3 and 5.9 MYA, respectively, based on the Mus–Rattus divergence at 12 MYA. Because more closely related species are missing in continental Asia, it is supposed that the formation of the Japanese Archipelago might have affected the generation of such highly divergent endemic species. Because the time estimates for the generation of A. argenteus (7.3 MYA) and A. speciosus (5.9 MYA) correspond to the Late Miocene age that saw the vegetation changes promoted by the environmental shift to a colder and drier climate (Cerling et al. 1997), some researchers supposed that these climate changes might have facilitated the diversification of the Apodemus lineages (Serizawa et al. 2000; Michaux et al. 2002; Suzuki et al. 2003, 2008). Regarding the intraspecific phylogeographic studies, Suzuki et al. (2004) examined the mitochondrial Cytb gene for A. speciosus and A. argenteus from all the major islands and neighboring small islands, and estimated, with the evolutionary rate of 2.4 %/lineage/myr (Suzuki et al. 2003), that the most basal divergence in A. speciosus and A. argenteus occurred 0.51 and 0.64 MYA, respectively. Notably, they detected two major clades in A. speciosus separating central (Honshu, Shikoku, and Kyushu Islands) and peripheral (Hokkaido, Sado, Izu, and Satsunan Islands) lineages. Tomozawa and Suzuki (2008) basically obtained supportive evidence from the nuclear gene (Irbp) for the central/peripheral trend, although the Hokkaido and Sado lineages were included in the central lineage, the Satsunan lineages were observed in both lineages, and the Oki and Tsushima lineages were added to the peripheral lineage. Tomozawa and Suzuki (2008) discussed that allopatric fragmentation might have caused the generation of the central and peripheral lineages, and the recent population expansion at 0.14 MYA (based on 2.4 %/lineage/myr as earlier) formed the current distribution pattern of the central lineage, where the male-biased dispersal might replace the nuclear genome of Hokkaido, Sado, and a part of Satsunan populations with the peripheral mtDNA types that remained because of female philopatry. It should be noted here that although all the dating estimates are fundamentally based on the frequently used divergence time between Mus and Rattus (12 MYA), the recent palaeontological (11.0–12.3 MYA; Benton and Donoghue 2007) and molecular phylogenetic (8.6–10.3 MYA; Steppan et al. 2004a) studies have gradually supported younger dates than 12 MYA. Therefore, the real divergence and population expansion times could also be more recent than obtained here based on the 12 MYA for the Mus–Rattus divergence.

Serizawa et al. (2002) conducted phylogeographic research with the mitochondrial Cytb gene for A. peninsulae and clarified the monophyly of the Hokkaido population to the continental one. The estimate for the divergence time of the Hokkaido lineage from the continental equivalent in the most recent study was 0.1 MYA in the Late Pleistocene, based on fossil calibrations that the divergence between A. mystacinus and the other Sylvaemus species and that between A. sylvaticus and A. flavicollis occurred 7 and 4 MYA, respectively (Sakka et al. 2010; also based on Cytb). This dating is consistent with the absence of this species in the other southern Japanese islands because of the presence of the Tsugaru Strait in the Late Pleistocene as in the case of Sorex minutissimus and Sorex gracillimus (Ohshima 1991). Both these studies indicated that the population from Siberia to the Russian Far East harbors a higher extent of genetic diversity, therefore implying the existence of refugia in the glacial periods in the Pleistocene. Probably the Hokkaido lineage is a result of the population expansion from the refugial region in Far East Asia (Serizawa et al. 2002; Sakka et al. 2010).

The harvest mouse Micromys minutus has a wide distribution from Europe to the easternmost islands of Japan (BP7; Table 3.1) and mainly favors open environments (e.g., grasslands; Ohdachi et al. 2015). Interestingly, the intraspecific genetic variation was inferred to be very small according to the phylogeographic research based on the mitochondrial Cytb gene and control region sequences (Yasuda et al. 2005), which is suggestive of the recent rapid population expansion across wide areas in Eurasia. Yasuda et al. (2005) estimated on the basis of 2.4 %/lineage/myr (Suzuki et al. 2003) that the divergence between the Japan/Korea and European lineages occurred 0.08 MYA in the Late Pleistocene and also suggested that the population expansion within the Japanese Archipelago occurred 0.03 MYA. Such a recent establishment of this species in Japan is in agreement with the lack of any fossil records of this species in Japan (Ohdachi et al. 2015). On the other hand, such estimates for the Late Pleistocene origin of M. minutus in Japan would not agree with the argument of Ohshima (1991) that the Korea (Tsushima) Strait was not available by land bridges in this age. I discussed the possible process of the origin of M. minutus in the text (oversea dispersal hypothesis). Here again, also note that dating with the 2.4 %/lineage/myr rate of Suzuki et al. (2003) based on the 12 MYA for Mus–Rattus divergence is likely to lead to overestimations of divergence and expansion times.

It has been repeatedly debated, and still remained to be resolved, whether the establishment process of the Ryukyu mouse Mus caroli in the Ryukyu Islands (BP10; Table 3.1) was natural or human induced (Motokawa 2000; Motokawa et al. 2003). Terashima et al. (2003) showed that M. caroli possesses a unique mitochondrial Cytb gene lineage that is not found in the other populations in Southeast Asia and concluded that it was established by a natural dispersal instead of an anthropogenic effect. On the other hand, Shimada et al. (2007) used the same genetic loci for individuals from much wider sampling localities and presented the opposite conclusion for human commensalism in the prehistoric age in Japan, because the M. caroli in Ryukyu Islands was shown to be closely related to the Laos lineage within a strongly supported clade of the southern Southeast Asian clade and did not show close affinity with the geographically close Taiwan lineage. Because of this ambiguous status, I did not consider this species in discussions.

The remaining Murinae species are included in the Ryukyu-endemic genera Tokudaia and Diplothrix. Three Tokudaia species, the Okinawa spiny rat T. muenninki, the Amami spiny rat T. osimensis, and the Tokunoshima spiny rat T. tokunoshimensis, are each distributed in Okinawajima, Amami-Oshima, and Tokunoshima Islands, respectively, in the Ryukyu region [BP9; Table 3.1; also see Endo and Tsuchiya (2006) and Ohdachi et al. (2015)]. Molecular phylogenetic studies have demonstrated that Tokudaia formed a clade with Apodemus (Michaux et al. 2002; Sato and Suzuki 2004), and the divergence time between these species was estimated to be 6.5–8.1 MYA by nuclear gene analyses (Irbp and Rag1; Sato and Suzuki 2004). To date, few studies have examined the chronology for interspecific divergence within Tokudaia. Suzuki et al. (1999) suggested with rDNA-RFLP data that the lineages in Amami–Oshima and Tokunoshima diverged 1.2 to 2.3 MYA based on the previously estimated evolutionary rate of 1–2 %/myr, whereas the time was 4.4 MYA based on a partial Cytb gene data although it depended on the evolutionary rate of Brown et al. (1979). Murata et al. (2010) demonstrated that T. muenninki was first branched off in the phylogeny among three Tokudaia species, meaning that two Y-chromosome-lacking Tokudaia species, T. osimensis and T. tokunoshimensis, formed a clade. According to the Cytb gene sequences provided by Murata et al. (2010), average pairwise difference between T. muenninki and the other two species was 12.4 % and that beween T. osimensis and T. tokunoshimensis was 5.4 %, which could be calculated to be 2.58 and 1.13 MYA, respectively, on the basis of the 2.4 %/lineage/myr substitution rate (Suzuki et al. 2003).

The Ryukyu long-furred rat Diplothrix legata is a monotypic species inhabiting the Amami–Oshima, Tokunoshima, and Okinawajima Islands in the Ryukyu region (BP9; Table 3.1; also see Ohdachi et al. 2015). Suzuki et al. (2000) examined the nuclear Irbp and the mitochondrial Cytb genes and supported that Diplothrix legata is closely related to a clade consisting of Rattus rattus and Rattus argentiventer to the exclusion of Rattus norvegicus, strongly implying the paraphyly of the genus Rattus to Diplothrix. They also showed that the extent of the divergence between Diplothrix and Rattus species corresponds to 20–30 % of that between Mus and Rattus. Although Suzuki et al. (2000) estimated the divergence time to be 3–4 MYA, it was based on 14 MYA for the Mus–Rattus split, which is not consistent with the current knowledge as already noted (8.6–10.3 MYA, Steppan et al. 2004a; 11.0–12.3 MYA, Benton and Donoghue 2007). If I adopt the average 10.55 MYA {([8.6 + 10.3]/2 + [11 + 12.3]/2)/2}, 20–30 % means 2.11–3.17 MYA for the divergence of the lineage for D. legata. This time estimate is consistent with the geological evidence in Kimura (2000) that the Ryukyu Islands were connected multiple times to the Eurasian continent in two stages, the former of which was 1.3 to 1.6 MYA.

1.4.4 Cricetidae

The family Cricetidae is the second most speciose family in the order Rodentia, comprising 130 genera and 681 species (Musser and Carleton 2005). Except for the introduced species (the muskrat Ondatra zibethicus), there are six cricetid species in the Japanese archipelago, all of which are classified in the subfamily Arvicolinae (Kaneko 2006; Ohdachi et al. 2015). Although confusion has surrounded the taxonomy within this subfamily, in this chapter I followed Musser and Carleton (2005) and Carleton et al. (2014) for the genus name, but basically adopted the common names and species numbers in Ohdachi et al. (2015). Thus, the Japanese cricetid species discussed in this chapter are the northern red-backed vole Myodes rutilus, the grey red-backed vole Myodes rufocanus, the dark red-backed vole Myodes rex, the Anderson’s red-backed vole Myodes andersoni, the Smith’s red-backed vole Myodes smithii, and the Japanese field vole Microtus montebelli.

The five Myodes species are adapted to the forested environments in the Hokkaido and HSK regions. M. rutilus, M. rufocanus, and M. rex similarly have their native ranges in Hokkaido, but show different distribution patterns outside Hokkaido, expanding to Holarctic (BP1), Palaearctic (BP2), and Sakhalin (BP3) regions, respectively (Table 3.1). The origins of these three species in Hokkaido are considered independent of each other. Kohli et al. (2015) conducted extensive phylogeographic analyses for M. rutilus across Eurasian and American continents by using the mitochondrial Cytb and three nuclear genes (Ets2, Irbp, and Mlr). They found that there are distinct phylogroups that would have been derived from at least three refugia located in western Eurasia, central Eurasia, and Beringia. The divergences among these phylogroups were estimated to have occurred during the last 0.1 million years in the Late Pleistocene. It was also suggested that the monophyletic Hokkaido lineage was possibly originated by the expansion from a northern refugium, Beringia. If the establishment of this species in Hokkaido originally occurred in the Late Pleistocene, it can explain the absence of this species in more southern Japanese islands because of the same reason for Sorex minutissimus, S. gracillimus, and Apodemus peninsulae (the hypothesis of the Tsugaru Strait barrier in the Late Pleistocene; Ohshima 1991). On the other hand, for M. rufocanus, Abramson et al. (2012) estimated with the Cytb gene and the palaeontological information (the differentiation among European haplotypes occurred 8,000 years ago; the divergence between M. rutilus and M. glareorus at 2.5 MYA) that the monophyletic Hokkaido lineage diverged from the clade mainly composed of the Far East Russian and Sakhalin lineages 0.27 MYA and experienced the population expansion 0.04–0.05 MYA in Hokkaido. Although Abramson et al. (2012) did not provide intra-lineage dating information, the most recent common ancestor of the Hokkaido individuals could be estimated to be present in the Late Pleistocene (<0.13 MYA), inferred from the branch length of the phylogeny that they provided. Therefore, the absence of M. rufocanus in Honshu and the more southern islands may be explained by the Tsugaru Strait barrier in the Late Pleistocene as in the case of M. rutilus. Abramson et al. (2012) further suggested that there should have been multiple refugia in the Russian Far East because the population possesses higher genetic diversity there, implying that the Hokkaido lineage originated from the expansion from the refugia in the Russian Far East, which is in sharp contrast to the original source for the population expansion of M. rutilus (Beringia), but consistent with the case of Apodemus peninsulae in the light of space (not time). M. rex can be regarded as the earliest colonizer among three Myodes species in Hokkaido because the divergence from the most closely related lineage (M. rufocanus) was estimated to have occurred in the Middle Pleistocene (0.81 MYA; Abramson et al. 2012). Based on the results of the mitochondrial DNA control region, Kawai et al. (2013) detected four phylogroups from the several fragmented populations in Hokkaido and concluded that such phylogroups were shaped by the genetic divergences across several refugia formed in the glacial periods since the Middle Pleistocene. This conclusion is because in their study the time to the MRCA of the M. rex phylogroups was estimated to be 0.12–0.58 MYA in the Middle Pleistocene based on the “mutation” rate of 3.6 %/myr (based on the divergence time of 7–8 MYA between Microtus and Clethrionomys (in this case M. rutilus, M. rufocanus, and M. rex are all included in this genus) (Matson and Baker 2001) and 17 %/myr (based on the divergence time of 1.8 MYA for the Siberian and the Nearctic brown lemmings; Fedorov and Stenseth 2001) estimated for the arvicoline control region diversity. However, despite that these rates were originally proposed as “divergence” rates considering two descendant lineages, Kawai et al. (2013) adopted them as per-lineage mutation rates. Therefore, the foregoing estimates (0.12–0.58 MYA) should be corrected to 0.24–1.16 MYA. The corrected estimates also largely include the Middle Pleistocene period and are not inconsistent with the divergence of the M. rex lineage as estimated above (0.81 MYA). It is not clear why M. rex did not reach more southern Japanese islands despite the presence of the land bridge in the Tsugaru Strait in the Middle Pleistocene, but a community ecological mechanism may explain the absence in Honshu and more southern islands.

Two Myodes species in the HSK region, M. andersoni and M. smithii, are endemic species in Japan (BP6; Table 3.1). The former was mainly found in the eastern Honshu and the latter in the western Honshu, Shikoku, and Kyushu, while their distributions are partly overlapped in the central Honshu. Their ancestral common lineage was inferred to have diverged 0.9 to 2.3 MYA from the clade composed of M. rufocanus and M. rex (Luo et al. 2004; Lebedev et al. 2007; Kohli et al. 2014). Although the time estimates are variable among studies, it is probable that the Myodes species in the HSK region might have originated in the Early to Middle Pleistocene, much earlier than that of the Myodes species in Hokkaido. However, the patterns of lineage differentiations within and between M. andersoni and M. smithii are quite complicated. M. simthii is not monophyletic in the phylogeny inferred by the Cytb gene. Namely, the lineage in Shikoku was shown to be the most divergent among all the lineages detected in M. andersoni and M. smithii, and the M. smithii lineage composed of haplotypes detected in Honshu and Kyushu was closely related to M. andersoni (Iwasa and Suzuki 2002). Time scales for the differentiations have not been fully assessed to date in a reliable manner. Assuming that the time to the MRCA of the red-backed voles was 1.8 MYA, Luo et al. (2004) suggested that the Shikoku lineage of M. smithii diverged from the rest 0.77 MYA, and the Honshu–Kyushu lineage of M. smithii diverged from M. andersoni 0.16 MYA. Kohli et al. (2014) estimated with the fossil calibration (2.6 MYA for the time to the MRCA of Myodes species based on the oldest fossil found in Russia) that M. smithii in Honshu diverged from M. andersoni 0.50 MYA. These time estimates might be too recent to be regarded as those for valid species differentiation. Iwasa and Suzuki (2003) detected “interspecific” hybridization between both species. Kaneko (2006) mentioned that it is difficult to identify the Myodes species in Honshu by morphological criteria only. Probably future taxonomic revision would be needed on the basis of more rigorous morphological and molecular phylogenetic studies for the Myodes species in the HSK region. Alternatively, it is also probable that the ancestral polymorphisms of both species have been retained since the divergence between M. andersoni and M. smithii. Female philopatry might have influenced the maintenance of such divergent lineage in the mitochondrial gene. Iwasa and Suzuki (2002) suggested that both Honshu and Shikoku populations have the same type of Y-chromosomal Sry gene, which is male specific and irrelevant for the female philopatry. Also in this case, all the populations may be assigned to one Myodes species.

Microtus is the most speciose genus within the subfamily Arvicolinae (62 species; Musser and Carleton 2005) and includes mainly herbivorous small rodents adapted to grasslands, taiga, steppe, and tundra. There is only one Microtus species in Japan, M. montebelli, which is endemic to and distributed in Honshu and Kyushu regions (BP6; Table 3.1). It is phylogenetically closely related to M. oeconomus, widely distributed in Holarctic regions in the Eurasian and American continents, and M. kikuchii, endemic to the Taiwan Islands (Conroy and Cook 2000; Jaarola et al. 2004; Bannikova et al. 2010). There have been few studies that examined chronological aspects of this species based on molecular data. Bannikova et al. (2010) only estimated with the mitochondrial Cytb gene and the time constraint of 2.2 MYA for the radiation of the basal Microtus lineages (based on fossils) that M. montebelli diverged from the aforementioned closely related species 0.95 MYA. The time estimate is similar to that of the Myodes species in HSK. In the fossil evidence, this species was found since the Middle Pleistocene (Kawamura et al. 1989). The estimated time scale suggests that there should have been some possibilities that M. montebelli could have expanded to all the Japanese islands, Hokkaido, Honshu, Shikoku, and Kyushu. The absence of this species from Hokkaido and Shikoku requires some explanation in light of phylogeography or ecology. Kaneko (2006) discussed that the interspecific competition with Apodemus species might be one reason for the absence of M. montebelli in Shikoku. Future studies in the phylogeography and comparisons of niche requirements between M. montebelli and species in the same guild would shed more light on the reasons for the peculiar distribution pattern of this species.

1.5 Lagomorpha

1.5.1 Leporidae

Leporidae is the most species-rich lagomorph family, encompassing 61 species belonging to 11 genera (Hoffmann and Smith 2005). In Japan, 3 species are present and their distributions are geographically partitioned. The mountain hare Lepus timidus is distributed in Hokkaido and expanded to the Palaearctic region in the Eurasian continent (BP2; Table 3.1), the Japanese hare Lepus brachyurus is endemic to Japan and only observed in the HSK region (BP6; Table 3.1), and the Amami rabbit Pentalagus furnessi exists only in Amami–Oshima and Tokunoshima Islands in the Ryukyu region (BP9; Table 3.1). Considering the difference in the extent of endemism in most Japanese mammals among these three major biogeographic regions, it could be predicted that P. furnessi, L. brachyurus, and L. timidus migrated into the Japanese Archipelago in this order. Matthee et al. (2004) conducted molecular phylogenetic analyses with two mitochondrial (Cytb and 12S rRNA) and five nuclear (Sptbn1, Prkci, Thy, Tg, and Mgf) genes and showed that the genus Pentalagus is closely related to the genera Caprolagus in India, Oryctolagus in Europe, and Bunolagus in South Africa, and the divergence of Pentalagus from the other genera was 9.44 MYA in the total data analysis and 8.63 MYA in the nuclear gene data analysis on the basis of the palaeontological records (Leporidae–Ochotonidae split, 20–40 MYA; the origin of the modern leporid, 12–20 MYA; the divergence of the genus Lepus, 4–6 MYA; the oldest divergence time of the ingroup, 60 MYA). Probably the former value (9.44 MYA) would be overestimated because the mitochondrial genes were included in the examined data matrix. The latter estimate (8.63 MYA) is not contradicted by the Late Miocene origin of the murine rodent genus Tokudaia, also endemic to the Ryukyu region (6.5–8.0 MYA; Sato and Suzuki 2004; also see above). Yamada et al. (2002) indicated by using the mitochondrial 12S rRNA and Cytb genes that the mountain hare (Lepus timidus) in Hokkaido has a distinct monophyletic lineage and is closely related to L. arcticus, L. townsendii, and L. othus in Greenland and North America, forming a well-supported clade [L. timidus group; L. arcticus, L. townsendii, and L. othus are suspected to be the same species as L. timidus (Wu et al. 2005)]. In Yamada et al. (2002), L. brachyurus was sister to the L. timidus group (albeit with a low support value), whereas the position of Pentalagus furnessi was not resolved on the earlier radiation among different leporid genera. Assuming the molecular clock of the Cytb gene evolution and setting 30 to 40 MYA for the divergence between Ochotonidae and Leporidae, the times for the divergence of P. furnessi, L. brachyurus, and the Hokkaido lineage of L. timidus from each sister lineage were estimated to be 12–16 MYA, 4–5 MYA, and 0.5–0.6 MYA, respectively (Yamada et al. 2002). In addition, Wu et al. (2005) examined four mitochondrial genetic loci (Cytb, Nd2, 12S rRNA, and Dloop) for the extensive Lepus species, and indicated that the L. brachyurus first branched off in the clade composed of Eurasian Lepus species. Wu et al. (2005) also estimated with the Cytb gene that the divergence time for the branching of the L. brachyurus lineage was 3.62 MYA, which is not in disagreement with the estimate within the Pliocene in Yamada et al. (2002). However, the estimate of Yamada et al. (2002) for the origin of Pentalagus (12–16 MYA) is much earlier than that of Matthee et al. (2004) (8.63 MYA), which mainly used nuclear genes. The difference would be attributed to the distinct properties between mitochondrial and nuclear genes for dating divergences, as repeatedly discussed earlier. Therefore, the divergence times of Yamada et al. (2002) and Wu et al. (2005) might be somewhat overestimated because of the exclusive use of the mitochondrial genes. If I conduct a simple calculation that 8.63 MYA is 62 % of 14 MYA (average between 12 and 16 MYAs) and apply this proportion to the other estimates of Yamada et al. (2002), 4–5 MYA for L. brachyurus and 0.5–0.6 MYA for L. timidus are calculated to be 2.48–3.10 MYA and 0.31–0.37 MYA, respectively.

On the basis of the phylogeographic analyses of the intraspecific variation in the Cytb gene for L. brachyurus, Nunome et al. (2010) detected major two lineages (northern and southern lineages) and suggested several regions for refugia during the Pleistocene glacial periods (Kanto, Chubu, Shikoku, and Kyushu), where the genetic diversity was estimated to be higher. These two major clades were inferred to have diverged 1.2 MYA following the lineage-specific evolutionary rate of 1.4 %/lineage/myr (therefore, 2.8 % divergence rate) on the basis of the assumption that L. timidus and L. brachyurus diverged 3.6 MYA (Wu et al. 2005). Their recent study on the nuclear male-specific Sry gene also indicated a similar but a little younger divergence time between northern and southern lineages (1.07 MYA; Nunome et al. 2014). Nunome et al. (2010) also showed that the initial divergence in each northern and southern clade occurred 0.33 MYA and 0.38 MYA, respectively. The ages raised here are in agreement with the palaeontological record that the fossil remains were found in Japan in the Middle Pleistocene (Kawamura et al. 1989). Additionally, the multiple locations for refugia suggested by Nunome et al. (2010) are consistent with those of the water shrew (Iwasa and Abe 2006), macaque (Kawamoto et al. 2007), and Japanese giant flying squirrel (Oshida et al. 2009b).

In the phylogeographic analyses with the Cytb gene, Kinoshita et al. (2012) elucidated that the Hokkaido lineage of L. timidus diverged from the most closely related Korean lineage, L. coreanus, 0.30 MYA and that the ancestral lineage including both L. timidus in Hokkaido and L. coreanus was separated from the other Eurasian L. timidus lineages 0.46 MYA. The adopted assumptions for calibration points were the divergence between Oryctolagus and Lepus, 11.8 MYA; the time to the MRCA of Lepus, 5.16 MYA; the divergence between L. timidus and L. townsendii, 1.13 MYA. The estimates obtained by Kinoshita et al. (2012) are consistent with the already calculated divergence time of the Hokkaido lineage of L. timidus (0.31–0.37 MYA). Because the Sakhalin lineage was demonstrated to be included in the Eurasian clade, the colonization of the eastern marginal islands off the Eurasian continent by the lineage of L. timidus would have occurred more than twice, in which one could reach Hokkaido but the other stopped at Sakhalin. The time to the MRCA of Hokkaido lineages was estimated to be 0.17 MYA in this study, suggesting that the origin of the L. timidus population in Hokkaido was in the Middle Pleistocene, although an extent of overestimation might be suspected because of the use of the mitochondrial gene and little old fossil assumption. Furthermore, they clarified that there were two distinct lineages in Hokkaido that would have been created through within-island vicariance into two southern refugia in Hokkaido.

To sum the foregoing discussions, despite the presence of some uncertainties in the chronological estimates for three leporid species in Japan, all the estimations here are consistent with the prediction that P. furnessi, L. brachyurus, and L. timidus colonized the Japanese Archipelago in this order.

1.5.2 Ochotonidae

The family Ochotonidae includes only one genus, Ochotona, composed of 30 pika species (Hoffmann and Smith 2005). Among the 28 Ochotona species distributed in Asia, only 1 species, the northern pika Ochotona hyperborea, exists in Hokkaido among the Japanese Archipelago, and this species is also distributed in the Palaearctic region in Eurasia (BP2; Table 3.1). Previous molecular phylogenetic analyses have indicated that the lineage of O. hyperborea itself is closely related to O. scorodumovi inhabiting continental East Asia, and the Hokkaido lineage of O. hyperborea is the most genetically distinct among the other Eurasian conspecific lineages (Lissovsky et al. 2007; Lanier and Olson 2009). The estimation of the divergence time with the Cytb gene conducted by Niu et al. (2004) implied that most among-species divergence within the genus Ochotona occurred in the Early Pleistocene on the basis of the divergence rate of what is stated as 2–5 %/myr of Irwin et al. (1991), although such a rate was not proposed in that study. The lineage of O. hyperborea was also estimated to have occurred in this period. In contrast, Lanier and Olson (2009) similarly estimated with the mitochondrial Cytb and Nd4 gene that O. hyperborea diverged from O. scorodumovi in the Late Pliocene based on the divergence time of 37 MYA for the split between Ochotonidae and Leporidae, but did not provide any information about the time scale for the origin of O. hyperborea in Hokkaido despite examining the sample from Hokkaido in their analyses. It is suspicious that these two dating trials did not grasp the realistic time scale because of using the divergence rate of Irwin et al. (1991) and a too-old calibration point (37 MYA) for the Cytb gene. In addition, Yu et al. (2004) used the 10 % silent divergence rate/myr of Irwin et al. (1991; this rate was noted in this study) and different time constraints. However, as repeatedly described here, the evolutionary rate extracted from the ungulate taxa is difficult to apply to other taxa because of the difference in the evolutionary rate among mammalian taxa. Lissovsky et al. (2007) showed that the pairwise distance of the Cytb gene between the Hokkaido and the continental lineages is 4.53 %. If I adopt the 2.8 %/myr divergence rate used for Lepus brachyurus (Nunome et al. 2010), the lineage of O. hyperborea in Hokkaido could be interpreted to have occurred 1.62 MYA, which is unusually old for the common species inhabiting only Hokkaido in Japan.

1.6 Carnivora

1.6.1 Felidae

Felidae is the second largest family in the order Carnivora, including 40 species (Wozencraft 2005). In contrast to the species-rich trend of this group, only 1 extant species is present in Japan, the leopard cat Prionailurus bengalensis. This species is widely distributed in South, Southeast, and East Asia (Wozencraft 2005), whereas in Japan it can only be observed in two isolated islands, Tsushima and Iriomotejima Islands (BP10; Table 3.1) (also see Ohdachi et al. 2015). They are designated as two different subspecies, P. b. euptilurus and P. b. iriomotensis, respectively (Ohdachi et al. 2015). Although traditionally these two subspecies were sometimes treated as different species or genus, molecular phylogenetic studies have demonstrated that they are phylogenetically so close to the continental conspecific lineages to the extent of the intraspecific variation (Masuda et al. 1994; Masuda and Yoshida 1995; Tamada et al. 2008). The phylogenetically closest species was inferred to be the fishing cat P. viverrinus, mainly inhabiting Southeast Asia, and the divergence time was estimated to be 2.55 MYA based on the nuclear and mitochondrial gene segments (~23 kb) with many fossil calibration points (Johnson et al. 2006; also see their Supporting Online Material for the calibration points). Tamada et al. (2008) conducted a phylogeographic study of the Asian leopard cats by using the two mitochondrial genetic loci (Cytb and control region) and showed that there are three major lineages (one northern lineage and two southern lineages), where two Japanese leopard cats were included in the northern lineage. Assuming the substitution rate of the Cytb gene to be 1.38 %/lineage/myr (2.76 %/myr divergence rate) based on the divergence time of 4.5 MYA between the domestic cat and the tiger (Masuda et al. 1994), the Tsushima leopard cat was estimated to have diverged 0.03 MYA from the continental Far East individual. However, the calibration point in Masuda et al. (1994) is not consistent with recent understandings for the divergence between the domestic cat and the tiger. If I adopt the information from Johnson et al. (2006) where divergence time between them was estimated to be 10.8 MYA (more than twice as much as 4.5 MYA), the divergence time (0.03 MYA) of the Tsushima leopard cat was corrected to be 0.072 MYA by using the 1.15 %/myr divergence rate [=2.76 divided by 2.4 (=10.8/4.5)]. It is not clear how the Tsushima leopard cat migrated into Tsushima Islands and not into the Kyushu Islands in the situation that the water depth is deeper in the sea between Tsushima and the continent than that between Tsushima and Kyushu (Park et al. 2000). Some ecological or environmental reasons should have caused the absence in the main islands in Japan (see Sect. 3.4.2 in the text). Although Tamada et al. (2008) did not infer the divergence time of the Iriomote leopard cat, the average genetic distance between the Iriomote leopard cat and the closest lineage (0.3 %) can be interpreted to be 0.264 MYA based on the corrected fossil assumption. This time estimate is not inconsistent with geological evidence of Kimura (2000) that the Ryukyu Islands were connected twice to the Eurasian continent; the latter connection occurred 0.2 to 0.025 MYA. Therefore, the geological factor could have affected the establishment of the Iriomote leopard cat.

1.6.2 Canidae

Canidae currently includes 35 dog-related species among 13 genera (Wozencraft 2005). Except for the extinct grey wolf Canis lupus, there are only 2 species in Japan, the red fox Vulpes vulpes and the raccoon dog Nyctereutes procyonoides (Ohdachi et al. 2015). They are both distributed across the major Japanese islands (Hokkaido, Honshu, Shikoku, and Kyushu) and the same species are also present in the Eurasian and American continents (therefore, BP = 4; Table 3.1). Although there have relatively been few studies on the molecular phylogeny of this family as a whole, Lindblad-Toh et al. (2005) conducted molecular phylogenetic analyses with nuclear gene sequences of approximately 15 kb and found that both species were placed in one of four major clades in Canidae (the red fox-like clade), where V. vulpes formed a clade with the Ruppell’s fox, Vulpes ruppellii, distributed from West Asia to Northern Africa, whereas N. procyonoides was closely related to the bat-eared fox Otocyon megalotis living in Africa. Bardeleben et al. (2005) also obtained consistent results on the basis of molecular phylogenetic analyses with data from 6 nuclear and 2 mitochondrial genes. Perini et al. (2010) examined 3 mitochondrial and 22 nuclear genes and estimated on the basis of the fossil-based minimum bound of 40 MYA for the divergence between Canidae and Ursidae, 7.5 MYA for the divergence between Ailuropoda and Ursus (this constraint were based on the oldest fossil of Ailurarctos at 7–8 MYA, but may have to be corrected as a consequence of the recent report that the oldest remain of the giant panda lineage was found at 11–12 MYA; Abella et al. 2012), and 8 MYA for the divergence between Canini and Vulpini that the time for occurrence of the N. procyonoides lineage was 7.7 MYA. Unfortunately, to date, no time frame has been proposed for the divergence between V. vulpes and V. ruppellii. Only Perini et al. (2010) inferred that the divergence between V. vulpes and the Corsac fox V. corsac, the closest relative of the V. vulpes–V. ruppellii clade, was about 2.0 MYA.

Concerning intraspecific phylogeographic studies, Inoue et al. (2007) detected two major mitochondrial DNA (Cytb) lineages of V. vulpes in Japan, where one is observed in both Honshu/Kyushu and Hokkaido (lineage I) and the other exclusively in Hokkaido (lineage II or Hokkaido II). Precise inspection with longer mitochondrial DNA sequences (Cytb, control region, and some tRNA sequences) revealed that the lineage I included two distinct clades of the Hokkaido individuals (Hokkaido Ia and Ib) and a Honshu/Kyushu-specific clade, suggesting that the establishment of V. vulpes in the HSK region took place by only one event, whereas multiple migrations might have generated the lineages in Hokkaido (Inoue et al. 2007). A recent phylogeographic study with more extensive samples from western Europe through East Asia including Japan to North America clarified that all the four major lineages observed in Japan (Hokkaido Ia, Ib, II, and Honshu/Kyushu) each have closest relatives in the Eurasian continent, confirming the multiple independent migrations into the Japanese Archipelago (Kutschera et al. 2013). Based on the variations in the mitochondrial control region with external and tip calibrations from fossil and ancient DNA data, respectively, Kutschera et al. (2013) estimated that the divergences of Honshu/Kyushu and Hokkaido II lineages from each continental relative occurred 0.021–0.064 MYA and 0.027–0.045 MYA, respectively, suggesting Late Pleistocene origins. In the same study, Hokkaido Ia and Ib lineages were inferred to have occurred during a more recent part of the Late Pleistocene, probably after the Last Glacial Maximum. However, these time estimates are not concordant with the fossil evidence that the fossil remains of V. vulpes were found in the HSK region in the Middle Pleistocene (Kawamura et al. 1989). Because Tsugaru and Korea (Tsushima) Straits were basically not available for dispersals in the Late Pleistocene period, as repeatedly noted (Ohshima 1991), the establishment of the Honshu–Kyushu lineage would have occurred in the Middle Pleistocene. This idea is in sharp contrast to the conclusion of Kutschera et al. (2013) that the Honshu–Kyushu lineage is a consequence of human introduction in the Late Pleistocene. One reason for these younger estimates of Kutschera et al. (2013) may be reflected in too much higher substitution rates obtained by their analyses (33.2–41.9 %/lineage/myr). If I adopt the 1.90–2.68 %/myr divergence rate of the control region obtained from the study of the Japanese weasel Mustela itatsi (Masuda et al. 2012; see the following Mustelidae section) and conduct a simple proportion calculation from the 33.2–41.9 %/lineage/myr substitution rate, the divergences times of Honshu–Kyushu and Hokkaido II (estimated earlier to be 0.021–0.064 MYA and 0.027–0.045 MYA) were corrected to 0.52–2.88 and 0.66–1.98 MYA, respectively. The age around the Last Glacial Maximum (~0.02 MYA) estimated for the origin of the Hokkaido Ia and Ib can be calculated to be 0.50–0.88 MYA.

One of the few phylogeographic studies treating the Japanese lineages of N. procyonoides is that by Kim et al. (2013), who examined the Cytb gene for individuals mainly from Far Eastern Asia including the Japanese Archipelago (Hokkaido, Honshu, and Shikoku). They clarified that N. procyonoides in Japan is monophyletic and the most divergent from the other Eurasian continental lineages. Although they did not infer the divergence time between lineages in the continent and Japan, the average pairwise difference in the transversional substitutions between the Japanese and continental lineages can be calculated to be 0.135 % based on the data of Kim et al. (2013), enabling me to estimate that the time for the origin of the raccoon dog lineage in Japan is 0.64 MYA based on the 0.21 % transversions/myr divergence rate estimated for the mustelid taxa in Sato et al. (2003), where the transversional substitutions were not shown to be affected by the saturation problem. The average pairwise distance between the most divergent lineages in Japan was also calculated to be 0.081 %, meaning that the earliest lineage diversification of N. procyonoides occurred 0.38 MYA in Japan. However, because the evolutionary rate in Sato et al. (2003) was based on the interspecific divergence, leading to the underestimation of the divergence rate and overestimation of the divergence time (Ho et al. 2005), the real divergence times might be younger than estimated.

1.6.3 Ursidae

The family Ursidae for bears and the giant panda is one of the large body-sized and charismatic taxa to which much attention has been paid for reasons of attractiveness and human–bear conflicts. There are only eight species in this family in the world and two of them exist in the Japanese Archipelago (Wozencraft 2005): one is the brown bear Ursus arctos in Hokkaido, and the other is the Asian black bear Ursus thibetanus in the HSK region. U. arctos is also distributed in the northern parts of Eurasia and America [therefore currently shown as a BP1 species, but treated in this study as a BP4 species because of the fossil evidence in Honshu (Kawamura 1994); Table 3.1], and U. thibetanus has a distribution in more southern areas in the Eurasian continent including the Oriental and Southwest Asian regions (BP7; Table 3.1). The phylogenetic relationships among the ursine species have been an evolutionary conundrum extensively debated to date and still remain to be clarified. Yu et al. (2007) and Krause et al. (2008) examined the whole mitochondrial genome and obtained the result with high reliability that U. arctos and the polar bear Ursus maritimus were the most closely related to each other, and that the two black bears, U. thibetanus and the American black bear Ursus americanus, were also sister species to each other. On the other hand, Pagès et al. (2008) also proposed a well-resolved and highly supported, but different, phylogeny estimated with 14 nuclear gene sequences, where in the clade of the genus Ursus, U. arctos and U. maritimus similarly formed a clade with U. americanus and U. thibetanus successively closely related to this clade, suggesting that the two black bears did not form a clade. Although such a mtDNA–nucDNA conflict in the ursid phylogeny should be addressed in a future study, the divergence times for the branching of U. arctos and U. thibetanus from the closest relatives were estimated to be 1.32 MYA and 5.19 MYA, respectively, in the analyses of the mitochondrial protein-coding gene sequences on the assumption of 12 MYA for the divergence between the giant panda and other bears (Yu et al. 2007). The time constraint of 12 MYA for the divergence between Ailuropoda and the other ursine lineage is consistent with the recent fossil finding in the lineage of the giant panda at 11–12 MYA (Abella et al. 2012). However, as I have noted repeatedly, the divergence times obtained here might be overestimates because of using the mitochondrial DNA with a calibration point too old for dating. Yu et al. (2007) also noted that their estimates are older than the fossil records and previous nuclear gene estimates. On the other hand, Krause et al. (2008) also estimated the dievergence times with the mitochondrial genome but different assumptions about the fossil calibrations (33.9 MYA for the divergence between Ursidae and Phocidae and 4.2–7.1 MYA for the basal ursine radiation). They showed 0.75–0.97 MYA and 3.66–4.69 MYA for the occurrence of lineages of U. arctos and U. thibetanus, respectively, which are not much different from but a little younger than estimates by Yu et al. (2007). Applying the same reasoning as in Yu et al. (2007), these dates are also results of overestimations because of the old calibration point (33.9 MYA), although the effect seems reduced by the younger calibration point (4.2–7.1 MYA). It should be noted that U. maritimus was demonstrated to be placed within the variation of U. arctos, so that the divergence times presented in both study does not reflect the emergence of U. arctos. Krause et al. (2008) estimated that the lineage including both U. maritimus and U. arctos diverged from the extinct cave bear U. spelaeus 2.41–3.11 MYA by using the ancient DNA technology.

It is still not clear from the dating information of only the interspecific divergence when U. arctos and U. thibetanus migrated into the Japanese Archipelago because the two species are not endemic to Japan. Therefore, the intraspecific variations should be assessed for clarifying the origins of these species in Japan. Using the mitochondrial Cytb and control region, Matsuhashi et al. (1999, 2001) detected three distinct lineages (central, eastern, and southern lineages) of U. arctos in Hokkaido that could have been established by independent migrations from the continent. The mitochondrial genome analyses also corroborated this result (Hirata et al. 2013). The central lineage is closely related to a widely expanded Holarctic lineage, the eastern lineage has close relatives in Russian Far East and Alaska, and the southern lineage shows phylogenetic affinity to the North American lineage (Korsten et al. 2009; Davison et al. 2011; Gus’kov et al. 2013; Hirata et al. 2013). On the basis of the mitochondrial genome and the ancient DNA (radiocarbon) calibration, Hirata et al. (2013) inferred that the divergence between the central lineage in Hokkaido and the clade composed of Eurasian, Sakhalin, and Alaskan lineages occurred 0.053 MYA. Unfortunately, they did not estimate the date for the origin of the eastern lineage in Hokkaido, only showing that the time to MRCA of the clade including the lineages in Hokkaido and Kuril (Etorofu and Kunashiri) islands was 0.042 MYA. Although a further analysis of the mitochondrial genome for the closely related Russian Far Eastern and Alaskan individuals would be needed, the Late Pleistocene origin for the eastern lineage in Hokkaido could be considered as suggested in other previous studies (Korsten et al. 2009; Davison et al. 2011). Hirata et al. (2013) also estimated that the southern lineage in Hokkaido diverged from the closely related American lineage 0.194 MYA, suggesting that the southern lineage first migrated into Hokkaido in the Middle Pleistocene when the Tsugaru Strait was not established. This time estimate is consistent with the palaeontological evidence that fossil remains of U. arctos were discovered from the Middle to Late Pleistocene layers in Honshu Islands, although currently no brown bears are found there (Kawamura et al. 1989; Kawamura 1994).

In contrast to U. arctos, the lineage of U. thibetanus in Japan would have migrated from the southern route via Korean Peninsula and Kyushu (albeit considered extinct in Kyushu now) as the distribution of the same species ranges in relatively southern areas in the Eurasian continent. Two papers could be mentioned to explore the origin of U. thibetanus in Japan (Ohnishi et al. 2009; Yasukochi et al. 2009). Ohnishi et al. (2009) examined the mitochondrial control region and found that there are three lineages (eastern, western, and southern lineages) among the black bear populations. Using the calibration point that the divergence time between U. arctos and U. thibetanus was 2.0–3.5 MYA, they estimated that U. thibetanus in Japan diverged from that in the continent 1.42–2.57 MYA and the time to the MRCA of three lineages detected in Japan was 0.295–0.583 MYA although they noted a concern about the estimated divergence time because of the uncertainty in the molecular clock. On the other hand, Yasukochi et al. (2009) estimated that U. thibetanus in Japan diverged from that in the continent 0.48 to 0.66 MYA in the Middle Pleistocene by using the mitochondrial Cytb gene and the calibration point of 1.2 MYA for the U. arctos and the cave bear U. splaeus divergence. However, the assumptive date for the calibration point (1.2 MYA) is not consistent with the recent estimate of their divergence (2.41–3.11 MYA; see above). If I take a brief proportion calculation, 0.48–0.66 MYA calculated with the 1.2 MYA for the U. arctos and U. splaeus can be translated into 0.96–1.71 MYA with 2.41–3.11 MYA for the calibration point, the range of which is overlapped with that in Ohnishi et al. (2009; 1.42–2.57 MYA). The corrected estimate (0.96–1.71 MYA) is more close to the fossil evidence that U. thibetanus was found around 0.50 MYA in Japan and also the land bridge is considered to have been formed in this age between Korean Peninsula and the Japanese Archipelago (Dobson and Kawamura 1998).

1.6.4 Mustelidae

Mustelidae is the most speciose family in the order Carnivora (59 species; Wozencraft 2005), showing a variety of ecological adaptations with aquatic, arboreal, fossorial, and terrestrial locomotors. Such ecological diversifications were demonstrated to have occurred in the Middle to Late Miocene after the mid-Miocene climatic optimum (ca. 15 MYA) and produced divergent lineages corresponding to the current subfamilies (Sato et al. 2012; Sato 2016; Sato and Wolsan 2016). There are seven indigenous, terrestrial, and extant mustelids in the Japanese archipelago (therefore, the exotic American mink Neovison vison, the aquatic sea otter Enhydra lutris, and the extinct Japanese otter Lutra nippon are not included), representing three subfamilies (Sato et al. 2012), arboreal Guloninae (two Martes species), fossorial Melinae (one Meles species), and terrestrial Mustelinae (four Mustela species). The distributions of mustelid species can be divided into four major biogeographic patterns (BP2, 4, 6, and 9; Table 3.1) (also see Sato 2013). The origin of the sable Martes zibellina in Hokkaido (BP2) has been inferred by using the mitochondrial DNA variations (Sato et al. 2011; Kinoshita et al. 2015) and a recombination rate in some nuclear genes (Ishida et al. 2013). Although there are some variations among the estimates from these studies, all supported the monophyly and the Late Pleistocene origin of the sable in Hokkaido. The estimate for the divergence time between the Hokkaido lineage and the closest continental equivalent spanning western to eastern Russia was recently estimated to be 0.09 MYA on the basis of the mitochondrial Nd2 gene sequences and the assumption that the divergence between Martes foina and the other Martes species occurred 3.045 MYA (the time to the MRCA of the Hokkaido individuals was estimated to be 0.04 MYA; Kinoshita et al. 2015). The time estimate and the distribution patterns are again in agreement with the hypothesis of the Tsugaru Strait barrier in the Late Pleistocene as Sorex minutissimus, Sorex gracillimus, Apodemus peninsulae, and Myodes rutilus. Excluding the artificially introduced populations in Hokkaido and Sado Islands, the Japanese marten Martes melampus is endemic to the HSK region (BP6; Table 3.1). It has been revealed through the phylogenetic analyses of the whole mitochondrial genome with detailed palaeontological information as calibration points that M. melampus diverged from the clade composed of M. zibellina and the pine marten Martes martes 1.0 to 1.1 MYA (Li et al. 2013; in particular, see this paper for the calibration details). It could therefore be stressed that M. melampus might have migrated into the Japanese Archipelago earlier than the congeneric species M. zibellina on Hokkaido, probably via Sakhalin, considering that the distributions of the closely related species (true martens) are all in northern Eurasia. Although the reason for the absence of M. melampus in Hokkaido is not clear, Ishida et al. (2013) implied the past interspecific hybridization in Hokkaido between M. zibellina and an entity closely related to M. melmapus and M. americana.

Establishment of the ermine or stoat Mustela erminea and the least weasel Mustela nivalis in the Japanese Archipelago has been considered a somewhat complicated. Although both species possess a similar distribution pattern (BP4; Table 3.1), the extent of intraspecific genetic variations of M. nivalis is much larger than M. erminea (Kurose et al. 2005). The former species also shows salient morphological diversity among local subspecies worldwide (Abramov and Baryshnikov 2000). Kurose et al. (2005) considered that such a contrast between M. erminea and M. nivalis resulted from the difference in their migration histories after the last glacial age. In their study, four haplotypes of M. erminea were found in Japan. Three of these haplotypes from Hokkaido and Honshu were closely related to Novgorod (western Russia) and Kazakhstan haplotypes, and the other one, from Honshu, to the Kamchatka (eastern Russia) haplotype. On the other hand, five haplotypes were detected from M. nivalis in Japan, four of which were clustered as a monophyletic Hokkaido lineage closely related to North American haplotypes, whereas the other one from Honshu showed close affinity with Russian haplotypes. Although almost all the relationships for these two species lineages were weakly supported, these results indicate multiple migrations of these two weasels into the Japanese Archipelago. Unfortunately, to date, the time scale has not been estimated for either species in Japan from molecular data. Dawson et al. (2014) examined intraspecific variations of M. erminea with two mitochondrial and four nuclear genes and showed with the Cytb gene variations that all the intraspecific diversifications occurred in the Late Pleistocene, although the Japanese lineages were not discussed in their study. Because they used the divergence rate calculated based on the geology of the Japanese islands, I could not use this information as independent evidence for the Late Pleistocene origin of M. erminea in Japan because of the circular argument. If I use the divergence rate of 1.90–2.68 %/myr obtained in the study of the Japanese weasel (Masuda et al. 2012), the extent of genetic differences between the Japanese and continental lineages in M. erminea and M. nivalis calculated for the data of Kurose et al. (2005) correspond to 0.13 to 0.39 MYA in the Middle Pleistocene and 0.45 to 1.02 MYA in the Early to Middle Pleistocene, respectively, meaning that M. nivalis first colonized the Japanese archipelago, followed by M. erminea. In contrast, it is confusing that fossil remains suggest that M. erminea was found in the Middle to Late Pleistocene, whereas M. nivalis is only in the Late Pleistocene in the HSK region (Kawamura et al. 1989).

The lineage of the Japanese weasel Mustela itatsi, endemic to the HSK region (BP6; Table 3.1), was inferred to have been generated 1.5–1.6 MYA (Sato et al. 2012) through a rapid diversification among M. itatsi, the Siberian weasel M. sibirica, and a clade of other mainly European species in the same genus (M. eversmanii, M. lutreola, M. putorius, and probably M. nigripes). Masuda et al. (2012) examined the intraspecific variations with the mitochondrial region and estimated using the divergence rate of 1.90–2.68 %/myr that the earliest divergence time between the most divergent lineages (Honshu vs. Shikoku-Kyushu) was 0.83–1.17 MYA. The fossil evidence suggests that M. itatsi was present in the Middle Pleistocene (0.43 MYA; Ogino et al. 2009). It is therefore considered that the origin and the diversification of M. itatsi in Japan lies in the Early Pleistocene. Except for the introduced populations, M. sibirica is only present in the Tsushima Islands in Japan (BP10; Table 3.1). Although the divergence time of the Tsushima population from the continental equivalent has not been assessed and remains to be elucidated, Masuda et al. (2012) showed that the Tsushima lineage diverged from the clade composed of the Korean and Russian haplotypes. The average pairwise sequence difference between these lineages (1.5 %) could be calculated into 0.56–0.80 MYA, suggesting that M. sibirica in Tsushima originated in the Middle Pleistocene.

The Japanese badger Meles anakuma is also endemic to the HSK region as are Martes melampus and Mustela itatsi (BP6, Table 3.1). The closest lineage was estimated to be the Asian badger Meles leucurus, inhabiting the eastern parts of the Eurasian continent (Marmi et al. 2006; Cerro et al. 2010; Tashima et al. 2011; Sato 2016), and their divergence time was calculated to be 0.5 MYA (Tashima et al. 2011), which is consistent with the fossil record (0.43 MYA; Ogino et al. 2009). In addition, Tashima et al. (2011) suggested that there were no genetic differentiations in the mitochondrial control region and the Y-chromosomal Sry gene among populations in Honshu, Shikoku, and Kyushu, probably formed by a rapid population expansion after the glacial period. They estimated that the differentiations among them occurred 0.11–0.23 MYA based on the divergence rate of 1.92 %/myr proposed in Marmi et al. (2006) and discussed that the time estimate of the diversification included the interglacial period before the last glacial period.

1.7 Cetartiodactyla

1.7.1 Bovidae

Bovidae is the most speciose artiodactyl family, including 40 genera and 143 species (Grubb 2005). Together with the chevrotain family Tragulidae, the pronghorn family Antilocapridae, the giraffe and okapi famlily Giraffidae, the deer family Cervidae, and the musk deer family Moschidae, they constitute Ruminantia, a species-rich clade in Artiodactyla (Hassanin et al. 2012). Only one bovid species, the Japanese serow Capricornis crispus, is indigenously and exclusively present in the HSK region of the Japanese archipelago (BP = 6; Table 3.1) (also see Ohdachi et al. 2015). Molecular phylogenetic research based on the mitochondrial genome sequences has suggested that C. crispus is the most closely related to the clade consisting of two congeneric species, the Chinese serow (or just the serow), C. milneedwardsii, and the Formosan serow C. swinhoei (Chikuni et al. 1995; Hassanin et al. 2012; Bibi 2013). It was suggested in the latter two studies that the divergence of C. crispus from the sister clade occurred around the Late Pliocene to the Early Pleistocene. Okumura (2004) examined the mitochondrial control region for the three Capricornis species (it should be noted that they used the name C. sumatraensis for the Chinese serow) and estimated on the basis of the divergence rate from the bovid homologous region (10.62 %/myr; Luftus et al. 1994) that C. crispus diverged from C. swinhoei in the age older than 1 MYA. Liu et al. (2013) also adopted the similar rate assumption of the mitochondrial control region and obtained an estimate of 0.85 MYA for the occurrence of the C. crispus lineage (Although their explanations for the evolutionary rate was so confusing because they noted both human and bovid evolutionary rates in their paper and in addition the citations for the bovid evolutionary rate were incorrect, I suppose that the rate 10.62 % indicated by their study should come from Luftus et al. 1994).

For the intraspecific phylogeographic analyses, Min et al. (2004) examined the mitochondrial Cytb gene for the Japanese serows collected from Honshu and Kyushu and detected five haplotypes. Although the extent of the sequence differences was demonstrated to be low, each haplotype showed a locality-specific trend to a certain degree. Unfortunately, they did not infer the time scale. Because no reliable evolutionary rate of the Cytb gene variations could be found for the bovid lineages from previous literature, I provided a tentative evolutionary rate by a brief proportion calculation. First, the average pairwise difference of the Cytb gene sequences between C. crispus and the continental Capricornis lineages can be calculated to be 6.5 % from the data of Min et al. (2004). Considering that the average pairwise genetic distances of the control region between C. crispus and the continental Capricornis lineages is 10.5 %, based on data of Hassanin et al. (2012), 6.5 % of the Cytb gene variation shows 61.9 % (6.5 × 100/10.5) of the control region variation. Simply applying the proportion to the 10.62 %/myr rate of the control region, the divergence rate of the Cytb gene could be calculated to be 6.57 %/myr. If I adopt the 6.57 %/myr divergence rate of the Cytb gene, the pairwise genetic distance between the two most divergent haplotypes (0.6 %) was calculated to be 0.09 MYA, suggesting that the MRCA of the Honshu and Kyushu lineages was present in the Late Pleistocene in the Japanese Archipelago. The diversification of the extant Japanese lineages in the Late Pleistocene period is consistent with the story that this species could not migrate into Hokkaido because of the presence of the Tsugaru Strait (Ohshima 1991).

1.7.2 Cervidae

Cervidae encompasses 51 species assigned into 19 genera, constituting the second largest family among artiodactyls (Grubb 2005). Two major subfamilial lineages were recognized as Cervinae and Capreolinae, and the Cervinae species were estimated to have diversified in Central Asia during the Miocene to Pliocene period, producing the most speciose Cervus lineages (Gilbert et al. 2006). There is only one species, Cervus nippon, in the Japanese Archipelago, naturally distributed across the major islands except for the Ryukyu region with the same species observed in the continental East and Southeast Asia and Taiwan (BP = 4; Table 3.1) (also see Ohdachi et al. 2015). Molecular phylogenetic studies have clarified that the species most closely related to C. nippon is the red deer, Cervus elaphus, distributed worldwide in the Holarctic region, and the white-lipped deer Cervus (Przewalskium) albirostris, found in continental East Asia (Gilbert et al. 2006; Hassanin et al. 2012). Although the exact phylogenetic relationships among them remains to be clarified, Gilbert et al. (2006) indicated that the diversification of these lineages occurred around 1.5 MYA in the analyses of two mitochondrial (Cytb and Co2) and two nuclear (αLalb and Prkci) genes.

As regards the intraspecific variation of C. nippon in the Japanese Archipelago, it has been demonstrated that there are two major mitochondrial DNA lineages geographically demarcated by a borderline in southern Honshu between Hyogo and Yamaguchi Prefectures (in the Chugoku district), which are called the northern and southern lineages (Tamate et al. 1998; Nagata et al. 1999). Assuming the 10.6 %/myr divergence rate of the mitochondrial control region sequences (Luftus et al. 1994), Nagata et al. (1999) estimated that divergence time among the Chinese conspecific, the northern Japanese, and the southern Japanese lineages was about 0.30 MYA in the Middle Pleistocene. Because the divergence time between northern and southern Japanese lineages was the oldest, they interpreted that the divergence occurred in the continent and that these two Japanese lineages independently migrated into the Japanese Islands. The Middle Pleistocene origin of C. nippon in the continent and subsequent migrations into the Japanese Archipelago are consistent with the fossil records of Kawamura et al. (1989) that the Japanese sika deer was first found in the Late Pleistocene.

1.7.3 Suidae

The family Suidae contains 19 species classified into 5 genera (Grubb 2005) and includes the wild boar Sus scrofa, which is an ancestor of the pig, an important domesticated animal strongly related to human life. Recent molecular phylogenetic analyses with the mitochondrial genome revealed that the Suidae and the closely related peccary family Tayassuidae constitute the most basal lineage within the Cetartiodactyla phylogeny (Hassanin et al. 2012). S. scrofa is a species widely distributed in the Eurasian continent and northern Africa (Grubb 2005) showing two major lineages in Europe and Asia, each having independently produced domestic pigs (Giuffra et al. 2000; Kijas and Anderson 2001). The occurrence of the S. scrofa lineage was estimated to have occurred 4.2 MYA in the Pliocene based on the autosomal genome sequences with millions of single-nucleotide polymorphism (SNP) data produced by the next-generation sequencer and on the fossil records, although some interspecific hybridizations were detected between S. scrofa and other closely related Southeast Asian species (Frantz et al. 2013). S. scrofa also exists in the major Japanese islands, except for Hokkaido (BP8; Table 3.1) (also see Ohdachi et al. 2015). HSK and the Ryukyu regions harbor different subspecies, the Japanese wild boar S. s. leucomystax and the Ryukyu wild boar S. s. riukiuanus, respectively (Ohdachi et al. 2015). It has been demonstrated with mitochondrial DNA variations that these two subspecies could have probably originated from different source populations in continental Eurasia (Okumura et al. 1996, 2001; Watanobe et al. 1999, 2003; Hongo et al. 2002). Watanobe et al. (2003) showed by examining the mitochondrial control region that the Japanese wild boar in HSK is phylogenetically closely related to the East Asian lineages in the continent and is not monophyletic because some northeast Asian haplotypes are included within this Japanese wild boar clade. It was estimated in the same study that the lineage of the Japanese wild boars in the HSK region diverged 0.14 to 0.25 MYA in the Middle Pleistocene from the continental East Asian lineage on the basis of the assumption that the divergence time between European and Asian wild boars was 0.5 to 0.9 MYA (Giuffra et al. 2000; Kijas and Anderson 2001). It should however be noted here that, as Giuffra et al. (2000) based their time estimations on the evolutionary rate from Brown et al. (1979; RFLP data for primates), the divergence time of 0.5 MYA between Asian and European wild boars may not be accurate. In addition, Frantz et al. (2013) estimated with genome sequencing that the divergence was much older (1.2 MYA) and relatively close to that of Kijas and Anderson (2001; 0.9 MYA). If I adopt the average (1.05 MYA) between 1.2 MYA and 0.9 MYA, the divergence of the Japanese wild boar would have taken place 0.29 MYA, also in the Middle Pleistocene. It has been demonstrated that there are at least three lineages among the Japanese wild boars (S. s. leucomystax), which would have been established multiple independent migration events (Ishiguro et al. 2008). On the other hand, using the mitochondrial control region, Hongo et al. (2002) and Ishiguro et al. (2008) suggested the close affinity between Ryukyu and Vietnamese individuals, which implies the migrations of the Ryukyu wild boar from the southern routes. However, the supportive value for the clade was extremely low and therefore such an affinity should be viewed as preliminary. Although Watanobe et al. (2003) did not examine the Vietnamese individuals, they estimated on the basis of 0.5–0.9 MYA for the European/Asian wild boar divergence that the Ryukyu lineage diverged 0.31–0.57 MYA from the other Asian lineage including the lineages of the Japanese wild boars in the HSK region. If I apply 1.05 MYA for the calibration points as above, the Ryukyu lineage was estimated to have branched off 0.66 MYA. There should, however, be a more closely related and still unidentified lineage in the continent such as the Vietnamese lineage because fossil evidence suggested that the wild boar is likely to have migrated into the Ryukyu Islands 0.018 MYA (Hasegawa 1980), which is also consistent with the geology of the Ryukyu Islands (Kimura 2000). Unfortunately, I could not calculate the divergence time between the Ryukyu and the Vietnamese lineages because there are no sequences of the Ryukyu endemic M16–M20 D-loop haplotypes used in Hongo et al. (2002) and Ishiguro et al. (2008) in the DNA database. Watanobe et al. (1999) showed that Amami–Oshima, Okinawajima, and Iriomotejima Islands in the Ryukyu region each possess an island-specific haplogroup (or haplotype) of the combined mitochondrial Cytb gene and control region, although no time scale was inferred.