To illuminate molecular phylogenetic relationships among Eurasian species of the genus Mustela (Mustelidae, Carnivora), we determined nucleotide sequences of the complete mitochondrial cytochrome b gene region (1,140 base pairs). Molecular phylogenetic trees, constructed using the neighbor-joining and the maximum likelihood methods, showed the common topology of species relationships to each other. The American mink M. vison first branched off and was positioned very remotely from the other species of Mustela. Excluding M. vison, the ermine M. erminea first split from the rest of the species. Two small body-sized weasels, the least weasel M. nivalis and the mountain weasel M. altaica, comprised one cluster (named “the small weasel group”). The other species formed another cluster, where the remarkably close relationships among the domestic ferret M. furo, the European polecat M. putorius, and the steppe polecat M. eversmanni were noticed with 87–94% bootstrap values (named “the ferret group”), supporting the history that the ferret was domesticated from M. putorius and/or M. eversmanni. The European mink M. lutreola was the closest to the ferret group. The genetic distance between the Siberian weasel M. sibirica and the Japanese weasel M. itatsi corresponded to differences of interspecific level, while the two species were relatively close to M. lutreola and the ferret group. These results provide invaluable insight for understanding the evolution of Mustela as well as for investigating the hybridization status between native and introduced species for conservation.
The family Mustelidae, which consists of more than 65 species, has been most diversified in the order Carnivora (Nowak, 1991). In this family, the genus Mustela including 17 species is a group of small or middle-sized weasels and distributed in a worldwide scale (Wozencraft, 1993; Abramov, 1999a). In Eurasia, 12 species belonging to Mustela are known: the mountain weasel M. altaica Pallas, 1811; the ermine M. erminea Linnaeus, 1758; the least weasel M. nivalis Linnaeus, 1766; the Siberian weasel M. sibirica Pallas, 1773; the European polecat M. putorius Linnaeus, 1758; the steppe polecat M. eversmanni Lesson, 1827; the European mink M. lutreola (Linnaeus, 1761); the Japanese weasel M. itatsi Temminck, 1844; the yellow-bellied weasel M. kathiah Hodgson, 1835; the back-striped weasel M. strigidorsa Gray, 1853; the Malaysian weasel M. nudipes Desmarest, 1822; the Indonesian weasel M. lutreolina Robinson and Thomas, 1917. In addition, the American mink M. vison Schreber, 1777 was introduced to the Old World from North America and naturalized in Europe, Siberia, and Hokkaido Island of Japan.
Some researchers reported the phylogeny of Mustela based on previous morphological data (Youngman, 1982; Anderson, 1989; Wozencraft, 1989; Dunstone, 1993). However, the overall phylogeny of Mustela is still controversial, because there are few comparative studies between European and Asian species. On the other hand, based on genetic data, some phylogenetic relationships of Mustela were reported as follows. Graphodatsky et al. (1976) studied karyotaxonomy among seven species of Mustela, and considered that M. vison first split from the other Mustela species. The close relationship between M. putorius and M. eversmanni was indicated, and their ancestral characters were seen in M. lutreola and M. sibirica. Mustela altaica and M. nivalis were closer to each other, and M. erminea retained ancestral characters for the other species except M. vison. Masuda and Yoshida (1994) reported the molecular phylogeny of Mustelidae living around Japan using partial sequences of the mitochondrial DNA (mtDNA) cytochrome b gene. Also there, M. vison was remote from the other Mustela species. The large genetic difference between M. itatsi and M. sibirica was found, supporting the original taxonomical treatment that M. itatsi Temminck, 1844 is a distinct species.
Especially, the phylogenetic position and genetic characteristics of M. vison are required to be elucidated. At present, the precise evolutionary distance between M. vison and other species is still unclear, although the common opinion is that M. vison is phylogenetically positioned out of the other Mustela species (Graphodatsky et al., 1976; Belyaev et al., 1980; Lushnikova et al., 1989; Taranin et al., 1991; Masuda and Yoshida, 1994).
The domestic ferret M. furo is generally thought to be domesticated from M. putorius, or the congener M. eversmanni which has a more similar cranial morphology (Blandford, 1987). Davison et al. (1999) investigated the mtDNA phylogeny among putorius, eversmanni, and furo, and found two geographically distinct lineages of putorius in Britain. The population expansion may be mediated by dispersing male putorius hybridizing with female feral ferrets (Davison et al., 1999). The hybridization between the ferret and polecats has been anticipated as a threat to the genetic integrity of native polecat populations in Europe. Thus, it is now much required to clarify the genetic relationships within Mustela.
In the present study, we determined the complete nucleotide sequences of the cytochrome b gene for ten Mustela species distributed in Eurasia, including the domestic ferret and the American mink. Based on sequence diversity, we here present the molecular phylogeny within this genus and discuss the genetic relationships between species.
MATERIALS AND METHODS
Samples and DNA extraction
Species of the genus Mustela examined in the present study were listed in Table 1. Sequence data of the Japanese marten Martes melampus and the sable Martes zibellina as outgroups were quoted from our previous report (Kurose et al., 1999). Muscle tissues from animals were preserved in 70–100% ethanol at room temperature until use. Total DNAs were extracted from muscles using the phenol/proteinase K/sodium dodecyl sulfate (SDS) method of Sambrook et al. (1989) with some simplified modifications as indicated by Masuda and Yoshida (1994). DNA extracts of M. lutreola (muscle tissues preserved in 70% ethanol for 30 years), and M. altaica (muscle tissues preserved in 70% ethanol for 108 years) were concentrated to approximately 100 fold using Centricon-30 microconcentrators (Amicon), because these tissues were preserved so long that the DNAs contained were fragmented. Hair DNA was extracted only from M. furo using the method of Walsh et al. (1991). An aliquot (1–10 μl) of each DNA extract was used as template of subsequent polymerase chain reaction (PCR).
Profiles of samples examined in the present study
PCR amplification and direct sequencing
The complete (1,140 base pairs, bp) or partial (approximately 450 bp) region of the cytochrome b gene was PCR-amplified using any pair of the following primers, all of which were newly designed in the present study (Fig. 1): Cb-M1 5′-CTCACATGGAATCTAACC-ATGAC-3′; Cb-M6 5′-AGACGTCAACTACGGCTGAAT-3′; Cb-M7 5′-GAATGAATCTGAGGCGGATTTTC-3′; Cb-M8 5′-CAACCCCCTCAA-CACACCT-3′; Cb-M9 5′-CTATTAGTATTATTCTCACCCGA-3′; Cb-M11 5′-CGATTCTTCGCTTTCCACTT-3′; Cb-MR1 5′-TCTTCCTTGA-GTCTTAGGGAG-3′; Cb-MR5 5′-GGCTAGGACTAGGATGGAGA-3′; Cb-MR6 5′-GCCGATGTTTCATGTTTCGG-3′; Cb-MR7 5′-GATCCTG-TTTCGTGGAGGAATA-3′; Cb-MR8 5′-GATCGTAGAATAGCATAT-GCGAA-3′. Procedures of symmetric PCR and direct sequencing were the same as those reported by Kurose et al. (1999).
Sequence alignment was done using GeneWorks computer software (Intelligenetics). The neighbor-joining tree (Saitou and Nei, 1987) using Kimura's (1980) two-parameter distance was constructed by Mega computer software (Kumar et al., 1993). The maximum likelihood tree (Felsenstein, 1981) was constructed using PHYLIPS computer software (Felsenstein, 1993). Numbers of transitions and transversions were calculated using Clustal V computer software (Higgins et al., 1992). The divergence time between species was estimated using the transversional substitution rate (0.5% per million years, Myr) at the third codon positions of the mammalian cytochrome b gene (Irwin et al., 1991).
Figure 1 shows nucleotide substitution frequencies in every 100-bp segment of cytochrome b among the Mustela species. It is obvious that nucleotide substitutions within Mustela appeared almost constantly throughout the gene region. Transitions occurred more dominantly than did transversions (Table 2).
Percentage differences (above diagonal) and numbers of transitions/transversions (below diagonal) for the complete cytochrome b gene sequence (1,140 bases) of genera Mustela and Martes (outgroup)
In order to assess multiple substitutions in cytochrome b of Mustela, transitions per site were plotted against transversions per site (Fig. 2). The relationships between transitions and transversions were almost proportional within the genus Mustela as well as between Mustela and Martes (outgroup) (Fig. 2), indicating that transitional substitutions among cytochrome b genes of these species have not yet been saturated and that genetic distances calculated from sequences comparisons certainly provide confident phylogenetic relationships. Percentage differences within Mustela were 0.3–13.8% and those between Mustela and Martes (outgroup) were 11.3–15.4% (Table 2).
Molecular phylogenetic trees constructed using the neighbor-joining (NJ) and the maximum likelihood (ML) methods showed the almost same topology of the species relationships (Fig. 3). M. vison first split from the other species. Then, M. erminea was separated from the rest of the species. M. altaica and M. nivalis were clustered with 57%/72% (NJ/ML) bootstrap values, and the cluster was named “the small weasel group”.
In another cluster, M. putorius, M. eversmanni, and M. furo were remarkably close to each other (named “the ferret group”) with 94%/87% bootstrap values, having small genetic distances (0.3–0.6% sequence difference, Table 2) among them. NJ tree showed that M. eversmanni was closer (65% bootstrap value) to M. furo than to M. putorius, while ML tree indicated that M. eversmanni was closer to M. putorius (44% bootstrap value) than to M. furo (Fig. 3), although the bootstrap values were not so high.
Mustela lutreola was closer to, but out of, the ferret group (100%/95% bootstrap values). The genetic distance (5.3% sequence difference, Table 2) between M. itatsi and M. sibirica corresponded to differences between other distinct species, while the two species were separated from the ferret group and M. lutreola.
The overall species relationships (Fig. 3) obtained in the present study are partially in concordance with previous morphological classification and karyotaxonomy (Graphodatsky et al., 1976; Wozencraft, 1989; Dunstone, 1993). Moreover, our data well answered some questions concerning the Mustela phylogeny as follows.
Phylogenetic positions of two minks, M. vison and M. lutreola
The phylogenetic relationship between M. vison and M. lutreola had long been unclear. Many researchers believed that the two minks were conspecific or at least closely related to each other. Then, the presence of some remarkable differences in cytogenetic and molecular characters between M. vison and the other Mustela species was reported (Graphodatsky et al., 1976; Belyaev et al., 1980; Lushnikova et al., 1989; Taranin et al., 1991; Masuda and Yoshida, 1994). For example, the diploid chromosome number of M. vison is 30 (2n=30, Table 1), while the other Mustela species have a range between 38 and 44 chromosomes (Ewer, 1973). The present study also demonstrated that M. vison is clearly separated from the other Mustela species. The genetic distances (11.3–13.8%) which M. vison had with the other Mustela species were much higher than interspecific differences among the other Mustela species, and nearly corresponded to the differences between the other Mustela species and another genus Martes (Table 2, Fig. 3). However, the frequencies of transversions at the third codon positions (Table 3) clearly show that the two species of Martes are remoter than Mustela vison from the other Mustela species. M. vison was estimated to have diverged from the other Mustela species approximately 8–11 Myr ago (Table 3).
Percentage differences of transversions at the third codon positions of the cytochrome b sequences (above diagonal) and the estimated divergence time (Myr, below the diagonal) using the transversional substitution rate (0.5%/Myr) at the third codon positions of mammalian cytochrome b reported by Irwin et al. (1991)
Previous morphological classification (Youngman, 1982; Wozencraft, 1989; Dunstone, 1993) as well as karyotaxonomy (Graphodatsky et al., 1976) supported the closer relatedness in the lutreola-sibirica lineage. The molecular phylogeny of the present study also exhibited that M. lutreola has close relationship with M. sibirica as well as with the ferret group (but not included in this group cluster). The comparative analysis of the antigenic structure of the immunoglobulin chain in Mustelidae (Taranin et al., 1991) showed that M. lutreola is closer to the ferret group. Meanwhile, Davison et al. (1999) reported that M. lutreola was included in the ferret group cluster comprising M. furo, M. putorius, and M. eversmanni, based on the molecular phylogeny using partial cytochrome b sequences (337 bp). The partial sequence (Accession No. AF068544) reported by Davison et al. (1999) was identical with the homologous region of the complete cytochrome b sequences obtained in the present study. The difference on M. lutreola's position may be ascribed to the shorter informative sequence (337 bp) used by Davison et al. (1999). Otherwise, the intraspecific variations of M. putorius and M. eversmanni might be so large that the position of M. lutreola was included in them.
Now, it is obvious that M. vison and M. lutreola are phylogenetically much differentiated from each other, although both animals are commonly called ‘mink’.
Parental species of the domestic ferret M. furo
Our results showed the closer relationships among the three morphological similar polecats, M. putorius, M. eversmanni, and M. furo (Table 2 and Fig. 3). The sequence differences (0.3–0.6%) were very similar to the level of intraspecific variations of the Martes species reported by Kurose et al. (1999). The divergence among these three species was estimated to have occurred within one Myr ago (Table 3). The diploid chromosome number is 2n=40 for M. putorius as well as M. furo, while 2n=38 for M. eversmanni (Table 1). The former two species have morphologically identical chromosome sets, but the karyotype of M. eversmanni differs from those of the former by a single Robertsonian rearrangement (Fredga and Mandhal, 1973; Volobuev et al., 1974). A series of embryological characteristics of M. furo was reported to be more similar to M. putorius than to M. eversmanni (Ternovsky, 1977). Experimental hybridization among M. furo, M. putorius, and M. eversmanni was found to be possible, and all the hybrids were fertile (Ternovsky, 1977). In the present study, we found the very close molecular phylogenetic relationships among the three species, supporting the parental species of the domestic ferret was the polecats. Which polecat is closer to M. furo was not concluded in the present study, because the relationships among them were not supported with high confidence (Fig. 3). Further analyses of karyotypes, mtDNA and nuclear DNA, and morphology of the ferrets with different lineages as well as the wild populations of polecats would illuminate more precise history of the ferret's domestication.
Phylogenetic positions of the Japanese weasel Mustela itatsi and the Siberian weasel M. sibirica
The taxonomical relationship between M. itatsi and M. sibirica has been obscure for a long time. Very often was itatsi considered as conspecific to sibirica (Ellerman and Morrison-Scott, 1951; Corbet and Hill, 1992; Wozencraft, 1993; Pavlinov et al., 1995). Imaizumi (1960) also regarded itatsi as a sub-species of M. sibirica, showing that the ratio of the tail length to the head-body length is more than 50% for sibirica of Korean Peninsula and around 40% for itatsi of Japanese islands. The present study followed this classification of the two weasels. Recently, Abramov (1999b) reported the presence of a clear craniological difference between the two species. Masuda and Yoshida (1994) also supported the idea that itatsi is an independent species, based on partial cytochrome b sequence data. Table 2 shows that itatsi has the relatively closer genetic distance (5.3%) with sibirica, while this value refers to the sequence difference between the other Mustela species. The two weasels are still phylogenetically closer to each other than to other species (Fig. 3). We recently found the karyotypical difference of G- and C-banding patterns between the two weasels (unpublished data), both of which share the identical diploid chromosome number (2n=38, Table 1). The transversional difference of the third codon positions (0.8%, Table 3) between itatsi and sibirica refers to approximately 1.6 Mry ago of the divergence time. The ancestor of itatsi might have been derived from sibirica probably in the continent in the early Pleistocene. After that, a certain ancestral population of itatsi might have immigrated to Japanese islands, and then it could have been isolated on the islands through the strait formation. The other ancestral populations remaining in the continent might have been extinct. Otherwise, from an ancestor common to sibirica, itatsi might have evolved independently on Japanese islands after separation from the continent.
Phylogeny of the small weasel group
The closer relationship between M. altaica and M. nivalis indicated in the present study was also supported by classic morphological classification (Heptner et al., 1967; Youngman, 1982; Wozencraft, 1989; Dunstone, 1993) and karyotaxonomy (Graphodatsky et al., 1976). M. altaica is distributed from the Altai mountains to the Himalayan region and some mountainous areas of Asia, while M. nivalis is widespread in Eurasia and North America. The two species were estimated to have diverged over five Myr ago (in the end of Miocene) (Table 3). The present study first reported mtDNA sequence data of M. altaica and exhibited the molecular phylogenetic position.
Mustela erminea is widespread in Eurasia and North America. The divergence time between M. erminea and the other Mustela species except M. vison was estimated around three to five Myr ago (Table 3). Thus, the species diversification within Mustela might have started from the end of Miocene to Pliocene. Our data that M. erminea first split from the other Mustela species is supported by the karyological study (Graphodatsky et al., 1976; Obara, 1991) that M. erminea is the more ancestral form among Mustela.
The present study revealed the phylogenetic relationships among Eurasian species of the genus Mustela. Our results provide not only invaluable insight to reconstruct the taxonomy of this carnivore group, but also useful information to survey genetic characteristics and hybridization problems between native and introduced populations for species conservation. In addition, further study involving other Mustela species of Eurasia and America would illuminate the worldwide evolution of Mustela.
We thank H. Abe (Hokkaido University), Y. Masuda (Shiretoko Museum), F. Sekiyama (Iwate Prefectual Museum), and S. Watanabe for supplying specimens. We are grateful to the Institute of Low Temperature Science of Hokkaido University for technical support. This study was supported in part by Grants-in-Aid for the Scientific Research from the Ministry of Education, Science, Sports, and Culture, Japan.