The Japanese spined loach, Cobitis takatsuensis, has some unique morphological and ecological features among Japanese Cobitis species. Mitochondrial DNA analyses were conducted to investigate the magnitude of intraspecific differentiation and phylogenetic relationships among Japanese congeners of C. takatsuensis. PCR-RFLP (polymerase chain reaction-restriction fragment length polymorphism) analysis of the ND1 region with 16 restriction enzymes was carried out for thirteen C. takatsuensis populations throughout the species' distributional range. Populations in each river system possessed their own haplotypes, with marked genetic differentiation between the populations from Chugoku and Kyushu (ChugokuKyushu group) and those from Shikoku (Shikoku group). The two allopatric groups also showed different color pattern. Subsequently, sequencing analysis of part (725 bp) of the cytochrome b gene was carried out for C. takatsuensis and six other closely-related Japanese species. The phylogenetic tree indicated the presence of two major mtDNA lineages within Japanese Cobitis. It was noteworthy that the Chugoku-Kyushu and Shikoku groups of C. takatsuensis were included in separate mtDNA major lineages, and each group was closely related to other species. It is inferred that the distinct mtDNA relationship between the two allopatric C. takatsuensis groups is a result of the parallel evolution or mtDNA introgression, rather than divergence by geographic isolations.
The Japanese spined loach, Cobitis takatsuensis (Mizuno, 1970), is small species inhabiting only stony bottoms in mountain streams of western part of Japan (Kimizuka et al., 1982; Shimizu and Mizuno, 1994). It is rare species listed as ‘Threatened I’B rank in the 1997 Red List of Threatened Animals of Japan (Environment Agency of Japan, 1997). This species is morphologically characterized by thin caudal peduncle, incomplete sexual dimorphism, and unique color pattern (Mizuno 1970, 1975). These ecological and morphological features are peculiar among Japanese Cobitis species.
In addition, C. takatsuensis exhibits considerable morphological variation among local populations (Shimizu and Mizuno, 1994; Mizuno, 1995). In particular, marked differences in color pattern exist between populations from Chugoku District (western Honshu) and Kyushu Island, and those from Shikoku Island. The former were roughly represented by striated, and the latter by spotted types (Shimizu and Mizuno, 1994; Fig. 1). Although they have been regarded as intraspecific geographical variations, their taxonomic status remains obscure.
In this study, we conducted restriction fragment length polymorphism (RFLP) and sequence analyses on some parts of the mtDNA to investigate the magnitude of genetic differentiation between the two types of C. takatsuensis and their phylogenetic relationships among Japanese congeners, C. biwae, C. matsubarai, and three undescribed species of Cobitis (C. sp. 1, C. sp. 2, and C. sp. 3), which taxonomically follow as Hosoya (1993) (Table 1). Recently, Saitoh et al. (2000) reported the presence of three distinct mtDNA lineages within C. biwae, corresponding to geographical proximity (eastern part of Honshu, central-western part of Honshu, and Pacific slope of Shikoku). We considered them as distinct groups in C. biwae here.
Distribution and chromosome number of Cobitis and Niwaella species in Japan, according to Hosoya (1993).
MATERIALS AND METHODS
Specimens were caught by hand net during 1995 and 1999. Total of forty-seven individuals of C. takatsuensis were obtained from thirteen locations in twelve river systems (Fig. 2, Table 2). In addition, nine individuals of other species (C. biwae, C. matsubarai, C. sp. 1, C. sp. 2 subsp. 2, C. sp. 3, and Niwaella delicata) were included for inter-specific phylogenetic analysis. Three groups of C. biwae (sensu Saitoh et al. 2000), are herein temporarily referred to as Eastern, Western (consisted of diploid and tetraploid races (Kobayashi, 1976)), and Kochi groups, respectively. Accordingly, four individuals of C. biwae representing the three groups and a tetraploid race were included (Table 2). The identification of diploid or tetraploid specimens was determined from blood-smear preparations (Sezaki and Kobayashi, 1978). The remaining five individuals represented four other Cobitis species and a sister genus species Niwaella delicata as an outgroup. Saitoh et al. (2000) reported close genetic relationships among subspecies within C. sp. 2, and also that karyotype races within C. matsubarai. For simplification in the analysis, a single subspecies (subsp. 2) and a race (2n=94, inferred from collecting locality) each represented (Table 2). Samples were frozen immediately following collection, and stored at −20°C until analysis. Voucher specimens were deposited in the collection of the Fisheries Research Laboratory of Mie University (FRLM: catalogue numbers were listed in Table 2).
Sampling localities, sample sizes, and catalogue numbers. Locality numbers correspond to those in Fig. 2.
Total DNA was extracted from approximately 100 mg of frozen muscle tissue, as described by Asahida et al. (1996).
PCR-RFLP for intraspecific analysis
Restriction analysis was conducted for all specimens of C. takatsuensis.
A segment of about 2.0 kbp, containing a complete NADH dehydrogenase subunit 1 (ND1) and part of the 16S ribosomal RNA (16SrRNA) gene of mtDNA, was amplified by the polymerase chain reaction (PCR). A pair of primers (forward: 5′ - ACC CCG CCT GTT TAC CAA AAA CAT - 3′ and reverse: 5′ - GGT ATG AGC CCG ATA GCT TA - 3′) described by Hall and Nawrocki (1995) was used. However, because these were not effective for fish from Shikoku, another pair of primers, modified for carp (forward: 5′ - GCC TCG CCT GTT TAC CAA AAA CAT- 3′ and reverse: 5′ - GGT ATG GGC CCG AAA GCT TT - 3′), were used. PCR consisted of 30 cycles of 1 min. at 94 °C, 1 min. at 50°C and 2 min. at 72°C. Subsequently, PCR products were digested with sixteen restriction enzymes that specifically recognize four or five base nucleotides (Table 3; from New England Biolabs, Beverly, MA; and Takara Shuzo, Kyoto), following the protocols recommended by the manufacturers. The restricted fragments were separated by horizontal electrophoresis in 3% agarose gel. Digested fragments were visualized and photographed on an ultraviolet transilluminator after ethidium bromide staining. Nucleotide sequence divergences between mtDNA haplotypes were estimated according to Nei and Li (1979), using the D program from REAP (MacElroy et al., 1992). Phylogenetic trees were generated by the UPGMA method using the Neighbor program from PHYLIP ver. 3.572 (Felsenstein, 1996).
Fragment patterns and approximate molecular size (in base pairs; bp) generated by sixteen restriction enzymes.
Sequencing for interspecific phylogenetic analysis
Sequence analysis was performed on mtDNA fragments comprising part of the cytochrome b gene. Seven individuals of C. takatsuensis, representing each RFLP cluster, and the nine individuals of closely-related species (Table 2) were sequenced. A pair of primers (forward: 5′ - TGA CTT GAA RAA CCA YCG TTG - 3′ and reverse: 5′ - RGC RAA KAR RAA RTA YCA TTC - 3′) described by Palumbi et al., (1991) were used. PCR conditions for the sequencing analysis were similar to those for RFLP analysis, with the annealing temperature set at 54°C. Amplified DNA was purified with a QIA quick spin column (Qiagen, Germany). Direct sequencing of purified products was achieved using the Ready Reaction Dye Terminator Cycle Sequencing Kit (Applied Biosystems) following the manufacturer's protocol. Primers used were the same as those for PCR. Sequences were analyzed on an Applied Biosystems Model 377 automated DNA sequencer. Each DNA strand was determined the sequence two or three times. DNA sequence data were edited using DNASIS programs (Hitachi Software Engineer. Co. Ltd.). The nucleotide sequence data reported here are available from DDBJ, EMBL and GeneBank accession numbers AB039337– AB039352.
Three different methods were conducted to infer the phylogenetic relationships. A distance matrix was calculated based on Kimura's two-parameter method (Kimura, 1980) and clustered by the neighbor-joining method (Saitou and Nei, 1987) using MEGA ver. 1.0 (Kumar et al., 1993), and a maximum likelihood analysis performed using DNAML in PHYLIP ver. 3.572 (Felsenstein, 1996). A maximum parsimony analysis was performed with the Heuristic Search program from PAUP ver. 3.1.1 (Swofford, 1993). The robustness of each phylogeny was assessed by bootstrap analyses consisting of 1,000 replicates (Felsenstein, 1985).
PCR-RFLP for intraspecific analysis
The digested mtDNA fragment patterns are listed alphabetically for each of the 16 enzymes (Table 3). The total length of the digested DNA fragments varied in some cases, since the detection of small-sized fragments (< 70 bp) was not possible with the method used. The composite genotypes (haplotypes) detected among C. takatsuensis populations and the location of each are given in Table 4. All 16 enzymes produced variable restriction patterns. Restriction patterns detected in the populations from Chugoku and Kyushu were so different from those from Shikoku as to have hardly any fragments in common. Therefore, length-difference method (Nei and Li, 1979) cannot be applied for the analysis between them (Nei, 1987). Accordingly, we referred to them as Chugoku-Kyushu and Shikoku groups, respectively. The sequence divergences of the two groups were calculated separately.
Composite PCR-RFLP haplotypes for 13 populations of C. takatsuensis.
Within the Chugoku-Kyushu group, ten of the sixteen enzymes were polymorphic. A total of eight haplotypes (C1-7 and K1) were observed among the 31 fish analyzed (Table 4). Populations in each river system were fixed for their own haplotype (Table 4). The resulted UPGMA tree (Fig. 3. A) revealed some sub-clusters corresponding to geographical proximity. The estimated range of pairwise sequence divergence between haplotypes varied from 0.13 to 1.33%.
Within the Shikoku group, twelve of the sixteen enzymes were polymorphic. A total of six haplotypes (S1-6) were observed among the 16 fish analyzed (Table 4). The estimated range of pairwise sequence divergence among all haplotypes varied from 0.24 to 2.14%. Each population possessed it's own haplotypes (Table 4). In the UPGMA tree (Fig. 3. B), haplotypes detected in the same populations (S3 and S4, S5 and S6) were clustered.
Sequencing for interspecific phylogenetic analysis
A 725 base pair region of the cytochrome b gene was successfully sequenced for all specimens. No insertions or deletions were observed.
The NJ tree (Fig. 4) constructed from all three codon positions (725 bp) on the cytochrome b sequence indicated two major lineages of Cobitis supported by high bootstrap value (98% and 99%, respectively). Nucleotide divergence between the two major lineages averaged 17.21%. It was noteworthy that the haplotypes of C. takatsuensis from the ChugokuKyushu (C2, C4, C7, and K1) and Shikoku (S1, S3, and S6) groups were included in separate major lineages. Similarly, haplotypes of C. biwae from the Western group (W2 and W4) were included in different major lineages to those from the Eastern (E) and Kochi (K) groups. Cobitis takatsuensis from the Shikoku group (S1, S3, and S6) were clustered very closely with C. biwae from the Western group (W2 and W4). On the other hand, C. takatsuensis from the Chugoku-Kyushu group (C2, C4, C7, and K1) was associated with C. biwae from the Eastern group (E) with a high bootstrap value (95%).
Maximum likelihood analysis was conducted for the two individuals of C. takatsuensis (C7 and S1) representing each group, and the nine individuals of other species which were analyzed by the NJ analysis. The transition / transversion ratio was set at 4.6, following the method of Oohara et al. (1999). Although a change in the transition / transversion ratio value to accommodate a wide range (from 1 to 10) was attempted, the topology of the tree was not entirely changed. The topology of the ML tree (Fig. 5. A) was concordant with that constructed by the NJ method (Fig. 4). The two major lineages of Cobitis were completely (100%) supported by bootstrap replications. The each close relationship between C. takatsuensis (S1) and C. biwae (W2 and W4), and C. takatsuensis (C7) and C. biwae (E) were also strongly supported (100% and 96%, respectively).
The nucleotide substitutions at the third position of codon between each taxon were counted from 10 to 106 of the 242 sites, indicating saturation between distantly-related taxa. There was a possibility that multiple substitutions could not be adequately corrected in maximum parsimony analysis (Nei, 1987). Therefore, only first and second codon positions data set was used for MP analysis. The latter produced a topology concordant with those obtained by the NJ (Fig. 4) and ML (Fig. 5. A) methods, with the exception of an ambiguity in the branching order deep within the lineage I (Fig. 5. B).
The present result indicated the marked mtDNA divergence between C. takatsuensis populations from ChugokuKyushu and those from Shikoku, corresponding to color pattern differentiation proposed by Shimizu and Mizuno, (1994) (Fig. 1). RFLP haplotype divergences within the ChugokuKyushu and Shikoku groups were 0.13–1.33 % and 0.24– 2.14%, respectively. Since present RFLP data were estimated from a part of mtDNA, we could not directly compare to other population studies that have used whole mtDNA. Nevertheless, these values seem to be relatively high compared with other intraspecific divergence in freshwater species summarized by Bernatchez and Wilson (1998) (the maximum intraspecific divergence value < 1%, in the half of 42 species). Moreover, no shared haplotypes among river systems were observed. Within the each group, such considerable differentiations among populations must be attributed to the isolations for their exclusive mountain stream habitats (Kimizuka et al., 1982; Shimizu and Mizuno, 1994). On the other hand, it is unlikely that the differentiation between the two groups (19.24% sequence divergence in cytochrome b gene sequence) has resulted from such local isolations. If a conventional 2% divergence per Myr (Brown et al. 1979) is applied to this data, the two groups of C. takatsuensis could have diverged about 9–10 million years ago. This time estimation implies that the divergence occurred much long before the geographical isolation among western Chugoku, northern Kyushu and Shikoku Regions (the isolation time is shorter than 20,000 years, after Kaizuka, et al. 1995).
The present phylogenetic results provided some hints for distinct mtDNA compositions between the two allopatric groups of C. takatsuensis. The phylogenetic trees revealed the presence of two major mtDNA lineages within Japanese Cobitis. It is noteworthy that the two divergent groups within C. takatsuensis (Chugoku-Kyushu and Shikoku) were included in separate major Cobitis lineages, as well as three distinct groups of C. biwae (Eastern, Western, and Kochi groups). A simple, general explanation for the distinctive mtDNA composition within one species is possible random lineage sorting of ancestral polymorphic mtDNA (Billington and Herbert, 1991). In the present case, however, the divergence time of the two lineages (approximately 8–10 million years ago) calculated from the conventional vertebrate mtDNA clock (2% divergence per Myr; Brown et al. 1979) is much too long to have survived ancestral mtDNA polymorphisms (Avise, 1994). Accordingly, this explanation should be rejected.
Two alternative possible hypotheses exist satisfying the distinct mtDNA composition between the two groups. The first is that the two C. takatsuensis groups may have evolved independently from distinct lineages (Fig. 6. A). If this is correct, the two groups of C. takatsuensis must be regarded as two distinct species.
The second is that mtDNA introgression from another species to one of the C. takatsuensis groups may have occurred. The most parsimonious scenario under the introgression hypothesis is that mtDNA exchange between ancestors of Chugoku-Kyushu group C. takatsuensis and Western group C. biwae (W2 and W4) may have occurred (Fig. 6. B). In this scenario, two groups of C. takatsuensis can be considered as having evolved monophyletically. On the other hand, since C. matsubarai and C. sp. 1 were inferred as allotetraploid species whose maternal origins are C. biwae (Saitoh et al. 2000), it is reasonable that C. biwae is also considered as a monophyletic group. In fish species, interspecific mtDNA introgression, including a reciprocal case as in the present hypothesis, have been reported in several species (Avise and Saunders, 1984; Billington and Herbert, 1991; Dowling and Hoeh,1991; Dowling and DeMarais, 1993; Bernatchez et al., 1995, etc.). Moreover, evolutions originated from interspecific hybridization, being highly suggestive the possibility of mtDNA introgressions, had reported among several Cobitis species (Vasil'ev et al., 1989; Kim and Lee, 1990; Sezaki et al., 1994; Ráb and Slavík 1996; Saitoh et al. 2000). Published karyological data (Table 5) indicates that C. takatsuensis exhibits little karyotypic variations even between the two divergent groups (Kimizuka et al., 1982). Although C. biwae shows some variations, their karyotypes are characterized by unusual compositions among Cobitis species (high proportion of meta-centric chromosomes; Ueno et al., 1980; Ueno, 1981). The above seems to support the monophyletic evolutions of C. takatsuensis and C. biwae lineages, respectively.
Brief comparison of published data of karyotypes of C. takatsuensis and C. biwae.
However, it is impossible to distinguish whether parallel evolution or mtDNA introgression (or a combination of both) is responsible for the distinct mtDNA compositions of the two groups of C. takatsuensis, based solely upon mtDNA data. The two above hypotheses are presently the subject of ongoing nucleic DNA analyses.
We thank I. Oohara (National Research Institute of Fisheries Science), K. Kawamura (National Research Institute of Aquaculture), S. Kimura (Fisheries Research Laboratory of Mie University), and K. Saitoh (Tohoku National Fisheries Research Institute) for variable advice and assistance. We are grateful to G. S. Hardy (Thames, New Zealand) and two anonymous reviewers for their critically reading and improving the manuscript. We also thank Y. Fujioka (Shiga Prefectural Fisheries Experimental Station), I. Tanaka (Masuda City, Shimane Prefecture), D. Inoue (Kitakyushu High School) and N. Uechi (Mie University) for providing specimens used in the present study.