Gene products of 22 protein coding loci from 29 populations of a uniquely sea-going cyprinid genus Tribolodon, T. hakonensis (anadromous and/or river-resident), T. brandtii (anadromous), T. ezoe (freshwater) and T. nakamurai (freshwater), were analyzed in order to infer the genetic divergence and dispersal of Tribolodon species around the Sea of Japan. According to the genetic distance, T. brandtii is located near the center between T. hakonensis and T. ezoe-T. nakamurai cluster. The outgroup root (Pseudaspius leptocephalus, Rhynchocypris lagowskii and Leuciscus waleckii) is attached between T. hakonensis-T. brandtii and the T. ezoe-T. nakamurai clusters. The origin of the genus Tribolodon and its salinity tolerance may date back to the Miocene Sea-of-Japan Lake and the Pliocene Paleo-Sea-of-Japan, respectively, as judged from the divergence time between Tribolodon and the outgroups. The genetic relationships and distribution pattern of Tribolodon species suggest they speciated on the Sakhalin-Japan side, and the continental populations of T. hakonensis and T. brandtii originated from anadromous colonizers. A new faunal element of Japanese freshwater fishes, the Sakhalin-Japan sub-element, is proposed for Tribolodon and some other fishes.
Fishes have currently the highest species diversity among vertebrates (about 30000 species), suggesting that they have experienced the most frequent speciation events (Nelson, 1994). Freshwater fishes such as cyprinids, cobitids and bagrids are among the ideal animal groups for research on speciation, because of the restricted gene flow between geographically isolated populations, which facilitates the evolution of reproductive isolation. The areas abutting the Sea of Japan have been thought to form one of such speciation center for freshwater as well as marine fishes throughout the Tertiary to Quaternary (e.g. Lindberg, 1972; Nishimura, 1974; Nakajima, 1986, 1987).
The Far Eastern daces, Tribolodon, comprise a unique group in the large family Cyprinidae not only because of their variation in life-history type from freshwater to anadromous mode of life, but also because of their geographic restriction to areas around the Sea of Japan (Nakamura, 1969; Nishimura, 1974). Therefore, the derivation of the genus, especially its anadromy, is thought to have involved the history of the Sea of Japan (Nishimura, 1974).
Tribolodon includes four species, “Ugui” T. hakonensis (Günter), “Maruta” T. brandtii (Dybowsky), “Ezo-ugui” T. ezoe Okada and Ikeda and “Ukekuchi-ugui” T. nakamurai Doi and Shinzawa. Tribolodon hakonensis populations exhibit freshwater as well as anadromous modes of life, T. brandtii has an anadromous life history, T. ezoe lives entirely freshwater life, and T. nakamurai probably has a freshwater life history (Nakamura, 1969; Kurawaka, 1977; Sakai, 1995).
Tribolodon hakonensis is the most widely distributed species, covering nearly the entire Tribolodon range around the Sea of Japan, but the freshwater form scarcely occurs in the continental side (Uchida, 1939). On the other hand, T. brandtii is absent from the southern half of Japan, and T. ezoe is restricted to the northern half of the Sakhalin-Japan part of the range. Tribolodon nakamurai is restricted to a narrow part of the southernmost area of the Tohoku District, Honshu Island, Japan, and it is syntopic with all three other species (Nakamura, 1969; Kurawaka, 1977; Sakai, 1995).
Due to the above context, it is very interesting and rewarding to probe into the process of their speciation involving an acquisition of anadromy. Hanzawa and Taniguchi (1982) and Sakai (1995) analyzed allozymes of the four species electrophoretically and suggested closer genetic relationships among the three of the species to the exclusion of T. hakonensis. However, their samples were obtained from restricted localities in Japan and no outgroup comparison was made in their analysis. Kartavtsev et al. (2001) compared allozymes between Tribolodon and two allied genera, but still surveyed only a small number of populations. Therefore, the results unsatisfactorily address the Tribolodon divergence and/or speciation problem.
In the present study, the allozyme genetic relationships of the four species of Tribolodon, with some outgroups, were studied based on populations from nearly the entire range, including Sakhalin, the Russian Maritime Territory, the Korean Peninsula and Japan, and derivation and speciation of Tribolodon along with genetic divergence are discussed.
MATERIALS AND METHODS
A total of 638 individuals from 29 populations of Tribolodon were surveyed by allozyme analysis (Fig. 1, Table 1). Two individuals of “Manshu-uo” Pseudaspius leptocephalus, 12 individuals of “Abura-haya” Rhynchocypris lagowskii and 5 individuals of “Kitanougui” Leuciscus waleckii were also examined for outgroup comparison in cluster analysis, because the pharyngeal teeth morphology of Amurian Pseudaspius has been suggested to be similar to that of Tribolodon (Tomoda, 1984, 1998), Far Eastern Rhynchocypris has been determined morphologically to be a sister genus of Tribolodon, and Eurasian Leuciscus to be an outgroup of the Tribolodon-Rhynchocypris sister clade (Cavender and Coburn, 1992; Coburn and Cavender, 1992). Among them, Leuciscus was set as the outermost-group because it was most distantly related to Tribolodon genetically (see the result).
Twenty-nine Tribolodon populations, and Pseudaspius leptocephalus, Rhynchocypris lagowskii and Leuciscus waleckii populations sueveyed
The fish were frozen immediately after collection and stored at −70°C until processed for starch-gel electrophresis (12% gel). The gene products of 22 allozyme or protein coding loci were investigated: aspartate aminotransferase (E.C. 184.108.40.206, Aat-1*, Aat-2*), alchohol dehydrogenase (E.C. 1.1.1, Adh*), fumarate hydratase (E.C. 220.127.116.11, Fh*), glycerol-6-phosphate isomerase (E.C. 18.104.22.168, Gpi-1*, Gpi-2*), L-iditol dehydrogenase (E.C. 22.214.171.124, Iddh*), isocitrate dehydrogenase (E.C. 126.96.36.199, Idhp-2*, Idhp-3*), L-lactate dehydrogenase (E.C. 188.8.131.52, Ldh-1*, Ldh-2*, Ldh-3*), malate dehydrogenase (E.C. 184.108.40.206, Mdh-1*, Mdh-2*, Mdh-3*), phosphogluconate dehydrogenase (E.C. 220.127.116.11, Pgdh*), phosphoglucomutase (E.C. 18.104.22.168, Pgm*), superoxide dismutase (E.C. 22.214.171.124, Sod*), xanthine dehydrogenase (E.C. 126.96.36.199, Xdh*) and general protein (Prot-2*, Prot-3*). Locus and gene nomenclature follows Shaklee et al. (1990). The most common allele at a locus of T. hakonensis is designated as *100, the others being named according to their relative mobility to the allele *100.
A neighbor-joining (NJ) dendrogram (Saito and Nei, 1987) based on Nei's (1972) genetic distance (D) between pairs of all populations was constructed with 1000 bootstrap replications using the computer package PHYLIP (version 3.57c, Felsenstein, 1995). Inter- and intra-generic divergence time was calculated with three calibrations; 1D=5 million years (Myr) (traditional one; Nei, 1975), 1D=12.5 Myr (for Tanganyikan cichlids; Nishida, 1997) and 1D=19.3 Myr (for Greek Leuciscus, Cyprinidae; Doadrio and Carmona, 1998).
Allelic frequencies are presented in the Appendix, all populations in all polymorphic loci (most common allele does not exceed 0.95) exhibiting no significant departure from the Hardy-Weinberg expectation.
Tribolodon hakonensis (at Pgm* and Prot-2*), T. ezoe (at Ldh-2* and Pgm*), and T. nakamurai (at Idhp-2*) have their own characteristic alleles. On the other hand, T. brandtii has no such diagnostic locus, at least one allele being common to another species at every locus.
Mean D with standard deviation and numbers of diagnostic loci between pairs of species are shown in Table 2. The distances between Tribolodon species and the out-groups, P. leptocephalus, R. lagowskii and L. waleckii, are D=0.450–0.895 (alleles being displaced at 6–11 loci), 0.782–1.087 (12–14 loci) and 1.347–1.541 (14–15 loci), respectively. Among Tribolodon species, all species other than T. brandtii are closely related to T. brandtii at similar distances (D=0.236–0.338, alleles being displaced at 3–4 loci). Tribolodon hakonensis is distantly related to the other species except for T. brandtii (D=0.464–0.509, alleles being displaced at 6 loci), the distances being larger than that between T. ezoe and P. leptocephalus (D=0.450).
Number of diagnostic loci (above diagonal) and inter-specific Nei's (1972) genetic distance (mean±SD) (below diagonal) between pairs of Tribolodon hakonensis, T. brandtii, T. ezoe, T. nakamurai, Pseudaspius leptocephalus, Rhynchocypris lagowskii and Leuciscus waleckii, and intra-specific genetic distance (on diagonal)
The relationships among them are exhibited in the NJ dendrogram (Fig. 2). Bootstrap probabilities (p) of each Tribolodon species cluster other than T. brandtii (p=36.3%) are fairly high (p=67.6–98.7%). Tribolodon hakonensis with T. brandtii (p=49.4%), and T. ezoe with T. nakamurai (p=40.1%) form two pairs of neighbors and the two clusters are rooted by the outgroups (p=68.9%). Tribolodon cluster is connected first with P. leptocephalus (p=67.6%).
Populations of T. hakonensis from Japan, Sakhalin and the northern Maritime Territory (the Tumnin River population) are connected in a cluster (p=55.8%, the locus Prot-2* is fixed for allele *100), populations from the southern Maritime Territory and Korean Peninsula form another cluster (p=56.9%, Prot-2* is fixed for *85), and the Tsushima population is intermediate between them (both alleles exist in Prot-2*) (Figs. 2 and 3).
Hokkaido, Yamagata and Niigata populations of T. brandtii are fixed for the same allele *86 at Gpi-1* with T. ezoe, and form a cluster with the Sakhalin population, which also possesses the allele *86 (p=51.5%, Figs. 2 and 4). The other populations have allele *100 in common with T. hakonensis at Gpi-1*, and form a cluster (p=31.9%, Figs. 2 and 4).
Tribolodon ezoe are roughly divided into two groups, Tumnin, Sakhalin and Hokkaido cluster (p=62.2%) and the Tohoku District populations (Fig. 2). Tohoku populations from Yamagata and Akita Prefectures are fixed or nearly fixed for the allele *–100 at the locus Gpi-2* (Fig. 5).
Mean inter-generic distances, D=0.670 (vs. Pseudaspius), D=0.964 (vs. Rhynchocypris) and D=1.440 (vs. Leuciscus), are calculated to be about 3.4, 4.8 and 7.2 Myr ago by Nei's (1975), 8.4, 12.0 and 18.0 Myr ago by Nishida's (1997), or 12.9, 18.6 and 27.8 Myr ago by Doadrio and Carmona's (1998) calibrations, respectively. Mean inter-specific distance among Tribolodon species (D=0.347) is about 1.7 Myr (Nei, 1975), 4.3 Myr (Nishida, 1997) or 6.7 Myr ago (Doadrio and Carmona, 1998).
In the resulting dendrogram (Fig. 2), Tribolodon brandtii is located near the center between T. hakonensis and the T. ezoe-T. nakamurai cluster with very short branch length. According to Sakai and Hamada (1985) and Sakai (1995), T. hakonensis and T. ezoe frequently hybridize with T. brandtii in Hokkaido (10–20% of wild offspring are hybrids), usually T. brandtii being the mother parent in these hybridization events (85%), and most of the hybrid offspring back-cross also with T. brandtii. In such hybridization, allozyme alleles may have introgressed from T. hakonensis and/or T. ezoe to T. brandtii, and mtDNA introgression may be less likely than allozyme allelic introgression. For example, some Gpi-1* alleles of T. brandtii are assumed to have introgressed from T. hakonensis and/or T. ezoe as discussed later. Tribolodon brandtii does not have its any own diagnostic loci, but has at least one allele in common with other species at every examined locus (Appendix). The allozyme allelic variation in T. brandtii may be due to the above-mentioned introgression. Such introgression may cause the genetic intermediacy of T. brandtii in the genus, may make the evolutionary rate of allozyme genes of T. brandtii slower, and may shorten the branch length of the T. brandtii cluster.
In the dendrogram, the outgroups are all freshwater dwellers, and Tribolodon is divided into two clusters, freshwater (T. ezoe-T. nakamurai) and anadromous (T. brandtii-T. hakonensis) clusters. Therefore, it seems reasonable to infer that the anadromous Tribolodon species had been derived from the genus founder which had the same primitive non-anadromous life history as the outgroups, the first anadromous species probably having two modes of life, anadromous and river-resident, like T. hakonensis.
Genetic similarity between Tribolodon and Pseudaspius
Pseudaspius leptocephalus is the nearest neighbor of Tribolodon in the dendrogram. Kartavtsev et al. (2001) have reported a similar result. Genetic distances between them, D=0.450–0.895 (mean 0.670), correspond to the inter-generic level applied to American cyprinids, D=0.11–1.35 (mean 0.73) (Buth et al., 1991). However, the genetic distance between P. leptocephalus and T. ezoe (D=0.450) is smaller than those between T. hakonensis and T. ezoe or T. nakamurai, fitting also into the inter-specific range determined for American cyprinid genera, D=0.04–0.97 (mean 0.20) (Buth et al., 1991). Taking the similarity of pharyngeal teeth morphology between Tribolodon and Pseudaspius (Tomoda, 1984, 1998) into consideration, it is highly probable that Pseudaspius is the nearest sister group of Tribolodon. Further analyses on morphology and DNA should be performed.
Intra-specific differentiation of Tribolodon hakonensis
In Tribolodon hakonensis, only the anadromous form was found in the Maritime Territory and Korean Peninsula before (Uchida, 1939; Gavrenkov and Ivankov, 1979; Dai et al., 1982; Jeon and Sakai, 1984). This presents a striking contrast with Western Japan, neighboring the Korean Peninsula, where the river-resident form prevails (Nakamura, 1969). The Samchokoship River population from southern Korea examined in the present study, however, is thought not to have been exposed to seawater based on its otolith Sr/Ca analysis (unpubl. data). The Tumnin River population from the northern Maritime Territory may also be river-resident because the collection site is located more than 100 km from the river mouth and some semi-adult fish larger than 20 cm SL were collected in non-spawning season. These may be the first records of the river-resident form of T. hakonensis on the continental side of its range.
The Korean river-resident population of T. hakonensis does not share the Prot-2*100 allele with the river-resident Tumnin and Japanese populations, but does the same allele (*85) with the anadromous Vladivostok population (Fig. 3). This suggests that the populations from the southern Maritime Territory to Korea originated in the anadromous colonizer, being different from Sakhalin-Japan side populations (plus population from northern Maritime Territory), and the Korean river-resident population was secondarily landlocked in the southernmost range.
The Tsushima population of T. hakonensis has both Prot-2* alleles (*85 and *100), probably caused by allelic introgression between the Japanese and continental populations. No further introgression is seen beyond the Tsushima population, indicating that the secondary population contact between the two regions might have occurred recently (maybe in the last glacial epoch), and the land connection or freshwater habitat connection at that time was so restricted as to minimize introgression between the two regions even in the anadromous T. hakonensis. In fact, based on geological evidences, Ohshima (1991) insisted there was no land connection between Japan and Korea in the last glacial epoch.
Intra-specific differentiation of Tribolodon brandtii
Japanese Tribolodon brandtii has been divided into two types, “Maruta” type from Tokyo Bay and “Jusan-ugui” type from Aomori Prefecture, based on their mean lateral-line scale number, 83 or less (mean 80, range 76–83) for the former and 83 or more (mean 88, range 83 -95) for the latter (Nakamura, 1969). Although both live an anadromous life, the “Jusan-ugui” type is slightly less tolerant to salinity than the “Maruta” type (Nakamura, 1969). The reported lateral-line scale numbers indicate that the Sendai Bay population (Iwate population in the present study) is included in the “Maruta” type (Zama, 1999), and the continental populations (Gavrenkov and Ivankov, 1979; Dai et al., 1982; Jeon and Sakai, 1984), Sakhalin populations (Gritsenko, 1974), Hokkaido populations (Sakai and Hamada, 1985) and Sea of Japan populations from the Tohoku District (Onodera and Honma, 1976) all fall in the “Jusan-ugui” type.
However, in the present study (Fig. 2), the continental populations (“Jusan-ugui” type) form a cluster with the Iwate and Tokyo populations (“Maruta” type), not with the Sakhalin, Hokkaido and Tohoku populations (“Jusan-ugui” type). This is due to Gpi-1*100 (in common with T. hakonensis) and *110 alleles (in common with T. ezoe) (Appendix, Fig. 4) connecting the continental, Iwate and Tokyo populations and the Sakhalin, Hokkaido and Tohoku populations into two different clusters, respectively (Fig. 2). As already mentioned, some male T. hakonensis and T. ezoe individuals hybridize with female T. brandtii in Hokkaido (Sakai and Hamada, 1985; Sakai, 1995). Through such hybridization events, inter-specific allelic introgression of Gpi-1* between T. brandtii and T. hakonensis or T. ezoe might have caused the complex relationships. The fact that the Sakhalin, Hokkaido and Tohoku populations of T. brandtii exhibit a similar distribution range to T. ezoe (Figs. 4 and 5) may also support the introgression.
In any way, taking the morphological and physiological differences (Nakamura, 1969) into consideration, the “Maruta” and “Jusan-ugui” types must be two natural intra-specific entities of T. brandtii. Mitochondrial DNA data (Sanada et al., in prep.) also support the existence of these two types. The continental populations and the Sakhalin, Hokkaido and Tohoku populations probably represent two colonizers of the “Jusan-ugui” type along the continental coast and along the Sakhalin-Japan coast, respectively. The divergence of the “Maruta” and “Jusan-ugui” types would have occurred geographically between the Pacific and Seaof-Japan slopes at an earlier time than the above-mentioned “Jusan-ugui” dispersals.
Intra-specific differentiation of Tribolodonezoe and T. nakamurai
The two recognized groups of Tribolodon ezoe perhaps resulted from a loss in genetic diversity in southern populations, especially in Gpi-2* (Fig. 5). Divergence of T. nakamurai between the Akita and Yamagata populations is unclear because only one individual represented the Akita sample.
Time of Tribolodon derivation
Lindberg (1972) and Nishimura (1974) hypothesized that Tribolodon, which is distributed around the Sea of Japan, originated in the middle to late Pliocene at a hypothesized freshwater lake, the Lake Sea-of-Japan, and acquired their salinity tolerance in the following epoch when sea water began to enter the Lake Sea-of-Japan. However, geological studies, such as lithologic and planktonic foraminiferal studies of piston cores, do not support the existence of such a freshwater lake in the Pliocene (Taira, 1977; Ujiie and Ichikura, 1977; Kaseno, 1989). Reliable evidence of such a freshwater lake dates back to the early Miocene or older time (e.g. Kaseno, 1989; Fujita, 1990; Taira, 1990). The early Miocene lake is supposed to have been a center of cyprinid diversity (Tomoda et al., 1977; Nakajima, 1986, 1987), and Tribolodon has possibly originated there.
In the present study, the divergence time between Tribolodon and the outgroups are estimated as 3.4–27.8 Myr ago, but only the estimations adopting the time scale for the divergence data of Leuciscus (Doadrio and Carmona, 1998), one of the outgroups used in the present study, can date back the early Miocene, 12.9 (vs. Pseudaspius), 18.6 (vs. Rhynchocypris) or 27.8 (vs. Leuciscus) Myr ago. Accordingly, it is hypothesized that the Tribolodon ancestor itself or a common ancestor of Far Eastern Tribolodon and Amurian Pseudaspius or Far Eastern Rhynchocypris originated in the early Miocene Sea-of-Japan Lake, after which Tribolodon derived and acquired salinity tolerance in the middle to late Miocene Sea of Japan, having overcome the entry of seawater.
Place of Tribolodon speciation
Adopting the cyprinid calibration (Doadrio and Carmona, 1998), the mean divergence time among Tribolodon species is calculated to be 6.7 (range 4.7–10.0) Myr ago, when the Paleo-Sea-of-Japan is thought to have been a semi-closed cool sea (Kaseno, 1989). Therefore, the genus might have speciated through the late Miocene to Pliocene, and intra-specific entities might have derived in the Pleistocene with repeating inter-glacial and glacial epochs. Early Pleistocene Tribolodon fossils from Tochigi Prefecture, Japan (Uyeno, 1967) do not contradict this hypothesis.
Although a river-resident T. hakonensis population is found in Korea, all of the remaining river-resident populations are distributed in the Sakhalin-Japan side of the range including the northern Maritime Territory facing to the Tatar Strait. Freshwater T. ezoe and T. nakamurai are also restricted to the same side. Tribolodon hakonensis from the southern Maritime Territory to Korea are inferred to have originated from the anadromous colonizer. Continental T. brandtii is also thought to represent a colonizer along the continental coast after the “Maruta” vs. “Jusan-ugui” divergence. All these things suggest that most of Tribolodon speciation events occurred on the Sakhalin-Japan side.
In focusing on the origin of the freshwater fish fauna of the Japanese Archipelago, many authors have included Tribolodon among so-called Siberian element (e.g. Aoyagi, 1957; Nishimura, 1974; Goto et al., 1978; Goto and Nakano, 1993). However, this designation is probably incorrect for the above-mentioned reasons as well as the fact that no Tribolodon species inhabit the Amur River except for anadromous T. hakonensis and T. brandtii entering the river mouth (Berg, 1948–1949).
Therefore, it might be better to call Tribolodon a “Sakhalin-Japan sub-element”. There also exist some fish groups present on the Sakhalin-Japan side but absent from Siberia (except for the region facing to the Tatar Strait in some species), which should constitute the same sub-elements: “Itou” Hucho perryi, “Ezo-tomiyo” Pugitius tymensis, “Hana-kajika” Cottus nozawae, “Ezo-hana-kajika” C. amblystomopsis, etc. (unpubl. data).
Aoyagi (1957) and Goto et al. (1978) stated that “Yachiugui” Rhynchocypris percnurus sakhalinensis, “Fuku-dojou” Noemacheilus barbatulus toni and “Ezo-hotoke” Lefua nikkonis typify the freshwater ichthyofauna of Hokkaido. These three each have sister species or subspecies in Siberia, and represent true Siberian element. The Sakhalin-Japan sub-element proposed above would also typify northern Japan freshwater ichthyofauna with the real Siberian element.
We thank Dr. K. Watanabe, Nara Women's University, for his invaluable discussion and Dr. M. D. Ohto for English correction. The following persons helped us in collecting materials; H.-K. Byeon, K. Fujita, K. Hashimoto, Y. Hida, C. Imai, Y. Ito, H. Kato, K. Katsura, H. Kohno, H. Mizutani, M. Motegi, S. Nakano, N. Okabe, D. L. Pitruk, S. N. Safronov, H. Sato, H. Sugiura, H. Sugiyama, K. Takata, S. Takeyama, V. V. Tsuiger, Y. Yamazaki, A. Zama, and F. Zolotukhin. This study was supported in part by Grant-in-Aid for Oversea Scientific Survey No. 06041004 and 09041138 from the Ministry of Education, Science, Sports and Culture, Japan.