Allozyme analysis for 41 populations of brown frog species, Rana dybowskii, R. huanrenensis, and R. amurensis from Korea and three reference species (Chinese R. chensinensis and Japanese R. dybowskii and R. tsushimensis), were performed to clarify taxonomic status of Korean brown frogs. The level of average genetic differentiation (Nei's D) among local populations of each species in Korea was very low (D<0.012) and Korean and Japanese R. dybowskii also showed conspecific level of differentiation (D=0.070). Whereas, much larger, discrete genetic differences were detected in the interspecific comparisons (D>0.370). In the genetic relationships among five species examined, the 24 chromosome brown frogs (R. dybowskii, R. huanrenensis, and R. chensinensis) did not form a monophyletic group. Rana dybowskii with the chromosome number of 2n=24 was grouped together with R. amurensis with the chromosome number of 2n=26. The hypothesis of reversal change from 24 to 26 in Korean R. amurensis seems to better explain the phylogenetic relationships of east Asian brown frogs than the assumption of parallel reduction in chromosome number from 2n=26 to 24 in R. dybowskii and in the common ancestor of R. huanrenensis and R. chensinensis. The genetic, morphological, and reproductive divergences between Korean R. dybowskii and R. huanrenensis were compared.
The Eurasian brown frogs are a morphologically conservative assemblage consisting of the Eurasian Rana temporaria and a large number of similar species considered to be related (Frost, 1985; Borkin and Kuzmin, 1988; Green and Borkin, 1993; Nishioka et al., 1992; Maeda and Matsui, 1999). The chromosome number of great majority of Rana species is 26 and most of brown frog species have the same number. Some of brown frogs, however, are unique in having diploid chromosomes of 2n=24 (Matsui, 1991; Green and Borkin, 1993; Xie et al., 1995). These 24 chromosome brown frogs include the European R. arvalis and several east Asian species allied to R. chensinensis, such as R. dybowskii, R. ornativentris, R. pirica, and R. huanrenensis (Kobayashi, 1962; Seto, 1965; Wu, 1982; Green, 1983; Luo and Li, 1985; Lee and Park, 1986; Ma, 1987; Wei et al, 1990; Liu et al., 1993; Green and Borkin, 1993; Xie et al., 1995; Lee and Lee, 1998). These east Asian brown frogs are quite similar in morphology, and are very difficult to identify (Nakamura and Ueno, 1963; Matsui et al., 1993, 1998; Xie et al., 1995; Yang et al., 2000). Indeed, most of them were originally described on the basis of slight morphological differences. Recently, taxonomic status of each species was made clearer by lines of additional information, such as considerable genetic divergences among them (Matsui, 1991; Green and Borkin, 1993; Tanaka-Ueno et al., 1998; Matsui et al., 1998; Kim et al., 1999; Yang et al., 2000). Although extensively studied in the laboratory (Kawamura et al., 1981), the direct evidence of reproductive isolation in the field among these allied species have never been reported because of their geographic isolation due to allopatric distribution.
Until recently, it has been reported that R. dybowskii (a 24 chromosome member) and R. amurensis coreana (a 26 chromosome member) are distributed in South Korea (Yang and Yu, 1978; Sengoku, 1979; Green and Borkin, 1993; Matsui et al, 1998). Most recently, we (Yang et al., 2000) reported a new Korean brown frog member (R. huanrenensis Fei, Ye and Huang, 1990) which was morphologically and karyologically (2n=24) very similar to R. dybowskii.
In this study, we investigate the degree of inter- and intraspecific genetic variation and to clarify the genetic relationships among three species of Korean brown frogs. For comparisons, Japanese R. dybowskii and R. tsushimensis and Chinese R. chensinensis are also incorporated to the analysis. In addition, we surveyed the levels of morphological, genetic, and reproductive divergence between sympatricsamples of R. dybowskii and R. huanrenensis from South Korea.
MATERIALS AND METHODS
Collection and field notes
Five brown frogs were collected from 41 localities in Korea, Japan, and China (Table 1). During most collecting trips in Korea and Japan, the notes and photographs on color pattern of each specimen and records of each breeding site were taken.
Collection localities, collection dates, and sample sizes (N) for electrophoretic and morphological analyses of Rana huanrenensis, R. dybowskii, R. chensinensis, R. amurensis, and R. tsushimensis from Korea, Japan, and China.
For an electrophoretic examination, a total of 849 specimens belonging to 41 populations of five species were employed. These include 10 populations of Rana huanrenensis from South Korea, 17 populations of R. dybowskii from South Korea and Japan, 2 populations of R. chensinensis from China, 11 populations of R. amurensis from South Korea, and 1 population of R. tsushimensis from Japan (Table 1).
Live samples were transported to the laboratory and were stored at −70°C until use. In the laboratory, the tissues of liver, heart and skeletal muscle were removed from each specimens and homogenized by glass homogenizer in an equal volume of distilled water and were centrifugated at 18,000 rpm for 30 min at 4°C to obtain the supernatant for electrophoresis. Voucher specimens were fixed in 10% formalin, preserved in 70% ethanol, and deposited in Yang's collection at Inha University. The supernatant was subjected to horizontal starch-gel (12%) electrophoresis and histochemical staining procedures (Yang et al., 1997: Appendix I). Multiple loci were numbered sequentially, and alleles were designated alphabetically with “a” being the fastest migrant.
Individual genotypes were used to calculate allele frequencies for each population, these in turn were used to calculate matrices of genetic similarity (Rogers, 1972) and genetic distance (Nei, 1978). Three different methods were employed to infer relationships among populations. First, Nei's (1978) distance was clustered according to the UPGMA algorithm (Sneath and Sokal, 1973). Then, modified Rogers' distance (Wright, 1978) was analyzed by the Neighbor-joining (NJ) method (Saitou and Nei, 1987), and finally, we employed Felsenstein's (1993) DNAML procedure with allele frequencies for the maximum-likelihood (ML) analysis. These analyses were performed by use of BIOSYS-1 (Swofford and Selander, 1981) and PHYLIP vers. 3.5 C computer packages (Felsenstein, 1993).
Genetic variation and relationships among brown frogs
Genetic variation — By-products of 18 loci were scored from 13 enzymes and general proteins. Observed allelic frequencies are given in Appendix II.
Based on allelic frequencies listed in Appendix II, the degree of genetic variation of each population was estimated (Table 2). The genetic variability of R. dybowskii was P=26.0% (22.2–33.3%), Ho=0.118 (0.070–0.183), and He=0.122 (0.078–0.153). The genetic variabilities of R. huanrenensis and R. amurensis were P=22.2% (16.7–27.8%), Ho=0.063 (0.046–0.073), He=0.067 (0.058–0.081) and P=22.2% (5.6–33.3%), Ho=0.080 (0.029–0.120), He=0.086 (0.035–0.124), respectively. In Korean brown frogs, Kanseong population of R. dybowskii had the highest genetic variability (P=36.4%, Ho=0.165, He=0.165) while Koseong population of R. amurensis showed the lowest variability (P=9.1%, Ho=0.048, He=0.042). On the other hand, Chinese R. chensinensis, a reference species, showed P=22.3%, Ho=0.078, He=0.075. Another reference species Japanese R. tsushimensis, was more variable, with P=27.8%, Ho=0.094, He=0.097, than in Korean brown frog species.
Genetic variation of 41 populations in Rana huanrenensis, R. dybowskii, R. chensinen-sis, R. amurensis and R. tsushimensis from Korea, Japan, and China.
Genetic relationships — Based on allelic frequencies listed in Appendix II, average genetic similarities (Rogers' S) and distances (Nei's D) among populations of five brown frog species were calculated (Appendix III). In the Korean brown frogs, R. huanrenensis, R. dybowskii and R. amurensis, the degree of genetic differentiation within a species was small (D=0.034: Appendix III), but differentiations among these Korean brown frogs were very distinct (D=0.584 between R. huanrenensis and R. dybowskii, D=0.788 between R. huanrenensis and R. amurensis, and D=0.500 between R. dybowskii and R. amurensis) due mainly to Gp-4, aGpd, Mdh, and Ldh-1 loci that were ascertained as diagnostic among these Korean species.
When populations of 24 chromosome species from outside of Korea were included, genetic dissimilarities between R. huanrenensis (populations 1–10; Appendix III) and R. dybowskii (pops. 11–30) included fixed allelic differences at Gp-4, Mdh, and Iddh loci and diagnostic differences at the 95% confidence level (Ayala and Powell, 1972) at Ldh-1 locus. Rana huanrenensis and R. chensinensis (pops. 28 and 29) included fixed allelic difference at Ldh-1 locus and diagnostic differences at Iddh, Aat-1, and Acoh. Fixed allelic differences at Gp-4, Ldh-1, Mdh, and Iddh and diagnostic differences at Aat-1 were found between R. dybowskii and R. chensinensis (Appendix II). Among populations of three brown frog species with 24 chromosomes (Appendix III), average genetic distances among local populations of a single species were low (D=0.008 in R. huanrenensis, D=0.005 in Korean R. dybowskii, D=0.070 in Korean and Japanese R. dybowskii, D=0.053 in R. chensinensis), whereas the average genetic distances among three species were distinctly high (D=0.584 between R. huanrenensis and R. dybowskii, D=0.386 between R. huanrenensis and R. chensinensis, and D=0.485 between R. dybowskii and R. chensinensis).
Between Korean R. amurensis (pops. 30–40) and Japanese R. tsushimensis (pop. 41), both with 26 chromosomes, genetic dissimilarities included fixed allelic differences at Got-1, Gp-4, Idh, Sod, Ldh-1, and Ldh-2 loci and diagnostic differences (at the 95% confidence level) at Mdh, Pgm-1, and Pgm-2 loci (Appendix II). The average genetic differentiation among these two 26 chromosome species were distinctly high (mean D=0.935).
Fig. 1A shows the UPGMA tree based on Nei's unbiased genetic distance. Although the bootstrap support for most of the nodes, except for monophyly of each species (not shown in the figure), was weak, Rana tsushimensis exhibited the earliest divergence among all populations examined. The remaining populations were divided into two distinct groups; One group included R. huanrenensis and R. chensinensis, and the other included R. dybowskii and R. amurensis. Topologies of NJ (Fig. 1B) and ML (Fig. 1C) trees based on modified Rogers' distance and allele frequenicies, respectively, were similar to that of UPGMA tree in that R. tsushimensis first diverged and R. amurensis and R. dybowskii, and R. chensinensis and R. huanrenensis, respectively, formed a separate subcluster.
Comparisons between R. dybowskii and R. huanrenensis
Morphology — Intraspecific morphological variation was much less notable than interspecific one. Rana huanrenensis was morphologically very similar to R. dybowskii, but differs from the latter in the ventral color pattern (Fig. 2). In males, R. dybowskii had a milky white ground (Fig. 2C), whereas the ground color of male R. huanrenensis was yellowish gray with minute black dots densely distributed over the throat and chest (Fig. 2A). In the breeding season, females of R. huanrenensis had throat and chest covered with yellowish green (Fig. 2B), whereas in females of R. dybowskii, the red color patched over the throat and chest, which color turned to black patches in alcohol (Fig. 2D). In addition to these differences in coloration, male R. dybowskii had paired internal vocal sacs, while male R. huanrenensis lacked vocal sacs.
Protein electrophoresis — R. huanrenensis and R. dybowskii showed a discrete genetic difference (Nei's D=0.585: Appendix III) and no evidence of gene flow between these two species was found in the sympatric areas surveyed (Jangseong, Inje, Kapyeong, and Donghae; see Table 1, Appendix II).
Ecological notes — R. huanrenensis is sympatric with R. dybowskii in some parts of South Korea such as Tonghae, Inje, Jangseong, and Kapyeong (see Table 1), and therefore, ecological comparison of the two species is pertinent. R. dybowskii altitudinary ranges very wide, from plains to montane regions, where they breed in still waters in rice fields and small pools in early spring. On the other hand, R. huanrenensis occurs only at valley in relatively high montane regions, where the species spawn on the rocks in streams. Eggs of the species laid in relatively small and tightly clustered egg mass, and each egg mass is attached on the submerged rock in small streams in early spring (Table 3).
Morphological and ecological diagnostic characters between Rana huanrenensis and R. dybowskii in breeding season
The Eurasian brown frogs are very difficult to classify (Matsui, 1991; Green and Borkin, 1993; Tanaka-Ueno et al., 1998). Especially, members with 24 chromosomes are morphologically quite similar to each other and have a complicate taxonomic history, but now, taxonomic status of each member is made more clear than before by the presence of distinct genetic divergences among them (Matsui, 1991; Green and Borkin, 1993; Tanaka-Ueno et al., 1998; Matsui et al., 1998; Kim et al., 1999). It has long been known that the frogs with 24 chromosomes include several east Asian species allied to R. chensinensis, such as R. ornativentris, R. dybowskii, R. pirica. However, it has been known recently that R. huanrenensis, originally described from China (Fei et al., 1990), is also a member of this group (Xie et al., 1995) and co-occurs with R. dybowskii in South Korea (Yang et al, 2000). Before this finding, R. huanrenensis has been known only from the type locality, Huanren County, Liaoning Province, China for nearly 10 years. The significant range extention to Korea was recorded from localities that were well-known for the presence of R. dybowskii (Yang et al., 2000).
In South Korea, R. huanrenensis has been misidentified as R. dybowskii because of difficulties in identification. However, as shown in the present study, R. huanrenensis is actually well differentiated morphologically from R. dybowskii chiefly by the ventral color pattern. Moreover, males of these two species clearly different in the presence or absence of vocal sacs.
Since the separation of gene pools is the essence of species formation, a study of speciation must involve the examination of the level of reproductive isolation between the taxa compared. Allozymic analysis has been used extensively for such an examination at the zones of sympatry, and the contact zones of amphibian species that are problematic in taxonomic status (Wake et al, 1980; Yang and Park, 1988; Yang et al, 1988, 1997; Good, 1989). In our result, genetic divergence between R. huanrenensis and R. dybowskii included fixed allelic differences at Gp-4, Mdh, and Iddh loci, and these three loci are diagnostic genetic markers to identify them. No evidence of gene flow between these two species was found at the zone of sympatry. R. huanrenensis and R. dybowskii are completely isolated reproductively by their microhabitats, especially of the spawning site, and breeding habits. Particularly, the different condition of vocal sacs in males of the two species means the presence of clear differences of mating signals between them.
The east Asian brown frogs include two chromosomal groups (Kuramoto, 1979; Nishioka et al., 1986; Matsui, 1991; Green and Borkin, 1993). R. dybowskii, R. huanrenensis, and R. chensinensis have 2n=24 chromosomes, while R. amurensis and R. tsushimensis have 2n=26 (Lee and Park, 1986; Nishioka et al., 1986; Xie et al., 1995; Yang et al., 2000). It is generally believed that the fundamental chromosome number in Rana is 2n=26 (Morescalchi, 1973; Wilson et al., 1974; Kuramoto, 1979, 1989; Schmid, 1980; Green, 1983; Park, 1990). From the study of R. dybowskii, Green (1983) proposed that the karyotypes with 24-chromosomes could have arisen in east Asia, based on the location of secondary constrictions and chromosome bands. Meanwhile, from the banding patterns of Eurasian and North American brown frogs, Nishioka et al. (1986, 1987) similarly suggested the chromosome number reduction from 2n=26 to 2n=24. Chromosome evolution through reduction in number resulted from inversion/fusion has also been reported in other anuran species (King, 1990; Bogart and Tandy, 1981; Blommers-Schlosser, 1978). Considering this pattern of chromosome evolution as a single event, it could be presumed that the species with putative derived chromosome number (2n=24) form a monophyletic group. However, our results indicate that R. dybowskii with 2n=24 is genetically closer to R. amurensis with 2n=26 (D=0.500) than to R. huanrenensis (D=0.584) or to R. chensinensis (D=0.584) both with 24 chromosomes. Reflecting this situation, R. amurensis did not form a cluster, but was included in a cluster containing other brown frogs with 24 chromosomes in all the three trees we obtained.
These results imply that the interspecies relationships incidental to the chromosomal evolution are not in accordance with relationship inferred from genetic analyses. In view of our results, two assumptions of chromosomal evolution in brown frogs around Korea would emerge. One possibility is that the chromosome number reduction has evolved independently at least two times (parallel reduction in chromosome number from 2n=26 to 24). Namely, R. tsushimensis first differentiated from the common stock of brown frogs around Korea with 2n=26 chromosomes. Subsequently, through a reduction of primary chromosome number, divergence of an ancestor of the R. huanrenensis and R. chensinensis lineage (2n=24) occured from an ancestral species (2n=26) common to the Korean R. amurensis and R. dybowdkii lineage. Finally, speciation of R. dybowskii (2n=24) and Korean R. amurensis (2n=26) occurred while also accompanying a secondary chromo-some number reduction in the R. dybowskii lineage.
Another possibility is that the common ancestor of all these four species, after diverged from R. tsushimensis, reduced the chromosome number from 26 to 24 before the separation of the R. huanrenensis and R. chensinensis lineage and the Korean R. amurensis and R. dybowskii lineage. Subsequent speciation of the latter lineage would have included the reversal change in chromosome number from 24 to 26 in Korean R. amurensis.
It is yet to be surveyed which of these two assumptions is more probable, but the first assumption parallels with the idea proposed by Green and Borkin (1993) or Nishioka et al. (1992) that R. arvalis with 2n=24 chromosomes is paraphyletic with east Asian brown frogs having the same 2n=24 chromosomes. However, there are strong disagreements between Green and Borkin (1993) and Nishioka et al. (1992). Green and Borkin (1993) suggested parallel reduction to 2n=24 in European R. arvalis and all east Asian species including R. dybowskii, but according to Nishioka et al. (1992), all east Asian brown frogs with 26 chromosomes, excepting R. tsushimensis but including R. amurensis and even European R. temporaria, have that number as a result of reversal change in chromosome number from 24 to 26.
The second assumption more conforms to Green and Borkin (1993) or Tanaka-Ueno et al. (1998). These authors considered Japanese R. ornativentris, with 24 chromosomes, represents the sister group of other east Asian species having 24 chromosomes. Including R. ornativentris, “the parallel chromosome number reduction” hypothesis needs three steps (reductions in R. ornativentris, R. dybowskii, and the R. huanrenensis and R. chensinensis lineage), but “reversal change in Korean R. amurensis” requires only two steps (one reduction in the common ancestor of all species with 24 chromosomes and one reversal in Korean R. amurensis).
Moreover, later divergence of R. amurensis among east Asian brown frogs, suggested by our result and Nishioka et al. (1992), strongly contradicts to the idea proposed by Green and Borkin (1993) from allozyme analyses and by Tanaka-Ueno et al. (1998) from the analyses of mitochondrial DNA. Both of these reports suggested the earliest divergence of Russian R. amurensis among east Asian brown frogs. Interestingly, Korean and Russian R. amurensis exhibit different degree of genetic differentiation between R. dybowskii; The genetic differentiation between Korean R. dybowskii and R. amurensis we obtained in the present study (D=0.500) was intermediate between those reported between Korean R. dybowskii and Russian R. amurensis (D=0.874) by Green and Borkin (1993) and between R. dybowskii from Tsushima and R. amurensis from Mongolia, China, and Russia (D=0.304-0.311) reported by Nishioka et al. (1992).
These genetic inconsistencies of Korean and Russian R. amurensis suggest a distinct taxonomic status of each population. In order to clarify the problem of chromosome number change, as well as the relationships of local populations of R. amurensis, more extensive studies including many more taxa from regions surrounding Korea are strongly required.
We thank two anonymous reviewers for improving an early version of the manuscript. This research was supported by a grant from Korean Ministry of Education (BSRI-97-4423). We thank Dr. T. J. Papenfuss (University of California, Berkely) for help in examining material and Prof. H. Y. Lee, Dr. J. H. Suh, Miss Y. J. Kang, Mr. D. E. Yang, Mr. H. Lee, and Dr. C. H. Jeong (Inha University, South Korea) for help in collecting specimens.
Buffer systems and enzymes for the analysis of horizontal starch gel electrophoresis
Allele frequencies of 41 populations in Rana huanrenensis, R. dybowskii, R. chensinensis, R. amurensis and R. tsushimensis from Korea, Japan, and China.