Fifteen enzymes and two blood proteins encoded by 24 presumptive loci were analyzed using starch-gel electrophoresis in 136 frogs of 16 populations of Rana ornativentris and 21 frogs of a sympatric population of Rana japonica, in order to elucidate the degree of geographic divergence of R. ornativentris in Honshu and its genetic relationships to R. japonica. The UPGMA dendrogram constructed from Nei's genetic distances showed that R. ornativentris from Honshu was divided into two distinct groups, western and eastern, and that the latter split into three subgroups, southern, central and northern. Genetic divergence was distinct between western and eastern populations of R. ornativentris at three loci, PEP-A, SOD-1 and Hb-1, with the Fst values of Wright of 0.624, 0.635 and 0.876, respectively. The average value of Fst (Fst), excluding the five invariant loci, was 0.306. Nei's genetic distances among the four western populations of R. ornativentris were 0.015∼0.061, 0.043 on average. Those among the 12 eastern populations were 0.011∼0.179, 0.063 on average, whereas those between the four western and 12 eastern populations were 0.128∼0.313, 0.225 on average. The genetic distances between the 16 populations of R. ornativentris and one population of R. japonica were 0.579∼0.956, 0.793 on average. The UPGMA dendrogram showed that R. ornativentris was distinctly separated from R. japonica.
INTRODUCTION
Rana ornativentris was first described as Rana japonica var. ornativentris by Werner (1904). Stejneger (1907) placed this taxon in the synonymy of R. japonica and Okada and Kawano (1923) described this subspecies as R. temporaria var. montana. Subsequently the name R. temporaria ornativentris was given by Stejneger (1924) and Okada (1931) adopted this combination. Kawamura (1962) elevated this subspecies as a valid species on the basis of its morphology and reproductive isolating mechanism and named it Rana ornativentris Werner, which was followed by Okada (1966) and Nakamura and Uéno (1972).
This mountain brown frog usually inhabits mountain regions of Honshu, Shikoku and Kyushu in Japan, and is occasionally found in sympatry with the related brown frog R. japonica at lower altitudes. Intraspecific divergence in R. ornativentris has not been studied extensively, whereas that of R. japonica has been examined reproductively, karyologically and biochemically by Sumida (1981,1994,1996) and Sumida and Nishioka (1991, 1994). The reproductive isolating mechanisms between these two species were studied by Kawamura (1950) and Kawamura et al. (1981), and a comparative study on the karyotypes of these two species and other related brown frog species was reported by Nishioka et al. (1987b).
This study was carried out to investigate the degree of geographic divergence in Rana ornativentris and the genetic relationships between R. ornativentris and R. japonica.
MATERIALS AND METHODS
A total of 136 adult Rana ornativentris, consisting of 38 females and 98 males, from 16 geographic locations throughout Honshu and 21 adult R. japonica, consisting of five females and 16 males, from Saiki-cho of Hiroshima Prefecture were used in this study (Table 1). Fifteen enzymes and two blood proteins were analyzed with horizontal starch-gel electrophoresis (Table 2). The details of the electrophoretic method have been reported by Nishioka et al. (1980). The detection of each enzyme was carried out by means of the agar-overlay method outlined by Harris and Hopkinson (1976). The detection of blood proteins was carried out by the amido-black staining method. Multiple loci except LDH were numbered so that the most anodal was designated “1”. The LDH loci were lettered because vertebrate homology was known. The electrophoretic bands corresponding to multiple alleles at each locus were named A, B, C, etc. in the order of mobility from fast to slow with A being fastest, and the alleles were indicated by a, b, c, etc.
Table 1.
Specimens used in the present study
Table 2.
Enzymes and blood proteins analyzed in the present study
The fixation index (Fst) of Wright (1978) was utilized as a standard to indicate the degree of genetic divergence found at a locus among local populations. When multiple alleles existed in a frequency of more than 1% at a locus, this locus was regarded as polymorphic. In order to quantitatively show the genetic variation in local populations, the mean proportions of heterozygous loci per individual, mean proportions of polymorphic loci per population and mean number of alleles per locus were calculated (Lewontin and Hubby, 1966; Lewontin, 1974). The genetic relationships among local populations were evaluated by calculating Nei's genetic distances (D) (Nei, 1972). The phenetic relationships among these local populations were conjectured by seven methods, the unweighted pair-group arithmetic average (UPGMA) clustering method, furthest neighbor method, flexible method, centroid method, median method, nearest neighbor method and Ward method (Sneath and Sokal, 1973; Nei, 1975, 1987) on the basis of genetic distances (D).
RESULTS
Electrophoretic patterns and allelomorphs
The electrophoretic patterns showed that the enzymes and blood proteins were controlled by the genes at 24 presumptive loch (Table 3, Fig. 1). ADA, MPI, PGM and Alb were monomeric and heterozygotes showed double-banded patterns. AAT, GPD, GPI, IDH, MDH, PEP-A and SOD were dimeric and heterozygotes showed triple-banded patterns. FH, LDH and ME were tetrameric and heterozygotes showed five-banded patterns. Hb also had a tetrameric structure, although one heterozygote showed the pattern of a monomer. AAT, IDH, LDH, MDH, ME, SOD and Hb were each coded by two separate genetic loci. The LDH isozymes of different loci produced several hybrid bands, although some of them were faint or missing. Atypical patterns in heterozygotes were observed at the IDH-2 and MPI loci at the expected relative intensity of bands. Several bands produced from probably post-translational modification were observed at the MDH-2 and PGM loci (Fig. 1).
Table 3.
Number and kind of alleles and phenotypes at 24 loci in 16 populations of Rana ornativentris and one population of R. japonica
Fig. 1.
Electrophoretic patterns of 13 enzymes and two blood proteins in 16 populations of Rana omativentris and one population of R. japonica. The genotypes are represented by aa, bb, cc, etc.
Three of the 24 loci (AK, CK and LDH-A) were invariant. The MPI locus was the most polymorphic and 27 phenotypes were produced by 12 alleles. At the other 20 loci, there were two to 13 phenotypes produced by two to six alleles (Table 3).
Frequencies of phenotypes and alleles
The numbers of individuals exhibiting each scored phenotype are shown in Table 4 as raw data according to Buth (1984). The allele frequencies at all variable loci are presented in Table 5. At four loci (AAT-2, GPD, IDH-2 and PGM) a single allele predominated in all populations, including the 16 populations of R. ornativentris and the one population of R. japonica, although another allele was found in low frequencies in several populations (Table 5). At seven loci (AAT-1, ADA, FH, GPI, MDH-1, MDH-2 and Hb-2) a single allele predominated in the 16 populations of R. ornativentris, whereas a different allele predominated in R. japonica (Table 5). At five loci (IDH-1, LDH-B, ME-1, MPI and Alb) the predominant allele was different among several groups of populations of R. ornativentris, whereas another allele predominated in R. japonica (Table 5, Fig. 2). At the other five loci (ME-2, PEP-A, SOD-1, SOD-2 and Hb-1) the predominant allele was different among several groups of populations of R. ornativentris, and one of these alleles also predominated in R. japonica (Table 5, Fig. 3).
Table 4.
Frequencies of phenotypes at 24 loci in 16 populations of Rana ornativentris and one population of R. japonica
Table 5.
Allele frequencies at 21 loci in 16 populations of Rana ornativentris and one population of R. japonica
Fig. 2.
Geographic distributions of multiple alleles at four loci, the IDH-1, LDH-B, MPI and Alb loci, in 16 populations of Rana omativentris and one population of R. japonica. Numbers represent populations designated in Table 1.
Fig. 3.
Geographic distributions of multiple alleles at four loci, the PEP-A, SOD-1, SOD-2 and Hb-1 loci, in 16 populations of Rana ornativentris and one population of R. japonica. Numbers represent populations designated in Table 1.
Genetic variation in R. ornativentris
The fixation index (Fst) was calculated according to Wright (1978) (Table 6). When the allele frequencies at a definite locus are the same in all the 16 populations, the fixation index is zero, whereas it is 1.000 when there is a characteristic allele at a definite locus in one or more populations. The higher the fixation index, the more advanced the divergence in the locus.
Table 6.
Fixation index at 24 loci in 16 populations of Rana ornativentris
The most advanced loci in divergence was the Hb-1 locus, being 0.876 in Fst. This was followed by the SOD-1, PEP-A, SOD-2, Alb, IDH-1, MPI and LDH-B loci, ranging from 0.635 to 0.312 in Fst. At these seven loci, except the SOD-2 locus, the genetic divergence was distinct between the eastern and western populations, whereas the genetic divergence was clear in the central regions of Honshu at the SOD-2 locus. The ME-2, ME-1, FH, IDH-2, GPD, ADA, AAT-1 and MDH-1 loci were from 0.261 to 0.030 in Fst, and showed various degrees of genetic divergence. The remaining five loci (AK, CK, LDH-A, MDH-2 and Hb-2) were zero in Fst (Table 6). Average value of Fst (
st) excluding the five invariant loci was 0.306.
The mean proportion of heterozygous loci per individual, mean proportion of polymorphic loci per population and mean number of alleles per locus in the 16 populations of R. ornativentris were 4.2%∼24.2%, 14.4% on average, 4.2%∼66.7%, 32.8% on average, and 1.04∼2.21, 1.45 on average, respectively (Table 7). In 11 populations of which the sample size was larger than three, the comparable figures were 12.0%∼24.2%, 15.9% on average, 20.8%∼66.7%, 39.8% on average, and 1.29∼2.21, 1.57 on average, respectively. There were no noticeable differences between these rates and the expected values, except populations consisting of one, two or three samples.
Table 7.
Genetic variabilities at 24 loci in 16 populations of Rana ornativentris
Genetic distances
The genetic distances among 16 populations of R. ornativentris ranged from 0.011 between the Itoigawa and Yamakita populations to 0.313 between the Yamakita and Nima populations, with a mean of 0.127 (Table 8). The genetic distances among the four western populations of R. ornativentris ranged from 0.015 between the Geihoku and Saiki populations to 0.061 between the Geihoku and Nima populations, with a mean of 0.043 and those among the 12 eastern populations ranged from 0.011 between the Itoigawa and Yamakita populations to 0.179 between the Shiojiri and Tsuruga populations, with a mean of 0.063 (Table 8). On the other hand, the genetic distances between the four western and 12 eastern populations ranged from 0.128 between the Tsuruga and Geihoku or Saiki populations to 0.313 between the Yamakita and Nima populations, with a mean of 0.225 (Table 8). The genetic distances between the 16 populations of R. ornativentris and one population of R. japonica ranged from 0.579 between the Yamaguchi population of R. ornativentris and the Saiki population of R. japonica to 0.956 between the Shiojiri population of R. ornativentris and the Saiki population of R. japonica, with a mean of 0.793 (Table 8). The genetic distances between the four western populations of R. ornativentris and the Saiki population of R. japonica were 0.579∼0.638, 0.608 on average, whereas those between the 12 eastern populations of R. ornativentris and the Saiki population of R. japonica were 0.707∼0.956, 0.855 on average.
Table 8.
Genetic identity (I) and genetic distance (D) among 16 populations of Rana ornativentris and one population of R, japonica
Dendrogram
The phenetic relationships were assumed from a dendrogram drawn by the UPGMA method, which is the most commonly used. The UPGMA dendrogram showed that R. japonica is clearly separated from R. ornativentris, which constitutes two clusters, the western and eastern (Fig. 4). In the eastern cluster, the Takatomi and Tsuruga populations constitute a subcluster and split into another 10 populations which form two subclusters, the northern and southern. The other six kinds of dendrograms did not remarkably differ from that drawn by the UPGMA method.
DISCUSSION
The question of how much genetic divergence occurs during the process of speciation is one of the most cardinal problems of evolutionary genetics. The genetic divergence between taxa of various levels of evolutionary divergence has been reviewed in many organisms (Ayala, 1975; Avise, 1976; Avise and Aquadro, 1982). The intraspecific genetic divergence has been estimated by calculating the genetic distances among populations in many amphibian species. The mean genetic distances among different populations were 0.007∼0.205 in 55 amphibian species (Sumida, unpublished). It is probable that species having large genetic distances among populations are divided into several local groups which are geographically isolated or diverged long ago.
The present study revealed that the genetic divergence between the western and eastern groups of Rana ornativentris was distinct. Although no samples from the region between Tsuruga and Nima were examined, it was inferred that the boundary between the eastern and western groups was around the Kinki area on the basis of allelic distribution at several loci. Considerable east-west divergence was evident for seven loci (IDH-1, LDH-B, MPI, PEP-A, SOD-1, Alb and Hb-1), 0.312∼0.876 in Fst, whereas at the SOD-2 locus, 0.578 in Fst, the divergence was distinct in the central region of Honshu (Figs. 2 and 3). There were distinct gradients from east to west in the alleles at the PEP-A, SOD-1 and Hb-1 loci. According to Ueda (Personal communication), the reciprocal hybrids between the Geihoku and Hirosaki populations of R. ornativentris are normal both in viability and reproductive capacity. Nei's genetic distances (D) correspond with the divergence time (T) from the common ancestor by the equation of T = 5 × 106 D (years) (Nei, 1975, 1987). Applying this equation to the present study, it was speculated that R. ornativentris diverged into the eastern and western groups about 1.1 million years ago and then the latter diverged into the three subgroups about 0.3 ∼ 0.5 million years ago. Japan was a part of continent up to the end of the Riss glacial stage in the Pleistocene (Minato et al., 1965). It seems probable that after the ancestors of R. ornativentris invaded Japan and widened their distribution all over Japan during the Pleistocene, they were isolated into eastern and western groups due to marine transgression during the interglacial stages in the middle Pleistocene (Minato et al., 1965). Before they established a reproductive isolating mechanism, they came into contact with each other during the marine regression in the Riss and Würm glacial stages. Thus the gradients from east to west were presumably formed in the alleles at several loci in the central regions of Honshu.
Studies of biochemical divergence in amphibians distributed in Honshu were carried out using Bufo japonicus (Kawamura et al., 1990), Rana nigromaculata (Nishioka et al., 1992b), Rana rugosa (Nishioka et al., 1993), Rana japonica (Sumida and Nishioka, 1991,1994) and Cynops pyrrhogaster (Hayashi and Matsui, 1988, 1990). Biochemical divergence between the western and eastern groups was also observed in B. japonicus, R. rugosa and R. japonica. The divergence of R. japonica into the eastern and western groups was evident at the PEP-A and Hb-1 loci, which were 0.722 and 0.781 in Fst, respectively (Sumida and Nishioka, 1994). These two groups were reproductively isolated by incomplete hybrid male sterility (Sumida, 1981, 1994,1996). In Bufo japonicus, there were distinct gradients from east to west in alleles at the AAT-2, IDH-1, LDH-B, ME-1 and MPI loci, which were 0.451, 0.730, 0.346, 0.420 and 0.383 in Fst, respectively. The UPGMA dendrogram showed that the toads distributed in Japan were divided into the eastern and western groups and then each group was divided into several subgroups (Kawamura et al., 1990). Matsui (1984) has reported that B. japonicus is divided into eastern and western types on the basis of morphological variation analyses with the central Honshu region as a dividing line, although the eastern and western groups of this species were not reproductively isolated (Kawamura et al., 1980). In Rana rugosa, the divergence between the eastern and western groups was distinctly recognizable at the CK, LDH-B, MDH-1, PEP-A and Hb-1 loci, which were 0.781, 0.901, 0.795, 0.979 and 0.863 in Fst, respectively. The UPGMA dendrogram showed that R. rugosa was first divided into the western and eastern groups, and that the latter was divided into three subgroups, northern, intermediate and southern (Nishioka et al., 1993). Levels of population subdivision were estimated by calculating the average values of Fst (st) for many amphibian species (Larson, 1980; Larson et al., 1984; Ragghianti and Wake, 1986). The present study revealed that the average value of Fst excluding five invariant loci was 0.306 in the 16 populations of Rana ornativentris, whereas that for eight variant loci including IDH-1, LDH-B, MPI, PEP-A, SOD-1, SOD-2, Alb and Hb-1 was 0.537. As is the case for several species mentioned above, Rana ornativentris is probably comprised of populations that are isolated from each other.
Genetic variabilities in allopatric populations were reviewed in anurans and urodeles by Nevo et al. (1984), Nevo and Beiles (1991) and Shaffer and Breden (1989). According to these researchers, the mean proportions of heterozygous loci per individual and polymorphic loci per population were 7.3±0.4% and 25.5±1.1%, respectively, in 188 amphibian species (two order and 13 families) including 123 urodeles (five families) and 65 anurans (eight families) (Nevo and Beiles, 1991). These two parameters were 8.7±0.5% and 26.4±1.4%, respectively, in 102 species from 19 genera and six families of urodeles (Shaffer and Breden, 1989). In the genus Rana, mean genetic indices for all 22 species were 7.5±1.3% and 23.3±2.9%, respectively (Nevo and Beiles, 1991). The two parameters for four Japanese Rana species, R. limnocharis, R. nigromaculata, R. porosa and R. rugosa, are 6.1∼9.9% and 23.1∼31.1%, respectively (Nishioka and Sumida, 1990; Nishioka et al., 1992b, 1993). These four species generally inhabit constant environments such as plains and lowlands, especially around rice fields, although Rana rugosa is also found in low mountains near water. On the other hand, in three Japanese Rana species, R. tagoi, R. japonica and R. ornativentris, the two parameters were high, 11.3∼16.1% and 39.8∼55.2%, respectively (Nishioka et al., 1987a; Sumida and Nishioka, 1994; present study). These three brown frog species chiefly inhabit variable and narrow habitats such as hillsides and mountain districts. Rana ornativentris is usually abundant in mountain regions up to 1900 m (Maeda and Matsui, 1989). Dessauer et al. (1975) and Nevo (1978) have suggested that the amounts of genetic polymorphism and heterozygosity are correlated with ecological heterogeneity, and may be regarded as an adaptive strategy for increasing population fitness in an ecologically variable environment. This interpretation may be applicable not only to R. tagoi and R. japonica but also to R. ornativentris in this study.
The genetic distances have been estimated among various species of the genus Rana. The genetic distances between Rana boylei and R. muscosawere 0.68∼0.77 (Case, 1978). Those among five species endemic to western North America, R. boylii (probably R. boylei), R. muscosa, R. aurora, R. cascadae and R. pretiosa, were 0.171∼0.733, whereas those between these five species and three related species, R. sylvatica, R. temporaria and R. dybowskii, were 0.506∼1.122 (Green, 1986). He found that the 24 chromosome species, R. dybowskii, was very distant genetically from all other species examined and had a Nei's genetic distance of over 0.940, on average, from all other samples. Fanglin et al. (1989) reported that the genetic distances among three Eurasian brown frog species, R. amurensis, R. chaochiaoensis and R. japonica, were 0.500∼0.876. Nishioka et al. (1992a) presented the genetic divergence among 30 populations of 12 brown frog species distributed in the Palearctic region. Among seven species having 26 chromosomes, R. japonica, R. tsushimensis, R. okinavana, R. longicrus, R. temporaria, R. asiatica and R. amurensis, the genetic distances were 0.294∼1.396, whereas those among four species having 24 chromosomes, R. ornativentris, R. pirica (Matsui, 1991; Matsui et al., 1993), R.dybowskii and R. arvalis, were 0.474∼1.018. Those between the seven species with 26 chromosomes and the four species with 24 chromosomes were 0.410∼1.715. Green and Borkin (1993) reported that the genetic distances among 11 Eurasian brown frog species including R. chensinensis, R. dybowskii, R. ornativentris, R. arvalis, R. japonica, R. tagoi, R. amurensis, R. temporaria, R.dalmatina, R. camerani and R. macrocnemis were 0.232∼1.127.
The present study revealed that the genetic distances between R.ornativentris and sympatric R. japonica, which have a diploid chromosome number of 24 and 26, respectively, were relatively large, 0.579∼0.956, 0.793 on average. The genetic distances between R. japonica and R. ornativentris were also reported by Matsui (1991), Nishioka et al. (1992a), Matsui and Wilkinson (1992) and Green and Borkin (1993). Applying Nei's equation to these studies, it was speculated that the lineage of two species diverged about 3∼5 million years ago. Rana ornativentris and R. japonica are occasionally found sympatrically, but are completely reproductively isolated from each other by hybrid sterility or gametic isolation (Kawamura, 1950; Kawamura et al., 1981). The hybrids between female R. japonica and male R. ornativentris developed normally but became sterile males, and no eggs of R. ornativentris cleaved by insemination with the sperm of R. japonica. According to Nishioka et al. (1992a) and Green and Borkin (1993), it seems likely that R. ornativentris is more closely related to R. dybowskii and R. chensinensis (sic), which have 24 chromosomes than to R. japonica, which has 26 chromosomes. Tanaka et al. (1994) investigated phylogenetic relationships among five Japanese brown frog species by the analysis of molecular sequences in the cytochrome b gene of mtDNA, and revealed that R. ornativentris and R. pirica having 24 chromosomes formed a subcluster and split from R. japonica having 26 chromosomes. It is probable that after the two brown frog lineages having 24 and 26 chromosomes entered Japan, they diverged into several valid species having 24 and 26 chromosomes, respectively.
Acknowledgments
The authors are grateful to Dr. N. Yoshizaki, Gifu University, Dr. H. Iwao, Yamaguchi University, Dr. M. Ryuzaki, Kitasato University, Mr. K. Kinebuchi, Niigata Senior High School, Dr. l. Miura and Mr. H. Hanada, Hiroshima University, for collecting and providing us valuable specimens, and to Dr. K. Sekiya, Niigata University, Dr. T. Fujii, Hiroshima Women's University, Messrs. T. Hongo and S. Nakagawa, Araya Senior High School, for their kind aid in collecting specimens. This study was supported in part by a Grant-in-Aid (No. 07640931) for Scientific Research from the Ministry of Education, Science, Sports and Culture of Japan to M. Sumida.
