DNA extraction

Total genomic DNAs for PCR were extracted from clipped toes using standard protocols of chemical digestion (0.1M Tris-HCl pH8.0, 0.2M EDTA, 1% SDS, 20 mg/ml proteinase K) followed by phenol/chloroform extraction (Sumida et al., in preparation). Air-dried DNA pellets were eluted in TE. The frog samples used were frozen and stored at the Institute for Amphibian Biology, Hiroshima University.

PCR primers

PCR primers (F51 and R51) were designed by Sumida et al. (2002) to cover a 546-bp segment of the 16S rRNA gene corresponding to sites 3976–4549 in the X. laevis sequence reported by Roe et al. (1985). The primer sequences were F51 (5′-CCC GCC TGT TTA CCA AAA ACA T-3′) and R51 (5′-GGT CTG AAC TCA GAT CAC GTA-3′) (Sumida et al., 2002). The primers (FS01 and R16) designed by Sumida et al. (1998, 2000a,b, 2001) were used for amplification and sequencing of a 411-bp segment of the 12S rRNA gene corresponding to sites 2816–3225 in the R. catesbeiana sequence (Yoneyama, 1987). The primer sequences were FS01 (5′-AAC GCT AAG ATG AAC CCT AAA AAG TTC T-3′) and R16 (5′-ATA GTG GGG TAT CTA ATC CCA GTT TGT TTT-3′) (Sumida et al., 1998, 2000a,b).

PCR and direct sequencing

PCR mixtures were prepared with the TaKaRa Taq^TM Kit as recommended by the manufacturer (TaKaRa) in a final volume of 50 μl. The 12S and 16S rRNA genes were amplified by 30 cycles, each cycle consisting of denaturation for 1 min at 94°C, annealing for 1 min at 55°C, and extension for 1 min at 72°C. The PCR products were purified by MicroSpin^TM S-300 HR Columns (Pharmacia Biotech) and used for sequencing. Purified DNA was sequenced by the DyeDeoxy^TM Terminator Cycle Sequencing method using a 373A DNA Sequencing System Ver. 1.2 (ABI). Nucleotide sequences were analyzed using DNASIS (Ver. 3.2, Hitachi Software Engineering). Alignments for DNA sequences were deterd based on maximum nucleotide similarity using CLUSTALW (Thompson et al., 1994). Nucleotide sequences were deposited in DDBJ, EMBL, and GenBank nucleotide sequence databases under the Accession Nos.  AB058852 ∼  AB058887 (Table 1).

Phylogenetic analysis

Phylogenetic relationships were estimated by the neighbor-joining (NJ) method (Saitou and Nei, 1987) on the basis of sequence divergences calculated by Kimura's two parameter method (Kimura, 1980), using the program included in Ver. 3.5c of PHYLIP (Felsenstein, J. Univ. of Washington, Seattle, 1993, unpubl.). Confidences in topology were assessed by 1,000 bootstrap replications (Felsenstein, 1985). The nucleotide sequences of X. laevis (Roe et al., 1985) were used as an outgroup to construct the phylogenetic tree.

Developmental capacity of hybrids

Tables 2 and 3 show the results of a total of 57 kinds of crossing experiments among 12 populations of 10 species distributed in the Palearctic and Oriental regions. In all of the control matings except for the Chinese R. chensinensis, the eggs developed normally and completed metamorphosis; 81.6∼98.6% of the total number of eggs cleaved normally, 46.0∼97.2% became normally hatched tadpoles, and 35.9∼90.7% metamorphosed normally (Tables 2 and 3). The eggs used in the Chinese R. chensinensis matings were considered to be in an unfavorable condition, but 6.8% of the total number of eggs completed metamorphosis (Table 2). On the other hand, various degrees of abnormalities were found in the offspring produced by the interspecific crosses. No eggs cleaved, normally or abnormally, in any of the following crosses: the eight crosses using female R. temporaria, the crosses between female R. ornativentris from Hiroshima and male R. chensinensis from Russia and China, and the crosses between female R. japonica from Hiroshima and male R. chensinensis from Russia (Tables 2 and 3). No eggs developed into normally hatched tadpoles in any of the following crosses: the 8 interspecific crosses using female R. asiatica, the crosses between female R. chensinensis from Russia and male R. okinavana, and the crosses between female R. chensinensis from China and male R. longicrus (Tables 2 and 3). No tadpoles metamorphosed normally in the interspecific crosses between female R. chensinensis from Russia and males of R. japonica from Hiroshima and R. tsushimensis, between female R. pirica and male R. temporaria, between female R. chensinensis from China and male R. tsushimensis, between female R. ornativentris from Hiroshima and male R. temporaria, between female R. tsushimensis and males of R. chensinensis from Russia and R. temporaria, and between female R. asiatica and male R. temporaia (Tables 2 and 3).

Table 2

Developmental capacity and sex of brown frog hybrids and the controls. I.

Table 3

Developmental capacity and sex of brown frog hybrids and the controls. II.

The tadpoles metamorphosed normally in only 16 out of the 48 interspecific crosses.

Sexes of the hybrids

The metamorphosed frogs from all of the control matings included both males and females, and the percentages of females ranged from 37.5 to 55.8%. On the other hand, all the hybrid frogs produced from interspecific crosses were males (Tables 2 and 3).

Testes of mature hybrids

The inner structures of testes of mature males derived from 8 control matings and 16 interspecific crosses were examined by histological observation (Tables 2 and 3). The testes of males derived from the control matings were completely normal in inner structure: the seminiferous tubules were filled with compact bundles of normal spermatozoa (Figs. 3A and 4A, D). On the other hand, the testes of all male hybrids produced from interspecific crosses were abnormal in inner structure: the seminiferous tubules contained no normal spermatozoa, and there were various quantities of large abnormal spermatozoa and pycnotic nuclei (Figs. 3B-F and 4B, C, E, F). When mature male hybrids were mated with the control females, all male hybrids were completely sterile in reproductive capacity.

Fig. 3

Cross-sections of the testes of mature male hybrids between female Rana chensinensis from Russia and males of three species and the control males. Scale bars equal 50 μm. (A) Control R. chensinensis from Russia, chen. R ♀ No.1 × chen. R ♂ No.1. (B) Hybrid between female R. chensinensis from Russia and male R. chensinensis from China, chen. R ♀ No.1 × chen. C ♂ No. 1. (C) Hybrid between female R. chensinensis from Russia and male R. chensinensis from China, chen. R ♀ No.1 × chen. C ♂ No. 2. (D) Hybrid between female R. chensinensis from Russia and male R. pirica from Sapporo, chen. R ♀ No.1 × piri. J ♂ No. 1. (E) Hybrid between female R. chensinensis from Russia and male R. pirica from Sapporo, chen. R ♀ No.1 × piri. J ♂ No. 2. (F) Hybrid between female R. chensinensis from Russia and male R. ornativentris from Hiroshima, chen. R ♀No.1 × ornat. J^H ♂ No. 1.

Fig. 4

Cross-sections of the testes of mature male hybrids between female Rana pirica and males of R. chensinensis from Russia and China, between female R. chensinensis from China and male R. pirica, and between female R. japonica from Hiroshima and male R. chensinensis from China and the control males. Scale bars equal 50 μm. (A) Control R. pirica, piri. J ♀ No.1 × piri. J ♂ No.1. (B) Hybrid between female R. pirica and male R. chensinensis from Russia, piri. J ♀No.1 × chen. R ♂ No. 1. (C) Hybrid between female R. pirica and male R. chensinensis from China, piri. J ♀No.1 × chen. C ♂ No. 1. (D) Control R. chensinensis from China, chen. C ♀No.1 × chen. C ♂ No. 1. (E) Hybrid between female R. chensinensis from China and male R. pirica, chen. C ♀ No.1 × piri. J ♂ No. 1. (F) Hybrid between female R. japonica from Hiroshima and male R. chensinensis from China, jap. J^H ♀No.1 × chen. C ♂ No. 1.

Molecular phylogeny

Nucleotide sequences were determined in a 411-bp segment of the 12S rRNA gene and 546-bp segment of the 16S rRNA gene in 18 frogs from 18 populations of 13 species distributed in the Palearctic and Oriental regions. Eighteen haplotypes were observed in these frogs (Table 1), and each population or species had a specific haplotype. The sequenced 411-bp segment of the 12S rRNA gene provided 116 variable sites including eight gaps between species, and the sequenced 546-bp segment of the 16S rRNA gene provided 121 variable sites including 13 gaps.

Table 4 shows the nucleotide sequence divergences and the numbers of nucleotide substitutions in all pairs of 18 haplotypes of the 12S and 16S rRNA genes in the Palearctic and Oriental frogs. The nucleotide sequence divergences of the 12S and 16S rRNA genes were 0.53%, 1.50% and 5.90% between two haplotypes of R. ornativentris, R. japonica and R. chensinensis, respectively, and 0.11%∼0.32% among three haplotypes of R. amurensis (Table 4). There were 5, 17 and 54 nucleotide substitutions between two haplotypes of R. ornativentris, R. japonica and R. chensinensis, respectively, and 1∼3 among three haplotypes of R. amurensis including gaps (Table 4). The nucleotide sequence divergences were 2.69%∼6.61% (x̄=4.15%), and there were 26∼62 nucleotide substitutions including gaps among seven haplotypes of 2n=24 brown frog species (Table 4). The nucleotide sequence divergences were 2.80%∼8.31% (x̄=6.21%), and there were 29∼80 nucleotide substitutions including gaps among 10 haplotypes of 2n=26 brown frog species (Table 4). The nucleotide sequence divergences were 3.56%∼7.77% (x̄=6.24%), and there were 33∼73 nucleotide substitutions including gaps between seven haplotypes of 2n=24 brown frog species and 10 haplotypes of 2n=26 brown frog species (Table 4). The nucleotide sequence divergences were 13.96%∼15.22% (x=14.68%), and there were 122∼133 nucleotide substitutions including gaps between outgroup R. latouchii and ingroup 12 brown frog species (Table 4). The nucleotide sequence divergences were 29.79%∼32.27% (x̄=30.71%), and there were 278∼293 nucleotide substitutions including gaps between outgroup X. laevis and ingroup 12 brown frog species (Table 4). The nucleotide sequence divergences were 29.02%, and there were 262 nucleotide substitutions including gaps between outgroup R. latouchii and X. laevis (Table 4).

Table 4

Percent divergences estimated by Kimura's two-parameter method (above diagonal) and the numbers of transition sites (Ts), transversion sites (Tv) and gaps (G) (Ts / Tv/G) (below diagonal) among haplotypes of nucleotide sequences of the mitochondrial 12S and 16S rRNA genes of the brown frogs distributed in the Palearctic and Oriental regions and R. latouchii (lat. T) and X. laevis (xeno.) as out-groups.

When the nucleotide sequence divergences in the 12S rRNA gene were plotted against those in the 16S rRNA gene in all pairwise comparisons of 19 mtDNA haplotypes of the ingroup and outgroup, the rates of sequence divergences for the 12S rRNA gene were approximately 1.2 times greater than those for the 16S rRNA gene in all pairwise comparisons of 19 haplotypes (Fig. 5A). The comparisons of the mtDNA haplotypes from the ingroup and out-group in the 12S and 16S rRNA genes showed more C ⇔ T than A ⇔ G transitions (C ⇔ T / A ⇔ G ratios = 1.66) (Fig. 5B). The frequencies of transitions in the 12S and 16S rRNA genes appeared to plateau in pairwise comparisons between the ingroup Palearctic and Oriental brown frog species and outgroup R. latouchii / X. laevis, where multiple substitutions occurred at the same site (Fig. 5C).

Fig. 5

(A) Dependence of percent sequence divergences of the 12S rRNA gene on the 16S rRNA gene. (B) Dependence of C ⇔ T (CT) transitions on A ⇔G (AG) transitions of the 12S and 16S rRNA genes. (C) Dependence of transitions on transversions of the 12S and 16S rRNA genes.

The NJ tree based on the nucleotide sequence divergences of the 12S and 16S rRNA genes showed that the outgroup R. latouchii was clearly separated from the brown frog species (Fig. 6). The ingroup brown frog species, the monophyly of which was strongly supported in 100% of 1,000 bootstrap iterations, were largely divided into six clades (Fig. 6). The first clade contained three populations of R. amurensis. The second clade included R. tsushimensis and R. okinavana, the third clade consisted of R. longicrus and two populations of R. japonica, the fourth clade contained R. asiatica, and the fifth clade contained R. temporaria and R. arvalis. The sixth clade included four species with 2n=24 chromosomes, R. chensinensis from Russia and China, R. pirica, R. dybowskii, and R. ornativentris from Hiroshima and Aomori (Fig. 6).

Fig. 6

Phylogenetic tree constructed by the neighbor-joining (NJ) method based on the nucleotide sequences of 411-bp segment of the 12S rRNA gene and 546-bp segment of the 16S rRNA gene from 18 haplotypes of the brown frogs distributed in the Palearctic and Oriental regions and Rana latouchii (lat.T) and Xenopus laevis (xeno.) as outgroup. Numbers on the tree are bootstrap probabilities (above 60%) based on 1000 replications. Scale bar represents branch length in terms of nucleotide substitutions per site. Abbreviations of haplotypes are given in Table 1.

Reproductive isolation of Palearctic and Oriental brown frogs

Mayr (1940) defined species as “groups of actually or potentially interbreeding natural populations which are reproductively isolated from other such groups.” Dobzhansky (1950) defined the species as “the largest and most inclusive … reproductive community of sexual and cross-fertilizing individuals which share in a common gene pool.” If two taxa are reproductively isolated from each other and cannot interbreed, they are considered different species.

Brown frog species and populations from Japan, Taiwan, China, and Russia were found to be reproductively isolated from one another by gametic isolation, hybrid inviability at the embryonic or tadpole stage, and male hybrid sterility. The reproductive isolating mechanisms among the Palearctic and Oriental brown frog species and populations are summarized from the present results and the data provided by Kawamura et al. (1981) (Fig. 7). This figure shows that there was complete gametic isolation in 15 hybrid combinations, from which no cleaved eggs were obtained. Gametic isolation is related to the difficulty with which the eggs are inseminated with frog sperm, that is, the difficulty with which sperm can penetrate jelly envelopes. This figure also shows that all the hybrids derived from 10 and 27 hybrid combinations died during embryonic and tadpole stages, respectively. The hybrid inviability is attributed to the incompatibility between the two genomes. Furthermore this figure clarifies that all viable hybrids derived from the 28 hybrid combinations became completely sterile males. The complete sterility of mature male hybrids has also been reported in other hybrid combinations among various brown frog species (Dürken, 1935, 1938; Kawamura, 1940, 1943, 1950; Kawamura and Kobayashi, 1959, 1960; Kawamura and Nishioka, 1962, 1973, 1977). The hybrid sterility is attributed to structural differences in chromosomes between the parental species. Thus, it may be that different brown frog species in the Palearctic and Oriental regions are first isolated from each other, either completely or incompletely, by gametic isolation: when the gametic isolation is incomplete, the two species are completely or incompletely isolated by hybrid inviability; when the hybrid inviability is incomplete, the two species are completely isolated by hybrid sterility.

Fig. 7

Reproductive isolating mechanisms found in various combinations among 11 brown frog species distributed in the Palearctic and Oriental regions. The mark of each section shows the reproductive isolating mechanism by which two species are completely isolated from each other. The data from Kawamura et al. (1981) are asterisked.

Relationship among R. pirica and Russian and Chinese R. chensinensis

Assignment of the brown frog from Hokkaido to R. chensinensis was made provisionally by Kawamura (1962). The original description of R. chensinensis (David, 1875) is brief and uninformative, and nothing is reported for the topotypic population. Liu and Hu (1961) and Hu et al. (1966) noted that R. chensinensis, the so-called Chinese brown frog, is itself highly polymorphic and assuredly comprises several distinct forms. Kawamura et al. (1985) proposed that Japanese R. chensinensis should be described as R. ezoensis, based on a preliminary report of crossing experiments. From the available morphological information about the nominate population of R. chensinensis (Hu et al., 1966), Matsui (1991) regarded the brown frog occurring in Hokkaido as an undescribed form and designated this frog as a new species, R. pirica. In a morphological comparison of R. chensinensis from China with three Japanese relatives, R. pirica, R. ornativentris, and R. dybowskii, Matsui et al. (1993) found that R. pirica, a species long regarded as conspecific with R. chensinensis, is more remote from topotypes of R. chensinensis morphologically than are R. ornativentris and R. dybowskii. The present results of the crossing experiments showed that R. pirica occurring in Hokkaido is reproductively isolated from R. chensinensis from Russia and China by complete hybrid sterility. The hybrids between these three taxa usually showed incomplete gametic isolation and hybrid inviability, but a small number of hybrids attained sexual maturity. While this seems to show that the three taxa are somewhat related to one another, all the hybrids became completely sterile males. As these three taxa are reproductively isolated by complete hybrid sterility and morphologically distinguished from each other (Figs. 2 and 7), each should be a different valid species. Thus, it was reasonable of Matsui (1991) to describe the brown frog from Hokkaido as a new species Rana pirica, even though the species name Rana ezoensis had been proposed by Kawamura et al. (1985) as a new species name based on preliminary data from crossing experiments. It is also very interesting to note that Russian R. chensinensis is a separate valid species which is morphologically distinguished and reproductively isolated from Chinese R. chensinensis.

Molecular phylogeny of brown frogs

Recently, molecular phylogenies of Asian brown frogs have been clarified by analyzing nucleotide sequences of the mitochondrial cytochrome b gene (Tanaka et al., 1994, 1996, 1998a, b, c, 1999; Matsui et al., 1998; Sumida and Ogata, 1998). The present study analyzed the nucleotide sequences of mitochondrial 12S and 16S rRNA genes to elucidate phylogenetic relationships among the brown frogs distributed in the Palearctic and Oriental regions. Regarding the relationships of Asian brown frogs, our results are partially consistent with those obtained in the previous studies of mitochondrial cytochrome b gene sequences by Tanaka et al. (1994, 1996, 1998a, b, c, 1999) and Matsui et al. (1998), and the inclusion of additional taxa revealed some new information. The present results showed that the out-group R. latouchii was remotely related to the brown frog species examined. R. latouchii apparently belongs to a different lineage from brown frogs according to Dubois (1992), and this proposition has been supported by allozyme analysis (Nishioka et al., 1992). The ingroup brown frog species, of which the monophyly was strongly supported in 100% of 1000 bootstrap iterations, was largely divided into six clades. The first clade includes three populations of R. amurensis (bootstrap 100%). The second clade includes R. tsushimensis and R. okinavana (bootstrap 95%). The association of R. tsushimensis and R. okinavana has also been reported based on mitochondrial cytochrome b gene data (Tanaka et al., 1996). The third clade includes R. longicrus and R. japonica. Rana longicrus was once regarded as con-specific with R. japonica based on the close morphological similarities between the two species. Karyological and ecological studies further supported their close phylogenetic relationships (Kuramoto et al., 1973, 1984), as did the present study (bootstrap 93%), although the clade was only weakly supported by mitochondrial cytochrome b gene data (bootstrap 47.7%, Tanaka et al., 1998a). The fourth clade includes R. asiatica. The fifth clade includes R. temporaria and R. arvalis (bootstrap 67%), both of which are European brown frog species. The sixth clade includes R. chensinensis from China and Russia, and R. ornativentris, R. pirica and R. dybowskii (bootstrap 87%), all of which have 24 chromosomes (Nishioka et al., 1987). The same groupings of Asian brown frogs with 2n=24 chromosomes were mostly supported by allozyme analyses carried out by Nishioka et al. (1992) and Green and Borkin (1993). Our molecular data, in which five clades comprising 7 species with 2n=26 chromosomes showed a polytomous relationship with the sixth clade comprising 4 species with 2n=24 chromosomes, could not elucidate the detailed phylogenetic relationships among brown frog species with 2n=26 chromosomes by analyzing the mitochondrial 12S and 16S rRNA gene sequences. Further examination will be necessary to infer the detailed phylogenetic relationships among Palearctic and Oriental brown frog species with 2n=26 chromosomes by analyzing other mitochondrial and nuclear genes.

Concerning the genetic variations within species, the sequence divergences among several three populations of R. amurensis from Russia, China and Mongolia were very small (0.11%∼0.32%). A similarly low genetic differentiation among several populations of R. amurensis has been demonstrated by allozyme analysis (Nishioka et al., 1992; Green and Borkin, 1993) and mtDNA cytochrome b gene analysis (Tanaka et al., 1998c). Thus, R. amurensis widely found from Mongolia and northeastern China to Far East Asia is considered to have diverged within a short time rather recently. It is noteworthy that within R. japonica, genetic differentiation between the western and eastern populations of Honshu was rather large (1.50%), whereas it was rather small (0.53%) within R. ornativentris distributed sympatrically with R. japonica. Indeed, genetic differentiation between western and eastern populations of R. japonica has also been reported by crossing experiments, spermatogenesis, allozyme data and mitochondrial cytochrome b gene sequence data (Sumida, 1981, 1994, 1996; Sumida and Nishioka, 1994; Sumida and Ogata, 1998).