An electrophoretic survey was conducted for the Southeast and East Asian populations of Rana limnocharis. Results indicated that specimens from a single locality in Java derived from two genodemes showing complete allelic displacement at eight out of 25 presumptive loci examined. Within each of these genodemes, genotype frequencies at all polymorphic loci were well concordant with the Hardy-Weinberg expectation. These results strongly suggest that those specimens actually represent two biological species. Comprehensive distance analyses suggested that one of the two species from Java is genetically most divergent among all samples examined, whereas the other is most similar to the Laos sample and then to a group consisting of samples from Hongkong, western China, and the southern Ryukyus. Large values of genetic distances obtained for each combination of allopatric samples imply the presence of additional cryptic species in this nominal species.
INTRODUCTION
The Indian rice frog Rana limnocharis is a lowland medium-sized ranid with an extraordinarily wide distributional range including East, Southeast, and South Asia (Frost, 1985; Maeda and Matsui, 1989). Recently, several populations in South Asia (Nepal and India), previously assigned to R. limnocharis, were described as separate species on the basis of acoustic and slight morphological differences (Dubois, 1975; Dutta, 1997). Moreover, Toda et al. (1997), in the allozyme study of R. limnocharis in the temperate to subtropical East Asia (i.e., Japan and central eastern and western China), demonstrated a considerable genetic divergence between the southern Ryukyu populations and the remaining genetically remarkably uniform populations, with Nei's (1978) genetic distance values falling within ranges of the interspecific level (Thorpe, 1982). Results of these studies predict that R. limnocharis in the remaining, still broad unsurveyed area (Southeast Asia and southern part of East Asia) is actually a composite, with substantial genetic and reproductive divergences. Indeed, several authors also have concerned morphological variation in East and Southeast Asian populations of R. limnocharis, but without making any taxonomic changes [see Kuramoto (1968) for the review].
Protein electrophoresis is a powerful tool to detect occurrences of cryptic species within morphologically conservative lineages, e.g., hynobiid salamanders (Matsui, 1987), dicamptodontid salamanders (Good, 1989), and typhlopid snakes (Hedges and Thomas, 1991). In this study, we adopted this approach to assess genetic variation and actual taxonomic diversity in Southeast and southern East Asian populations of R. limnocharis. Resultant data, while confirming considerable genetic differentiations among allopatric populations as predicted above, indicated the occurrence of sympatric cryptic species in Java.
MATERIALS AND METHODS
Specimens examined in this study were collected from New Territories, Hongkong (N = 10), and Malinping, Jawa Barat (western Java), Indonesia (N = 18), and were also obtained from a commercial market in Vientiane, Laos (N = 9) (Fig. 1). Liver and muscles were removed from each specimen, and their extracts were subjected to horizontal starch gel electrophoresis. Twenty-five presumptive loci controlling 19 enzymes were assayed by using six kinds of buffer systems (Table 1). For comparisons, we also incorporated Toda et al's (1997) allozyme data for samples from Wenjiang, western China, and Ishigaki, the southern Ryukyus, into the analyses as representatives of the two divergent lineages of temperate to subtropical East Asia.
Table 1
Enzymes, presumptive loci, tissues and buffer systems used in the analysis of genetic divergence among Southeast and East Asian populations of Rana limnocharis
In order to estimate overall genetic differentiations among samples, Rogers' (1972) distance and Nei's (1978) unbiased genetic distance (D) were calculated for all pairwise comparisons of samples. The Rogers' distance coefficient was clustered by the Neighbor-Joining (NJ) method (Saitou and Nei, 1987). For the out-group rooting of the NJ network, we incorporated data for Rana cancrivora from Toda et al. (1997) into the analysis. This species is supposedly closely related to R. limnocharis (see Toda et al., 1997). We also calculated the mean expected heterozygosity (Hexp) (Nei, 1978) for each sample. The NJ procedure was performed by N. Saitou's original computer program, and the other computations were made by the BIOSYS-1 computer program (Swofford and Selander, 1981).
RESULTS
Of 25 presumptive loci assayed, all but four (Ada, Ck, Gdh and Sod) were polymorphic. Genotype frequencies at these loci were well concordant with the expectation of the Hardy-Weinberg equilibrium in the Hongkong, the Vientiane, and the remaining two East Asian samples. In the Malinping sample, however, the frequencies showed significant deviations at ten loci (Aat-2, Ldh-1, Ldh-2, Mdh-1, Mdh-2, Me-1, Mpi, Pep-lv, Pgd, Pgm) due to the deficiency or complete absence of heterozygotes. Based on genotypes at four of these loci (Aat-2, Ldh-2, Mdh-2, Pgd) in which heterozygotes were completely lacking, two groups of individuals were recognized in the Malinping sample. One of these groups consisted of eight individuals having genotypes cc at Aat-2, aa at Ldh-2, aa at Mdh-2, and aa at Pgd, whereas the other group consisted of ten individuals having genotypes dd, bb, bb, and cc at respective loci. Moreover, these groups possessed different alleles at other four loci. The former group possessed alleles c, d, and e at Ldh-1, b and c at Mdh-1, b and d at Pep-lp, and b and c at Pep-lv, whereas the latter group possessed alleles a and b, a and d, a and c, and only d at these loci, respectively. These results strongly suggest that the Malinping sample is actually a mixture of two genetically divergent and reproductively isolated demes. We, therefore, redesignated these two groups of specimens as two separate samples, Malinping-A (N = 8) and Malinping-B (N = 10). With this modification, frequency distributions of genotypes became well concordant with the Hardy-Weinberg expectation at all polymorphic loci in all samples.
Allele frequencies at polymorphic loci examined in this study are given in Table 2. Matrices of Nei's and Rogers' distances among samples are presented in Table 3. The largest value of the Nei's D (0.597) was obtained between the Malinping-B and Ishigaki samples. Each of these two samples was genetically differentiated also from the other samples with large values of Nei's D [0.301-0.458 (x̄ = 0.339) between the Malinping-B and the other samples, 0.312-0.440 (x̄ = 0.373) between the Ishigaki and the other samples]. On the other hand, D values obtained among the Hongkong, Vientiane, and Malinping-A samples were relatively small [D = 0.170-0.287 (x̄ = 0.225)]. The Wenjiang sample, while showing a relatively small D value with the Hongkong sample (D = 0.250), was also considerably differentiated from the remaining samples [D = 0.371-0.414 (x̄ = 0.398)].
Table 2
Allele frequencies and mean expected heterozygosities (standard errors in parentheses) in six samples of Rana limnocharis from Southeast and East Asia and a sample of R. cancrivora (No. 7). Reference numbers of R. limnocharis samples (Nos. 1-6) correspond to those in Fig. 1
Table 3
Matrices of Rogers' (1972) (above diagonal) and Nei's (1978) genetic distance coefficients (below diagonal) for Southeast and East Asian samples of Rana limnocharis and a sample of Rana cancrivora
NJ phenogram based on the Rogers' distance is shown in Fig. 2. In this phenogram, the Malinping-B sample constituted one of the two major clusters by itself. Remaining samples composing the other major cluster were grouped into three sub-clusters: one consisting of the Malinping-A and Vientiane samples, another of the Ishigaki sample alone, and the other of the Hongkong and Wenjiang samples.
DISCUSSION
Present results clearly indicate that specimens from Malinping, though consistently sharing morphological features diagnostic of R. limnocharis, actually represent two genetically divergent and reproductively isolated entities. Considering the fact that all specimens of the sample were collected from a limited paddy field area (less than 90 m × 70 m), it is obvious that these “entities” are most appropriately referred as two biological species. This finding, along with the fact that the type locality of R. limnocharis is restricted to Java (e.g., Stejneger, 1907), raises a serious taxonomic problem: Because of the close morphological similarity between the two Javanese species, it is expected to be extremely difficult to determine which of these strictly corresponds to the nominal species. Further detailed morphological comparisons among specimens of these cryptic species and the type specimen of R. limnocharis are strongly desired.
Result of NJ clustering seems to indicate that the Malinping-B sample is genetically most divergent among all samples analyzed. This suggests that the population was first differentiated from the remaining Southeast and East Asian populations of R. limnocharis (sensu lato). We therefore consider that this population actually represents one distinct species by itself.
Even among populations other than Malinping-B, levels of genetic differentiations are considerably high. Especially, distances of Wenjiang from Vientiane and Malinping-A (D = 0.371 and 0.414, respectively), and those of Ishigaki from all other samples (D = 0.312-0.440) were large, overlapping with those between the Malinping-B and the other samples (D > 0.301) (Table 3). Furthermore, Nei's D values between the Malinping-A sample and the two representatives of East Asian populations (i.e. Ishigaki and Wenjiang samples: D = 0.440 and 0.414, respectively) were larger than the value obtained between the latter two samples (D = 0.410), which most likely represent two distinct species (Toda et al., 1997). The NJ phenogram also suggests the distant position of the MalinpingA sample in relation to the Ishigaki and Wenjiang samples. These results further suggest that the Malinping-A population has been diverged from both of the two cryptic species of East Asia, as well as from the other species represented by the Malinping-B population. Thus, it is highly likely that there are at least four cryptic species within R. limnocharis (sensu lato).
Appropriate taxonomic allocations of the remaining populations, (i.e., those from Hongkong and Vientiane) are still unclear. In the NJ phenogram, the Hongkong sample was clustered with the Wenjiang sample, whereas the Vientiane sample was clustered with the Malinping-A sample. Paradoxically, the Hongkong sample seems to be also closely related to the Vientiane sample and vice versa with a relatively small D value (0.170). As such, although our results suggest the occurrences of at least four species in the nominal species, Rana limnocharis, in Southeast and East Asia, their boundaries remain to be resolved on the basis of additional samples.
Acknowledgments
We wish to express our sincere thanks to H. Hasegawa for providing specimens from Vientiane. We also thank N. Saitou for providing NJ computer program. H. Ota is much indebted to A. Bogadek and M. W. Lau for helping with collection of materials, and to St. Louis School for provision of comfortable accommodation during the fieldwork in Hongkong. Sampling in Java was made under the permission of Lembaga Ilmu Pengetahuan Indonesia (LIPI). This research was partially supported by Monbusho International Scientific Research Program (Field Research No. 03041065, project leader: I. Miyagi).