Pelophylax nigromaculatus, P. porosus porosus, and P. p. brevipoda are three pond frog species distributed in Japan. Their distributions overlap at two basins in central Japan (P. nigromaculatus and P. p. porosus in the Matsumoto basin, and P. nigromaculatus and P. p. brevipoda in the Ina basin), and hybrid descendants have been found in these areas. To clarify the distribution areas and hybrid zones of the frogs, and to understand the mode of introgressive hybridization and its impact on the frog populations, we conducted exhaustive sampling at each basin and performed allozyme and mtDNA analyses of 233 individuals. Analysis using genetic markers clearly detected nine F1 hybrids and 94 hybrid descendants of P. nigromaculatus and P. porosus from the overlapping areas of both basins. Allozyme and mtDNA data suggest directional hybridization between female P. p. porosus and male P. nigromaculatus in the Matsumoto basin. Over the past 30 years, the distribution of P. p. porosus has been narrowed and fragmented by the invasion of P. nigromaculatus, seemingly because of directional hybridization in the Matsumoto basin. In the Ina basin, the “pure” P. p. brevipoda (n = 8) population was extremely reduced by gene introgression from P. nigromaculatus, yet its distribution was barely changed compared to the Matsumoto basin. Consequently, this study shows that P. porosus populations are threatened by interspecific hybridization with P. nigromaculatus, and that introgressive hybridization damaged P. porosus populations by different means in each basin.
INTRODUCTION
Pelophylax nigromaculatus and its relatives P. porosus porosus and P. p. brevipoda are species (or subspecies) of pond frogs distributed in Japan. They commonly inhabit areas around rice paddies and watersheds in flat lands. Pelophylax nigromaculatus is widely distributed across Kyushu, Shikoku, and Honshu (except in the Kanto and Sendai plains) and has been introduced to Hokkaido. Pelophylax p. porosus occurs in Eastern Honshu from the Kanto Plain to the Sendai Plain, in central and southern Niigata Prefecture, and in northern and central Nagano Prefecture, and it has also been introduced to Hokkaido. Pelophylax p. brevipoda is distributed in Western Honshu, Tokai, central Kinki, San-yo districts, and northeastern Shikoku (Maeda and Matsui, 1999) (Fig. 1). All three species are known to hybridize with each other in overlapping areas (e.g., Moriya, 1959; Nishioka et al., 1992; Sumida and Ishihara, 1997). The distribution of P. nigromaculatus and P. p. brevipoda overlaps widely, while the distribution of P. p. porosus rarely overlaps with that of the others (Fig. 1). Remarkably, in Nagano Prefecture, the distributional areas of the frogs are close to each other. In particular, the Matsumoto and Ina basins are known as hybrid zones of P. nigromaculatus and P. p. porosus and of P. nigromaculatus and P. p. brevipoda (Shimoyama, 1986; Nishioka et al., 1992). Although the F1 hybrid male of P. nigromaculatus and P. porosus is sterile, the hybrid female is fertile (Moriya, 1960), indicating gene introgression between P. nigromaculatus and P. porosus populations through F1 females.
Introgression is a phenomenon that imports genes from one species (or population) into a gene pool of another through repetitive backcrossing of interspecific hybrids. Introgression occasionally plays an important role in biological events, such as gene pollution, genetic diversification, speciation, adaptive evolution, and extinction, and has thus attracted the attention of researchers in a variety of disciplines, including population genetics and evolutionary and conservation biology (e.g., Rhymer and Simberloff, 1996; Allendorf et al., 2001; Burke and Arnold, 2001; Mooney and Cleland, 2001; Mallet, 2005; Evans et al., 2006; Baack and Rieseberg, 2007; Song et al., 2011). Several studies have reported historical, or possibly ongoing, introgressions among ranid frogs, including pond frog populations (e.g., Berger, 1968; Kim et al., 2004). For example, Liu et al. (2010) reported frequent occurrences of mtDNA introgression from P. plancyi to P. nigromaculatus populations throughout their distributional areas, which cover a wide range of eastern Eurasia. However, difficulties in locating and sampling introgressants and specifying hybrid zones based on morphological observation and/or small numbers of usable loci to diagnose the parent species of an introgressant often prevent surveying introgression details in wild populations (Allendorf et al., 2001). This is also true for the Japanese pond frogs. Although several studies based on allozyme markers have demonstrated the occurrence of introgressions between P. nigromaculatus and P. porosus in their parapatric distributional areas, such as Hakone, Maibara, Igaueno, Higashiosaka, and Konko (Nishioka et al., 1992; Sumida and Ishihara, 1997), detailed survey of hybrid zones using DNA markers, especially maternal mitochondrial genes, has not been conducted. To date, a variety of knowledge on these frogs has accumulated and covers areas such as diagnostic allozyme loci, possible hybrid zones, differences in breeding period and reproductive behavior, and reproductive abilities of hybrids (Moriya, 1960; Nishioka et al., 1992; Sumida and Ishihara, 1997; Shimoyama, 1986, 2000). More importantly, information is available on the past distribution of the pond frogs in Nagano Prefecture, including the Matsumoto and Ina basins (Shimoyama, 1986), which is the only area where all three Japanese pond frogs occur parapatrically or sympatrically. In addition, the possible occurrence of two distinct hybrid zones (P. nigromaculatus × P. p. porosus in the Matsumoto basin, and P. nigromaculatus × P. p. brevipoda in the Ina basin) has been suggested. These two basins seem to be the most relevant areas for studying the hybridization of pond frogs and the impact of hybridization in relation to other biological events, such as gene introgression, outbreeding depression, niche competition, and distributional change.
Fig. 1.
Distribution map of the three pond frog species in Japan as described by Maeda and Matsui (1999). Nagano Prefecture, where the Matsumoto and Ina basins are located, is shown by a solid line. Pelophylax nigromaculatus and P. p. porosus were introduced to Hokkaido.
Fig. 2.
Dorsal view of the pond frogs and a hybrid individual. (A) P. nigromaculatus. (B) P. p. porosus. (C) P. p. brevipoda. (D) F1 hybrid between P. p. porosus female and P. nigromaculatus male. Scale bars: 1 cm.
To clarify (1) the accurate distribution of three pond frog taxa and their hybrid zone areas in the Matsumoto and Ina basins, and (2) the degree and pattern of two introgressive hybridizations (i.e., P. nigromaculatus × P. p. porosus, and P. nigromaculatus × P. p. brevipoda), we performed exhaustive sampling of pond frogs from each basin and conducted morphological observation and allozyme and mtDNA (cytb gene) analyses. In addition, we compare the present distributional areas with those of 30 years ago, and discuss the observed distributional change and its cause.
MATERIALS AND METHODS
Sampling localities and specimens
A total of 233 specimens of pond frogs were collected from 30 localities of the Matsumoto and Ina basins in Nagano Prefecture in June 2010, which is their breeding season. First, all specimens were identified on the basis of dorsal morphological characters according to Moriya (1954) and Maeda and Matsui (1999). Dorsal images of representative individuals of P. nigromaculatus, P. p. porosus, P. p. brevipoda, and their hybrid (P. p. porosus female × P. nigromaculatus male) are shown in Fig. 2. Sampling localities and sample size of each locality are shown in Table 1, Fig. 3A and Supplementary Table S1 (10.2108.zsj.29.351.s1.pdf) online. Localities 1–16 and 17–30 belong to the Matsumoto and Ina basins, respectively. Skeletal muscles and the toes of specimens were sampled for allozyme and mtDNA analyses, respectively.
Allozyme analysis
In addition to the 233 collected individuals, three other individuals corresponding to P. nigromaculatus, P. p. porosus, and P. p. brevipoda (used in Nishioka et al., 1992) were used in the allozyme analysis. Four enzymes, α-GDH, IDH-B, LDH-B, and MDH-B, known as diagnostic loci for distinguishing between P. nigromaculatus and P. porosus (Nishioka et al., 1992), were examined by horizontal starch gel electrophoresis, following the procedure reported by Nishioka et al. (1980). Detection of each enzyme was performed by the agar overlay method of Brewer (1970) and Harris and Hopkinson (1976) with a slight modification, as described by Nishioka et al. (1992).
DNA extraction and mtDNA sequence analysis
Total DNA was extracted from the clipped toes using DNeasy Tissue Kit (QIAGEN) according to the manufacturer's instructions. Total DNA was used to amplify a partial sequence of the mitochondrial cytb gene with a set of primers by polymerase chain reaction (PCR). Primers used were L14850 (5′-TCT CAT CCT GAT GAA ACT TTG GCT C-3′) and H15410 (5′-GTC TTT GTA GGA GAA GTA TGG-3′), designed by Tanaka et al. (1996). PCR mixtures were prepared with the Ex-Taq Kit (TaKaRa) according to the manufacturer's instructions. DNA sequencing was performed with an automated sequencer (ABI 3130xl, Applied Biosystems). Resultant cytb sequences and the three other cytb sequences from previous studies (P. nigromaculatus from Hiroshima, P. p. prosus from Tsuchiura, and P. p. brevipoda, Maibara; GenBank Accession Numbers AB029937, AB029938, and AB029940, respectively; Sumida et al., 1998, 2000) were aligned using ClustalW (Thompson et al., 1994). Based on the aligned data of 491 nucleotide sites, an unrooted neighbor-joining (NJ) tree was reconstructed using MEGA ver. 4 (Tamura et al., 2007).
Table 1.
Morphological and genetic characteristics of pond frogs in each locality. N, P, B, and H indicate the characters of P. nigromaculatus, P. p. porosus, P. p. brevipoda and their hybrids, respectively. F1 in the allozyme column indicates individuals with heterozygous alleles of P. nigromaculatus and P. porosus at all four loci examined, and introgressant indicates individuals that possessed alleles specific to both P. nigromaculatus and P. porosus, except F1 hybrids.
RESULTS
Identification and distributional survey based on external morphology
Species (subspecies) and hybrid identification of the 233 specimens based on dorsal characters identified 148 as P. nigromaculatus, 38 as P. p. porosus, 35 as P. p. brevipoda, and 12 as hybrids (Table 1). Distribution of each taxon based on morphological observation is shown in Figs. 3A and 5A.
In the Matsumoto basin, P. nigromaculatus was distributed on the shore of Kizaki Lake (population 1) and in the central to southern areas of the basin and watershed of the Sai River (populations 2, 4–7, 9, 12, 13, 15, and 16). Pelophylax p. porosus was distributed in six populations in three isolated areas of the northern (populations 3 and 4), western (populations 8 and 11), and eastern (populations 10 and 14) areas. Hybrid individuals were found in the central parts of the basin (populations 6, 8, and 14). In the Ina basin, P. nigromaculatus was distributed across a wide range from the northern to southern parts of the basin (populations 17–30), while the distribution of P. p. brevipoda was limited to the northern part of the Tenryu River watershed (populations 20, 21, 23, 24, and 26). Unlike the distribution in the Matsumoto basin, the entire distributional area of P. p. brevipoda overlapped with that of P. nigromaculatus.
Fig. 3.
Sampling localities, distributions, and results of allozyme and cytb gene analyses. (A) Map showing sampling localities and morphology-based distribution. N, P, B, and H indicate occurrence of P. nigromaculatus, P. p. porosus, P. p. brevipoda, and hybrid individuals assigned by dorsal morphology, respectively. (B) Summary of the results by allozyme and cytb analyses. Chart numbers correspond to sampling localities shown in A. The charts show the allelic condition of eight alleles at four analyzed loci, each vertical column in the charts corresponds to each individual, and black and white bars in each column correspond to P. nigromaculatus- and P. porosus-specific alleles, respectively. The N, P, and B marks below each column indicate the cytb haplotypes specific to P. nigromaculatus, P. p. porosus, and P. p. brevipoda, respectively.
Allozyme analysis
Of the four allozyme loci (α-GDH, IDH-B, LDH-B, and MDH-B) α-GDH and IDH-B are known to have alleles specific to P. nigromaculatus and P. porosus, respectively (designated N and P). The LDH-B locus has three alleles specific to P. nigromaculatus, P. p. porosus, and P. p. brevipoda, respectively (designated N, Pp, and Pb); and the MDH-B locus has three alleles consisting of two intraspecific polymorphic types in P. nigromaculatus and one type in P. porosus, respectively (N1, N2, and P) (Nishioka et al., 1992). The results of our allozyme analysis are summarized in Table 1 and Fig. 3B. The chart numbers in Fig. 3B correspond to the sampling localities in Fig. 3A, and the charts show the allelic condition of the eight alleles of the four loci. Vertical columns correspond to individuals, and black and white bars correspond to P. nigromaculatus-and P. porosus-specific alleles, respectively (N, P, and B below each column indicate the cytb haplotypes specific to P. nigromaculatus, P. p. porosus, and P. p. brevipoda, respectively). For example, chart 2 in Fig. 3B shows that four individuals were collected from population 2, and three individuals have only P. nigromaculatus-specific alleles at all four observed loci (total eight alleles), but one individual has a P. porosus-specific allele at one locus. Chart 2 also shows that all four individuals from population 2 have the P. nigromaculatus-specific cytb haplotype. Detailed allelic information is displayed in Supplementary Table S1 (10.2108.zsj.29.351.s1.pdf) online.
Out of the 233 specimens, 115 had only P. nigromaculatus-specific alleles at the four analyzed loci, and 55 frogs possessed homozygous genotypes consisting of only P. porosus-specific alleles (Supplementary Table S1 (10.2108.zsj.29.351.s1.pdf) online). Based on the allozyme analysis, these individuals can be regarded as “pure” P. nigromaculatus and P. porosus. However, the remaining 63 individuals had both P. nigromaculatus and P. porosus alleles, and nine of these showed heterozygous genotypes of P. nigromaculatus and P. porosus at all four loci (in populations 4, 6, 8, 10, and 24; Fig. 3B). These nine individuals can be regarded as F1 hybrid individuals, for the following two reasons: (1) The F1 hybrid must have heterozygous genotypes of parent species at all loci; and (2) pond frogs with sterile F1 males (F2 of F1 male × F1 female) cannot occur. Consequently, the probability of a hybrid descendant arising with heterozygous alleles at all four loci is very low (6.25% and 0.40% for first and second backcross generations, respectively). The remaining 54 individuals possessing alleles specific to the two species at more than one locus are predicted to be hybrid descendants (i.e., introgressants) emerged via backcrossinq of female F1 hybrids.
Of the 148 individuals assigned as P. nigromaculatus on the basis of dorsal characters, 114 (77.0%) possessed only P. nigromaculatus-specific alleles (“pure” P. nigromaculatus), but 34 (23.0%) had alleles from both P. nigromaculatus and P. porosus (five F1 hybrids and 29 introgressants). Likewise, of the 38 individuals regarded morphologically as P. p. porosus, 34 (89.5%) were regarded as “pure” P. p. porosus, and the remaining four (10.5%) consisted of one F1 hybrid and three hybrid descendants. Of the 35 frogs initially regarded as P. p. brevipoda, 19 (54.3%) and 16 (45.7%) were revealed to be “pure” P. p. brevipoda and hybrid descendants, respectively. Of the 12 individuals morphologically regarded as hybrids, three (25.0%) and six (50.0%) were revealed to be F1 hybrids or their descendants, and the remaining one (8.3%) and two (16.7%) were shown to be P. nigromaculatus (mtDNA replaced by that of P. p. porosus) and “pure” P. p. brevipoda, respectively. In total, 63 F1 hybrids and their descendants were detected by allozyme analysis.
Cytb gene analysis
Partial cytb regions (491 bp) from the 233 frogs were sequenced. In the alignment matrix of the cytb sequences, 76 variable sites were found. Sequence data of observed 23 haplotypes were deposited into GenBank (Accession Nos. AB686619–AB686641). Our NJ tree, reconstructed from the cytb data, clearly showed three major clades corresponding to the haplogroups of P. nigromaculatus (N), P. p. porosus (P), and P. p. brevipoda (B), respectively (Fig. 4). Average uncorrected “p” distances between these haplogroups were 0.085 (N vs. P), 0.090 (N vs. B), and 0.045 (P vs. B), respectively. The cytb haplotypes of each individual are shown in Fig. 3B and Supplementary Table S1 (10.2108.zsj.29.351.s1.pdf) online.
Interestingly, some “pure” individuals revealed by allozyme analysis possessed the cytb haplotype of another species. Of the 115 “pure” P. nigromaculatus identified by allozyme analysis, 89 (77.4%) carried the P. nigromaculatus haplotype, but 24 (20.9%) and two individuals (1.7%) carried P. p. porosus and P. p. brevipoda haplotypes, respectively. Likewise, of 36 “pure” P. p. porosus individuals, one (2.8%) carried the P. nigromaculatus haplotype, and of 21 “pure” P. p. brevipoda individuals, 13 (68.4%) carried the P. nigromaculatus haplotype. In total, the mtDNA of 40 individuals was replaced with that of another species. These individuals showed complete mismatch between nuclear genotypes (i.e., allozyme), and mitochondrial haplotypes must have emerged by mtDNA introgression through a number of backcrossing generations and/or self-crossing between hybrid descendants. Consequently, species and hybrid identification based on genetic markers indicated that our morphological observation underestimated the occurrence of hybrids and introgressants. But it is clear that distribution maps made on the basis of morphology and genetic markers are nearly congruent (Fig. 5A, B).
Fig. 5.
The distribution maps of pond frogs. (A) Present distribution map of pond frog taxa based on morphological observation. (B) Distribution map based on allozyme and cytb data. (C) Distributions of F1 hybrids and hybrid descendants assigned by genetic markers. (D) Past distribution based on the morphological survey from 1978–1985 by Shimoyama (1986). In (A), (B), and (D), dotted, vertically-striped, and horizontally-striped areas indicate the distributional areas of P. nigromaculatus, P. p. porosus, and P. p. brevipoda, respectively. Distribution of hybrids is not shown in (A), (B), and (D), but is illustrated in (C), which shows the areas where F1 hybrids or introgressants were found by genetic markers (shaded).
Of the F1 hybrids and introgressants (n = 63) revealed by allozyme analysis, 44 (69.8%), 11 (17.5%), and eight (12.7%) carried the P. nigromaculatus, P. p. porosus, and P. p. brevipoda cytb haplotypes, respectively. Furthermore and interestingly, of nine possible F1 hybrids of P. nigromaculatus and P. porosus, seven carried the P. p. porosus haplotype (from the Matsumoto basin) and only two carried the P. nigromaculatus haplotype (from the Matsumoto and Ina basins).
DISCUSSION
Distribution and hybridization of P. nigromaculatus and P. p. porosus in the Matsumoto basin
1. Shift of distributional areas and hybrid zones over the past 30 years
About 30 years ago, from 1978 to 1985, Shimoyama (1986) surveyed the distributions of the three pond frog species in the Matsumoto and Ina basins (Fig. 5D). In his report, the distribution of P. nigromaculatus in the Matsumoto basin was limited to the southern area, while the range of P. p. porosus was from the northern to central areas of the basin, including the Sai River watershed, and their distributions overlapped in the central area (Fig. 5D). Comparing past and present distributions (Fig. 5B, D), it is obvious that P. nigromaculatus extended its distributional areas north, especially around the Sai River. In addition, the distributional areas of P. p. porosus have decreased over the past 30 years. Pelophylax p. porosus individuals identified on the basis of external morphology are found only in three fragmented areas (populations 3 and 4; 8 and 11; 10 and 14), and “pure” P. p. porosus individuals identified by genetic markers have the same distribution pattern (Fig. 5A, B).
The present study detected many hybrid descendants of the two species from the past overlapping areas (Fig. 5C, D). Possible F1 hybrids of P. nigromaculatus and P. p. porosus were found in populations 4, 6, 8, and 10 (Figs. 3B, 5C), which suggests that the hybridization between these species is ongoing in the present populations. Remarkably, populations 4 and 6 are clearly distant from P. nigromaculatus of the past distributional area (Fig. 5D). Furthermore, in populations 2, 5, and 7, where only P. p. porosus was distributed in the past, P. nigromaculatus was found alongside hybrid descendants. Comparison of past and present distribution and the hybrid areas detected in this study clearly indicates that the hybrid zone of P. nigromaculatus and P. p. porosus has shifted northward in response to P. nigromaculatus invasion.
It is noteworthy that the area around population 16 located in the most southern part of the Matsumoto basin supported the P. nigromaculatus population 30 years ago (Fig. 5D), and all 13 individuals collected from the population in this study have only P. nigromaculatus-specific alleles and P. nigromaculatus external morphology (Figs. 3B, 5A). However, in 12 of the 13 frogs, mitochondrial haplotypes were replaced by those of P. p. porosus (Fig. 3B). This observation suggests that P. p. porosus was distributed in these areas before Shimoyama's survey (1986); subsequently, P. nigromaculatus invaded the population, and mtDNA from P. p. porosus was introgressed into the invading P. nigromaculatus (referred to as the “ghost of hybrids past” by Wilson and Bernatchez, 1998). Presumably, the situation of population 16 means that the hybrid zone has gradually shifted from the southern area around population 16 to the northern area in the Matsumoto basin.
2. Cause of distributional change
The distribution of P. nigromaculatus in the Matsumoto basin has expanded, whereas that of P. p. porosus has narrowed and fragmented. A gender bias in the hybridization (directional hybridization), as expected from our allozyme and cytb analyses, seems to be a major cause of this distributional change.
Of the eight possible F1 hybrids of P. nigromaculatus and P. p. porosus of the Matsumoto basin, seven carried the P. p. porosus haplotype (Fig. 3B and Supplementary Table S1 (10.2108.zsj.29.351.s1.pdf) online). The abundance ratio of P. nigromaculatus to P. p. porosus haplotypes in these F1 hybrids is significantly biased toward the P. p. porosus haplotypes (binominal test, P = 0.0352). Furthermore, except for the F1 hybrids, 25 out of 34 hybrid descendants carried the P. p. porosus haplotypes, and this is also significantly biased toward the P. p. porosus haplotype (P = 0.0045). Considering that no differences in viability of reciprocal hybrids of P. nigromaculatus and P. p. porosus were reported from crossing experiments (Moriya, 1960), and that the maternal inheritance manner of mtDNA is maintained, even in ranid hybrids (e.g., Sumida, 1997), the observed bias toward the P. p. porosus mtDNA in the hybrids (including introgressants) seems to be caused by directional hybridization between P. p. porosus females and P. nigromaculatus males, rather than by postzygotic factors. The occurrence of hybridization in the same direction between these frog taxa was suggested by Sumida and Ishihara (1997). They surveyed a population in the hybrid zone of P. nigromaculatus and P. p. porosus using allozyme and mtDNA analyses in Kanagawa Prefecture in Japan and found four possible F1 hybrids, all of which possessed the P. p. porosus mitochondrial haplotypes. Furthermore, Shimoyama (1999, 2000) reported three breeding features of P. nigromaculatus based on his field observations. First, compared with the P. porosus male (including both P. p. porosus and P. p. brevipoda), the P. nigromaculatus male has the advantage of clasping P. porosus females, because the body size of P. nigromaculatus is larger than that of P. porosus. Second, the P. nigromaculatus male does not distinguish between heterospecific mating calls and conspecific ones. In contrast, it is known that P. p. brevipoda can distinguish between different and conspecific mating calls. Consequently, the P. nigromaculatus male often joins the mating chorus of P. p. brevipoda, but the P. p. brevipoda male usually does not join that of P. nigromaculatus. Third, the reproductive pair formation of P. nigromaculatus is started by males forcibly clasping females, while that in P. p. brevipoda is initiated by females approaching males in their territory. The distinguishability and pair formation characters have not been studied in P. p. porosus. However, if P. p. prosus has the same characters as in P. p. brevipoda, these characters will lead to high frequent directional hybridization between P. p. porosus females and P. nigromaculatus males, compared with reverse pairs.
Interspecific hybridization in a hybrid zone generally brings disadvantages to both parent species because of wastage of reproductive efforts (except when the resultant hybrids have normal or advanced fitness), and reproductive efforts in females are usually greater than those in males (e.g., Mallet, 2005). Thus, in a hybrid zone where directional hybridization occurs, mother species will suffer more damage than father species (e.g., Konishi and Takata, 2004). Removing females by directional hybridization also leads to less opportunity for normal mating in mother species and brings about serious damage to mother species, especially in a species in which the sex ratio or operational sex ratio are biased toward the male (the latter being the case with Japanese pond frogs; Shimoyama, 1996). Consequently, directional hybridization occasionally causes species replacement by father species in hybrid populations. For example, Konishi and Takata (2004) showed species replacement in hybrid populations caused by a directional hybridization between the Japanese minnow species Pseudorasbora pumila and P. parva. Similarly, the directional hybridization between P. nigromaculatus and P. p. porosus detected in the present study seems to be a major factor for the range expansion of the father species (P. nigromaculatus) and the range contraction and fragmentation of the mother species (P. p. porosus).
Introgression between P. nigromaculatus and P. p. brevipoda in the Ina basin
In the Ina basin, the distribution areas of P. nigromaculatus and P. p. brevipoda do not seem to have changed over the past 30 years (Fig. 5B, D), although many hybrid descendants (and a possible F1 hybrid) were found in overlapping populations (northern part, populations 17–26; Fig. 5C). Therefore, in contrast to the Matsumoto basin, the pond frog species in the Ina basin have hybridized with each other without changing their distributions. The difference in the distributional aspect between the two basins more likely comes from the lower frequency of hybridization between P. nigromaculatus and P. p. brevipoda in the Ina basin than that between P. nigromaculatus and P. p. porosus in the Matsumoto basin. Between P. nigromaculatus and P. p. brevipoda, many premating isolation mechanisms (e.g., differences in breeding sites, breeding period, and advertising calls) and other ecological differences that are unknown or not well known between P. nigromaculatus and P. p. porosus have been reported (e.g., Shimoyama, 1986; Matsui, 1996; Maeda and Matsui, 1999). These isolation mechanisms would have prevented the occurrence of high frequent hybridization between P. nigromaculatus and P. p. brevipoda. In agreement, only one possible F1 hybrid between P. nigromaculatus and P. p. brevipoda was found in the Ina basin (population 24), in contrast to the Matsumoto basin where eight possible F1 hybrids of P. nigromaculatus and P. p. porosus were found (significant difference; P = 0.0359, Fisher's exact test). Consequently, the low frequent hybridization seems to have allowed the long-term (at least over 30 years) co-occurrence of these frogs without distributional change.
Pelophylax p. brevipoda is now listed in the Red Data Book as a critically endangered species in Nagano Prefecture (where the Matsumoto and Ina basins are located), and in fact, distribution of P. p. brevipoda is limited in the northern part of the basin (Fig. 5B, D). Furthermore, our allozyme and mtDNA analysis detected only eight individuals that can be assigned as “pure” P. p. brevipoda. Many individuals with the P. p. brevipoda morphology (29/35) are assigned as introgressants derived from backcrosses between P. p. brevipoda and hybrids. This observation indicates that large-scale gene introgression occurred between these two species in the Ina basin, seemingly contradicting the low frequent hybridization of these species mentioned above. However, considering the long-term co-occurrence of these species, the presence of many hybrid descendants is of little wonder. Namely, the normal fitness of females of F1 hybrids and the gradual recovery of fitness in hybrid descendants can allow these hybrids to engage in continuous backcrossing and self-crossing of introgressants, and the long term backcrossing (at least 30 years) can accumulate introgressed genes in populations of each species, even if the initial hybrid number is small. Furthermore, small population size generally leads to easy fixation of introgressed genes via random genetic drift (Frankham et al., 2002); the small population size of P. p. brevipoda would have enhanced the rampant introgression from P. nigromaculatus into P. p. brevipoda and caused extreme reduction of the “pure” P. p. brevipoda in this basin. In addition, it remains a possibility that the eight individuals regarded as “pure” P. p. brevipoda are actually introgressants carrying P. nigromaculatus alleles at loci not analyzed here. To figure out the accurate survival and genetic situation of endangered P. p. brevipoda, for example, in the context of conservation, an immediate and more detailed survey with additional genetic markers, such as microsatellite loci, seem to be necessary.
It is noteworthy that P. nigromaculatus-like individuals with a P. porosus-specific allele or P. p. brevipoda-specific cytb haplotype are found in southern population 29, which is outside the area of P. porosus distribution revealed by both Shimoyama (1986) and the present study. This may be the legacy of P. p. brevipoda originally distributed in this area.
Two individuals of the P. p. porosus-specific haplotype were found in populations 25 and 30 in the Ina basin (Fig. 3B), where the occurrence of P. p. porosus has not been known. As a cause of these introgressants, two possible explanations —past occurrence of P. p. porosus in this area and artificial transports of P. p. porosus (or its hybrids)—can be suggested.
CONCLUSION
In the present study, we surveyed multiple hybrid zones of P. nigromaculatus and P. porosus in the Matsumoto and Ina basins and showed that different modes of introgressive hybridizations have occurred in each basin. In both basins, P. porosus populations are commonly threatened by interspecific-hybridizations and subsequent gene introgressions from P. nigromaculatus. Among the basins, however, introgressive hybridizations damaged P. porosus populations in different ways. Populations of P. p. porosus have been narrowed and fragmented by P. nigromaculatus invasion of the Matsumoto basin. In contrast, in the Ina basin, “pure” P. p. brevipoda has been decreased by the introgression without distributional change. We also postulated the cause of these differences by combining our results and information of past distribution as well as other biological features.
In the near future, the distributional area of P. p. porosus will be further narrowed in the Matsumoto basin. Furthermore, “pure” P. p. brevipoda individuals might completely disappear from the Ina basin due to continuous gene introgression from P. nigromaculatus and random genetic drift in scarce populations where many hybrid descendants already exist. The real impact of the introgressive hybridization between pond frog species should become clearer with further continuous monitoring of the pond frogs in these basins.
ACKNOWLEDGMENTS
We are grateful to T. Yoshida, T. Motoki, A. Ichikawa and S. Kitano for providing fundamental information on the pond frogs in Nagano Prefecture. We thank A. Iwasawa for his advice about experimental plan and K. Kobayashi for his field assistance and comment. We also thank the following people for their valuable comments: K. Sekine, T. Kano, M. Ogitani, Y. Tanaka, Y. Morii, A. Nagakubo, T. Suzuki, N. Yoshiyama, E. Kanke and T. Igawa.
