Origin of Japanese Populations of Reeves' Pond Turtle, Mauremys reevesii (Reptilia: Geoemydidae), as Inferred by a Molecular Approach
Abstract
Reeves' pond turtle, Mauremys reevesii, is an aquatic geoemydid species that is broadly distributed in East Asia. Like many other Asian turtles, this species is facing an extinction crisis in most countries where it occurs. In Japan, however, this turtle commonly occurs in various freshwater habitats. Although the Japanese populations have generally been considered to be native, a few recent studies have yielded circumstantial evidence that suggested that they had originated from relatively recent artificial introductions from abroad. To evaluate the validity of such a view, we analyzed sequence variations of the mitochondrial cytochrome b gene and the control region in M. reevesii samples from various localities in Japan and adjacent countries. The results revealed the presence of 3 distinct genetic groups (groups A, B, and C) in the Japanese samples, of which groups A and B included haplotypes that were almost identical with some haplotypes from the Korean sample and the Chinese and Taiwanese samples, respectively. Sequences from M. reevesii shell products commercially traded in Taiwan included one that was almost identical to the group C sequence. The current Japanese populations of M. reevesii seem to have been derived from multiple artificial introductions from adjacent countries. This finding implies a need to be concerned about the effect of M. reevesii on the Japanese native wildlife, particularly on the Japanese endemic pond turtle, Mauremys japonica, with which M. reevesii is phylogenetically closely related and reportedly frequently hybridizes in the wild. However, the Japanese population may be useful as a genetically variable stock for conservation of this species, which is highly endangered in most of its range outside Japan.
Reeves' pond turtle, Mauremys reevesii (previously Chinemys reevesii: see Lovich et al. [in press] for problems relevant to generic status of this species), is a moderate-sized aquatic species of the family Geoemydidae, and is widely distributed in East Asia, ranging from southeastern China through the Korean Peninsula and Taiwan to mainland Japan (Lovich et al. 1985, in press; Iverson 1992). As with many other turtles and tortoises in Asia that are facing an extinction crisis basically due to various anthropogenic factors, M. reevesii is categorized as endangered on the Red List by International Union for Conservation of Nature and Natural Resources (IUCN) and is officially protected in most countries where it occurs (Bhupathy et al. 2000; van Dijk 2000). The only exception is in Japan, where this species is common in various freshwater habitats in the lowlands of central and western Honshu, Shikoku, Kyushu, and in some adjacent islets (Ota 2001). In Japan, therefore, M. reevesii is not assigned to any threatened category in the national Red Data Book (Matsui et al. 2000).
Except for putative feral populations recently found on a few central Ryukyu islands (Ota et al. 2004), Japanese populations of M. reevesii have long been regarded as native (e.g., Nakamura and Uéno 1963). However, a few recent studies have yielded results that circumstantially suggest a non-native origin of the current Japanese populations. For example, Yabe (2003) reported an almost complete absence of M. reevesii bones in the collection from a survey of chelonian skeletal remains at 9 archeological sites in central Honshu that correspond to a period from the Jomon Era (no earlier than 4000 years before present) to the 16th century. The single shell identified as M. reevesii remains problematic in date because of its location in the mixed sediment immediately below the ground surface (Excavation Group of the Ikawatsu Archaeological Site 1988). Hirayama et al. (2007) also highlighted the complete absence of fossil records of M. reevesii or its ancestral form in Japan despite frequent fossil discoveries of other freshwater turtles, including possible ancestors of the extant endemic Japanese pond turtle, Mauremys japonica, from this region (also see Hirayama and Isaji [2010] for further confirmation of the absence of M. reevesii fossils in additional Pleistocene chelonian materials from Japan). Hirayama et al. (2007), therefore, argued that M. reevesii was probably introduced from the Korean Peninsula after the Jomon Era. Likewise, Hikida and Suzuki (2010), in a review of articles and statements in medieval Japanese documents that referred to turtles in Japan, surmised that M. reevesii was introduced to Japan from Korea in the late 18th century. However, because all these works depended on rather weak circumstantial evidence, their claim of the entirely nonnative origin of current Japanese populations of M. reevesii still needs careful verification, it is hoped, on the basis of different lines of evidence.
Recent artificial spread of invasive species, either intentional or accidental, has become one of the major threats to the native biodiversity in many areas all over the world (e.g., Sakai et al. 2001). As in many other countries, there are a number of alien species in Japan, of which many have been causing devastating damage to the native biodiversity through the predation of, competition with, and introgression to various native species (Ecological Society of Japan 2002). This last issue is particularly serious in such an assemblage of moderately isolated continental islands as Japan, where endemic taxa that frequently have vicariant counterparts in adjacent regions are abundant (e.g., Kawamura et al. 2001; Kawamoto 2005).
In various freshwater habitats of Japan, M. reevesii is very frequently found in syntopy with M. japonica. In the past few decades, putative hybrids that exhibit morphologic intermediacy between M. reevesii and M. japonica have also been observed occasionally in such habitats (Kosuge et al. 2003; Haramura et al. 2008; Ota and Suzuki, unpubl. obvs., 1998 and 2006-2009), and Kato et al. (2010) confirmed the hybrid nature for several individuals with such morphologic intermediacy from central Honshu by investigating random amplified polymorphic DNA. Moreover, occurrences of F1 and F2 hybrids of the 2 species have been observed in captivity (Ota and Suzuki, unpubl. data, 1979, 1992, 2008, and 2009). The similarities in body size and courtship behavior (Yasukawa et al. 2008; Lovich et al. in press), and close phylogenetic affinity of M. reevesii and M. japonica (Honda et al. 2002; Barth et al. 2004; Spinks et al. 2004; Feldman and Parham 2004; Sasaki et al. 2006) may enable their successful reproduction and the emergence of fertile hybrids. If the Japanese populations are native, with a long history of sympatry with M. japonica, we can interpret the recent occasional occurrences of putative hybrids as an ephemeral process, which has actually been repeated numerous times but never led to the formation of a complete hybrid swarm as in the case of some sea turtle taxa (Karl et al. 1995; Bowen and Karl 2007), or otherwise, as a consequence of recent artificial habitat alterations, which has led to a breakdown of isolation mechanisms between the 2 species in the wild (Riley et al. 2003; Toda et al. 2006). Conversely, if M. reevesii in Japan actually originated from artificial introductions, then the recent occasional appearances of putative hybrids in the wild, along with the occurrence of F2 hybrids, at least in captivity, may reflect an ongoing process of introgression to the endemic native M. japonica.
The objective of this study was to examine whether the Japanese populations of M. reevesii are native or of introduced origin. We took a molecular approach, because this method is often very advantageous in achieving unequivocal conclusions regarding the geographic pattern and timing of organisms' dispersal as enhanced by recent human activities (e.g., Moritz et al. 1993; Austin 1999; Kuraishi et al. 2009; Johnson et al. 2011).
METHODS
Assumptions
A large part of current mainland Japan, which initially composed the eastern coastland of the Eurasian Continent, was supposedly involved in the Miocene tectonic movement that led to insularization of this region through the formation of the Sea of Japan (Maruyama et al. 1997). Based on the fossil records of mammals, such as elephants, it is generally considered that mainland Japan repeated several cycles of connections to and isolations from the Eurasian Continent until 0.3 million years before present (MYBP), when the latest land bridge between the 2 regions broke and submerged (Kawamura 1998). There might have been “ice bridges” that offered some terrestrial organisms further opportunities of dispersal among northern areas, such as Siberia, Sakhalin, Hokkaido, and northern Honshu, during the subsequent ice ages (or glacier periods: Kawamura 1998). However, when considering that the range of such “ice bridges” was distinctly north of the current northern extent of the geoemydid distribution in East Asia (Iverson 1992; Zhao and Adler 1993; Ota 2001) and that the air temperature in this region was lower during those ice ages than at present, dispersals of geoemydids, including M. reevesii, through the ice bridges were quite unlikely. Moreover, the ratio in the native terrestrial reptiles of mainland Japan that occur in the continent, including the Korean Peninsula is low (5/17 [29.4%]; calculated from the list of Japanese reptile fauna in Hikida [2002]). Among them, the natricine snake Rhabdophis tigrinus, common to mainland Japan and the continent, including the Korean Peninsula, showed considerably high-sequence divergence in the mitochondrial cytochrome b gene (cyt b) between the populations of the 2 regions (0.058–0.097 in Kimura's 2-parameter distance and 5.39%–8.88% in simple p-distance; interpreted as reflecting isolations for 4.10–10.56 MYBP, i.e., Early Pliocene to Middle Miocene; Takeuchi et al. in press), which strongly suggests that the duration of effective isolation for reptiles between the 2 regions should have actually been much longer than for mammals, most likely due to the lower temperatures during each of the Pleistocene glaciations in and before 0.3 MYBP, because this should have driven most reptiles that occur in the Korean Peninsula to more southern regions of the continent or to extinction before the land bridge connection between the Korean Peninsula and mainland Japan through sea level lowering (Hikida 2002; Ota, unpubl. data, 2011).
As such, if Japanese populations of M. reevesii possess haplotypes with 5% or greater p-distances from those of non-Japanese conspecific populations, then we consider that M. reevesii is native to Japan. Alternatively, if the sample from Japan has haplotypes that show much smaller p-distances (<1%) from non-Japanese haplotypes, it may possibly be indicative of recent artificial origin of the current Japanese population of M. reevesii.
Population Sampling
A total of 134 live individuals of M. reevesii were sampled, including 132 individuals from 19 localities in Japan, 1 individual from Korea (Kum-Gang, Janghang-eup, Seocheon-Gun, Chungcheongnam-do), and 1 from Taiwan (Kwanto, Taipei) (Fig. 1). Of these, the last specimen, originally purchased from a local turtle dealer, had long been in captivity. From this specimen, a small piece of tail-tip skin and a small amount of blood were collected. All other turtles had been captured by ourselves in rivers and ponds. They were narcotized to death by following the guideline for Animal Experiments of Kyoto University and then were subjected to sampling of muscle tissues. All tissues were stored at −80°C or in 99% ethanol until molecular processing. Our target sequences were the mitochondrial cyt b and control region (CR). These regions are known to be appropriate for analyses of intraspecific variations in turtles (e.g., Serb et al. 2001; Fritz et al. 2006; Fritz et al. 2008; Suzuki and Hikida 2011).
Citation: Chelonian Conservation and Biology 10, 2; 10.2744/CCB-0885.1
Additional genetic material of M. reevesii also was sampled. The cyt b sequences of 2 Chinese M. reevesii, deposited in GenBank under the accession numbers AJ519497 and AJ519499 (the latter listed as Chinemys megalocephala, a junior synonym of M. reevesii, at present; Barth et al. 2004), and those of 4 shell samples traded in the form of commercial products in Taiwan (FJ026830, FJ06852, FJ026853, and FJ026854: Lee et al. 2009) also were incorporated into the analyses. Most voucher specimens directly examined by us were deposited in the zoological collection of the Kyoto University Museum (see Appendix 1).
Molecular Methods
Total genomic DNA was extracted by the methods of Wada et al. (1992) and Honda et al. (1999). We used polymerase chain reaction to amplify cyt b and CR by using an Ex Taq polymerase kit (TAKARA BIO Inc, Otsu, Japan) with a polymerase chain reaction system GeneAmp 2700 Thermal Cycler (Applied Biosystems, Lincoln, California, USA). The primers used were GLU–5'eeg (5′-TGATATGAAAAACCACCGTTG-3′) (modified from Palumbi 1996) and M (5′-TCATCTTCGGTTTACAAGAC-3′) (Shaffer et al. 1997) for cyt b, and Mauremysd-loopF (5′-TCTCCCGTGCCCAACAGAGAAATGTC-3′) and Mauremysd-loopR (5′-GTTGCTCTCGGATTTAGGGGTTTGACG-3′) (Suzuki and Hikida 2011) for CR. The thermocycling parameters were 94°C for 5 minutes, followed by 35 cycles at 94°C for 1 minute, 48°C (for cyt b) or 58°C (for CR) for 2 minutes and 72°C for 2 minutes, and a final extension at 72°C for 7 minutes. Before sequencing, unincorporated primers were removed from polymerase chain reaction products by polyethylene glycol (PEG) precipitation by adding 0.6 volumes of PEG solution (20% PEG6000, 2.5 M NaCl). The sequencing primers were the above 4 primers and 2 additional primers, L15192cbEu (5′-TGAGGCGCAACCGTAATTACAAACCT -3′) (Okamoto and Hikida 2009) and H15263cbMauremys (5′-TGAAAGGTGAAGAATCGGGTTAGGG-3′) (modified from Okamoto and Hikida 2009) for cyt b. Cycle sequencing was performed by using 40 cycles of 10 seconds at 96°C, 5 seconds at 48°C, and 4 minutes at 60°C for cyt b, and at 10 seconds at 96°C and 4 minutes at 60°C for CR. Products were sequenced by using an ABI PRISM 3130 Sequencer (Applied Biosystems). The final data set for subsequent phylogenetic and population genetic analyses consisted of 1130 base pairs (bp) of cyt b and 714 bp of CR.
Analysis
The cyt b and CR sequences were aligned separately with CLUSTALX version 2.0.5 (Larkin et al. 2007). All alignment parameters were held constant at the default values. To investigate the relationships among haplotypes, we constructed a haplotype network of the combined data set of cyt b and CR with TCS version 1.21, which is based on statistical parsimony (Templeton et al. 1992). To incorporate available GenBank data into the analyses, we also obtained 2 other haplotype networks, both for cyt b only: one was based on sequence data for the 1008-bp portion used by Barth et al. (2004), who deposited data for 2 Chinese specimens; and the other was for the 340-bp portion by Lee et al. (2009), who deposited data for 4 specimens commercially traded in Taiwan.
To confirm that all M. reevesii individuals sampled in this study were not originated from hybridization between this and other species, especially M. japonica, genetic distances in cyt b sequence from M. reevesii were obtained for its close relatives by molecular evolutionary genetics analysis (Tamura et al. in press). The distances also were calculated among haplotype groups of M. reevesii (see Results) and between haplotype groups of M. reevesii and other species: M. japonica (AB559016; Suzuki and Hikida 2011), Mauremys nigricans (AJ519500; Barth et al. 2004), Mauremys sinensis (AY434615; Spinks et al. 2004), and Mauremys mutica (AJ564459; Barth et al. 2004). Furthermore, we also examined the distance between M. reevesii and Trachemys scripta elegans (FR717131; Fritz et al. 2011) because the latter species has already broadly established feral populations in Japan (Ecological Society of Japan 2002).
RESULTS
For the Japanese, Korean, and Taiwanese samples sequenced by us, we detected 7, 11, and 15 haplotypes from cyt b (1130-bp portion), CR (714-bp portion), and the combined data set, respectively. The sequence data have been submitted to GenBank database (AC no. AB561906–562151 and AB571605–571626; see Appendix 1). In the combined analysis, the Japanese haplotypes could be largely classified into 3 groups, groups A, B, and C, which included 10 haplotypes, 4 haplotypes, and 1 haplotype, respectively (Fig. 2a). Group A contained the majority of the Japanese specimens (102: A-1–9) and the one specimen from Korea (A-10). In this group, haplotype A-2, represented by 84 individuals, occupied the central position with all other group A haplotypes being derivable from this haplotype by substitution at one site (A-1, A-5–10), 2 sites (A-4: through A-5), or 3 sites (A-3: through A-9 and one undiscovered haplotype). The p-distances between the Japanese haplotypes that belong to group A and the Korean haplotype ranged from 0.05% to 0.22% (1–4/1844-bp).
Citation: Chelonian Conservation and Biology 10, 2; 10.2744/CCB-0885.1
Group B consisted of 28 Japanese specimens (B-1–4) and the one Taiwanese specimen (B-1) examined by us. Of the 4 component haplotypes of this group, haplotype B-1, shown by 26 specimens, occupied the central position, with the other 3 group B haplotypes being derivable from this haplotype by substitution at only one site. Thus, p-distances between the Japanese and Taiwanese group B haplotypes ranged from 0.00% to 0.05% (0–1/1844 bp). From the group A haplotypes, group B haplotypes differed by substitutions at 12 (between B-1 and A-7 or B-1 and A-10) or more sites. Because the only haplotype composing group C (shown by 2 Japanese specimens) was so distant from the other 2 groups, it could not be connected with any other particular haplotype or haplotypes with certainty in this analysis (Fig. 2a).
In the analysis of sequence data for 1008 bp of cyt b with incorporation of GenBank data of Barth et al. (2004) for 2 Chinese specimens, the 3 groups recognized above were still recognizable. In each of groups A and B, however, most haplotypes (A-2, A-3, A-5–7, A-9, and A-10 in group A; and haplotypes B-1–3 in group B) were identical, and the remainder were derivable by one site substitution. Interestingly, the sequences of the 2 Chinese specimens were identical with those of the haplotypes B-1–3 and p-distances of the Japanese group B haplotypes from these Chinese haplotypes ranged from 0.00% to 0.10% (0–1/1008 bp). However, haplotype C-1 was located 7 or more steps of substitutions from the group B haplotypes and 8 or more steps of substitutions from the group A haplotypes (Fig. 2b).
In the analysis of sequence data for 340 bp of cyt b that incorporated GenBank data from Lee et al. (2009), the 3 groups remained fairly recognizable but with substitution at only 1 site (among all group A haplotypes and most group B haplotypes) or 2 sites (among all group A haplotypes and haplotype B-4 or haplotype C-1). Of the 4 samples obtained from commercial trade specimens in Taiwan (Lee et al. 2009), 2 showed a sequence identical with that in haplotypes B-1–3, whereas another showed a sequence identical with group A haplotypes. Interestingly, the fourth sample showed a sequence closest to that of haplotype C-1, from which it differed by substitution at only one site (p-distance 0.29% [1/340 bp]: Fig. 2c).
The geographic distributions of the 3 genetic groups are shown in Fig. 3. Group A was recognized from all sites except site 1 and was predominant in western Honshu (sites 6–16) and in Shikoku (sites 17 and 18), whereas group B showed considerably discrete distribution in east–central Honshu (sites 1–5) and in Kyushu (site 19), with interruption among these sites by group A (Fig. 3). Except for the site 1 sample in which all haplotypes belonged to group B, haplotypes of this group always occurred along with those that represent group A (sites 2–4 and 19) or both of the other 2 groups (site 5). Group C was confined to site 5, where haplotypes of all 3 groups were found together.
Citation: Chelonian Conservation and Biology 10, 2; 10.2744/CCB-0885.1
Genetic distances in pairwise comparisons of the 3 haplotype groups of M. reevesii and other species are shown in Table 1. Genetic distances between different haplotype groups of M. reevesii (0.4%–0.8%) were distinctly smaller than those between these haplotype groups of M. reevesii and other species (4%–18.1%). In the pairwise interspecific comparisons within the genus Mauremys, distance values were distinctly smaller among M. reevesii, M. nigricans, M. japonica, and M. sinensis (approximately 4%–5%) than between each of them and M. mutica (7%–8%). Distances were much greater between each of the Mauremys species and T. s. elegans (17%–18.1%).
DISCUSSION
If M. reevesii is native to Japan, then its Japanese population has most likely been isolated from other conspecific populations no later than the Pliocene (see above), a period long enough to accumulate sequence divergences from the latter in such mitochondrial DNA domains as cyt b and CR that exhibit relatively high evolutionary rates (e.g., see Avise 2000). In most Japanese specimens whose haplotypes were classified as group A, however, the cyt b sequences were completely identical with that of the Korean haplotype. The CR sequence of the Korean sample (A-10) differed from that of the Japanese group A haplotypes but by only 1–4 sites (Fig. 2a, b). In addition, the major haplotype (A-2), which was genetically closest to the Korean haplotype (Fig. 2a) was found in all sampling sites within Japan, except site 1, which suggests that this haplotype has spread over a large part of Japan quite rapidly.
Hikida and Suzuki (2010), in a review of medieval Japanese documents on the wildlife of Japan, including turtles, mentioned that the oldest reference to M. reevesii had appeared in Ono's pharmacopias published in 1805 (Ono 1805), in which its distribution was mentioned as being confined to Fukuoka Prefecture, northern Kyushu. In this region, overseas trading with the Korean Peninsula was active at the time (Asada 2000), and Hikida and Suzuki (2010), therefore, suspected that M. reevesii had initially been introduced to Japan from Korea in the late 18th century. Hikida and Suzuki (2010), on the basis of information from old literature, also surmised that the distribution of M. reevesii had been restricted to western Japan until the late 19th century, and then has gradually extended eastward to the present. Because the haplotype group A was common to the Korean (site 20) specimen (Fig. 2) and specimens from western Japan, including localities in the vicinity of Fukuoka Prefecture (i.e., sites 15, 16, and 19: Fig. 3), this group of haplotypes may have possibly been possessed by those introduced individuals that had given rise to the first feral population in Japan and then gradually spread over western Honshu.
Group B contained 1 Taiwanese sample and both of the 2 Chinese samples (Fig. 2a, b) that showed no sequence divergence from the Japanese samples at all. Fong and Chen (2010) argued that M. reevesii may possibly be nonnative to Taiwan and that the individuals occasionally found there have derived from artificial introduction from continental China. The pattern of divergence and geographic distribution of the group B haplotypes demonstrated above (Figs. 2a, b, and 3) support the hypothesis of Fong and Chen (2010) and further suggest that the presence of this haplotype group in the current Japanese populations also is the consequence of recent artificial introduction, most likely from continental China. Yabe (2009) reported that males of 3 M. reevesii populations in Chiba Prefecture, eastern Honshu (near sites 1–3 here) could be discriminated from those of 5 conspecific populations in central Honshu and eastern Shikoku by a few morphologic characteristics, such as smaller body size. He suspected that the populations of eastern Honshu, including Chiba Prefecture, had originated from artificial introductions from continental China (Yabe 2009). The present results support this view.
In our survey, group B haplotypes notably predominated in the 3 easternmost sites (sites 1–3) and the 1 westernmost site (site 19) but with consistent absence in most other sites located between these extremes (Fig. 3). Group B haplotypes might have colonized there before the range extension of group A. In considering the complete absence of group B in western Honshu and Shikoku, this haplotype group, initially colonized to eastern Honshu, has apparently had no opportunities yet for westward dispersal, and this further suggests that the group B haplotypes in Kyushu (site 19) have resulted from an independent entry from continental China. Since the early 1970s, numerous M. reevesii have been imported by pet dealers from China to Tokyo near sites 1–3 and also to Kyushu where site 19 is located (Aoki, 1990; R. Aoki, pers. comm., 1992), and this may possibly be responsible for the present extremely disjunct distribution of the group B haplotypes in Japan.
Lee et al. (2009) molecularly evaluated the specific identity of 4 turtle shells obtained from commercial traders in Taiwan. We found that the partial cyt b sequence from one of their shells (FJ026853) is almost identical with that of the group C haplotype (Fig. 2c). Because the exact localities from which these turtle products were obtained are unknown (H.-M. Hsieh, pers. comm.), the origin of group C is yet to be confirmed. However, when considering that most, if not all, shells of M. reevesii sold in Taiwan originate from continental China rather than from within Taiwan or from Japan (Chen et al. 2009; Chen and Lue 2010), it is almost certain that the samples examined by Lee et al. (2009) also originated from continental China. Therefore, it is likely that the group C haplotype found in Japan has also originated from turtles artificially introduced from continental China.
Our study has generated conclusions and hypotheses that remain incompletely supported due to the small and relatively insufficient non-Japanese samples, those from the continent, in particular. Further genetic surveys of M. reevesii in Korea and China, therefore, are strongly recommended to verify our hypotheses regarding the probable nonnative origin of current Japanese populations.
Conservation Implications
Turtles are often transported artificially for various purposes, such as food materials, medicines, and pets (Kraus 2009; Lee et al. 2009). As a result, nonnative turtle populations have frequently been established (e.g., Sato and Ota 1999; Fong and Chen 2010). From the viewpoint of conservation biology, elucidation of the origin of a given turtle species in a given locality (i.e., native or nonnative) is essential for deciding its management direction (conservation, eradication, etc.). However, it often is not so easy to determine the origin of a turtle population, especially when apparently native populations of the same species occur in adjacent regions (Sato and Ota 1999). The population genetics approach in the present study, along with old literature records and archeological data, yielded largely plausible evidence for the nonnative nature of the Japanese M. reevesii populations.
As mentioned above, several researchers have reported recent occurrences of putative and genetically confirmed hybrids between M. reevesii and M. japonica (Kosuge et al. 2003; Haramura et al. 2008; Kato et al. 2010; Ota and Suzuki, unpubl. obvs., 1998 and 2006–2009). Furthermore, it has been observed, in captivity, that the F1 hybrids of the 2 species are fertile, being capable of producing F2 hybrids through a backcross with one of the parental species (Ota and Suzuki, unpubl. data, 1979, 1992, 2008, and 2009). Because M. reevesii has long been regarded as a species native to Japan, its hybridization with M. japonica has not been of concern in the context of conservation issues. However, because the Japanese populations of M. reevesii are now likely to be nonnative, the observed hybrids may, in reality, represent a first stage of genetic introgression from this species to M. japonica. Indeed, several previous studies have already shown that M. reevesii easily hybridizes with other geoemydid species (Wink et al. 2001; Galgon and Fritz 2002; Buskirk et al. 2005; Fong and Chen 2010). It, therefore, is probable that some M. japonica populations have already lost genetic features characteristic of this endemic Japanese species via introgression from M. reevesii. To clarify the present genetic condition and trend of M. japonica populations in the field, extensive surveys and subsequent monitoring by using nuclear gene markers are strongly recommended.
Results of the present study, along with those of some previous studies (e.g., Hirayama et al. 2007; Fong and Chen 2010; Hikida and Suzuki 2010), strongly suggest that the native range of M. reevesii is confined to the continental part of East Asia, where its natural populations are highly endangered by extensive human exploitation for food and medicine and habitat loss through artificial environmental changes (Bhupathy et al. 2000; Chen et al. 2009; Lovich et al. in press). In this situation, we may be able to recognize a special value for M. reevesii of Japan, which is exceptionally common in the wild in Japan and exhibits high genetic diversity that may correspond to the species' original diversity to some extent. For a species almost extinct in the wild, the reintroduction of ex situ bred individuals may offer an effective way to population recovery. In considering such an option, however, the following two, usually rather difficult, issues are essential: production of ex situ bred individuals in good numbers that are genetically as close to native individuals as possible, and maintenance of genetic diversity in the reintroduced population at a viable level through appropriate monitoring and management (Syed et al. 2007). Current Japanese populations of M. reevesii may possibly be the only source of individuals for such practices to conserve and reinforce continental populations of M. reevesii in the future. It, therefore, is desirable to keep under appropriate captive conditions genetically well-managed Japanese populations of M. reevesii, even as future studies likely verify their nonnative nature.
Map of East Asia, showing sampling localities of Mauremys reevesii. See Appendix 1 for further details.
Parsimony networks (TCS 1.21) of mitochondrial DNA haplotypes of Mauremys reevesii. The size of a symbol for a given haplotype reflects the frequency of individuals bearing this haplotype. Lines between symbols, between a symbol and a small open circle, and between small open circles (see below) represent one mutation step. Small open circles represent hypothetical intermediate haplotypes not sampled in this study. Networks in (a), (b), and (c) were based on 1844 base pair (bp) of combined cytochrome (cyt) b and control region, 1008 bp of cyt b, and 340 bp of cyt b, respectively. Haplotype C-1 was genetically much more distant from other haplotypes and thus was connected to neither of the other haplotypes in (a). In (a), numerals in parentheses just below their reference numbers denote sample sizes.
Map of East Asia, showing geographic distributions of haplotype groups of Mauremys reevesii. The ratios of white, black, and gray areas in each circle represent the frequencies of groups A, B, and C haplotypes, respectively. The total area of each circle reflects sample size as follows: small, 1 or 2 specimens; medium, 3–5 specimens; and large, 6 or more specimens.
