Species belonging to the genus Corbicula attain an almost worldwide distribution and inhabit in estuaries, lakes, and rivers [1] -[3] . In Japan, 3 endemic Corbicula species are reproductively and developmentally heterogeneous [4] -[7] . Although C. leana and C. sandai appear fresh waters only in Japan, C. japonica inhabits blackish waters throughout eastern Asia from Kyushu Island in Japan to Sakhalin Island in Russia between approximately 27˚N and 55˚N [8] .

C. japonica has been one of the most important species for inland fisheries in Japan [9] , and occupied 99% of the total catch of Corbicula [8] . The catch of Corbicula from 2002 to 2011 reached average 12,000 tons, which corresponded to approximately 30% of the total catch of inland fisheries [10] . As fisheries of C. japonica have so far prospered in a number of lakes and marshes, Lake Shinji in Shimane Prefecture, Lakes Jusan and Ogawa- ra in Aomori Prefecture, Lake Abashiri in Hokkaido Prefecture, and Kiso Three Rivers in Mie, Gifu and Aichi Prefectures are recent major fishing areas [11] . However, the catch of C. japonica decreased since 1970s due to overfishing, habitat destruction and coastal pollution caused by reclamation work and estuary barrage [12] . For the purpose of an effort to promote stock restoration and increase fisheries production, seedlings and/or adults of C. japonica suffered to frequent transplantation among fishing areas. In addition, a bulk of anonymous Corbi- cula individuals was successively imported from eastern Asia [12] . It has been therefore concerned with the ge- netic disturbance of C. japonica in Japan, but its genetic diversity and reproduction structure are still not well understood.

Mitochondrial DNA (mtDNA) has been extensively used for both population genetic and systematic genetic studies on various taxonomic levels [13] . The mtDNA cytochrome c oxidase subunit I (COI) gene has often been adopted as a tool for determining the genetic diversity of fisheries valuable bivalves in Japan [14] -[16] . The aim of this study was to evaluate the genetic variability and reproduction structure of C. japonica popula- tions among 4 major fishing brackish lakes in Japan using the mtDNA COI gene. The genetic information con- tributes to stock management of C. japonica resources for long-term sustainable fisheries and aquaculture.

The average shell length and shell height of C. japonica individuals collected in Lakes Jusan, Ogawara and Ab- ashiri ranged from 22.34 ± 1.43 mm to 26.73 ± 2.51 mm and from 20.42 ± 1.31 mm to 23.90 ± 2.24 mm, re- spectively. In contrast, the respective values of those collected in Lake Shinji were determined to be 18.16 ± 2.17 mm and 15.87 ± 1.76 mm, and relatively small shell size compared with the other lakes (Table 1). Fishe- ries cooperatives have defined minimum shell size of the catch for stock management in individual lake, and 10, 12, 15 and 14 mm are in Lakes Shinji, Jusan, Ogawara and Abashiri, respectively.

After editing nucleotide sequences of PCR product encoding the mtDNA COI gene, a portion of 556 bp were aligned from 188 individuals collected in 4 lakes, and 24 variable sites and no insertion or deletion were ob- tained (Table 2). Of 25 detected haplotypes identified to be C. japonica (Figure 2), only the haplotype HT01 was shared among all lakes with a high frequency accounting for 78.8% of 188 individuals, and remaining 24 haplotypes from H02 to H25 were unique to individual lake. The haplotype diversity h values ranged from 0.2026 ± 0.0778 in Lake Abashiri to 0.6040 ± 0.0566 in Lake Shinji (Table 3). Although the frequency of the haplotype H01 was good correlation with the haplotype diversity, the numbers of haplotypes were poor correla- tion with the h values. The nucleotide diversity π values of Lakes Jusan, Ogawara and Abashiri ranged from 0.0591% ± 0.0671% to 0.0601% ± 0.0678%, showing relatively low level diversity compared with that value of Lake Shinji to be 0.2207% ± 0.1586% (Table 3).

Haplotype network obtained out of 25 haplotypes revealed star-like shape with a partly bush-like clade (Fig- ure 3). Haplotype networks consisted of almost unique haplotypes with 1-2 substitution difference radiating from the haplotype H01, and 3 haplotypes further radiated from the haplotype H02 with 1 substitution difference only in Lake Shinji. They supposedly had a recent history of continuous reproduction, and Lake Shinji was more stable in the reproduction structure than the other lakes. These results were consistent with a markedly high level of the π value in Lake Shinji compared with the other lakes (Table 3).

The pairwise F_ST and Φ_ST values among 4 lakes ranged from −0.0074 to 0.2011 and from 0.0027 to 0.2715, respectively, and both values between Lake Shinji and the other lakes were evaluated to be significant (Table 4). In addition, the NJ trees based on the pairwise F_ST and Φ_ST values separated Lake Shinji from the other lakes (Figure 4).

In Lakes Jusan and Ogawara, the mismatch distribution appeared to be unimodal clearly, which closely matched the simulated values, whereas that in Lakes Shinji and Abashiri appeared to be bimodal (Figure 5). The

Dot indicates nucleotide identical to that of HT01 sequence.

^*p < 0.05, ^**p < 0.01 following Bonferroni correction (k = 6).

sum of square deviations (SSD) ranged from 0.0005 in Lake Jusan to 0.0563 in Lake Shinji with no statistical significance (p > 0.05), and the Harpending’s raggedness index (Hri) ranged from 0.2071 in Lake Shinji to 0.5291 in Lake Abashiri with no statistical significance (p > 0.05) (Table 5). The Tajima’s D and Fu’s F_S values for all lakes were estimated to be negative, and both values for Lake Shinji were not significant excepting the other lakes (Table 5). These results indicate that C. japonica occurred a recent expansion and formed the current geographical distribution in Japan.

C. japonica is an endemic species of East Asia, but little is known about its biological and ecological properties in the native range, because it had been one of the most common species with abundant resources in various parts of Japan. However, its stock level declined over 40 years since 1970s, as its habitat lost due to water pollu- tion and estuary modification [12] . To solve resources reduced, C. japonica individuals had been frequently transplanted mainly from Lake Shinji to every fishing area in Japan. These situations strengthened concerns over the genetic loss in C. japonica populations. In this study, we selected 4 major fishing brackish lakes with no in- troduction record to determine the genetic diversity of C. japonica in its native range (Figure 1).

Recent developments of genetic identification keys have helped to resolve questions surrounding bivalve ge- netic studies, and mtDNA markers such as 16S ribosomal RNA (rRNA) gene and COI gene were specifically developed for Corbicula species. Komaru et al. [30] reported genetic discrimination of recognized Corbicula species based on the 16S rRNA gene, which was verified as a sensitive molecular marker for species identifica- tion of marine bivalves [31] [32] . Suzuki et al. [33] reported population genetic structure of C. japonica in Japan based on the COI gene, which was also verified as a sensitive molecular marker for population differentiation of marine bivalves [16] [34] .

In this study, the mtDNA COI gene was adopted to evaluate the genetic diversity and reproduction structure of C. japonica populations in major fishing brackish lakes (Table 2). The genetic diversity in Lake Shinji was higher than that in the other lakes (Table 3), and it therefore provided evidence of highly significant genetic dif- ferentiation between Lake Shinji and the other lakes (Figure 4, Table 4). This genetic differentiation was cau- sed by unique population genetic structures on haplotype networks, which were radiated from the haplotypes H01 and H02 in Lake Shinji and from the haplotypes H01 alone in the other lakes (Figure 3). Otherwise, the mismatch distribution analysis indicated unimodal profile for Lakes Jusan and Ogawara and bimodal profile for Lakes Shinji and Abashiri, and the respective profiles show recent histories of population growth and continuous reproduction (Figure 5). These reproduction structures could depend on different seedling systems that were ar- tificial mass seedling production in Lakes Jusan and Ogawara and natural seedling production in Lakes Shinji and Abashiri. All results obtained in this study allowed us to characterize the genetic diversity and reproduction structure in 4 brackish lakes as follows; continuous reproduction with a high diversity in Lake Shinji, rapidly grown reproduction with a low diversity in Lakes Jusan and Ogawara, continuous reproduction but irregular ex- pansion with a low diversity in Lake Abashiri.

Passive dispersal of planktonic larvae may be strongly mediated by water movements, and the effect of water movements on the genetic diversity was commonly suggested in aquatic invertebrates [35] [36] . From spawning to Juveniles stage of C. japonica lasted approximately 12 days followed by planktonic stage to larval settlement for 54 hours to 6 days [37] [38] . Larvae reached the size of 4 - 5 mm shell length in1 year and 14 - 15 mm in 3

Parameters of the spatial expansion model and goodness of fit test to the model are shown with the respective significance for each clade. SSD, sum of squared deviations; Hri, Harpending’s raggedness index. ^*p < 0.05, ^**p < 0.01.

years, and those with >15 mm matured in Lakes Shinji [39] , Ogawara [40] and Abashiri [41] . The spawning season of C. japonica in Lake Jusan was reported to be mid-July to late September [39] , and that in Lake Oga- wara was to be late July to early September [42] . It is noteworthy that C. japonica in Lake Abashiri failed to spawn from year to year due to low temperature and salinity [43] . The spawning season in Lake Shinji was, however, to be late March to early November and remarkably longer than that in the other lakes [42] . Iidzuka et al. [15] reported that the long spawning season might produce the complicated population genetic structure in Lake Shinji. It is therefore inferred that the difference of spawning in the season and length reflects in not only the genetic diversity but also the reproduction structure of C. japonica population in individual lake.

This study thoroughly compared population genetic structure of C. japonica among 4 major fishing brackish lakes in Japan by means of nucleotide sequence analysis of a 556 bp portion of the mtDNA COI gene. Although C. japonica is supposed to occur a recent expansion and form the current geographical distribution in Japan, its genetic diversity and reproduction structure has been developed specifically to the lakes examined in this study. In order not only to conserve C. japonica populations as a genetic resource but also to manage them as a fishe- ries stock, haphazard transplantation and import of seedlings and/or adults of C. japonica and its related species should be absolutely avoided.

This work was supported in part by the Budget for Strategic Operation from Shimane University.