Accumulation of Microcystin (LR, RR and YR) in Three Freshwater Bivalves in Microcystis aeruginosa Bloom Using Dual Isotope Tracer

Stable isotope tracers were first applied to evaluate the Microcystis cell assimilation efficiency of Sinanodonta bivalves, since the past identification method has been limited to tracking the changes of each chl-a, clearity, and nutrient. The toxicity profile and accumulation of MC-LR, -RR and -YR in different organs (foot and digestive organs) from the three filter-feeders (Sinanodonta woodiana, Sinanodonta arcaeformis, and Unio douglasiae) were assessed under the condition of toxigenic cyanobacteria (Microcystis aeruginosa) blooms through an in situ pond experiment using 13C and 15N dual isotope tracers. Chl-a concentration in the manipulated pond was dramatically decreased after the beginning of the second day, ranging from 217.5 to 15.6 μg·L−1. The highest amount of MCs was incorporated into muscle and gland tissues in U. douglasiae during the study period, at nearly 2 or 3 times higher than in S. woodiana and S. arcaeformis. In addition, the incorporated 13C and 15N atom % in the U. douglasiae bivalve showed lower values than in other bivalves. The results demonstrate that U. douglasiae has less capacity to assimilate toxic cyanobacteria derived from diet. However, the incorporated 13C and 15N atom % of S. arcaeformis showed a larger feeding capacity than U. douglasiae and S. woodiana. Our results therefore also indicate that S. arcaeformis can eliminate the toxin more rapidly than U. douglasiae, having a larger detoxification capacity.


Introduction
Eutrophication in freshwater environments is a worldwide problem resulting in extensive blooms of potentially toxigenic cyanobacteria [1]. The development of cyanobacteria blooms has become a serious problem all over the world because many cyanobacteria have produced a wide variety of bioactive compounds including neurotoxins, hepatotoxins and cytotoxins in the past decades [2,3]. source (cyanobacteria) into consumers (filter feeders), in contrast to past studies where only changes in compositions of chl-a, clearity, and nutrients were taken as the evidence for these processes. Furthermore, the distribution patterns of three common MCs (MC-LR, MC-RR, and MC-YR) concentrations in differential organs (foot and digestive gland) of the bivalves are examined to discuss the possible mechanisms underlying these patterns, with comments on the potential risk to bivalve health when these bivalves consume toxic cyanobacteria (M. aeruginosa).

Characteristics of Ponds
To evaluate the effect of biocontrol on large cyanobacteria bloom of M. aeruginosa in the freshwater ecosystem, we conducted a biomanipulation test on in situ ponds using freshwater organisms, such as filter feeders (bivalves: U. douglasiae, S. woodiana and S. arcaeformis from the unionid family). The in situ pond tests were carried out from 13 October to 3 November 2008 to simulate a freshwater environment. The experimental site is located in the Sukmoon wetland (36 • 57 4 N, 126 • 37 5 E) in the mid-western region of South Korea (Figure 1). Ponds were classified into two types (see Figure 1): reference pond (RP) and manipulated pond (MP). Each pond enclosure was open at the surface and sealed at the bottom. They were constructed with transparent polyethylene and were suspended from frames made from PVC tubing. The enclosures were filled with reservoir water. Both ponds consisted of 100-ton tanks (10 × 5 × 2 m) and contained muddy sediment at the bottom, including reference pond (RP) and manipulated pond (MP): (1) Reference Pond (RP; in situ water containing predominantly M. aeruginosa).
Mar. Drugs 2017, 15, 226 3 of 15 discuss the possible mechanisms underlying these patterns, with comments on the potential risk to bivalve health when these bivalves consume toxic cyanobacteria (M. aeruginosa).

Characteristics of Ponds
To evaluate the effect of biocontrol on large cyanobacteria bloom of M. aeruginosa in the freshwater ecosystem, we conducted a biomanipulation test on in situ ponds using freshwater organisms, such as filter feeders (bivalves: U. douglasiae, S. woodiana and S. arcaeformis from the unionid family). The in situ pond tests were carried out from 13 October to 3 November 2008 to simulate a freshwater environment. The experimental site is located in the Sukmoon wetland (36°57′4″ N, 126°37′5″ E) in the mid-western region of South Korea (Figure 1). Ponds were classified into two types (see Figure 1): reference pond (RP) and manipulated pond (MP). Each pond enclosure was open at the surface and sealed at the bottom. They were constructed with transparent polyethylene and were suspended from frames made from PVC tubing. The enclosures were filled with reservoir water. Both ponds consisted of 100-ton tanks (10 × 5 × 2 m) and contained muddy sediment at the bottom, including reference pond (RP) and manipulated pond (MP): (1) Reference Pond (RP; in situ water containing predominantly M. aeruginosa).
(2) Manipulated Pond with bivalves (MP; in situ water containing predominantly M. aeruginosa + bivalves (U. douglasiae, S. woodiana and S. arcaeformis)). The filter feeders were captured from agricultural reservoirs using hand nets (mesh size <2 × 2 mm) and trawling. The mean shell length and width of the U. douglasiae (6.74 ± 0.51 cm and 3.82 ± 0.34 cm, respectively), S. woodiana (6.94 ± 0.46 cm and 3.97 ± 0.51 cm, respectively), and S. arcaeformis (6.78 ± 0.58 cm and 3.57 ± 0.42 cm, respectively) were slightly different; however, their standard errors overlapped. They were kept in aquaria filled with reservoir water in a layer of sand at 17~20 °C under a 16L:8D regime. Individuals were fasted for 48 h prior to the experiment. For the in situ pond experiment, the individual numbers of bivalves were determined through the preliminary experiments; bivalve U. douglasiae, S. woodiana and S. arcaeformis (5 ind. m −2 ) [32]. The filter feeders were captured from agricultural reservoirs using hand nets (mesh size <2 × 2 mm) and trawling. The mean shell length and width of the U. douglasiae (6.74 ± 0.51 cm and 3.82 ± 0.34 cm, respectively), S. woodiana (6.94 ± 0.46 cm and 3.97 ± 0.51 cm, respectively), and S. arcaeformis (6.78 ± 0.58 cm and 3.57 ± 0.42 cm, respectively) were slightly different; however, their standard errors overlapped. They were kept in aquaria filled with reservoir water in a layer of sand at 17~20 • C under a 16L:8D regime. Individuals were fasted for 48 h prior to the experiment. For the in situ pond experiment, the individual numbers of bivalves were determined through the preliminary experiments; bivalve U. douglasiae, S. woodiana and S. arcaeformis (5 ind. m −2 ) [32].

In Situ Artificial Pond Experiment
The tracer experiment was conducted when the massive toxic cyanobacteria bloom occurred in up to 89% of the phytoplankton biomass. As isotope tracers, NaHCO 3 (Isotech; 13 C > 99%) and (NH 4 ) 2 SO 4 (Isotech; 15 N > 99%) were added to each pond. The 13 C content of the dissolved inorganic carbon (DIC) pool was increased up to about 5% and the 15 N content of the dissolved inorganic nitrogen (DIN) pool was also stabilized by enriching with at most 10% every 48 h. Next, filter feeders (U. douglasiae, S. woodiana and S. arcaeformis) were added into the MP. The samples of water and bivalves were collected at the 1st, 2nd, 3rd, 4th, 5th, 6th, 8th, 10th, 13th, 16th and 21st days after the addition.

Enumeration and Biomass Determination of Phytoplankton
Water samples for the enumeration of phytoplankton cells were sedimented for 24 h, and a known volume of the concentrated sample was placed in a Sedgewick-Rafter counting chamber, where at least 300 cells were counted under ×200-400 magnification. Taxa (for example, Microcystis sp.) were enumerated only as larger units. Phytoplankton cell density was then determined on the basis of geometric solids that closely approximated each cell or colony shape [34].

Analysis of Stable Isotope Ratios
To analyze the stable isotope ratios of particulate organic matter (POM; mostly phytoplankton), water samples were passed through 20 µm mesh to remove zooplankton, and the remaining water was filtered through pre-combusted (450 • C, 24 h) glass fiber filters (Whatman GF/F, Sigma-Aldrach, Yong In, Korea) using gentle vacuum filtration. The filter samples were fumed for 24 h with saturated HCl to remove inorganic carbon and were dried completely using a freeze drier. The bivalve sampling was carried out using hand nets (mesh size: 2 × 2 mm) in the small pond. They were dissected in order to separate the digestive gland, stomach, muscle (abductor), gill, and mantle (three samples for each tissue). All tissue samples were freeze-dried and then ground into a fine powder using a grinder (FRITSCH-planetary mono mill, Pulverisette 6, Idar-Oberstein, Germany). The freezing and storage processes do not affect δ 13 C and δ 15 N values of biota tissue [35]. Homogenized powder samples of each tissue were decalcified with 1N HCl for at least 24 h to remove possible carbonates. However, subsamples for δ 15 N analysis were not treated with acid because it has been reported that HCl treatment affects δ 15 N values [36]. After the acid treatment, the samples were re-dried using a freeze drier and were ground to a fine powder, which was thoroughly mixed prior to analysis. Measurements of stable carbon and nitrogen isotopic ratios were performed with a continuous flow isotope ratio mass spectrometer (Isoprime; GV Instrument, Manchester, UK) coupled with an elemental analyzer (Euro EA 3000-D, Milan, Italy). Isotopic ratios were presented as δ values (‰) related to the Vienna PeeDee Belemnite (VPDB) standard and to atmospheric N 2 for carbon and nitrogen, respectively. The reference materials were IAEA-CH6 (δ 13 C = −10.45 ± 0.04‰) and IAEA-N1 (δ 15 N = 0.4 ± 0.2‰).
The analytical precision was within 0.2‰ and 0.3‰ for carbon and nitrogen, respectively. Isotope ratios were reported in per mil (‰) using standard delta notation, using Equation (1): where X = 13 C or 15 N, R = 13 C/ 12 C or 15 N/ 14 N, and std (standard) = Vienna Peedee Belemnite (VPDB) for carbon or air N 2 for nitrogen. The δ-values were converted to atom %, which is more appropriate for labeled samples. Conversion was performed according to Equation (2): where the A ns for carbon is 1.118 × 10 −2 and that for nitrogen is 3.677 × 10 −3 .

Determination of MCs Concentration and Production Rate
A bivalve sample was screened for the presence of neurotoxins and hepatotoxins, but since these were never found only results on MCs will be presented here. We choose the three most common MC compounds (-LR, -RR, -YR); this is also the case in the great majority of studies dealing with the presence and effects of toxicity on many aquatic organisms, thus enabling comparison with other results. The purification and analysis of MCs were carried out using the methods developed by Harada et al. [37]. From each sample of freeze-dried Microcystis cell material and filter feeders tissue, the MCs were extracted twice with 20 mL of 5% (v/v) acetic acid for 1 h while shaking at 140 rpm. The extract was centrifuged at 12,000× g, and then the supernatant was applied to a C18 cartridge (Sep-Pak; Waters Association, Milford, MA, USA). The cartridge was rinsed with water and 20% methanol in water. The eluate from the cartridge with 90% methanol in water was evaporated to dryness, and the residue was dissolved in 100% methanol. The sample solution was analyzed on an HPLC (Agilent Technologies 1200 series, Santa Clara, CA, USA). The separation was performed on an ODS column (Cosmosil 5C18-AR, 4.6 mm × 150 mm) reverse-phase column and the mobile phase was a 0.1% formic acid:water solution with constant flow at 1 mL min −1 . The MCs were identified by their UV spectra and retention times, and by spiking the sample with purified standards of MC-LR (Sigma, Sigma-Aldrach, Yong In, Korea), MC-YR (Sigma, Sigma-Aldrach, Yong In, Korea) and MC-RR (Sigma, Sigma-Aldrach, Yong In, Korea). MCs concentrations were determined by comparing the peak areas of the test samples with those of the standards available (MC-LR, MC-YR and MC-RR, Sigma, Sigma-Aldrach, Yong In, Korea).

Statistical Analysis
All data were tested for normality and homogeneity of variances (Levene's median test). A paired-difference t test was used to analyze the relationship of environmental factors between two ponds. A p-value of less than 0.05 was considered as statistically significant. All statistics were performed using SPSS software (version 22.0, SPSS Inc., New York, NY, USA).

Water Quality Condition in an Artificial Enclosed Pond
Biocontrol with bivalves ponds (MP) resulted in significant differences in water quality parameters such as Chl-a (p < 0.001) and DIN (p < 0.001) concentration in the cyanobacteria blooming environment in comparison with those of the RP pond ( Table 1). The DIN concentration of the MP by the third day was maintained at the range of 1.5 and 1.8 mg L −1 , which was higher than the RP since the phytoplankton exhausted the nutrients from the water column through photosynthetic activity in RP, while the filter feeders could release nutrients as fecal pellets into the water column in MP ( Figure 2a). The DIP concentration of the RP was higher until the fifth day, but fluctuation was observed from the sixth day to the end of the experimental period, probably due to resuspension of the phosphorus from sediment through windy turbulence and organism movement ( Figure 2b). The Chl-a concentration in MP was dramatically decreased compared with that of the reference pond after the beginning of the second day, with a range of 15.6-217.5 µg L −1 (Figure 2c). This indicates that phytoplankton containing toxic cyanobacteria (M. aerugionsa and A. spiroides) were eliminated by filter feeders as their prey. The water temperature showed the same variation in both ponds, ranging from 16 to 24 • C ( Figure 2d). Table 1. Statistical t test results for environmental parameters between the RP and MP pond (n = 5), (* p < 0.05). RP: Reference pond, MP: Manipulated pond.

Parameters (Unit)
p Value filter feeders as their prey. The water temperature showed the same variation in both ponds, ranging from 16 to 24 °C (Figure 2d).

The Composition and Biomass Variation of Phytoplankton
Biocontrol with bivalves ponds (MP) resulted in significant differences in phytoplankton cell density (p < 0.005) in the cyanobacteria blooming environment in comparison with those of the RP pond ( Table 1). The relative proportions of phytoplankton species at the beginning of experiment showed that most dominant species were cyanobacteria (87.8%) and bacillariophycae (7.9%), followed by crytophycee (0.7%), chlorophyceae (3.0%) and dinophyceae (0.5%) ( Table 1). Species of cyanobacteria were composed of M. aerugionsa, A. spirodides, Synechocystis pevalekii, Aphanocapsa elachista, and Chroococcus sp. in both ponds at the beginning of the experiment ( Table 2). The cell density of phytoplankton in the RP maintained a constant biomass with small fluctuations during the entire experiment, but dramatically decreased in the MP (Figure 3). At the end of the experiment, the relative proportion of cyanobacteria species in the RP showed a 6% difference, from 88% to 82%, but the MP had a 67% difference, from 88% to 21% (Figure 3). These results supported the biomanipulation effect of multiple filter feeders in reducing cyanophyceae biomass.

The Composition and Biomass Variation of Phytoplankton
Biocontrol with bivalves ponds (MP) resulted in significant differences in phytoplankton cell density (p < 0.005) in the cyanobacteria blooming environment in comparison with those of the RP pond ( Table 1). The relative proportions of phytoplankton species at the beginning of experiment showed that most dominant species were cyanobacteria (87.8%) and bacillariophycae (7.9%), followed by crytophycee (0.7%), chlorophyceae (3.0%) and dinophyceae (0.5%) ( Table 1). Species of cyanobacteria were composed of M. aerugionsa, A. spiroides, Synechocystis pevalekii, Aphanocapsa elachista, and Chroococcus sp. in both ponds at the beginning of the experiment ( Table 2). The cell density of phytoplankton in the RP maintained a constant biomass with small fluctuations during the entire experiment, but dramatically decreased in the MP (Figure 3). At the end of the experiment, the relative proportion of cyanobacteria species in the RP showed a 6% difference, from 88% to 82%, but the MP had a 67% difference, from 88% to 21% (Figure 3). These results supported the biomanipulation effect of multiple filter feeders in reducing cyanophyceae biomass.

Measurement of 13 C and 15 N Atom % of Filter Feeders
Most filter feeders showed continuous apparent enrichment of 13 C and 15 N tracers in their tissues during the experimental period ( Figure 4). Distinguished by filter feeder species, in the MP, U. douglasiae showed 1.08~1.11 13 C and 0.36~0.61 15 N atom % respectively, S. woodiana showed a 1.08~1.15 13 C and 0.36~0.99 15 N atom % respectively, and S. arcaeformis showed a 1.08~1.20 13 C and 0.36~1.28 15 N atom % respectively, demonstrating that S. arcaeformis has higher assimilation efficiency for toxic cyanobacteria cells through feeding activity ( Figure 5). The different atom percentages might indicate different assimilation capacities among the freshwater filter feeder species. The filter feeders showed different isotopic enrichment among the tissues, showing more enrichment in the digestive

Measurement of 13 C and 15 N Atom % of Filter Feeders
Most filter feeders showed continuous apparent enrichment of 13 C and 15 N tracers in their tissues during the experimental period ( Figure 4). Distinguished by filter feeder species, in the MP, U. douglasiae showed 1.08~1.11 13 C and 0.36~0.61 15 N atom % respectively, S. woodiana showed a 1.08~1.15 13 C and 0.36~0.99 15 N atom % respectively, and S. arcaeformis showed a 1.08~1.20 13 C and 0.36~1.28 15 N atom % respectively, demonstrating that S. arcaeformis has higher assimilation efficiency for toxic cyanobacteria cells through feeding activity ( Figure 5). The different atom percentages might indicate different assimilation capacities among the freshwater filter feeder species. The filter feeders showed different isotopic enrichment among the tissues, showing more enrichment in the digestive gland and stomach but relatively low incorporation rates in mantle and gill, indicating that each organ had differential turnover rates (Figure 4). gland and stomach but relatively low incorporation rates in mantle and gill, indicating that each organ had differential turnover rates (Figure 4).

MC Concentration of POM and Three Kinds of Bivalves
The MC concentrations of the POM in the RP and MP showed remarkable variation ( Figure 6). Within two days of beginning the experiment, the MC concentration showed a remarkable decrease of the POM in the MP while the RP maintained a relatively constant MC concentration through the end of the experiment. However, those values did increase slightly, possibly due to release from filter feeders as fecal pellets or expelled cyanobacteria cells in water column at last 7 days. gland and stomach but relatively low incorporation rates in mantle and gill, indicating that each organ had differential turnover rates (Figure 4).

MC Concentration of POM and Three Kinds of Bivalves
The MC concentrations of the POM in the RP and MP showed remarkable variation ( Figure 6). Within two days of beginning the experiment, the MC concentration showed a remarkable decrease of the POM in the MP while the RP maintained a relatively constant MC concentration through the end of the experiment. However, those values did increase slightly, possibly due to release from filter feeders as fecal pellets or expelled cyanobacteria cells in water column at last 7 days.

MC Concentration of POM and Three Kinds of Bivalves
The MC concentrations of the POM in the RP and MP showed remarkable variation ( Figure 6). Within two days of beginning the experiment, the MC concentration showed a remarkable decrease of the POM in the MP while the RP maintained a relatively constant MC concentration through the end of the experiment. However, those values did increase slightly, possibly due to release from filter feeders as fecal pellets or expelled cyanobacteria cells in water column at last 7 days. DW in gland, respectively. The highest amount of MC was incorporated into muscle and gland tissues in U. douglasiae during the study period, nearly two or three times higher than shown in S. woodiana and S. arcaeformis. Also, the MC concentration in gland tissue among the filter feeders was substantially higher compared to that in muscle tissue, due to differential turnover rates.

Dynamics of Feeding Activity and MCs Concentration in Filter Feeders
In this study, t tests of environmental variables means such as DIN (p < 0.001), Chl-a (p < 0.001) concentration and phytoplankton cell density (p < 0.005) showed that there were significant differences between the RP and MP ponds (Table 1). Chl-a and DIN concentration were less in both MP ponds compared with those in the RP pond. This suggests that three kinds of bivalves preyed on DW in gland, respectively. The highest amount of MC was incorporated into muscle and gland tissues in U. douglasiae during the study period, nearly two or three times higher than shown in S. woodiana and S. arcaeformis. Also, the MC concentration in gland tissue among the filter feeders was substantially higher compared to that in muscle tissue, due to differential turnover rates. DW in gland, respectively. The highest amount of MC was incorporated into muscle and gland tissues in U. douglasiae during the study period, nearly two or three times higher than shown in S. woodiana and S. arcaeformis. Also, the MC concentration in gland tissue among the filter feeders was substantially higher compared to that in muscle tissue, due to differential turnover rates.

Dynamics of Feeding Activity and MCs Concentration in Filter Feeders
In this study, t tests of environmental variables means such as DIN (p < 0.001), Chl-a (p < 0.001) concentration and phytoplankton cell density (p < 0.005) showed that there were significant

Dynamics of Feeding Activity and MCs Concentration in Filter Feeders
In this study, t tests of environmental variables means such as DIN (p < 0.001), Chl-a (p < 0.001) concentration and phytoplankton cell density (p < 0.005) showed that there were significant differences between the RP and MP ponds (Table 1). Chl-a and DIN concentration were less in both MP ponds compared with those in the RP pond. This suggests that three kinds of bivalves preyed on cyanobacteria (M. aeruginosa), which resulted in greater water clarity. However, reduced Chl-a concentration may not be direct evidence of bivalves assimilation of M. aeruginosa because bivalves may expel toxic cyanobacteria without digesting them in the form of feces or pseudofeces that is released into the water column or sinks to the bottom. Therefore, the grazing efficiency of bivalves was evaluated according to the 13 C and 15 N atom % incorporated into the biota.
This study is the first to directly compare the assimilation rates and MC concentrations of three different kinds of filter feeders in M. aerugionsa blooms. The inorganic forms of 13 C and 15 N tracers were assimilated into the cyanobacteria body as organic forms through photosynthesis, and synthesized new phytoplankton cells. The atom % increase of filter feeders indicates that newly synthesized phytoplankton cells were assimilated into the organism's cells through feeding activity. The 13 C and 15 N atom % of three kinds of filter feeders showed continuous apparent enrichment ( Figure 4) and apparent increases of MC concentration in their body tissues (Figure 7). Also, the Chl-a concentration and phytoplankton biomass in the MP decreased throughout the study period (Figures 2 and 3). These results indicate that filter feeders have a large capacity to assimilate a cyanobacteria-derived diet. Filter feeders such as bivalves can filter a broad size range of particle diameters, significantly reduce phytoplankton populations, and greatly increase water clarity [14]. In particular, unionid bivalves can filter large amounts of the water column and decrease phytoplankton densities in cases of large biomass [26,38]. Therefore, active filtration was exhibited by unionid filter feeders in this study, and led to declines in phytoplankton biomass, even in a large water body.
However, filter feeders cannot avoid the accumulation of toxins in their body from ingestion of toxic cyanobacteria during a bloom period. The distinction between 13 C and 15 N atom % among the filter feeders shows contrasting ingestion rates due to differential incorporation of Microcystis cells in their body tissues. To elucidate the different 13 C and 15 N atom % and MC concentrations among the three filter feeders (U. douglasiae, S. woodiana and S. arcaeformis for the family of unionid) (Figures 4 and 7), toxic MCs bioaccumulation and its depuration rate need to be considered because toxic Microcystis impacts the prey ingestion rate of bivalves [9]. Yokoyama and Park [39] have reported that the concentration of bioaccumulated MC in S. woodiana bivalves was very low compared to that in U. douglasiae bivalves, claiming that S. woodiana might reject toxic Microcystis as pseudofeces prior to ingestion. However, in our study, U. douglasia bivalves might expel more toxic Microcystis, because 13 C and 15 N atom % values in S. woodiana and S. arcaeformis were higher in comparison with those of U. douglasiae bivalves. These results suggest that S. woodiana and S. arcaeformis would continuously ingest newly photosynthsized Microcystis cells into their body, but they may have an immunological system for depuration of toxic MCs, unlike U. douglasiae. U. douglasiae assimilated Microcystis cells, but those were expelling bioaccumulated Microcystis cells into the water column, suggesting its limited ability to incorporate Microcystis cells ( Figure 5).
In the present study, we observed a rapid uptake of MC by filter feeders, since toxins were detected in muscle and gland tissues shortly after the beginning of the cyanobacteria bloom (Figure 7). Direct toxin transference to the filter feeders through oral ingestion can be interpreted from the existence of cyanobacteria in the water column. Among the filter feeders, the highest amount of MC was found in U. douglasiae, nearly two or three times higher than shown in S. woodiana and S. arcaeformis (Figure 7), reporting studies of MC bioaccumulation in bivalves implemented in both laboratory [40][41][42] and field [14,39,43]. Similar results have been found in other studies: Yokoyama and Park (2003) reported extremely high accumulations of MC-LR in U. douglasiae through laboratory experiment. Chen et al. [44] reported that toxic cyanobacteria application to S. woodiana resulted in lower accumulated hepatopancreas concentrations of MC, in comparison with three other kinds of filter feeders. Watanabe et al. [43] reported that S. woodiana, U. douglasiae and Cristaria plicata accumulated Microcystis at a range of 1 ± 5 mg/mussel during a large Microcystis bloom. Hepatopancreas tissues for S. woodiana showed lower MC-LR values (0.21 µg·g −1 DW) in Lake Suwa in Japan, compared to MC-LR values (2.7 µg·g −1 DW) in U. douglasiae. Moreover, Yokoyama et al. [39] reported that MC was detected at much higher concentrations in U. douglasiae (420 µg·g −1 DW) than in S. woodiana (12.6 µg·g −1 DW).
We found similar tendencies to past studies: it is clear that MC were detected at much higher concentrations in U. douglasiae, compared to in S. woodiana and S. arcaeformis (Figure 7). MC bioaccumulation in U. douglasiae was influenced not only by intracellular MC in toxic cyanobacteria (M. aeruginosa) but also by their depuration rate. These different bioaccumulation patterns among the filter feeders can be interpreted as interspecific differences in selective ingestion, MC metabolism, and depuration rate. First, selective ingestion by filter-feeding bivalves is a well-known feeding mechanism. The bivalves were capable of filtering a wide size range of sestonic particles. Ingestible particles are directed straight into the mouth, whereas unwanted particles are sorted on the gills and palps, then expelled as pseudofeces in the surrounding water [17]. However, we expect that these results were caused not only by selective grazing but also by the prey density effect, because of thẽ 90% composition of toxic cyanobacteria in the total phytoplankton abundance. Based on our study, unionid families could not ingest the toxic cyanobacteria as a prey even though they could selectively ingest prey, because of the impact on their physiological condition by harmful compounds (MCs).
Second, based on our results, it is clear that S. woodiana and S. arcaeformis bivalves differ from U. douglasiae in MC metabolism, which is involved with uptake and depuration. Possibly U. douglasiae depurated toxins more slowly over the same time, since its MC concentration was higher than in S. woodiana and S. arcaeformis. This differs from our results, which suggest MC may be more resistant to degradation in the U. douglasiae, as suggested by higher MC concentration in both organs. This kind of depuration process may be a result of cycles in the production and degradation of a protein phosphatase-MC adduct [45]. The activity of detoxification enzymes such as glutathione-Stransferase and glutathione peroxidase has been shown to increase in embryos of zebra fish (Danio rerio) exposed to MC-LR [46]. Pflugmacher et al. [47] demonstrated that MC was detoxified as a result of conjugation with glutathione in various aquatic organisms including zebra mussels, fish, and daphnid. Therefore, our results demonstrated that S. woodiana and S. arcaeformis would ingest toxic cyanobacteria, but they may have an immunological system that allows them to depurate toxic MCs in contrast to U. douglasiae. Also, MC-LR concentration was lower in muscle tissue in both S. woodiana and S. arcaeformis bivalves during the 14 days from the beginning of the experiment, while those values increased after 14 days. The increased MC-LR concentration might indicate that the detoxification activity of MC-LR decreased, leading to excretion or physiological degradation. It is known that MC-LR can conjugate with glutathione (GSH) or cysteine and ultimately degrades to MC-LR-Cys, consequently reducing the toxicity and enhancing the excretion of MC-LR [48]. Therefore, conjugation of MC-LR with cysteine or GSH in muscle tissue might be an important pathway for the detoxification and excretion of MC-LR in S. woodiana and S. arcaeformis. For future study, to better understand the role of MC-GSH and MC-Cys conjugates in the detoxification of MCs in bivalves, quantitative evaluation of such conjugates in filter feeder tissues is needed.

Stable Isotope Fractionation in Differential Organs
The filter feeders showed different isotopic enrichment among their tissues, showing more enrichement in their digestive glands and stomach organs but relatively low incorporation rates in mantle and gill (Figure 4). A distinction of isotopic ratios and turnover rates for each tissue has been observed in mussels Pecten masimus [49] and Mytilus galloprovincialis [50]. Incorporation reflects the metabolic breakdown of old tissue and subsequent replacement by tissue synthesized from the new diet. Raikow and Hamilton [51] demonstrated that stable isotope enrichment of the tissues (gland and muscle) of unionid bivalves reflected short-term and long-term assimilation of nutrients. Different isotopic enrichment of tissues from three bivalves showed more enriched atom % in the digestive glands and stomach organs during the entire experimental period because those organs have relatively faster turnover rates and reflect short-term assimilation records [51,52] (Figure 4). The muscle organs of the bivalves showed lower enriched atom % because of their slow turnover and relatively long-term integration of energy source [51]. Mantle and gills tissues also indicated relatively low incorporation rates, because these organs have slow turnover rates, integrating diet isotopic signatures over a longer time period than other tissues, such as digestive glands, that have faster turnover rates [4,49]. Hill et al. [53] reported that mantle tissue replaced its carbon isotopes faster than gill parts, while gill tissue was faster than adductor muscle in the mussel Perna perna, because mantle is the main energy source for sustaining gonadal development. However, a differential tendency of a faster turnover rate was observed in gills compared to mantle organs (Figure 4). Mantle organs in unionoid bivalves in Korean lakes and reservoirs metabolically synthesized a little active gonad, due to the warm season for spawning from April to June. Therefore, incorporated 13 C and 15 N atom % in mantle organ showed lower values than in gill organs, since our studied species rarely breed during the study period.

Management and Implications for Water Quality
Among the freshwater bivalves, zebra mussels (D. polymorpha) showed higher clearance rates in single cells of the Microcystis [54], with a reverse gradient between mussel densities and the cyanobacteria biomass [55]. Therefore, mussels could be used as a potential tool in the biomanipulation of shallow lakes suffering from harmful cyanobacteria blooms [21,56]. However, Vanderploeg et al. [14] have argued that D. polymorpha promoted the return of Microcystis blooms, because they selectively ejected colonies of Microcystis. These mussels make a seston selection but respond differently to single cells or colonies of cyanobacteria. Non-native mussel species (D. polymorpha) are a threat to native mussels because invasive species often differ greatly from native species in resource use and trophic interactions; they have great potential to negatively affect ecosystems [57,58]. Therefore, unionids are better adapted to shallow lakes or reservoirs with soft substrates than zebra mussels, which prefer hard substrate for settlement. Native bivalve species (unionids) are recommended instead of invasive species (D. polymorpha) for biomanipulation on harmful cyanobacteria blooms in lakes or reservoirs. Unionid mussels can filter large amounts of the water column, as long as their biomass is large [38]. However, the research on common native freshwater bivalves (unionid mussels, like Sinanodonta sp. and Unio sp.), especially under toxic cyanobacteria bloom conditions, is still lacking. Therefore, this study provided the responses of bivalve species in mesocosm to evaluate their bio-control effects. Xie et al. [59] and Yokoyama [39] have revealed that Sinanodonta bivalves persist intoxic cyanobacteria bloom and show lower bioaccumulation patterns of MC. Sinanodonta's grazing capacity per individual is equal to or higher than that of D. polymorpha [9]. For example, Sinanodonta bivalves showed × 12 higher efficiencies of clear unit per square meter per unit of water volume than Dreissena mussels [9,17]. Sinanodonta bivalves had a larger effect on the primary productivity of toxic cyanobacteria, demonstrating a more realistic achievement for biomanipulation. Also, Sinanodonta (S. arcaefomis and S. woodiana) should incorporate more toxic cyanobacteria cells than U. douglasiae, probably due to its larger detoxification capacity. As a result, S. arcaefomis and S. woodiana could be useful organisms to reduce massive toxic cyanobacteria blooms in eutrophic agricultural reservoirs and lakes. This is the first study using stable isotope tracers to evaluate the Microcystis cell assimilation efficiency of Sinanodonta bivalves.

Conclusions
We conclude from our study that S. arcaefomis is the most efficient grazer of toxic cyanobacteria. S. arcaeformis can eliminate the toxin more rapidly than U. douglasiae, probably due to its larger detoxification capacity. However, U. douglasiae has a smaller capacity to assimilate toxic cyanobacteria derived from diet, and is affected by the occurrence of toxin-producing cyanobacteria bloom. For biomanipulation purposes, we suggest that the use of Sinanodonta deserves more attention since this is a native (endemic) bivalve, contrary to U. douglasiae, which probably is not suitable for biomanipulation that involves the removal of toxic cyanobacteria.