Application of eDNA for monitoring freshwater bivalve Nodularia nipponensis and its glochidium larvae

Abstract

Habitat loss, the introduction of invasive species, and climate change due to human activity have threatened many freshwater unionid bivalves worldwide. Unionid bivalves represent important members of freshwater ecosystems, providing bitterling fish with spawning grounds and contributing to water clarification via the filtration of suspended solids. This study examined an environmental DNA (eDNA) approach for monitoring the unionid bivalve Nodularia nipponensis in a lake environment. We developed oligonucleotide primers for the specific detection of mitochondrial DNA from N. nipponensis and evaluated the decay and release rates of eDNA in a laboratory tank experiment under two bivalve densities (49 and 164 individuals m⁻²) selected to ensure sufficient eDNA yields. The eDNA release rate was weakly and positively correlated with bivalve density but strongly and positively correlated with inorganic nitrogen and phosphorus excretion from the bivalve. In a eutrophic lake, Lake Hachiro, 67% of the samples (16 out of 24) collected from June to July yielded detectable amounts of eDNA and were not detected at all in May and August (0 out of 20). This difference was probably due to the release of N. nipponensis glochidium larvae. During these months, unionid glochidia adhered to bundles of polyvinylidene chloride fibers placed on the lake bottom for 8 h, confirmed by microscopic inspection. We successfully obtained N. nipponensis eDNA from all bundles examined over the months, demonstrating the efficacy of bundle capture in quantifying unionid bivalve eDNA. Our eDNA‐based technique with specific oligonucleotide primers and fiber bundles may be a promising tool for monitoring N. nipponensis in freshwater environments. Fiber bundles are useful for tracing eDNA derived from glochidium larvae, even at low unionid bivalve densities, which may be conducive to conservation efforts.

INTRODUCTION

In freshwater environments, unionid bivalves interact with various aquatic organisms and play an important role in the ecosystem. However, many unionids are classified as threatened species (Lopes-Lima et al., 2014). The bitterling fish, such as Rhodeus ocellatus kurumeus (Kitamura, 2005) and Acheilognathus typus (Sakata et al., 2017), rely on unionid bivalves as an oviposition site, and the bivalves depend on specific fish species as larval hosts (Kondo, 2008; Marshall et al., 2018). Invertebrates and algae use unionid shells (Vaughn & Spooner, 2006). Furthermore, unionid bivalves act as powerful filter feeders that remove particulate organic matter from both the water column and interstitial sediments (Marroni et al., 2014; Vaughn et al., 2008), contributing to the clarification of lake water. Unionid bivalves are employed as a potent tool for controlling cyanobacterial blooms (Kim et al., 2018; Sugawara et al., 2021). Unionid bivalve populations have recently declined worldwide because of habitat loss and fragmentation, changes in river flow, pollution, overexploitation, the introduction of non-native species, and climate change (Lopes-Lima et al., 2014; Strayer, 2008). Water quality management by maintaining an aquatic ecosystem inhabited by fish, invertebrates, and algae requires the conservation of unionid bivalves (Vaughn, 2018).

Unionid bivalves have an unusual larval stage, which yields glochidia that function as ectoparasites on specific fishes (Barnhart et al., 2008; Kondo, 2008). During their life cycle, males discharge sperm spheres into the water (Ishibashi et al., 2000), and females take in the sperm spheres, which fertilize the eggs. Although the reproductive season varies among unionid bivalve species, one female bivalve produces 10⁴–10⁶ eggs, which are grown in marsupia until they develop into glochidium larvae (Kondo, 2008). Female bivalves release glochidium larvae in stages and glochidium larvae parasitize the fins or gills of the host fish (Aldridge & Mclvor, 2003; Marshall et al., 2018). After being exposed to tactile stimulation, glochidium larvae rapidly close their valves and attach to the specific host fish (Kondo, 2008). Glochidia sometimes attach to objects other than fish but do not grow in such situations. When glochidia attach to a suitable host fish, they metamorphize into juveniles, which eventually detach (Neves et al., 1985).

The distribution and abundance of unionid bivalves have been extensively studied through their capture, which requires a considerable sampling effort and often results in habitat disturbance (Lamand & Beisel, 2014). Environmental DNA (eDNA) techniques have attracted attention as alternative approaches for monitoring various bivalves. Since Deiner and Altermatt (2014) successfully detected the eDNA of the unionid bivalve Unio tumidus in river water, some studies have employed this approach in their survey of unionid bivalves in freshwater environments (e.g., Carlsson et al., 2017; Stoeckle et al., 2016). Recent studies have successfully assessed the diversity of unionid bivalves in rivers via eDNA metabarcoding techniques (Klymus et al., 2021; Prie et al., 2021). Additionally, eDNA from several unionid bivalves was specifically detected at sites with high densities of Margaritifera margaritifera (Stoeckle et al., 2016; Wacker et al., 2019), M. monodonta (Lor et al., 2020), and U. crassus (Stoeckle et al., 2021). However, these studies have also suggested that the seasonal decrease in metabolic activity of unionid bivalves and increased flow rates complicate eDNA detection even at high densities. Currier et al. (2018) concluded that it was difficult to estimate the abundance of unionid bivalves, namely Lampsilis fasciola, Ligumia nasuta, Ptychobrancus fasciolaris, and Quadrula quadrula, present at low densities (0.01–2 individuals m⁻²) using the eDNA approach because of the non-linear relationship between eDNA concentration and density. Togaki et al. (2020) did not detect the eDNA of Sinanodonta sp. living at a low density (0.5 individuals per hand collection for 10 min) in a habitat pond.

Unionid bivalves release DNA into the water column in the form of mucus, shell materials, and excrement (Geist et al., 2008; Henley et al., 2006; Sansom & Sassoubre, 2017). However, their soft tissues are barely exposed to the water column, and their shells decompose slowly, allowing only limited DNA release. Thus, factors such as the density of unionid bivalves, their metabolic activity, and flow rates seem to limit the application of eDNA detection for monitoring these species. Notably, Wacker et al. (2019) reported that M. margaritifera eDNA in the Drakstelva River increased more than 20-fold from less than 1000 copies L⁻¹ in May to approximately 20,000 copies L⁻¹ in August, when the bivalves released their glochidia. Thus, unionid bivalve eDNA might be detected by targeting glochidium larvae, even at low bivalve densities; however, little attention has been paid to glochidium larvae as an eDNA resource for unionid bivalves.

Our working hypothesis was that the glochidium larvae released from Nodularia nipponensis would serve as an effective eDNA source for monitoring bivalves. This species is endemic to the northern Honshu and Hokkaido Islands in Japan (Lopes-Lima et al., 2020), and its conservation potentially contributes to the control of cyanobacterial blooms in Lake Hachiro, Japan (Sugawara et al., 2021). We first developed oligonucleotide primers to specifically detect the mitochondrial DNA of N. nipponensis and evaluated the characteristics of N. nipponensis eDNA, such as decay and release rates, in a laboratory tank experiment. Furthermore, we quantified eDNA from the water of a eutrophic lake (Lake Hachiro, Akita, Japan). We used bundles of polyvinylidene chloride fibers to capture glochidium larvae at the lake bottom.

MATERIALS AND METHODS Mollusk specimens

The mollusk specimens used for examining oligonucleotide primer specificity comprised 14 bivalves—nine species from the family Unionidae (order Unionida), three species from Corbiculidae (Veneroida), and one species from Veneridae (Veneroida)—as well as four snails, three species from the family Viviparidae (Architaenioglossa) and one species from Pleuroceridae (Sorbeoconcha; Table 1). Mollusk specimens were collected from lakes, paddy fields, rivers, and seas in Japan (Table 1). These specimens were identified based on their morphological traits (Kondo, 2008). First, mollusk species that live in and around Lake Hachiro were selected for primer development (Table 1). Among the selected mollusks, however, only two species of Unionidae were closely related to N. nipponensis (i.e., Cristaria plicata and Sinanodonta sp.). Second, seven Unionidae species, including N. nipponensis from Lake Biwa, Yoshida River, and Omono River, were used to ensure primer and probe specificity. For the target species N. nipponensis, two specimens collected from Lake Hachiro and Lake Biwa were tested (Table 1). The brackish and seawater bivalves (Corbicula japonica and Ruditapes philippinarum) were also examined. For non-target species, only one specimen of each species was tested (Table 1).

TABLE 1 Specificity of oligonucleotide primers developed for detection of N. nipponensis

Order	Species	PCR Amplification	Locality
Unionida (bivalves)	Nodularia nipponensis	+	Lake Hachiro
	Nodularia nipponensis	+	Lake Biwa
	Beringiana gosannensis	−	Omono River
	Cristaria plicata	−	Lake Hachiro
	Inversidens brandtii	−	Lake Biwa
	Inversiunio jokohamensis	−	Yoshida River
	Lanceolaria oxyrhyncha	−	Lake Biwa
	Pronodularia japanensis	−	Lake Biwa
	Sinanodonta calipygos	−	Lake Biwa
	Sinanodonta sp.	−	Lake Hachiro
Veneroida (bivalves)	Corbicula fluminea	−	Lake Biwa
	Corbicula sandai	−	Lake Hachiro
	Corbicula japonica	−	Lake Ogawara
	Ruditapes philippinarum	−	Kumamoto Prefecture
Architaenioglossa (snails)	Cipangopaludina chinensis	−	Paddy fields near Lake Hachiro
	Cipangopaludina japonica	−	Paddy fields near Lake Hachiro
	Sinotaia quadrata	−	Paddy fields near Lake Hachiro
Sorbeoconcha (snail)	Semisulcospira sp.	−	Paddy fields near Lake Hachiro

Note: +, PCR amplicon detected; −, not detected.

Oligonucleotide primers and PCR

Specific oligonucleotide primers were designed by compiling sequences of mitochondrial cytochrome c oxidase subunit I (COI) genes from N. nipponensis (GenBank accession numbers KJ434522 and GQ451863) and four other unionid bivalve species: Corbicula sandai (KC211280), Corbicula japonica (AB498808), Cristaria plicata (EU698950), and Sinanodonta woodiana (GQ451867). These five bivalves inhabit Lake Hachiro. Based on the COI gene sequences, a primer and probe combination were designed “by eye” using MEGA ver.6 (Tamura et al., 2013). The forward and reverse primers, designated as NN_F and NN_R, had the sequences 5′-GTTACTTGTTCCTGCTTTG-3′ and 5′-CAAAACAGCAGTTACTGTA-3′, respectively. The oligonucleotide probe used for the quantitative PCR (qPCR), the NN_probe, was (FAM)_AATGTCGCTCATTCTGG_(TAMRA). The predicted amplicon was 274-bp.

The foot tissues of mollusk specimens were dissected and stored at −30 °C until DNA extraction. Genomic DNA was extracted from the tissues using the DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany), according to the manufacturer's instructions. PCR amplification of the mitochondrial COI gene from bivalve genomic DNA was performed on a SimpliAmp™ Thermal Cycler (Applied Biosystems, Life Technologies, Carlsbad, CA, USA) with a 20-μl reaction mixture containing 8.7–19.4 ng of genomic DNA, 5 μl 1 × Buffer, 2 mM MgCl₂, 0.2 mM dNTPs, 0.8 μM each primer (NN_F and NN_R), and 0.5 U TaKaRa Ex Taq HS (Takara Bio, Shiga, Japan). Thermal cycles included an initial denaturation step of 5 min at 94°C, followed by 30 cycles of denaturation at 94°C for 30 s, annealing at 56°C for 30 s, extension at 72°C for 25 s, and a final extension period of 7 min at 72°C. PCR amplification was verified by the appearance of 274-bp DNA bands on microchip electrophoresis by using a MultiNA system (Shimadzu, Kyoto, Japan), operated according to the manufacturer's protocol.

Tank experiment

Nodularia nipponensis bivalves with shell lengths of 51–66 mm were collected from Lake Hachiro in November and were maintained in a 30 L aquarium tank filled with dechlorinated tap water for 1 month to thoroughly eject fecal materials before use. In our prior study, a long-term culture experiment of over 80 days with N. nipponensis demonstrated that many individuals remain alive even under non-feeding conditions (Sugawara et al., 2021). Therefore, 1 month's incubation was considered to not substantially lower bivalve activity. The aquarium water was changed weekly. The release of eDNA from N. nipponensis and its decay in tank water were investigated in a dark room maintained at 20°C. Fifteen water tanks (32 × 19 × 23 cm) were used: The inside of all tanks was sterilized with 0.1% (w/v) sodium hypochlorite solution overnight and rinsed with dechlorinated tap water before use. Three or 10 bivalves were placed in each tank (n = 5), resulting in aerial densities of 49 and 164 individuals m⁻², respectively, and the remaining five tanks were used for the control experiment without bivalves. These densities were selected to ensure sufficient DNA for the target species. The bivalves were maintained for 7 days in a tank filled with 10 L dechlorinated tap water. During this period, these bivalves were unfed to minimize the deterioration of the tank water quality. Water samples (500 ml) were collected just before (−24 h) and after the bivalves were thoroughly removed from the tank. Subsequently, water samples were collected from the tanks at 0, 2, 4, 8, 16, 24, 48, 72, 96, and 168 h after the bivalves were removed. Before sampling, the tank water was stirred using a sterilized plastic stick.

The decay of eDNA was evaluated based on the decrease in eDNA concentration after removal of the bivalves. The decay rate constant was calculated using an exponential decay model: C_t = C₀ × e^{–k × t} (Sansom & Sassoubre, 2017), where C_t is the eDNA concentration at time t (copies mL⁻¹), C₀ is the eDNA concentration at time 0 (immediately after bivalve removal), and k is the decay rate constant (h⁻¹). We assumed that the eDNA concentrations reached the equilibrium state (i.e., release rate equivalent to decay rate) within 7 days because this state was achieved in a similar experiment on a shorter time scale (20 h) that used a unionid bivalve (Sansom & Sassoubre, 2017). The rate of eDNA release from the bivalves was calculated from the eDNA concentration at time 0 (C₀) and the decay rate constant (k) by using the following equation (Sansom & Sassoubre, 2017): S = C₀ × V × k, where S is the eDNA release rate in the tank (copies h⁻¹), and V is the tank water volume (ml).

The collected water samples were used to determine eDNA, dissolved inorganic nitrogen (DIN), and dissolved inorganic phosphorus (DIP). The water samples were immediately amended with 500 μl of a cationic surfactant (10% benzalkonium chloride solution) to prevent the degradation of eDNA (Yamanaka et al., 2017). The samples were then filtered with a GF/F glass fiber filter (pore size, 0.7 μm; GE Healthcare). The filters were stored at −80 °C before eDNA analysis. The filtrates (50 ml) were stored at 4 °C for the analysis of DIN and DIP by using nutrient autoanalyzer QuAAtro2-HR (BLTEC). DIN and DIP were calculated from the sum of NH₄ − N, NO₃ − N, and NO₂ − N contents or the PO₄ − P content, respectively. Although benzalkonium chloride contained NH₄ − N, these levels were negligible (<10 ng L⁻¹ based on content) when the reagent was added at a final concentration of 0.1% (v/v). Disposable nitrile gloves were used during the sampling process and replaced between the samples. All experimental tools were regularly disinfected using 0.1% sodium hypochlorite solution before analysis (Yamanaka et al., 2017).

Field sampling

The study site was on the southern shore of Lake Hachiro, Akita Prefecture, Japan (Figure S1 and Table S1). Lake Hachiro is a shallow eutrophic lake, with a mean depth of 2.8 m, maximum depth of 12 m, and area of 47.32 km². Cyanobacterial blooms occur in this lake during the summer (Araki et al., 2018). Nodularia nipponensis occurs only in a certain part of the lake at low densities. (0.39–0.56 individuals m⁻²) (Takada & Kurosawa, 2014). To recover N. nipponensis eDNA, 10 sampling points were set within the area, with a particular focus on the area around points E–H, where the highest densities were reported (Figure S1; Table S1); however, we did not collect samples from several stations each day because of bad weather conditions. The experimental period was May 31–August 26, 2020, when glochidium larvae were released during the early summer in Lake Hachiro (June and July; Yoshida et al., 2019). Lake water samples were collected from near-bottom water (depth ca. 2 m) using a Van Dorn water sampler (3 L volume; Rigo). In this study, surface water samples were not collected because bivalve eDNA was predicted to be higher in near-bottom water (Lor et al., 2020; Xia et al., 2018). The collected water was poured into a pre-bleached 1 L polyethylene bottle, and 1 ml of benzalkonium chloride was added to prevent the degradation of eDNA (Yamanaka et al., 2017).

To capture unionid glochidium larvae floating in lake water, we used bundles of polyvinylidene chloride fibers with a specific surface area of 1.2 m² m⁻¹ (Figure 1; Bio-cord PV-45, TBR Co.). At each point, 1 m of bundle was sunk on the lake bottom in the morning (at approximately 07:00), and it was collected after 8 h in a pre-bleached 2L polyethylene bottle containing 1 L of deionized water and 0.1% benzalkonium chloride. Each bundle was collected in a different bottle. The lake water samples were collected at this time. Several bundles recovered from the lake bottom on June 21 were also placed in a 2-L polyethylene bottle containing 1 L of surface lake water collected from the same site to collect live glochidia larvae for the laboratory. The bundles were also collected in different bottles at each site. In this case, 1-ml benzalkonium chloride was added to the bottle immediately before removing the glochidia larvae from the bundles.

FIGURE 1. Bundles of polyvinylidene chloride fibers used in this study

The samples collected at the field sites were brought back to the laboratory within 1 h and subjected to eDNA extraction. During the filtration of water samples with a GF/F glass fiber filter (GE Healthcare), the volume was limited to approximately 200 ml because of the turbidity of the lake water. The 2-L polyethylene bottle containing a bundle was shaken vigorously to tear off the glochidia, and the entire solution (1 L of total volume) was thoroughly filtered with a GF/F glass fiber filter (GE Healthcare). All filters were stored at −80 °C before eDNA extraction.

For microscopic observation of glochidium larvae, the contents of the aforementioned shaken bottle were poured into a plankton net (mesh size, 63 μm; Rigo, Tokyo, Japan) to concentrate it to 30 ml and was then centrifuged at 1500 g for 10 min at room temperature (ca. 20–25°C). Thereafter, the obtained glochidia were counted, sized under an optical microscope, and added to a 2-ml tube at each sampling point. The 2-ml tubes containing glochidia were stored at −80°C before DNA extraction.

eDNA extraction

DNA extraction was performed in spaces dedicated for isolation. Each filter was cut in half (half for analysis and half for archival storage) and shredded using bleached scissors to increase the surface area for eDNA extraction. The shredded filter was placed in a 2-ml screw cap tube, and 80-μl proteinase K (Qiagen) and 800-μl Buffer AL (Qiagen) were then added to fully submerse the filter. The tube with the shredded half filter and extraction liquid (i.e., proteinase K and Buffer AL) was held at 56°C for 1 h, and eDNA was recovered. The bottom of the tube was cut with bleached scissors and placed in a Salivette tube (16.8 mm inner diameter, 95 mm length; Sarstedt). The Salivette tubes with a half filter containing the 2-ml screw cap tube were centrifuged at 3000 g for 3 min, followed by the addition of 440-μl TE buffer (10 mM Tris–HCl and 1 mM EDTA, pH 8.0) onto the filters and another centrifugation at 3000 g for 3 min. The DNA in the eluted solution was added to 800 μl of ethanol and purified using the DNeasy Blood & Tissue Kit (Qiagen), according to the manufacturer's protocol. The total DNA was eluted in 200-μl buffer AE following the manufacturer's protocol, and the samples were stored at −30 °C until qPCR.

The genomic DNA was extracted from glochidium samples using the DNeasy Blood & Tissue Kit (Qiagen) and was also stored at −30 °C until qPCR analysis.

Quantification of eDNA using qPCR

To avoid contamination, reaction mixtures were prepared, and qPCR cycling was conducted in separate rooms. All qPCR assays were performed using the LIGHTCYCLER® 96 System (Roche Diagnostics, Basel, Switzerland). The total volume of the analyzed samples was 20 μl and comprised 10 μl 2× Master Mix, 0.1-μl each primer at 500 nM concentration, 0.02-μl TaqMan probe at 100 nM concentration, and 5-μl DNA template prepared using FastStart Essential DNA Probes Master (Roche). Quantitative PCR was performed under the following conditions: 600 s at 95°C, 45 cycles of 10 s at 95°C, 10 s at 56°C, and 28 s at 72°C.

For obtaining plasmid DNA, a preliminary PCR using total DNA from N. nipponensis was performed to amplify the 274 base-pair target region. Cloning was performed using a TOPO® TA Cloning® Kit for Sequencing (Invitrogen, Waltham, MA, USA) with a PCR 4-TOPO vector, including competent cells (Escherichia coli), according to the manufacturer's recommended protocol. Bacterial colonies were grown on agar plates using ImMedia™ Amp Blue (Invitrogen). Some non-blue colonies were inoculated in a liquid medium containing 40-ml LB medium (MP Biomedicals) and 50 μg mL⁻¹ ampicillin. The plasmids were isolated using the QIAprep® Spin Miniprep Kit (Qiagen), and PCR was conducted to confirm the incorporation of the mitochondrial cytochrome c oxidase subunit I target gene sequence into the plasmid (Galluzzi et al., 2008). Copy numbers for plasmid DNA standards were calculated based on the DNA concentration and base-pair length (Lorenz, 2012).

We performed a qPCR by using a dilution series of standards containing 10¹–10⁷ copies of plasmid DNA per microliter. A standard sample was run alongside the samples in each well of a 96-well plate. The cycle quantification (Cq) obtained from the amplification of the lowest standard (i.e., 10 copies μL⁻¹) was Cq = 34.9 ± 0.7 (mean ± SD). Thus, a Cq value greater than 35 cycles was regarded as below the limit of quantification, although qPCR was extended up to 45 cycles. The Cq values of each sample were obtained from triplicate qPCR measurements, and the average value was used to calculate the eDNA concentration. If the eDNA concentration was below the limit of quantification (<10 copies μL⁻¹), the data were presented. In the tank experiment, the eDNA concentration was conveniently regarded as zero to calculate the DNA decay and release rates. For several samples, no PCR amplification was observed for two or three of the triplicate measurements; in these cases, the eDNA content was denoted as “not detected.” Although dilution of DNA templates can reduce the inhibition of PCR amplification (Cao et al., 2012), it did not affect the efficiency of PCR amplification in this study.

Data analysis

The eDNA decay and release rates, DIN and DIP release rates, and biomass in each tank were log-transformed after adding one to fulfill the assumption of normality. Significant differences in decay and release rates at different densities were evaluated using the F test followed by the Student's t-test. The correlation between the bivariates was evaluated using the Pearson product–moment correlation coefficient. For the field sampling data, analysis of variance (ANOVA) was used to determine whether there was a significant difference in eDNA concentration in each sampling event. We used R version 3.6.3 for all statistical analyses (R Core Team, 2020).

RESULTS Specificity of PCR primers

The oligonucleotide primers constructed in this study enabled the PCR amplification of two N. nipponensis specimens collected from different locations (Table 1). No amplification was observed for other Unionidae, Corbiculidae, Veneridae, Viviparidae, and Pleuroceridae snails. Based on these results, oligonucleotide primers specific for N. nipponensis were found to be effective for the detection of N. nipponensis eDNA in aquatic environments.

Decay and release of N. nipponensis eDNA

First, qPCR results were validated using known copies of the N. nipponensis COI gene. The slopes of the standard curves for gene copy numbers had R² values >0.98. The PCR efficiency calculated from the slopes ranged from 93% to 110% (100 ± 5.6%; mean ± standard deviation). No PCR amplification was observed in the tested mixtures without N. nipponensis DNA (negative control). The quantification limit of the qPCR was determined as 10 copies in the PCR mixtures, which corresponded to 8.0 copies mL⁻¹ of tank water.

In the tank experiment, apparent steady-state concentrations of eDNA were determined at time 0 when all bivalves were removed from the tank (Figure 2). Subsequently, the eDNA concentration decreased exponentially during the first 24 h, after which the decay rate constants were calculated (Table S2). eDNA was detected in 30 of the 109 samples, which was below the limit of detection; thus, the concentration was regarded as 0 (Table S3). In tanks with 3 and 10 individuals, the eDNA concentrations at time 0 were (2.9 ± 2.3) × 10² and (4.1 ± 8.2) × 10⁴ copies mL⁻¹, respectively (Table S2). The concentrations at time 0 differed considerably among the series of tanks, leading to a large dispersion of concentration values (Figure 2, Table S2). When the average values of eDNA concentration at a steady state in the tanks containing 3 and 10 individuals were divided by their density (49 and 164 individuals m⁻², respectively), the concentration values obtained were 6.1 and 250 copies mL⁻¹, respectively.

FIGURE 2. Decay curves for eDNA in experimental tanks with 3 and 10 N. nipponensis individuals. Decay curve equations of experimental tanks with 3 and 10 individuals were estimated as Ct = (2.9 × 102) × e–0.074 × t and Ct = (4.1 × 104) × e–0.147 × t, respectively. Diamond and square dots indicate eDNA concentrations per milliliter of tank water at each time point. The first sample was collected at the time point of −24 h. Error bars show the standard deviation

The eDNA decay constants in tanks with 3 and 10 individuals were 0.074 ± 0.021 h⁻¹ and 0.147 ± 0.125 h⁻¹, respectively (Table S2). Although the eDNA decay rate constants tended to be higher in tanks with 10 individuals than in tanks with 3 individuals, there was no significant difference in the eDNA decay rate constants between the two groups (p > 0.1, Table S2). Based on these values and the apparent steady-state concentrations, the eDNA release rates were estimated to be (2.4 ± 2.1) × 10⁵ and (1.8 ± 2.4) × 10⁷ copies h⁻¹ for the tanks with 3 and 10 individuals, respectively (Figure 3, Table S2). The data plots indicated that the eDNA release rate depended on the number of bivalves because it partially correlated with the total biomass of bivalves per tank (r = 0.71, p = 0.02; Figure 3a). The slope shown in Figure 3a indicates that N. nipponensis released eDNA at a rate of (3.2 ± 6.1) × 10⁴ copies h⁻¹ g⁻¹ of biomass.

FIGURE 3. Relationship between the N. nipponensis eDNA release rate and the biomass (a) and the DIN (b) and DIP (c) release rates. Diamond and square points represent data from tanks with 3 and 10 individuals, respectively

In the tank experiment, DIN and DIP excretions were analyzed to assess their relationship with the eDNA release rate (Figure 3b, c). The DIN and DIP release rates largely differed among the tanks, despite the same number of bivalve individuals being studied (Figure 3b, c, Table S2). By contrast, some tanks with 3 and 10 individuals released the most equivalent amounts of DIN and DIP, respectively (Figure 3b, c, Table S2). These results suggest that DIN and DIP excretion differed considerably among individuals. Furthermore, the eDNA release rate was significantly correlated with both DIN and DIP release rates, with correlation coefficients of 0.81 (p = 0.004) and 0.86 (p = 0.001), respectively.

eDNA quantified from lake water samples

The qPCR was also validated for the water samples collected from Lake Hachiro (Table S4). The slope of the standard curves for gene copy numbers was accompanied by R² values of >0.98. The PCR efficiency calculated from the slopes ranged from 92% to 110% (98 ± 6.2%). In the tested mixtures without N. nipponensis DNA (negative control), no PCR amplification was observed under the reaction conditions. The quantification limit of qPCR was determined to be 10 copies in the PCR mixtures, which corresponded to 10 copies mL⁻¹ of lake water.

Nodularia nipponensis eDNA was quantified at various concentrations ranging from 0.017 × 10³ to 1.1 × 10³ copies [5-μl template DNA]⁻¹ (corresponding to 0.034 × 10⁵ to 2.2 × 10⁵ copies per liter of lake water) in several water samples collected from stations A–J (Table 2; Figure S1; Tables S1 and S4). The number of stations that yielded a quantitative result was relatively high on July 5 (4 out of 10) and lower on June 21 (2 out of 9) and July 26 (2 out of 5). The stations where eDNA was quantified differed by the sampling date. The eDNA was not quantified at any of the stations on May 31 and August 26 (Table 2).

TABLE 2 Concentrations of eDNA determined for water samples collected at the near bottom of Lake Hachiro

Date	Sampling station (unit: Copies per 5-μl template DNA)
	A	B	C	D	E	F	G	H	I	J
May 31	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
June 21	n.d.	(4.8 ± 8.2) × 102	n.d.	-	(0.9 ± 1.5) × 101*	n.d.	(0.6 ± 1.1) × 103	n.d.	n.d.	n.d.
July 5	(0.4 ± 4.0) × 101*	(6.1 ± 7.3) × 102	(0.2 ± 0.4) × 101*	n.d.	(4.1 ± 6.5) × 102	(1.8 ± 3.2) × 102	(0.2 ± 0.3) × 101*	n.d.	(0.3 ± 0.3) × 101*	(1.1 ± 1.4) × 103
July 26	(0.1 ± 0.1) × 101*	(1.7 ± 3.0) × 101*	-	-	(0.8 ± 1.4) × 103	(0.1 ± 0.1) × 101*	-	-	-	(0.7 ± 1.1) × 103
August 26	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.

Note: n.d., not detected; *, below limit of quantification (<50 copies per 5-μl template DNA).

The eDNA concentrations in the water samples, when quantified, were not significantly different (Table 2). The averaged values (per 5-μl template DNA) were (5.6 ± 1.1) × 10² copies on June 21 (n = 2), (5.7 ± 3.8) × 10² copies on July 5 (n = 4), and (7.3 ± 0.1) × 10² copies on July 26 (n = 2). However, the number of stations that yielded quantitative results differed considerably by the sampling date (Table 2).

eDNA quantified from the bundle samples

When fiber bundles were used to capture glochidium larvae, N. nipponensis eDNA was quantified on May 31, June 21, July 5, and July 26. N. nipponensis eDNA was quantified at one station (A; 5.6 × 10³ copies [5-μl template DNA]⁻¹) on May 31 and at all stations from June 21 to July 26 (Tables 3 and S4). The eDNA concentrations quantified (per 5-μl template DNA) varied among the stations and the range was (0.013–1.9) × 10⁴ copies [(3.9 ± 6.5) × 10³, mean ± standard deviation] on June 21, (0.076–7.8) × 10³ copies [(2.3 ± 2.4) × 10³, mean ± standard deviation] on July 5, and (0.092–1.1) × 10⁴ copies [(4.8 ± 3.5) × 10³, mean ± standard deviation] on July 26 (Table 3). On these three dates, the average eDNA concentration was not significantly different. N. nipponensis eDNA was not quantified at any of the stations on August 26.

TABLE 3 Concentrations of eDNA determined for bundle samples collected at the near bottom of Lake Hachiro

Date	Sampling station (unit: Copies per 5-μl template DNA)
	A	B	C	D	E	F	G	H	I	J
May 31	5.6 × 103	1.7 × 101*	0.4 × 101*	n.d.	3.0 × 101*	0.9 × 101*	1.5 × 101*	1.1 × 101*	n.d.	n.d.
June 21	9.0 × 103	2.1 × 102	6.9 × 102	-	1.5 × 102	1.9 × 104	1.1 × 103	1.1 × 103	3.0 × 103	1.3 × 102
July 5	1.2 × 103	5.2 × 102	7.6 × 101	2.2 × 103	4.8 × 103	1.7 × 103	4.3 × 102	1.8 × 103	7.8 × 103	3.1 × 103
July 26	1.1 × 103	2.2 × 103	6.0 × 103	3.1 × 103	1.0 × 104	1.1 × 104	5.8 × 103	9.2 × 102	4.4 × 103	3.4 × 103
August 26	n.d.	0.5 × 101*	0.7 × 101*	0.2 × 101*	n.d.	n.d.	n.d.	0.1 × 101*	n.d.	n.d.

Note: * and n.d.; see Table 2.

Microscopic observation revealed that several glochidia were attached to bundles of polyvinylidene chloride fibers (Figure 4). Their sizes ranged from 171 to 198 μm (n = 23), which is consistent with the size of the glochidium larvae of N. nipponensis described by Kondo (2008). In addition, we succeeded in quantifying N. nipponensis DNA from glochidium samples via qPCR, which indicated that the glochidium larvae attached to the bundle samples were indeed N. nipponensis (Figure 4, Table S5). In separate observations on June 21, we recovered 10, 1, 10, and 2 larvae from fiber bundles at stations A, B, F, and J, respectively; when the recovered specimens were subjected to qPCR, they yielded DNA copy numbers of 5.3 × 10⁴, 2.0 × 10¹, 4.2 × 10⁴, and 3.8× 10³ copies [5-μl template DNA]⁻¹, respectively.

FIGURE 4. Microscopic image of glochidium larva collected on June 21. Yellow and red arrows indicate the fiber and larva attached to the fiber, respectively

DISCUSSION

This study is the first to report an efficient quantification of eDNA derived from glochidium larvae. To characterize eDNA release and quantify eDNA in the field, we developed a novel primer set for N. nipponensis. The oligonucleotide primer set quantified the eDNA in the tank experiments and made it possible to calculate the release and decay rates of N. nipponensis eDNA. The primers also served to quantify the N. nipponensis eDNA in water samples collected around the habitats in a eutrophic lake, and the use of fiber bundles to capture the bivalve's glochidium larvae enabled the detection of increased amounts of DNA in the environment, increasing the sensitivity of eDNA monitoring.

The primers were specific for N. nipponensis; thus, they did not amplify genetic material from other mollusks, including various bivalves and snails, which potentially share habitats with N. nipponensis. We confirmed the utility of our primers for eDNA detection in tank experiments and in field bivalve monitoring.

The tank experiment showed eDNA decay curves that included the eDNA decay rate constants for calculating eDNA release rates (Figure 2). For freshwater Dreissenidae bivalves Dreissena polymorpha and Dreissena rostriformis bugensis, the eDNA decay rate was faster at a higher population density (Marshall et al., 2021). The fish Trachurus japonicus indicated a similar phenomenon (Jo, Arimoto, et al., 2019a). The higher eDNA decay rates associated with the larger biomass could reflect the activity and abundance of microbes and exonucleases in the water (Jo, Arimoto, et al., 2019a; Jo, Murakami, et al., 2019b). The eDNA decay rate constants obtained in N. nipponensis seemed higher than those reported for the unionid bivalve Lampsilis siliquoidea, which ranged from 0.0097 to 0.0530 h⁻¹ for 2, 10, and 20 mussels per 40 L of rearing water (Sansom & Sassoubre, 2017), and those reported for D. polymorpha and D. rostriformis bugensis, which ranged from 0.008 to 0.115 h⁻¹ for 2–48 mussels per 15 L of rearing water (Marshall et al., 2021). The eDNA decay in the tank experiments depends on the water temperature (Jo, Arimoto, et al., 2019a; Jo, Murakami, et al., 2019b; Strickler et al., 2015), biomass (Jo, Arimoto, et al., 2019a; Jo, Murakami, et al., 2019b), microbes (Jo, Arimoto, et al., 2019a; Jo, Murakami, et al., 2019b; Strickler et al., 2015), water volume (i.e., degree of microbial concentration), and pH (Strickler et al., 2015) in the experimental tanks, which are different from those in the literature. In further research, tank experiments with a wider density gradient than that used in this study under the same conditions could allow for the identification of factors influencing eDNA decay.

The release rates of N. nipponensis eDNA were weakly correlated with the number of individuals per tank, indicating that eDNA release partially reflected bivalve population density. However, eDNA release rates varied substantially, even among tanks with the same number of bivalve individuals. As shown in Figure 3, two tanks with 10 individuals exhibited lower eDNA release rates (0.0066 and 0.33 × 10⁶ copies h⁻¹ individuals⁻¹; Table S2) than the other tanks did. Sansom and Sassoubre (2017) conducted similar tank experiments with different densities of unionids, L. siliquoidea. The release rates of L. siliquoidea eDNA were 2.4 × 10⁷ and 5.6 × 10⁶ copies h⁻¹ tank⁻¹, respectively, with bivalves at half the density (56 vs. 111 individuals m⁻²) releasing a 4.3-fold higher amount of eDNA. Hence, the eDNA concentration detected in the environment may not accurately reflect the abundance of bivalves. Currier et al. (2018) found that eDNA copy estimates obtained for freshwater pearly mussel species were positively correlated with quadrat-derived estimates of mussel density at a field site while observing that some sites exhibited greatly varying eDNA concentrations. The eDNA release rates in N. nipponensis were strongly correlated with the DIN and DIP release rates (Figure 3b,c). These results suggest that bivalve eDNA is derived from excretion (Henley et al., 2006; Sansom & Sassoubre, 2017). Differences in the amount of excretion between individual bivalves were probably responsible for the variable eDNA concentrations between tanks. In this context, the eDNA approach for monitoring bivalves may not consider the presence of bivalves with low activity, resulting in the underestimation of bivalve abundance. In the tank experiment, it was unlikely that some bivalves released sperm or glochidium larvae, which were responsible for the dispersion of eDNA among tanks. The bivalves were maintained at 20 °C for 1 month and used in the tank experiment at the same temperature; under such conditions, the unionids could not proceed with their reproduction process (Ishibashi et al., 2000).

Based on the quantification limit of the qPCR (8 copies mL⁻¹) and the apparent steady-state eDNA concentrations in the tank experiment [(3.0 ± 3.5) × 10² copies mL⁻¹ for tanks with 3 individuals: Figure 2], our approach may be useful for the monitoring of habitats with a bivalve density of at least >1.3 individuals m⁻². In the lake environment, however, released eDNA is dispersed in the water column; therefore, high densities of bivalves are required for eDNA recognition by qPCR. However, Currier et al. (2018) quantified unionid bivalves eDNA at densities of 1.0 individuals m⁻² or less. In addition to the abundance of target species, amplicon length could lower detection and quantitative sensitivities. Jo et al. (2017) indicated that the degradation rates of longer DNA fragments (719 bp) were significantly faster than those of shorter DNA fragments (127 bp). The length of the amplicon amplified by the primer set developed in this study was 274 bp, which was longer than that used by Currier et al. (2018) (99–129 bp). Therefore, our qPCR system could be useful for the detection and quantification of fresh DNA immediately released from bivalves, although it may be less effective than others in amplifying shorter DNA fragments. For monitoring the decline of species, examining the threshold density in the environment is necessary and can be determined using our qPCR system.

Nevertheless, N. nipponensis eDNA was detected in several water samples collected from the near-bottom layer of eutrophic Lake Hachiro. The concentration range was (0.034–2.2) × 10⁵ copies per liter of lake water, and the quantified eDNA concentrations corresponded to the amounts released from bivalves at 0.56 to 35 individuals m⁻², based on estimates in the tank experiment. These estimates appeared considerably high with respect to the bivalve density reported for the study site, which was estimated to be as high as 0.56 individuals m⁻² via the quadrat method (Takada & Kurosawa, 2014). The presence of N. nipponensis eDNA in the lake waters was transient because no eDNA was quantified before or after the period from June 21 to July 26. These results suggest that the N. nipponensis eDNA floating at the study site increased transiently from June 21 to July 26. When sediments roll up in the overlying water, the feces and pseudofeces of bivalves could be sources of eDNA (Xia et al., 2018). However, because of the inability to quantify eDNA from bundle samples in May and August, sediment-derived supplies appeared to be nonsignificant. Based on microscopic observation and qPCR analysis of isolated glochidium larvae samples (Figure 4), we argue that glochidia probably comprised the bulk of eDNA detected during this period (Table 3). The period in which N. nipponensis eDNA was quantified was consistent with that of glochidium release by N. nipponensis in Lake Hachiro (Yoshida et al., 2019). Glochidium larvae adhere to the surfaces of available solids, including inanimate materials (Jansen et al., 2001). Once the valves are closed, they do not open again (Jansen et al., 2001; Paling, 1968). The results clearly indicate that the larvae captured by the fiber bundles were derived from N. nipponensis. In this study, however, it was difficult to show the relationship between DNA copy number and larval number because few data were available.

The amount of DNA extracted from the isolated glochidia was slightly higher than that of the eDNA quantified in the field. The reason for this result is the difference in the extraction methods. For isolated glochidia, tissue DNA extraction was adopted, and the extraction time was 3 h. By contrast, the glochidia attached to the bundles were subjected to the eDNA extraction method (extraction time 1 h) for comparison with water samples. In this study, the use of polyvinylidene chloride fiber bundles enabled the capture of glochidium larvae from near-bottom lake waters. The fiber bundles facilitated the quantification of N. nipponensis eDNA (Figure 4; Table 3), which was quantified at all stations from June 21 to July 26 and at one station on May 31.

This study is the first to provide evidence of the efficacy of eDNA quantification for monitoring the unionid bivalve N. nipponensis. Although N. nipponensis is a widespread unionid bivalve across freshwater environments in Japan, its habitats have recently been deteriorating (Negishi et al., 2008). Therefore, the eDNA approach examined here may be useful for the conservation of N. nipponensis. A comparison with the results obtained through a conventional approach (i.e., concentration and extraction of water samples) suggests that the glochidium larvae-based technique is a promising tool for monitoring unionid bivalves in the environment. Nevertheless, further research is necessary to determine whether the gradient of glochidium-derived eDNA concentration matches the density of spawning bivalves.

In conclusion, we developed a qPCR approach for the specific detection of N. nipponensis and employed it to determine eDNA release and decay rates. We employed the assay to analyze the waters of Lake Hachiro, where N. nipponensis habitats have been confirmed, and eDNA was successfully quantified from lake water and fiber bundle samples during the glochidium larvae release period. The most substantial finding of this study was the utility of the fiber bundle for tracing glochidium larvae even at low densities of unionid bivalves. Adopting eDNA quantification as a method to complement and reinforce traditional sampling may improve quantification sensitivity and thus be conducive to conservation efforts.

ACKNOWLEDGMENTS

We thank Seiko Furuta (Lake Biwa Environmental Research Institute, Shiga) and Hideki Sugiyama (Akita Prefectural University) for providing bivalve specimens. This study was partly supported by the Grant-in-Aid for JSPS Fellows, Grant Number JP21J13090 (to K. Sugawara).

CONFLICT OF INTEREST

None declared.

AUTHOR CONTRIBUTIONS

K.S., K.O., and N.M. designed experiments. K.S. and Y.S. performed tank experiments. K.S., Y.S., K.O., and N.M. performed field experiments. K.S., Y.S., and K.O. performed the molecular analyses. K.S., Y.S., K.O., M.W., and N.M. analyzed the data and wrote the first draft of the manuscript. All the authors edited and provided feedback on the manuscript.

DATA AVAILABILITY STATEMENT

The raw data for the qPCR experiments are included in the Supporting Information.