Assessment of the Appearance and Toxin Production Potential of Invasive Nostocalean Cyanobacteria Using Quantitative Gene Analysis in Nakdong River, Korea

Invasive nostocalean cyanobacteria (INC) were first reported in tropical regions and are now globally spreading rapidly due to climate change, appearing in temperate regions. INC require continuous monitoring for water resource management because of their high toxin production potential. However, it is difficult to analyze INC under a microscope because of their morphological similarity to nostocalean cyanobacteria such as the genus Aphanizomenon. This study calculates the gene copy number per cell for each target gene through quantitative gene analysis on the basis of genus-specific primers of genera Cylindrospermopsis, Sphaerospermopsis, and Cuspidothrix, and the toxin primers of anatoxin-a, saxitoxin, and cylindrospermopsin. In addition, quantitative gene analysis was performed at eight sites in the Nakdong River to assess the appearance of INC and their toxin production potential. Genera Cylindrospermopsis and Sphaerospermopsis did not exceed 100 cells mL−1 at the maximum, with a low likelihood of related toxin occurrence. The genus Cuspidothrix showed the highest cell density (1759 cells mL−1) among the INC. Nakdong River has potential for the occurrence of anatoxin-a through biosynthesis by genus Cuspidothrix because the appearance of this genus coincided with that of the anatoxin-a synthesis gene (anaF) and the detection of the toxin by ELISA.


Introduction
Climate change, caused by global warming, has negative effects on aquatic ecosystems. In particular, harmful cyanobacterial algal blooms (cyanoHABs) that cause harmful substances (such as toxins and off-flavors) are increasing in extent and frequency due to climate change, negatively impacting water resources [1,2]. Representative cyanobacterial genera that cause cyanoHABs are Microcystis, Dolichospermum, and Aphanizomenon [1,3,4]. CyanoHABs and the damage associated with them have been globally reported in many lakes and rivers [5][6][7]. Microcystis spp. is the most common cause of cyanoHABs that produces microcystins that cause liver damage, and adversely affects human and livestock health in the short or long term, so microcystins are managed in many countries [4]. Furthermore, in addition to cyanobacteria that cause cyanoHABs, cyanobacteria of the order Nostocales, first reported in tropical regions, are now spreading to temperate regions such as North America and Europe [8][9][10]. These cyanobacteria are referred to as invasive or exotic cyanobacteria. Invasive nostocalean cyanobacteria (INC) include genera Cylindrospermopsis, Sphaerospermopsis, Cuspidothrix, Raphidiopsis, and Chrysosporum [11][12][13][14][15][16].

Genus-Specific Gene Copy Numbers per Cell of INC
The gene copy numbers of three species of INC were proportional to the cyanobacterial cell count, with a high correlation (r 2 = 0.91 or higher) ( Figure 1). Quantitative analysis of the rpoB gene for 27-205 cells of the trace analysis sample of Cuspidothrix indicated 32-237 copies. The result of quantitative gene analysis for 933-9950 cells mL −1 of the sample diluted according to cell density ranged between 2312 and 13,480 copies mL −1 , and the copy number of rpoB per cell was 1.4 copies cell −1 . The copy numbers of rpoB per cell for the 11 strains of Cuspidothrix isolated from Nakdong River [24] were in the range of 1.0-1.4 copies cell −1 , which were roughly similar to those for the NIVA-711 strain isolated in Germany, with an average of 1.1 copies cell −1 (Table 1).
Quantitative analysis of the rbcLX gene of Sphaerospermopsis for 24-211 cells in the trace analysis sample indicated 40-254 copies. The quantitative gene analysis result for 1068-9990 cells mL −1 of the analysis sample diluted by cell density indicated 1613-23,287 copies mL −1 , and the copy number per cell of rbcLX was found to be 1.6 copies cell −1 (Figure 1). The strains of Sphaerospermopsis were isolated from Nakdong River. The copy numbers per cell of rbcLX for the eight remaining strains were in the range of 1.6-4.3 copies cell −1 . Each strain showed a different copy number per cell, with an average of 3.0 copies cell −1 (Table 1).    For Cylindrospermopsis, the result of quantitative rpoC1 gene analysis for 29-204 cells of the trace analysis sample indicated 106-253 copies. The result of quantitative gene analysis for 463-7420 cells mL −1 of the analytical sample per density indicated 1130-16,593 copies mL −1 , and the copy number of the rpoC1 gene per cell was 2.3 copies cell −1 (Figure 1). In the case of the four strains of Cylindrospermopsis isolated from Nakdong River, rpoC1 gene copy numbers per cell were in the range of 1.2-1.5 copies cell −1 . Strains isolated from Nakdong River did not show significant differences, but showed a difference of approximately 1 copy cell −1 from the CS-1101 isolated in Australia. Lastly, the copy number per cell for molecular ecological monitoring was calculated on the basis of the average copy number for each genus ( Table 1). The average copy number for Cylindrospermopsis was based on the average copy number per cell for the four strains isolated from the Nakdong.

Toxin Gene Copy Numbers per Cell and Concentrations of Toxin-Producing Cyanobacteria
The gene copy numbers and concentrations of the three toxins (anatoxin-a, cylindrospermopsin, and saxitoxin) were measured. Results showed that, when the cell density of the NIVA-711 strain of Cuspidothrix, which produces anatoxin-a, was in the range of 933-9950 cells mL −1 , the copy number of the anaF gene was 3080-14,580 copies mL −1 , and the toxin concentration was 0.020-0.134 µg L −1 . When the cell density of the CS-1101 strain of Cylindrospermopsis, which produces cylindrospermopsin, was in the range of 463-7420 cells mL −1 , the copy number of the cyrA gene was 967-15,933 copies mL −1 , the copy number of the cyrJ gene was 1664-21,400 copies mL −1 , and the toxin concentration was 0.012-0.043 µg L −1 . When the cell density of the NIVA-851 strain of Aphanizomenon, which produces saxitoxin, was in the range of 1826-14,620 cells mL −1 , the copy number of the sxtA gene was 4280-23,866 copies mL −1 , the copy number of the sxtI gene was 4347-22,800 copies mL −1 , and the toxin concentration was 0.001-0.016 µg L −1 . The three toxin genes showed proportional relationships with the cell counts and toxin concentrations (r 2 = 0.96 or higher) ( Figure 2). For anatoxin-a, the copy number of the anaF gene was 2.1 copies cell −1 , and the toxin concentration was 0.012 pg cell −1 . For cylindrospermopsin, the average copy number of the cyrA and cyrJ genes was 2.0 copies cell −1 , and the toxin concentration was 0.013 pg cell −1 . For saxitoxin, the average copy number of the sxtA and sxtI genes was 2.1 copies cell −1 , and the toxin concentration was 0.001 pg cell −1 ( Table 2).

Cell Density of INC in the Nakdong River
The genus Cuspidothrix was detected at a density of 0-1305 cells mL −1 during microscopic analysis. Quantitative gene analysis using a genus-specific primer (rpoB) showed a copy number of 0-1935 copies mL −1 , which when converted into cell count, indicated 0-1759 cells mL −1 (Figure 3). Cuspidothrix showed the highest cell density among the three INC genera. Cell density was the highest in June at the upstream sites (SJ, ND, GM), and in August at the midstream and downstream sites (GG, DS, CH, HC). The highest cell density among all sites was detected at the ND site in the second week of June. During the same period, cell density was also high at the GM site, at 956 cells mL −1 . At the ND site, density was 756 cells mL −1 even in the first week of June. In the second week of August, Cuspidothrix was detected at a concentration of approximately 200 cells mL −1 at the GG, DS, HC, and CH sites. At other sites, concentrations were less than 100 cells mL −1 . Microscopic analysis showed that Cuspidothrix first appeared at the SJ and ND sites in the fourth week of May. However, quantitative gene analysis detected it at all locations in the first week of March. Microscopic and quantitative gene analysis results showed that Cuspidothrix could be observed at most sites until November.
The genus Cylindrospermopsis appeared in the range of 0-60 cells mL −1 in morphological microscopic analysis. The result of quantitative gene analysis using a genus-specific primer (rpoC1) showed a copy number in the range of 0-90 copies mL −1 , corresponding to 0-69 cells mL −1 when converted into cell count. By site, the highest cell density was detected at GG in the third week of July. Cylindrospermopsis was detected at a concentration of 22-57 cells mL −1 at the GG and GM sites in the same period, at the ND site in the first week of April and the fourth week of October, and at the CH site in the second week of April. All other sites showed cell densities of less than 20 cells mL −1 (Figure 3).      The genus Cylindrospermopsis appeared in the range of 0-60 cells mL −1 in morphological microscopic analysis. The result of quantitative gene analysis using a genus-specific primer (rpoC1) showed a copy number in the range of 0-90 copies mL −1 , corresponding to 0-69 cells mL −1 when converted into cell count. By site, the highest cell density was detected at GG in the third week of July. Cylindrospermopsis was detected at a concentration of 22-57 cells mL −1 at the GG and GM sites in the same period, at the ND site in the first week of April and the fourth week of October, and at the CH site in the second week of April. All other sites showed cell densities of less than 20 cells mL −1 (Figure 3). The genus Sphaerospermopsis was detected at a concentration of 0-18 cells mL −1 under microscopic morphological analysis. The result of quantitative gene analysis using a genus-specific primer (rbcLX) showed a copy number in the range of 0-130 copies mL −1 , corresponding to 0-44 cells mL −1 when converted into cell count. By site, the highest cell density was detected at the CH site in the third week of June, and other sites showed cell densities less than 20 cells mL −1 (Figure 3). Microscopic analysis results showed that the cell density of Sphaerospermopsis was low (less than 20 cells mL −1 ) at all sites, and that it first appeared at the GG site in the second week of July. However, the results of quantitative gene analysis showed that it was first detected at a concentration of less than 5 cells mL −1 in the first week of March at the SJ, ND, and DS sites, and then at low densities at all locations until the second week of June. Furthermore, Sphaerospermopsis was detected until the third week of September according to microscopic analysis. However, in gene analysis, it was only detected at the CH site in the fourth week of October, whereas in November, it was not detected at any site.

Cyanobacteria with Toxin Genes in the Nakdong River
The anatoxin-a synthesis gene (anaF) copy number was in the range of 0-295 copies mL −1 , and when converted into a cell count, they were 0-140 cells mL −1 . By site, it was the highest at the ND site in the second week of June. Cyanobacteria with the anaF gene were detected in the range of 26-74 cells mL −1 at the SJ and GM sites during the same period. Furthermore, it was detected in the range of 25-77 cells mL −1 at the HC and CH sites in the second week of April, and at the GG, DS, HC, and CH sites in the second week of August. In contrast, it was lower than 10 cells mL −1 at other sites. The copy numbers of saxitoxin synthesis genes (sxtA, sxtI) were in the range of 0-101 copies mL −1 , and when converted into cell count, they were 0-48 cells mL −1 . By site, cell density was the highest at the ND site in the second week of July. Densities greater than 20 cells mL −1 were detected at the GM, HC, and CH sites from April to June. In contrast, densities were lower than 10 cells mL −1 at other sites. The copy numbers of the cylindrospermopsin synthesis genes (cyrA, cyrJ) were in the range of 0-46 copies mL −1 , corresponding to 0-23 cells mL −1 when converted into cell count. By site, cell density was the highest at the GM site in the third week of July; except when 17 cells mL −1 was detected at the SJ site in the fourth week of June, the cell density was generally lower than 10 cells mL −1 at other sites ( Figure 4).

ELISA-Based Toxin Analysis
The three types of toxins were analyzed using an ELISA kit at eight sites in the Nakdong River. Results showed that saxitoxin was detected in trace amounts in the range of 0.023-0.043 ng L −1 at the CG, GG, DS, and CH sites in April and May. Anatoxin-a was detected in the range of 0.154-0.284 ng L −1 at the ND and GM sites in the first and second weeks of June, but not at any other sites. Cylindrospermopsin was not detected at any site during the survey period (Table 3). Table 3. Result of toxin analysis using the enzyme-linked immunosorbent assay (ELISA) kit for samples collected at eight weir sites in Nakdong River (µg L −1 ).

Appearance Characteristics of INC
All three genera of INC were detected at all eight sites in Nakdong River in March, and they remained until November (Figure 3). Regarding the appearance trend according to the water temperature range, Cylindrospermopsis was first detected at water temperatures below 10 • C, and showed the highest appearance frequency at water temperatures of 20-30 • C; cell Toxins 2022, 14, 294 9 of 18 density was the highest at a water temperature of approximately 20 • C. Sphaerospermopsis was first detected at water temperatures below 10 • C and showed the highest appearance frequency at water temperatures of 20-25 • C; the cell density was the highest at a water temperature of approximately 25 • C ( Figure 5).

Appearance Characteristics of INC
All three genera of INC were detected at all eight sites in Nakdong River in March, and they remained until November (Figure 3). Regarding the appearance trend according to the water temperature range, Cylindrospermopsis was first detected at water temperatures below 10 °C, and showed the highest appearance frequency at water temperatures of 20-30 °C; cell density was the highest at a water temperature of approximately 20 °C. Sphaerospermopsis was first detected at water temperatures below 10 °C and showed the highest appearance frequency at water temperatures of 20-25 °C; the cell density was the highest at a water temperature of approximately 25 °C ( Figure 5). Cuspidothrix appeared in the full water temperature range (from below 10 to 30 °C). Cuspidothrix was also observed during phytoplankton monitoring analysis using a microscope at the eight weir sites in the Nakdong River in 2018-2019 [24]. Cell density was the Cuspidothrix appeared in the full water temperature range (from below 10 to 30 • C). Cuspidothrix was also observed during phytoplankton monitoring analysis using a microscope at the eight weir sites in the Nakdong River in 2018-2019 [24]. Cell density was the highest at water temperatures of 20-25 • C and at the upstream sites of SJ, ND, and GM ( Figure 6). At the ND site, where cell density was high, Cuspidothrix appeared at a concentration of 2524 cells mL −1 at water temperatures above 25 • C in June 2019. It began to appear at concentrations greater than 500 cells mL −1 after May, when the water temperature was maintained at 20 • C or higher.

Discussion
In this study, we developed a quantitative gene analysis method on the basis of genus-specific primers and toxin detection primers for INC, and they could be detected down to one trichome (approximately 25 cells). Additionally, INC could be detected down to the level of 1 cell mL −1 in field samples. Quantitative gene analyses in most prior studies targeted toxin genes of cyanobacteria that proliferate in large numbers, such as those in the genus Microcystis [38,39]. A few quantitative gene analysis studies were specifically

Discussion
In this study, we developed a quantitative gene analysis method on the basis of genusspecific primers and toxin detection primers for INC, and they could be detected down to one trichome (approximately 25 cells). Additionally, INC could be detected down to the level of 1 cell mL −1 in field samples. Quantitative gene analyses in most prior studies targeted toxin genes of cyanobacteria that proliferate in large numbers, such as those in the genus Microcystis [38,39]. A few quantitative gene analysis studies were specifically conducted on C. raciborskii, for which a species-specific primer was developed [27,28,40]. However, these studies calculated the copy number per milliliter or liter for the target gene, and compared it with the analyzed cell count from field samples with a microscope [27,28] or calculated the cell count for a culture sample with a higher density than 1.2 × 10 7 cells L −1 [40]. Therefore, it was difficult for them to accurately determine the cell density of target cyanobacteria that actually appear in rivers and lakes. Moreover, previous studies did not consider detecting target cyanobacteria in low-density samples.
Orr et al. [27] conducted a C. raciborskii monitoring study using real-time PCR. The copy number per cell of rpoC1 was in the range of 1.3-19.0 copies cell −1 , showing a large error. They analyzed samples from a lake rather than from cultured samples. Furthermore, morphological analyses using a microscope may underestimate the cell count of Cylindrospermopsis spp. when differentiated cells are not identified, such as the formation of terminal heterocytes and akinetes, which are classification keys, because of their morphological similarities with Aphanizomenon spp. and Cuspidothrix spp. [25]. This scenario seems to have caused differences in the copy number per cell. The difference in the copy number per cell between the Cylindrospermopsis strains isolated from Nakdong River and the CS-1101 strain isolated in Australia was not significant. However, there is a possibility of geographical heterogeneity [41]. Therefore, it was considered reasonable to calculate the cell count on the basis of the copy number per cell for strains isolated from the Nakdong River water system.
Quantitative gene analysis results could not be compared for Cuspidothrix and Sphaerospermopsis owing to the lack of prior studies. However, for Cuspidothrix, there were no significant differences in the copy numbers per cell for the genus-specific gene rpoB among strains. Hence, the analytical method used in this study may have helped in analyzing the cell density of Cuspidothrix. In the case of Sphaerospermopsis, genus-specific gene rbcLX showed differences in the copy numbers per cell for each strain. However, the average copy number per cell for the isolated strains was used because there were no primers that could specifically detect the target genus. Thus, additional development or improvements of the genus-specific primers is required.
Quantitative gene analysis results using toxin-detecting primers showed a proportional relationship between the gene copy numbers of the three toxins and the concentrations of each toxin, and cyanobacterial cell counts. However, the concentration of toxins is likely to change depending on environmental conditions, such as the growth stage of the target alga, the availability of nutrients, and water temperature [27,28]. Hence, accurate toxin concentration cannot be calculated using the quantitative toxin gene analysis results. However, this can be used as a proxy of the toxin production potential.
From the quantitative gene analysis for samples collected at eight sites in the Nakdong River, three genera of INC were detected at all study sites from March to November. Their appearance in the low-water-temperature period (March to April), which could not be verified by microscopic analysis owing to morphological similarities and the detection of only trace amounts, was confirmed using quantitative gene analysis. In the case of C. raciborskii, the minimal germination water temperature of the akinete was 15 • C [42]. However, in a lake in Poland, vegetative cells of C. raciborskii were detected at water temperatures below 7 • C [30]. In this study, rpoC1 (Cylindrospermopsis) was detected at a water temperature of 8.3 • C in the second week of March. There is a possibility that the actual water temperature for the germination of the akinete could be lower than this. Genera Cuspidothrix and Sphaerospermopsis were detected at water temperatures below 7 • C at the time of the first survey (March 2). The vegetative cells of S. aphanizomenoides were observed at water temperatures below 10 • C in a previous study [24]. Furthermore, the lowest water temperature during the survey period was 6.9 • C, and the possibility of INC presence from December to February, which was not investigated in the present study, cannot be excluded.
In 2020, the highest temperature of Nakdong River rarely exceeded 30 • C because rainfall was 1.5-1.9 times greater than the average. Thus, the generation trend and appearance of INC at water temperatures higher than 30 • C could not be verified. However, among the three genera of INC, Cuspidothrix showed relatively higher cell density, and the difference between the microscopic analysis result and the converted cell count from the quantitative gene analysis was not substantial when more than 200 cells mL −1 appeared. Therefore, it was possible to compare results with past monitoring data on the basis of microscopic analysis. In Nakdong River, blooms are mainly generated by Microcystis spp., and large blooms of up to 800,000 cells mL −1 that are generated in summer and autumn [43], mainly occur in the midstream and downstream sections. The cell density of the genus Cuspidothrix was higher in upstream sites (SJ, ND, GM), where the cell density of Microcystis was smaller than that in the mid and downstream sites ( Figure 6). During the period when the highest level of Cuspidothrix was detected in 2018-2019, cell density was the highest at a water temperature of 25 • C, similar to the result in 2020. This seemingly occurred because, although Cuspidothrix can grow at low water temperatures and actively proliferates at water temperatures above 20 • C, in summer, it is eliminated because of competition with Microcystis spp., a dominant cyanobacterium in the Nakdong River.
Quantitative analysis of the toxin genes showed that the cell count of the cyanobacteria with saxitoxin and cylindrospermopsin synthesis genes did not exceed 50 cells mL −1 at the maximum, and their appearance frequency was also low. In addition, the detection of the two toxin genes did not agree with the appearance trends of Cylindrospermopsis or Sphaerospermopsis. Thus, the toxin production potential of these two genera was considered to be low in the Nakdong River water system. Cylindrospermopsin was not detected in toxin measurements using ELISA, whereas saxitoxins were detected in trace amounts in the midstream and downstream sections in April and May. Although Cuspidothrix were detected during this period, there was no saxitoxin-producing strain among those isolated from the Nakdong River. Cyanobacteria with the anaF gene showed the highest cell count among the cyanobacteria with the three types of toxin genes; however, it was not found at concentrations exceeding 200 cells mL −1 . The appearance of cyanobacteria with the anaF gene agreed with the appearance trend of Cuspidothrix, and with the time of detection of trace amounts of anatoxin-a by ELISA (Figures 3 and 4; Table 3). A previous study found anaF gene expression and anatoxin-a production in Cuspidothrix strains (NRERC-654, 661) isolated from Nakdong River [24]; thus, Cuspidothrix was lastly verified as the alga that produced anatoxin-a in Nakdong River.
Although the appearance of INC and the amount of produced toxins were low, quantitative gene analysis using toxin-and genus-specific primers in Nakdong River revealed the potential production of anatoxin-a by Cuspidothrix at eight sites in Nakdong River. INC are currently increasing globally due to climate change, and many countries, including Korea, manage only microcystin, so it is necessary to manage toxins that INC can produce. Therefore, INC must be continuously monitored as part of the safe water quality management of Nakdong River, which is used as an important water source.

Conclusions
Quantitative gene analysis using toxin-and genus-specific primers enabled the precise analysis of the appearance times, cell densities, and toxin production potential of target algae. The toxin production potential of Cuspidothrix, rather than Cylindrospermopsis, was verified in Nakdong River, a temperate climate zone. The presence of Cuspidothrix must be continuously monitored for managing the water quality of Nakdong River, a crucial water source. Furthermore, the isolation of more INC strains and comprehensive analyses of gene copy numbers per cell for each strain are warranted to improve the accuracy of future quantitative gene analyses. If primers are developed and applied to cyanobacteria that produce various toxins, we expect that harmful cyanobacteria and related toxins can be monitored quickly and effectively using molecular biological techniques rather than microscopic analysis.

Selection of INC and Cultivation of Toxin-Producing Cyanobacteria
Three species of INC, C. raciborskii, S. aphanizomenoides, and C. issatschenkoi, were selected as the target species. In addition, S. reniformis, which had been found in the Nakdong River, was also included. However, Raphidiopsis and Chrysosporum, which are tropical and invasive cyanobacteria genera, were excluded from the target INC species in this study because their appearance in Nakdong River was not verified. For the cyanobacterial toxins, three toxins produced by INC, anatoxin-a, cylindrospermopsin, and saxitoxin, were selected: toxin-producing C. issatschenkoi strain NIVA-711 (Oslo, Norway), which contains toxin synthesis gene anaF, was used for anatoxin-a; C. raciborskii strain CS-1101 (ANACC, CSIRO, Clayton South, Australia), which contains toxin synthesis genes cyrA and cyrJ, was used for cylindrospermopsin; and toxin-producing Aphanizomenon gracile strain NIVA-851 (Oslo, Norway) strain, which contains toxin synthesis genes sxtA and sxtI, was used for saxitoxin. The INC of 4 species and 24 strains [24] isolated from Nakdong River, and positive controls of 3 species and 3 strains, were cultured under controlled conditions at 20 • C, 40 µEm −2 s −1 , and 14/10 h (Light/Dark) in MLA [44] medium. Strains were proliferated until the logarithmic growth phase before being used for quantitative gene analysis (Table 4).

Isolation of Genomic DNA
To extract genomic DNA, cultured algal and field samples were filtered with a 0.45 µm MicronSep nitrocellulose membrane disk (GVS Life Sciences, Findley, OH, USA) and stored at −80 • C until use. Genomic DNA was extracted using the QIAGEN DNeasy Plant Mini Kit (QIAGEN, Hilden, Germany). For the gene extraction process, a piece of filter paper stored at −80 • C was left at room temperature (15-25 • C) for 5 min. Then, a lysis buffer was added to the sample-containing filter paper, which was then sonicated (Vibra-Cell, Sonics & Materials, Newtown, CT, USA) following the protocol provided by the manufacturer. The concentration and purity of the extracted genomic DNA were measured using an Infinite M200 PRO Microplate Reader (Tecan Austria GmbH, Grödig, Austria). The extracted genomic DNA was stored at −20 • C.

Design of Toxin and Genus-Specific Primers for Quantitative Gene Analysis
In this study, quantitative gene analysis was performed using digital droplet PCR (ddPCR). This method offers the advantage of precisely detecting target genes in a complex mixture. It has been recently used to quantitatively analyze harmful marine algae [45] and microbes in water treatment plants [46], and to detect toxic cyanobacteria in plant leaves [47]. For quantitative gene analysis, the primer set was designed to include probes in addition to forward and reverse primers based on primers for detecting genes (anaF, cyrA, cyrJ, sxtA, and sxtI) related to the production of anatoxin-a, cylindrospermopsin, and saxitoxin [24,[34][35][36][37] and genus-specific primers [33] of INC (Cylindrospermopsis, Sphaerospermopsis, and Cuspidothrix) (Tables 5 and 6).

Quantitative Gene Analysis
The QX200 ddPCR system (Bio-Rad Laboratories Inc., Hercules, CA, USA), comprising a T100 Thermal Cycler measures the target gene copy number in single cells. When ddPCR was performed, the final sample volume was 20 µL. Each reaction mixture was composed of 10 µL of ddPCR ™ Supermix for Probes (no dUTP) (Bio-Rad Laboratories Inc., Munich, Germany), 0.9 µL each of 10 pmol forward and reverse primers, 0.25 µL of 10 pmol probe, and 1 µL of sample DNA. The final volume was adjusted using sterile distilled water. Reaction mixtures were separately inserted into each well of a 96-well plate. Each reaction mixture was divided into 20,000 droplets of approximately 1 nm in the Bio-Rad QX200™ Droplet Generator (Bio-Rad Laboratories Inc., Hercules, CA, USA). Droplets were then transferred to a new 96-well plate. PCR was performed in a T100 Thermal Cycler (Bio-Rad Laboratories Inc.). After PCR, the plate was moved to a droplet reader (Bio-Rad Laboratories Inc.), and droplets were quantified by measuring their positive or negative signals. Signals were analyzed using QuantaSoft™ software version 1.7.4 (Bio-Rad Laboratories Inc.). Values were used as actual measurements only when at least 12,000 droplet signals were read [48]. The threshold values of the fluorescent signals were directly set according to the signal values of the negative and positive controls. All reagents related to ddPCR were purchased from Bio-Rad Laboratories, Inc.

Calculation of Genus-Specific Gene Copy Number per Cell for the Target Cyanobacteria
Quantitative gene analysis of INC was performed by converting the ddPCR result into a cell count. To this end, the copy number of genus-specific genes (Cylindrospermopsis: rpoC1, Sphaerospermopsis: rbcLX, Cuspidothrix: rpoB) per cell of the three genera of INC were analyzed. For the analysis of the genus-specific gene copy number per cell, representative strains were selected for each species (NIVA-711, NRERC-600, and CS-1101). Cultured cyanobacteria were diluted to a density of 1 × 10 3 , 3 × 10 3 , 5 × 10 3 , 7 × 10 3 , and 10 × 10 3 cells mL −1 , and filtered through a membrane disk to extract genomic DNA before performing ddPCR. Each diluted sample was preserved by adding Lugol's iodine solution (final concentration 0.3%), and cell density was analyzed using a phase-contrast microscope (Imager M2; Carl Zeiss, Germany). Trace samples were analyzed by separating trichomes, the smallest unit of the target cyanobacteria, into 25, 50, 100, and 200 cells using a Pasteur pipette, and attaching them to a membrane filter. The gene copy number per cell (copies cell −1 ) was determined for each sample on the basis of cell density and quantitative gene analysis results. In addition to the representative strains, the copy number per cell for strains isolated from Nakdong River was calculated by performing density analysis.

Calculation of Toxin Gene Copy Number per Cell for Toxin-Producing Cyanobacteria
To calculate the toxin gene copy number per cell for toxin-producing cyanobacteria, strains that had been verified to produce toxins (CS-1101, NIVA-711, and NIVA-851) were cultured and diluted. Genomic DNA was extracted by filtering strains through a membrane disk, and ddPCR was performed for the toxin genes (anatoxin-a: anaF; cylindrospermopsin: cyrA and cyrJ; saxitoxin: sxtA, and sxtI). Each diluted sample was preserved by adding Lugol's iodine solution, and cell density was analyzed using a phase-contrast microscope. Lastly, the toxin gene copy number per cell (copies cell −1 ) was calculated using the quantitative gene analysis results.

Analysis of INC at Eight Sites in Nakdong River
To analyze the cell density of the INC, 2 L of surface water was collected 500 m upstream from eight weir sites (Sangju, SJ; Nakdan, ND; Gumi, GM; Chilgok, CG; Gangjeong-Goryeong, GG; Dalseong, DS; Hapcheon-Changnyeong, HC; and Changnyeong-Haman, CH) from upstream to downstream of Nakdong River biweekly, from March to November (weekly from June to September) (Figure 7). The climate of the Nakdong River system is a temperate zone with four distinct seasons, which drop below 0 in winter, 10-15 • C in spring and autumn, and 25-30 • C in summer. In summer, typhoons and rainy seasons also bring much rainfall.
Collected samples were kept at 4 • C while being moved to the laboratory. For microscopic analysis, 1 L water samples were preserved in glass bottles with Lugol's solution (0.3%). After leaving the sample for 48 h or more for natural precipitation, the supernatant was removed with a siphon and concentrated five to eight times. For quantitative gene analysis, 25 mL of a water sample was filtered using a 0.45 µm MicronSep™ nitrocellulose membrane disk to extract the genomic DNA, which was stored at −20 • C. For toxin analysis, water samples were used to fill 50 mL conical tubes, which were then stored at −20 • C. Before analysis, samples were thawed in a refrigerator at 4 • C, sonicated, and moved to separate tubes to individually analyze the three toxins. Anatoxin-a and saxitoxin were treated by adding the preservative solution included in the ELISA kit.
For morphological analysis of the INC using a microscope, 1 mL of the concentrated sample was analyzed at 100-400× magnification under a phase-contrast microscope using a Sedgewick Rafter counting chamber [25,49]. To identify the INC, only those cells that allowed for a clear distinction of the species by the characteristics of the terminal cells, the formation and position of the heterocytes, and the position of the akinete, which are important classification indicators for these species, were identified. Cell count per milliliter was determined by converting the concentration ratio.
For quantitative analysis using the genus-specific primers for INC, ddPCR was performed for genomic DNA using genus-specific primers (Table 5). Regarding the copy number of the target genes, the cell count of the target cyanobacteria per milliliter was calculated using the copy number per cell derived from quantitative gene analysis. For the quantitative analysis of cyanobacteria harboring toxin-related genes, ddPCR was performed for genomic DNA using the toxin primer set (Table 6). For the copy number of the target genes, the cell count per milliliter of cyanobacteria with toxin genes was calculated using the copy number per cell determined in this study. For the saxitoxin and cylindrospermopsin synthesis genes, cells were counted only when both pairs of genes (sxtA and sxtI, and cyrA and cyrJ) were detected. Regarding the anatoxin synthesis gene, it was considered that the toxin could only be produced if the anaF gene was present. For quantitative analysis using the genus-specific primers for INC, ddPCR was performed for genomic DNA using genus-specific primers (Table 5). Regarding the copy number of the target genes, the cell count of the target cyanobacteria per milliliter was calculated using the copy number per cell derived from quantitative gene analysis. For the quantitative analysis of cyanobacteria harboring toxin-related genes, ddPCR was performed for genomic DNA using the toxin primer set (Table 6). For the copy number of the target genes, the cell count per milliliter of cyanobacteria with toxin genes was calculated using the copy number per cell determined in this study. For the saxitoxin and cylin-

Toxin Analysis Using ELISA Kit
Toxin analysis of the cultured cyanobacterial strains and water samples collected near the eight weirs was performed using anatoxin-a (520060), saxitoxin (52255B), and cylindros-permopsin (522011) ELISA kits (Eurofins Abraxis, Inc., Warminster, PA, USA). For analysis, cultured cyanobacteria and water samples were placed in 50 mL tubes and sonicated on ice, followed by ELISA according to the manufacturer's instructions. Absorbance was measured at 450 nm using an Infinite M200 PRO microplate reader (Tecan Austria GmbH, Grödig, Austria).