Multiple cyanotoxin congeners produced by sub-dominant cyanobacterial taxa in riverine cyanobacterial and algal mats

Benthic cyanobacterial proliferations in rivers are have been reported with increasing frequency worldwide. In the Eel and Russian rivers of California, more than a dozen dog deaths have been attributed to cyanotoxin toxicosis since 2000. Periphyton proliferations in these rivers comprise multiple cyanobacterial taxa capable of cyanotoxin production, hence there is uncertainty regarding which taxa are producing toxins. In this study, periphyton samples dominated by the cyanobacterial genera Anabaena spp. and Microcoleus spp. and the green alga Cladophora glomerata were collected from four sites in the Eel River catchment and one site in the Russian River. Samples were analysed for potential cyanotoxin producers using polymerase chain reaction (PCR) in concert with Sanger sequencing. Cyanotoxin concentrations were measured using liquid chromatography tandem-mass spectrometry, and anatoxin quota (the amount of cyanobacterial anatoxins per toxigenic cell) determined using droplet digital PCR. Sequencing indicated Microcoleus sp. and Nodularia sp. were the putative producers of cyanobacterial anatoxins and nodularins, respectively, regardless of the dominant taxa in the mat. Anatoxin concentrations in the mat samples varied from 0.1 to 18.6 μg g-1 and were significantly different among sites (p < 0.01, Wilcoxon test); however, anatoxin quotas were less variable (< 5-fold). Dihydroanatoxin-a was generally the most abundant variant in samples comprising 38% to 71% of the total anatoxins measured. Mats dominated by the green alga C. glomerata contained both anatoxins and nodularin-R at concentrations similar to those of cyanobacteria-dominated mats. This highlights that even when cyanobacteria are not the dominant taxa in periphyton, these mats may still pose a serious health risk and indicates that more widespread monitoring of all mats in a river are necessary.

Recently toxic benthic cyanobacteria have been recorded in Californian streams [16,17,18]. Anatoxins, MCYs, NODs and CYNs have all been recorded in the mats and in samples collected using solid-phase absorption toxin tracking (SPATT) samplers [16,17]. A number of sites on the Eel and Russian rivers in Northern California regularly experience benthic cyanobacterial and chlorophyte proliferations [16]. The mats at these sites contain assemblages comprised of a variety of cyanobacterial genera known to produce cyanotoxins. Extensive mats dominated by the green alga C. glomerata, are often also present at these locations, with low levels of cyanobacteria in the mats. There is uncertainty about which genera in these systems are producing toxins, although Microcoleus has been identified as an anatoxin producer in the Eel River using assembled metagenomes [19], and cultures of Microcoleus isolated from the Russian River also produce anatoxins [18]. Both toxic and non-toxic genotypes occur in the genus Microcoleus, and these genotypes can be present at varying relative abundances in environmental mats [15,20], resulting in considerable spatial variability in anatoxin concentrations of mats within short distances.
Identifying cyanotoxin producers with culture-based studies can be prohibitively costly considering the range of potential toxic taxa present and the spatial variability in toxin production; however, molecular techniques (e.g., PCR and quantitative PCR) can be used to screen for the presence of genes involved in cyanotoxin biosynthesis. These techniques, coupled with DNA sequencing, can provide a strong indication of the toxin producers in environmental samples and this information can then be used to guide culturing of selected species. Molecular assays to determine the concentration of anaC gene copies using quantitative PCR (qPCR) and droplet digital PCR (ddPCR) have recently been developed and allow the toxin quota (amount of toxin per toxic cell) to be determined [15,20]. Comparisons of toxin quota, combined with data on the physicochemical conditions at the time of sampling, may provide insights into the drivers of anatoxin production.
To gain a better understanding of the diversity of cyanotoxins and cyanotoxin producers in the Eel and Russian rivers of Northern California, this study aimed to: (1) identify potential benthic cyanotoxin producers, (2) assess spatial variability in anatoxin concentrations and quotas within and between mats, sites and rivers, and (3) determine cyanotoxin levels in Cladophora-dominated mats. Periphyton samples were collected from the Eel and Russian rivers. PCR amplification of genes involved in toxin production combined with DNA sequencing of the amplified genes were used to identify potential cyanotoxin producers and liquid chromatography tandem-mass spectroscopy (LC-MS/MS) was used to quantify anatoxin congeners, nodularin-R (NOD-R) and MCYs. Spatial variability in anatoxin concentrations and anatoxin quotas were investigated by combining toxin concentrations determined by LC-MS/MS with ddPCR quantification of anaC gene copies.

Environmental sample collection
Sampling sites were chosen in the Eel and Russian rivers in Northern California. Sampling sites were either publicly accessible, or permission to sample was granted by landowners. No additional permits were required to collect samples. The riverbeds at the sampling sites were comprised of gravel, cobble or boulders. The four sites in the Eel River watershed were sampled on 29 July 2018 and spanned upstream drainage areas of 17 km 2 to 495 km 2 . Site 1_ELD was on Elder Creek and sites 2_SFE-4_SFE were on the South Fork Eel River (Fig 1). The single Russian River site (5_RUS) was sampled on 31 July 2018 and had an upstream drainage area of 793 km 2 (Fig 1). At each site, ten samples with visible cyanobacterial mats were selected and attached cyanobacteria collected by scraping a small sample (ca. 2 cm diameter) into a 15 mL centrifuge tube. At 4_SFE, fine-scale samples were collected by taking five samples from each of three additional rocks. Additional samples of C. glomerata mat material were collected at 3_SFE on the South Fork Eel River. Samples were stored on ice, frozen (-20˚C) on return to the laboratory (within 8 hours), and subsequently lyophilised prior to further analysis. At each site dissolved oxygen, temperature, pH, and conductivity were measured in the water with handheld probes. Filtered water samples were also collected (0.7 μm Whatman GF/F) and dissolved inorganic nitrogen and dissolved reactive phosphorus were measured using EPA methods 300.1, 351.2 and 365.2. All ten samples from each site were analysed for cyanotoxin concentrations. Three samples from each site were screened for cyanotoxin genes, which were sequenced if detected. Anatoxin quotas (the amount of anatoxin per toxigenic cell) were determined for the five samples with the highest anatoxin concentrations from sites 3_SFE, 4_SFE and 5_RUS, as well as the fine-scale samples from 4_SFE (S1 Fig). These sites were selected because anatoxin concentrations were markedly higher at these locations, and practical constraints limited the number of samples that could be processed.

Toxin analysis
Lyophilised environmental samples were homogenised with a sterile metal spatula and subsampled for cyanotoxin analysis by suspending ca. 10 mg dry weight of mat material in 1 mL of 0.1% formic acid in water. A previous assessment of microcystin extraction using an aqueous solvent compared to an organic extraction solvent demonstrated comparable extraction efficiency [21]. As NOD-R is slightly more polar than MCY-LR, under acidic conditions, it was assumed that the extraction efficiency would be similar. The samples were frozen at -20˚C and subsequently thawed in a bath sonicator for 30 min (53 kHz; LHC bath sonicator, Kudos, Shanghai, China). Freeze-thaw and sonication steps were repeated two more times for each sample. Cyanotoxin extracts were clarified by centrifugation (14,000 × g, 5 mins) and the supernatants transferred to septum-capped glass vials for analysis of anatoxins, CYNs, MCYs and NOD-R using liquid chromatography tandem-mass spectrometry (LC-MS/MS).
Anatoxins (ATX, HTX, dhATX and dhHTX) and CYNs (CYN and deoxycylindrospermopsin; doCYN) were analysed using the LC-MS/MS methodology described in Wood et al. [22]. Multiple-reaction monitoring (MRM) channel values for the anatoxins are provided in Wood et al. [22] and quantitation MRM channels for CYN and doCYN were 416.3 > 194.15 and 400.3 > 194.15, respectively. The anatoxins and cylindrospermopsins were quantified using a mixed external five-point calibration curve (0.5 ng mL -1 -18 ng mL -1 in 0.1% formic acid) made from certified reference materials for ATX and CYN (National Research Council, Canada) and a standard for dhATX calibrated by quantitative nuclear magnetic resonance spectroscopy. The concentrations of ATX and HTX were determined using the ATX calibration curve and the concentrations of dhATX and dhHTX were determined using the dhATX calibration curve and the concentration of CYN and doCYN were determined using the CYN standard. The analytical limit of detection (LoD) for ATX, dhATX and CYN was 0.05 ng mL -1 , which equates to an approximate LoD of 0.005 μg g -1 in the sample extracts (dependent on the amount of sample weighed out for extraction). The limit of quantitation (LoQ) was 0.02 μg g -1 dw.
Microcystins and NOD-R were analysed as described in Wood et al. [23] using a mixed external four-point calibration curve (2 ng mL -1 -100 ng mL -1 in 50% methanol) made up of standards for NOD-R, MCY-RR, -YR and -LR (DHI Lab Products, Denmark). The analytical LoD for MCY-RR, -YR, -LR and NOD-R was 0.02 ng mL -1 , which equates to an approximate LoD of 0.002 μg g -1 in the sample extracts (dependent on the amount of sample weighed out for extraction). The LoQ was 0.006 μg g -1 dw.

Molecular analyses
A sub-sample (ca. 15 mg dry weight) was placed into the first tube of a PowerSoil1 DNA Isolation Kit (Qiagen, CA, USA) and DNA extracted according to the manufacturer's protocols. The extracted DNA was quantified (NP80 NanoPhotometer, Implen GmbH, Munich, Germany) and stored at -20˚C until further molecular analyses.
For sequencing, positive PCR reactions were purified using a Nucleospin PCR clean-up kit (Machery-Nagel, Düren, Germany), according to the manufacturer's directions. Purified PCR product was then quantified (NP80 NanoPhotometer, Implen GmbH, Munich, Germany) and diluted to a concentration of 5 ng μL -1 (mcyE) or 4 ng μL -1 (anaC). Amplicons were sequenced bi-directionally with gene-specific primers using the BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, USA). Sequences were compared for similarity to reference sequences using Blastn (NCBI). Sequences obtained in this study were deposited in GenBank under accession numbers MK821061 to MK821086.
The absolute number of copies of anaC gene in samples that were anaC-positive using PCR were quantified by ddPCR, using the Phor-AnaC primers and probe (S1 Table) according to the methods of [15]. Briefly, the ddPCR reactions consisted of 1 × BioRad ddPCR supermix for probes (no dUTP), 450 nM of each primer and the probe, 1 μL of template DNA diluted to 1:20 and sterile water for a final reaction volume of 22 μL. Nanodroplets were generated using a BioRad QX200 droplet generator, which combined 70 μL of BioRad droplet oil with 20 μL of the reaction mixture, resulting in 40 μL of nanodroplets. Amplification was conducted on a PCR plate using a cycling protocol of a 95˚C for 10 min followed by 40 cycles of 94˚C for 30 sec, 60˚C for 1 min and a final hold at 98˚C for 10 min. The QX200 instrument was used to analyse the number of positive and negative droplets in the samples. The anaC copies per μL from the ddPCR was converted to copies of anaC per mg of sample using the formula: anaC copies per μL × 22 μL (the initial reaction volume) × 100 μL (the elution volume of the DNA extraction) / weight of the lyophilized material (in mg). Anatoxin quota were calculated as the summed anatoxins mass per mg dw (from the LC-MS/MS) divided by the anaC copy number per mg dw from the ddPCR as described by Kelly et al. [20]. Results of anatoxin quota are expressed as pg per cell as the anaC gene is single copy.

Statistical analysis
All statistical analyses were conducted in the software R Studio (R Version 3.5.1). Mean toxin concentrations, anaC gene copy numbers and anatoxin quota were tested for normality using the Shapiro-Wilks test. All variables failed to meet parametric test assumptions, so nonparametric tests were used. Comparisons of mean toxin concentrations and anatoxin quota were undertaken using Kruskal-Wallis tests and pairwise Wilcoxon rank sum tests with Benjamini-Hochberg adjustment for multiple comparisons.

Cyanotoxin presence and variability
All five sites surveyed had visible cyanobacterial biomass present. Sites 1_ELD, 2_SFE and 4_SFE were riffle habitats, and cyanobacterial mats were cohesive, attached to benthic cobbles, and were a black/brown colour characteristic of Microcoleus-dominated mats in these rivers. Site 3_SFE was a slow-moving pool and was dominated by senescing C. glomerata and spires of Anabaena spp., with Nostoc spp. also present upstream of the survey site. Site 5_RUS was a run habitat on the Russian River. It was also dominated by C. glomerata and epiphytized by Anabaena spp. In general, environmental conditions at the sites were similar, although there was some variation in temperature and nutrient concentrations among sites (Table 1). Water column nitrogen concentrations ranged from 0.13-0.15 mg L -1 and phosphorus concentrations ranged from 0.009-0.020 mg L -1 .
Anatoxins were detected by LC-MS/MS at all five sites, though levels were low (< 0.2 μg g -1 dw) at sites 1_ELD and 2_SFE (Fig 2A). Total anatoxin concentrations at the other three sites were considerably higher, with samples ranging up to 18.6 μg g -1 dw. The mean anatoxin concentration at sites 3_SFE, 4_SFE and 5_RUS were significantly different both from each other and from 1_ELD and 2_SFE (Fig 2A; p < 0.01; Wilcoxon test). Hepatotoxins comprised NOD-R, MCY-LR, and dmMCY-LR and were detected at all sites, though the concentrations in 4_SFE and 5_RUS were very low and only detected in one (4_SFE) or two (5_RUS) samples ( Fig 2B). Site 3_SFE had the highest hepatotoxin concentrations (mean 1.4 μg g -1 dw), with all other site concentrations between the LOD of 0.002 μg g -1 and LOQ of 0.006 μg g -1 dw. Microcystin/nodularin concentrations were below quantitation limits at most sites. Mean concentrations (± standard error) of anatoxins in the C. glomerata mats were 7.4 μg g -1 dw (± 0.15 μg g -1 dw) and mean nodularin concentrations were 0.33 μg g -1 dw (± 0.27 μg g -1 dw).
The anatoxins detected in the samples were almost exclusively comprised of ATX and dhATX, with low levels (< 0.01 μg g -1 dw) of HTX and dhHTX only detected at 1_ELD and 2_SFE (Fig 3). The composition of anatoxin congeners was similar within each site (S2 Table). The exception was site 4_SFE, which had more variable proportions of ATX and dhATX among samples. Hepatotoxins (MCYs and NODs) were almost exclusively comprised of NOD-R, with only low levels (< 0.01 μg g -1 dw) of MCY-LR and dmMCY-LR detected. Nodularin-R was the only hepatotoxin detected at 1_ELD and 5_RUS.

Potential cyanotoxin producers
At sites 3_SFE, 4_SFE and 5_RUS the PCR reactions were positive for anaC, but results were negative for samples from sites 1_ELD and 2_SFE. All the anaC sequences (n = 17) were identical and most closely matched those of Oscillatoria sp. from the Pasteur Culture Collection (Oscillatoria sp. PCC 10601, PCC 9240; GenBank accession: JF803652, JF803653; 100% cover and 99.7% identity) and Phormidium autumnale (now Microcoleus autumnalis; CAWBG618; GenBank accession: KX016036; 93% cover and 99.6% identity) from the Cawthron Institute Culture Collection of Microalgae (http://cultures.cawthron.org.nz/). Comparison of sequences from this study to draft metagenomes from [18] revealed 100% nucleotide sequence identity with the anaC gene from the draft M. autumnalis genomes assembled in their study. The closest Anabaena sp. sequence using BLASTn (Anabaena circinalis, GenBank accession: JF803647) yielded only 84.4% nucleotide sequence identity. Samples from site 3_SFE had PCR detections for the mcyE/nduF genes. Sequences from these reactions (n = 9) were identical and shared the closest sequence similarity to Nodularia spumigena (GenBank accession: CP020114; 100% Multiple cyanotoxin congeners produced by sub-dominant cyanobacterial taxa in riverine periphyton cover and 99.7% identity). All sequences obtained from the attached and floating samples (n = 3) of C. glomerata were identical to the above sequences from the cyanobacteria-dominated samples. Saxitoxin and cylindrospermopsin genes were not detected in any of the samples.

Within-site and within-mat cyanotoxin variability
Three sites were assessed for within site variability (3_SFE, 4_SFE and 5_RUS). Low levels of variability in anatoxin concentrations (< 5-fold) were observed within each site (Fig 4A). When anatoxin concentrations were normalised to the concentration of anaC gene copies ( Fig  4B), the variability was reduced to < 2-fold in each site ( Fig 4C). When samples had high anatoxin concentrations relative to others from the same site (e.g., 5_RUS sample 5; Fig 4), normalisation to the number of toxic cells reduced the anatoxin quota to levels comparable with the other samples. The mean anatoxin quota of 3_SFE was 0.98 pg cell -1 , while 4_SFE and 5_RUS had quotas of 0.23 pg cell -1 and 0.12 pg cell -1 , respectively. The mean anatoxin quota among all three sites were significantly different (p < 0.05; Kruskal-Wallis and Wilcoxon tests). To examine the variability of anatoxin concentrations and anatoxin quota within a single site, Microcoleus-dominated mats were investigated at site 4_SFE. Among the five within-mat samples collected from each of three rocks at 4_SFE, total anatoxins (the sum of the four congeners) per mg dw varied by 7-fold. There were no obvious patterns among mats and no significant differences between anatoxin gene copies or quota (p > 0.05; Fig 5). Individual samples (i.e. rock 1; Fig 5) with high anatoxin concentrations had comparable anatoxin quota upon normalisation to the abundance of toxic cells.

Cyanotoxin producers
The first aim of this study was to use molecular techniques to identify potential cyanotoxin producers in benthic mats from the Eel and Russian rivers. The anaC gene sequences strongly indicate that Microcoleus is the anatoxin producer, both in the Microcoleus and Anabaenadominated mat samples. There was no evidence that Anabaena was an anatoxin producer, with the anaC gene nucleotide sequence identity less than 85% between our sequences and the closest Anabaena sp. match. This is consistent with results from Anabaena cultures isolated, and Anabaena draft-genomes assembled from the Eel and Russian rivers (unpublished observations; K. BG.). This study detected the anaC gene at sites where [18] did not find anatoxin biosynthesis gene clusters in their metagenomes; however our results correspond with previous anatoxin detections in the upper reaches of the Eel watershed [16]. Bouma-Gregson et al. [18] found that only one of four Microcoleus strains detected in the metagenomes contained the anatoxin biosynthesis gene cluster. The lack of anaC detections with the ddPCR at 1_ELD and 2_SFE despite anatoxins being present at very low concentrations likely reflects the difficulty in completely homogenising samples and highlights the sensitivity of LC-MS/MS as an analytical detection method for these toxins.
Sequencing identified a nodularin producer, with the gene most closely resembling that from N. spumigena. This species is generally found in brackish water, though it has been reported in freshwater lakes in Europe based on morphometric identification [28]. Nodularia has been observed in environmental samples from the rivers in this study, however as we did not taxonomically identify the cyanobacterial communities in this study it remains a possibility that another cyanobacterium is responsible for the production of nodularin. Two other genera are known to produce nodularin: Iningainema in Northern Australia [29] and a Nostoc endosymbiont in cycads [30]. Culturing of the causative species and sequencing of other genes involved in nodularin biosynthesis is required to determine whether a novel nodularin producer is present. Sequences for mcyE were not detected in the samples, likely due to low abundances of mcyE genes compared with ndaF genes. Despite this, the presence of microcystins indicates that a yet-to-be-identified microcystin producer is also present in the Eel River.

Concentrations and variability in anatoxin, microcystin and nodularin
The concentrations of anatoxins, NOD-R and MCYs were consistent with data in previous studies from these sites [16,31], with anatoxin concentrations five times greater than microcystin/nodularin concentrations. The majority of NOD-R and MCY concentrations detected in this study were less than 1 μg g -1 , which is comparable to those detected in benthic cyanobacteria in Pennsylvania [32]. Solid-phase adsorption toxin tracking samplers (SPATTs) in the Eel River accumulated ATX and MCYs in 53%-54% and 41%-76% of samples for each toxin, respectively [16]. The site with the highest concentrations of both anatoxins and NOD-R was dominated by Anabaena, a finding that contrasts with that of Bouma-Gregson et al. [16], who found no differences in ATX concentrations between Anabaena and Microcoleus dominated mats, although the authors only measured ATX, not dhATX so may have underestimated the total anatoxin concentrations. This difference also demonstrates the important inter-annual variability of periphyton proliferations which highlights the need to further understand within mat, within site, within season and inter-annual variation. Multiple cyanotoxin congeners produced by sub-dominant cyanobacterial taxa in riverine periphyton All samples were dominated by ATX and dhATX, and not HTX and dhHTX, which contrasts with observations from Microcoleus-dominated mats elsewhere: dhATX, dhHTX and HTX typically dominate in New Zealand [8,15], and in France ATX dominates [33], although dhATX has also been detected [34]. Both ATX and HTX have been detected in benthic cyanobacteria from Pennsylvania, with HTX being the dominant cyanobacterial anatoxin congener [32] although the producing taxa were not investigated in that study. Dihydrohomoanatoxin-a was also putatively identified in the Pennsylvania samples using LC-MS/MS but was not quantified [32]. Similarly, Phormidium mats responsible for dog deaths in the Netherlands contained ATX, trace detections of HTX and suspected dhATX [35]. As the authors could not quantify the dhATX due to a lack of reference material [35], it is not clear which congener was dominant in those samples and it is possible that dhATX comprises a major component of the total cyanobacterial anatoxins in that system as well. Anderson et al. [19] used cyanobacterial anatoxins in crude extracts of Microcoleus cultures isolated from the Russian river in their toxicological study and dhATX comprised > 99% of the cyanobacterial anatoxins. This suggests there is likely large strain to strain variability in the variants produced. Homoanatoxin has comparable intraperitoneal toxicity to ATX [36,37,38]. Whilst the toxicity of dhHTX and dhATX is presently not known, reports of dog deaths following consumption of cyanobacterial material containing high concentrations of dhATX and dhHTX [22] suggest that the dihydroanatoxin congeners are toxic to mammals. The high proportion of dhATX identified in this study highlights the need to use analytical methods that detect this congener (along with HTX and dhHTX) to prevent underestimation of the risk posed by benthic cyanobacteria in these systems.
Spatial variability of anatoxin content occurred both among and within sites. The magnitude of this variability was relatively low, and upon normalisation to anatoxin quotas, the variability was reduced, a result that is consistent with other studies on Microcoleus-dominated mat samples [15,20]. Anatoxin quotas in this study (0.12 pg cell -1 -0.98 pg cell -1 ) were similar to those observed in the Cardrona River in New Zealand (mean 0.44 pg cell -1 [15]), however, anatoxin quotas in excess of 7.5 pg cell -1 have been reported [20]. Anatoxin quotas can vary both within and between rivers, with 16-to 42-fold differences in anatoxin quotas observed for samples collected at the same site on the same day [15]. It is possible that environmental parameters are linked to anatoxin concentrations or quota although there is insufficient data in this study to investigate this. In addition, there are significant challenges with accurately measuring environmental parameters at scales that are relevant to the cyanobacteria, for example the changes in the anatoxin concentration and quota all occurred within-rock. A solution to this would be to use metatranscriptomics to identify genes that are co-regulated with anatoxin production, which may provide information as to what conditions the cyanobacteria are responding to when they upregulate toxin production. Variability in the anatoxin quotas observed in Microcoleus-dominated mats may be the result of different toxic genotypes with varying capacities of toxin production [15,39]. Our results, similarly, indicate that anatoxin concentrations in benthic cyanobacterial mats from the Eel and Russian rivers are largely driven by the abundance of toxic cells. The low variability in anatoxin quotas at both site and within-rock scales observed in this study may, therefore, result from lower diversity in anatoxin-producing strains in the Eel and Russian rivers. Isolation and culture of Microcoleus strains would enable the characterisation of toxic genotypes in order to confirm this.

Cyanotoxins and producers in green alga dominant mats
Mats of the green alga C. glomerata at 3_SFE yielded concentrations of anatoxins and NOD-R comparable with those found in the cyanobacteria-dominated samples. Cladophora glomerata is globally distributed and supports complex epiphytic algal and microbial assemblages [11]. Previous analysis of similar mats from these rivers has shown the presence of a variety of cyanobacterial taxa including Oscillatoriacae [10]. The presence of cyanotoxins at such high concentrations in the C. glomerata-dominated mats in this system raises important questions about how extensive this phenomenon is in other non-cyanobacterial dominated mats. Further studies of cyanotoxins in periphyton mats should be a priority for future research to assist with increasing knowledge on the risks associated with cyanotoxins in river systems.

Conclusions
The data from this study provides further evidence that Microcoleus is the anatoxin producer in the Eel and Russian rivers. Genetic evidence also suggests that the nodularin producer in these rivers is most similar to Nodularia, which highlights the need for targeted culturing of this taxon from these rivers in future studies for definitive confirmation. Our data also show that when these organisms are not macroscopically visible, they can still produce toxin concentrations equivalent to those mats where they are macroscopically dominant. The spatial variability in toxin concentrations among samples highlights the need for careful sampling design when assessing toxin levels at a site for risk assessment purposes. Normalisation of anatoxin concentration data to quotas reduced variability among samples indicating that a key driver in toxin variability is the abundance of toxic genotypes present in a sample. Cyanotoxins were present at concerning concentrations in mats dominated by the green alga C. glomerata. Further research to investigate the extent of this phenomenon should be prioritised.