Screening the presence of cyanotoxin in selected reservoirs in Sri Lanka and employing molecular markers as an advanced alert system

The widespread occurrence of cyanotoxin-producing cyanobacteria in lakes and reservoirs is a worldwide concern (Manage et al. 2009; Lawton et al. 2011; Graham et al. 2020). Increased eutrophication and climate change have triggered more frequent toxic cyanobacterial blooms around the globe (Ranjbar et al. 2022). Dolichospermum circinale (formerly Anabaena circinalis), Aphanizomenon flos-aquae, Raphidiopsis raciborskii (formerly Cylindrospermopsis raciborskii), and Microcystis aeruginosa are some commonly occurring cyanotoxin producers (Ramos et al. 2021; Vidal et al. 2021). These toxic molecules can be categorised as hepatotoxic (e.g., microcystin (MC) and nodularin (NOD)), neurotoxic (e.g., anatoxin-a (ANA-a) and saxitoxin (SAX)), cytotoxic (e.g., cylindrospermopsin (CYN)), genotoxic (e.g., MC and CYN), and carcinogenic (e.g., MC) due to their diverse physiological interactions with humans and animals (Van Apeldoorn et al. 2007; Piyathilaka et al. 2015). For example, the death of 60 patients in a dialysis centre in Brazil due to MCs in water and the intoxication of 148 people following a bloom of R. raciborskii in a Queensland reservoir (Australia) have been linked to cyanobacterial toxins (Jochimesen et al. 1998; Moreira et al. 2021).

Harmful algal blooms (cyano blooms), primarily caused by toxin-producing cyanobacteria, present a significant threat to drinking water sources across Canada, China, and the United Kingdom. According to a report in 2019, nearly 150 cases of waterbodies affected by cyano blooms were documented (Moreira et al. 2021). Studies by Wijewickrama & Manage (2019) indicated that the levels of MC-LR in edible tissue of Oryza sativa and Ipomoea aquatica varied from 18.19 to 132.86 μg/kg, surpassing the World Health Organization's (WHO) tolerable daily intake of Microcystin LR (MC-LR) in some instances. This highlights a potential health hazard to humans (Wijewickrama & Manage 2019). Therefore, to prevent such unfortunate health hazards in future, it is essential to establish appropriate methods for the early detection of toxic blooms and screening for the occurrence of genes related to the production of cyanotoxins in aquatic systems.

With the knowledge of the gene sequences, several studies have exploited molecular approaches such as polymerase chain reaction (PCR) to detect and quantify potentially toxic cyanobacteria (Hisbergues et al. 2003). Notwithstanding the sensitive nature of the PCR approaches for detecting the occurrence of toxin-producing genes, it offers little insight into active mcy gene transcription, which is required to predict the concentrations of the toxins in the water sample more accurately (Pacheco et al. 2016). Therefore, PCR methods must be accompanied by reliable and rapid cyanotoxin quantification methods. Liquid chromatography-tandem mass spectrometry (LC-MS/MS)-based methods are currently considered the gold standard for analysing and quantifying multiple toxin variants in environment samples due to their high specificity and sensitivity (WHO 2020). However, despite their high-throughput analysis, such methods remain unaffordable for most parts of the world due to high capital, operational, and maintenance costs. Thus, DNA-based detection methods combined with reliable but more affordable methods such as enzyme-linked immunosorbent assay (ELISA) to evaluate whether toxins are present in water samples could be a solution for locations where funds are inadequate for advanced analytical solutions (Rantala-Ylinen et al. 2011). The integrated PCR and ELISA approach would help determine the occurrence of toxic cyanobacteria in water and provide an affordable screening method for enumerating cyanotoxin, an early warning system to take necessary safeguarding precautions.

Sri Lanka is an island in the Indian Ocean with a total land area of 65,610 km² and a population of 22.16 million. The country's economy mainly relies on agriculture. Due to its long-standing agro-economy-based culture, most people in Sri Lanka depend on natural and manmade water bodies for irrigational purposes and drinking water supply. According to recent studies, most of Sri Lanka's irrigation tanks and drinking water reservoirs reportedly contain elevated levels of M. aeruginosa and R. raciborskii, which could produce MC and CYN, respectively (Idroos & Manage 2015). Recent droughts, increased farming, and eutrophication have triggered frequent cyanobacterial blooms in Sri Lanka (Wijewickrama & Manage 2019). In addition, due to the lack of access to treated tap water, 80% of irrigation tanks are used as drinking water sources by the rural populations of Sri Lanka (Indika et al. 2022). Therefore, cyanobacteria and cyanotoxins in intensely used water bodies have recently raised island-wide concerns (Ganegoda et al. 2019; Abeysiri & Manage 2022a). In addition, in 2013, a WHO report highlighted that cyanobacterial toxins have been identified as one of the possible causes of widespread (mainly in rural areas) chronic kidney disease in Sri Lanka (Piyathilaka et al. 2015; Abeysiri & Manage 2022b). Further, MC-LR is a potent hepatotoxin, and increasing evidence suggests that it might also induce kidney injury.

The chronic exposure of CYN in the Wistar rat model has shown the increment of serum and urine creatinine compared to the control group suggesting the chronic effect of MCs on the kidney (Abeysiri & Manage 2022b). However, the limited access to high-throughput techniques such as LC-MS/MS has hindered the robust investigation and routine monitoring of cyanotoxins in Sri Lankan freshwater systems for decades. Therefore, this study aimed to survey frequently used (drinking, recreation, and irrigation) water bodies across the island to determine the occurrence of toxin-producing cyanobacteria and the concentrations of MC-LR, CYN, NOD, ANA-a, and SAX using the integrated PCR and ELISA techniques.

Commercial test kits (Beacon, USA) were used to quantify total (intracellular and extracellular) MC-LR, CYN, NOD, ANA-a, and SAX. Water samples (n = 43, 25 mL) were subjected to three freeze–thaw cycles to lyse any available cyanobacterial cells and liberate the intracellular toxin. The extracts were centrifuged (1,500 × g, 10 min, room temperature), and the supernatant containing toxins was decanted for the assay. Following the manufacturer's instructions, samples were analysed in the ELISA assay. The absorbance of total cyanotoxin was measured using a plate reader set to 450 nm (ELISA MC-108, USA) following the manufacturer's guide.

DNA extractions were carried out in a clean laboratory where no prior PCR or DNA extractions were performed to prevent contamination. Aliquots (500 mL) of collected water samples were filtered individually through a 47-mm-diameter polycarbonate filter (0.22 μm, Millipore Corporation, Bedford, MA, USA); filter papers were preserved in 70% ethanol and stored at −20 °C until analysis. DNA required for the PCR analysis was extracted from the filter paper using the MOBIO Power Water DNA Extraction Kit (Qiagen, USA) following manufacturers' instructions.

PCR mixtures containing 0.5 μL of target primer, 5 μL of GoTaq reaction buffer, 0.5 μL of dNTPs, 2.0 μL of 25 mM MgCl₂, and 0.1 μL of GoTaq DNA polymerase were adjusted to a total volume of 25 μL to identify toxin-producing genes. Purified DNA (5 μL) was utilised as a PCR template.

The primer combination mcya-cd1F/mcya-cd1R and the following PCR conditions were used to amplify Microcystis mcyA: initial denaturation at 95 °C for 2 min, followed by 35 cycles of 95 °C for 90 s, 56 °C for 30 s, and 72 °C for 50 s. Microcystis-mcy BCDEG-specific PCR amplifications were carried out using the following conditions: initial denaturation at 94 °C for 5 min, followed by 35 cycles of 95 °C for 60 s, 52 °C for 30 s, and 72 °C for 60 s. The strain LEGE 00063 was utilised as a positive control in all MC-related PCR experiments (Moreira et al. 2020).

PCR amplification was performed to screen for cyrA, cyrB, cyrC, and cyrJ genes. Initial denaturation at 94 °C for 3 min was followed by 30 cycles at 94 °C for 10 s, 50 °C for 20 s, and 72 °C for 60 s in the PCR for cyrA. Initial denaturation at 94 °C for 10 min was followed by 30 cycles at 94 °C for 30 s, 45 °C for 30 s, and 72 °C for 60 s for cyrB and cyrC, while initial denaturation at 94 °C for 3 min was followed by 30 cycles at 94 °C for 10 s, 57 °C for 20 s, and 72 °C for 60 s for cyrJ. A CYN-producing strain LEGE 97047 was utilised as a positive control in all CYN PCR experiments (Moreira et al. 2020).

The genetic markers stxA, stxG, and stxI were used to identify the SAXs. For sxtA and sxtG, the PCR conditions were as follows: 30 s of denaturation at 98 °C, followed by 35 cycles of 98 °C for 5 s, 62 °C for 5 s, and 72 °C for 10 s. Primers for sxtI were amplified using the following PCR conditions: 94 °C for 3 min, then 35 cycles of 94 °C for 10 s, 52 °C for 20 s, and 72 °C for 60 s. The LMECYA 040 was used as the positive control for all the SAXs PCR experiments (Moreira et al. 2020).

Initial denaturation at 94 °C for 2 min, followed by 35 cycles of 94 °C for 30 s, 58 °C for 30 s, and 72 °C for 30 s were used to identify anaC and anaC-anab. LEGE X-002 was used as a positive control in both PCR experiments (Moreira et al. 2020).

The following conditions were used for the amplification of nodA and nodC: an initial denaturation at 94 °C for 1 min, followed by 35 cycles at 94 °C for 30 s, 68 °C for 30 s, and 72 °C for 30 s. LEGE N-005 was used as the positive control (Moreira et al. 2020).

All PCR amplifications were completed with a final extension at 72 °C for 5–7 min. The BIOLAB PCR system thermal cycler was used for all PCR amplifications (BYQ6078E-757, China).

After each PCR, the amplified DNA products were visualised using gel electrophoresis. A 1% agarose-TAE gel with 4% ethidium bromide was used to resolve DNA in TAE buffer at 40 V continuous voltages for 1 h. A 100 bp DNA ladder was used to measure the amplicon fragment sizes of each PCR (Liyanage & Manage 2016).

The average toxin concentrations and toxin-producing genes were compared using the analysis of variance (ANOVA) test at various sample sites. Pearson's analysis tested the association between toxin concentration and toxin-producing gene values (all variables satisfied the normality assumption). Differences were judged significant at p = 0.05. The statistical analysis was carried out using Primer V7 software.

MC seemed to be the most dominant toxin, and 16 out of 43 reservoirs (37%) sampled during the study contained varying concentrations of MC. For example, the Beira Lake in Colombo, a recreational waterbody in the most populated city in Sri Lanka, contained 3.31 μg/L of MC (Figure 2). However, compared to Beira Lake, the rest of the water resources, which contained MC (Basawakkulama – 0.10 μg/L, Padaviya – 0.13 μg/L, Ambagaswewa – 0.14 μg/L, Parakrama Samudraya – 0.11 μg/L, and Rathkinda – 0.10 μg/L), showed comparatively low concentrations of the toxin. In contrast, water collected from the reservoirs in Trincomalee, Kurunegala, and Hambanthota districts showed no MC (Figure 2, Table 1).

CYN was less frequent than MC in the surveyed reservoirs. Water samples collected from Balalu, Nachchaduwa, Padaviya, and Unnichchi were contaminated with CYN, and the toxin concentration ranged from 0.05 to 0.07 μg/L. Interestingly, 3 out of 10 samples collected from the district of Anuradhapura contained CYN, with one location (Padaviya) reaching 0.07 μg/L (Figure 2, Table 1).

The highest concentration of SAX was detected from the Padaviya reservoir (0.09 μg/L), followed by Beira Lake (0.08 μg/L), Parakrama Samudraya (0.08 μg/L), and Ulhitiya (0.08 μg/L) reservoir, which were below the WHO minimum permissible limits (WHO 2020).

Among the 43 sampling locations, NOD was only found in four water bodies (Balalu – 0.05 μg/L, Padaviya – 0.08 μg/L, Unnichchi – 0.06 μg/L, and Ulhitiya – 0.08 μg/L), while ANA-a was found in Nuwara (0.17 μg/L), Padaviya (0.19 μg/L), Parakrama Samudraya (0.15 μg/L), and Unnichchi (0.16 μg/L), respectively (Figure 2, Table 1).

The extracted DNA was amplified in a PCR using gene-specific primers to investigate the occurrence of the selected cyanobacterial genes in the water samples. Table 1 depicts the outcomes of the PCR amplification of each genetic marker screened for the study. Sixteen of the 43 reservoirs analysed tested positive for the mcyA gene, and it was the most prevalent gene among the mcy genes detected in the water samples. Especially in the Anuradhapura district, 70% of the reservoirs, including Kala, Nachchaduwa, Tissa, Basawakkulama, Nuwara, Padaviya, and Yan Oya, were reported to contain mcyA, indicating a potential threat to the area consumers of the water bodies.

It was interesting to note that the mcyB and mcyC genes were less frequent than the mcyA gene. The mcyB gene was only detected in Beira Lake in Colombo; Kala, and Balalu in Anuradhapura; Rathkinda in Badulla; Maduru Oya and Kondawatuwana in Ampara; Castlereagh, Gregory Lake, and Kothmale in Nuwara Eliya districts. Meanwhile, Ambagaswewa, Parakrama Samudraya, and Minneriya from Polonnaruwa district and Ulhitiya reservoir in Polonnaruwa district showed the occurrence of mcyC (Table 1). Several reservoirs also harboured mcyD, mcyE, and mcyG genes, but none of the reservoirs had all six genes (mcyA, mcyB, mcyC, mcyD, mcyE, and mcyG) detected together in this study (Table 1). Overall, 24 out of 43 water samples tested positive for at least one mcy gene.

The CYN synthesising genes cyrA, cyrB, cyrC, and cyrJ were reported in the reservoirs in the Anuradhapura, Polonnaruwa, Batticaloa, and Nuwara Eliya districts. In addition, at least two CYN genes were found in the Balalu, Nachchaduwa, Basawakkulama, Nuwara, and Padaviya reservoirs in the Anuradhapura District, although no CYN was reported in Basawakkulama and Nuwara (Table 1).

In various reservoirs in the Anuradhapura, Polonnaruwa, Batticaloa, Badulla, Ampara, Kandy, and Nuwara Eliya districts, SAX-producing genes (stxA, stxG, and stxI) were detected. Except for Ulhitiya (0.08 μg/L), Parakrama Samudraya (0.08 μg/L), and Padaviya (0.09 μg/L), 10 out of 43 samples were positive for at least one of the SAX-producing genes screened (Table 1).

ANA and NOD-producing genes were widespread within the reservoirs of Anuradhapura, Polonnaruwa, Kurunegala, Badulla, Batticaloa, Ampara, Nuwara Eliya, and Kandy. For example, the ANA-producing genes anaA-anaB and anaC were present in the Padaviya, Kurunegala reservoir, Unnichchi, and Rantabe reservoirs. Senanayake Samudraya in the Ampara district tested positively only for anaA gene. In contrast, the anaC gene was detected in Parakrama Samudraya in Polonnaruwa, Ulhitiya Tank in Badulla, Jayanthi weva in Ampara, and Kotmale reservoir in Nuwara Eliya (Table 1).

Balalu and Padaviya in Anuradhapura, Ambagasweva and Girithale in Polonnaruwa, Ulhitiya in Badulla, Unnichchi in Batticaloa, Senanayake Samudraya in Ampara, and Gregory Lake in Nuwara Eliya were found to contain the NOD genes, nodC and nodA. In addition, the Kurunegala and Kothmale reservoirs tested positive for nodC, although no detectable toxin was present in the water samples.

During this study, 43 freshwater samples were analysed by ELISA and PCR, focusing on five major cyanotoxins (MC, CYN, NOD, SAX, and ANA-a) and the toxin-producing genes. Due to the long-existing agricultural economy and tradition, most of the population depends on natural water resources such as lakes and rivers. Changes to the seasonal variations in the tropical climate and the increased farming-related eutrophication seemed to be the main source of the recurrent occurrence of cyanobacterial blooms in Sri Lankan freshwaters. Freshwater habitats contaminated with cyanotoxins have various effects on drinking water quality, irrigation performance, recreation activities, and tourism (Gärtner et al. 2021). Nevertheless, the toxicity associated with cyanobacterial blooms has not been studied adequately in Sri Lanka.

The screening of the water samples was focused on the major water resources in the country, which have a substantial impact on the economy (e.g., irrigational) and recreation (e.g., fishing), as well as supplying drinking water to surrounding villages/towns (Dharmadasa et al. 2017; Abeysiri & Manage 2022a). Furthermore, the water samples were collected from November to July, when most freshwater cyanobacteria tend to proliferate rapidly, reaching bloom conditions in June–August (Piyathilake & Manage 2017). Thus, it is expected that the molecular techniques applied would act as an early warning system about possible upcoming toxic bloom conditions.

Molecular techniques such as PCR and ELISA have been widely used worldwide to investigate the occurrence of toxicogenic cyanobacteria and quantify the level of different toxins (Pacheco et al. 2016; Moreira et al. 2020). Most genetic analyses have focused on exploring the MC-producing genes by amplifying the mcy gene cluster, which spans 55 kb, along with two operons, mcyA-C and mcyD-J. The mcy gene cluster is distributed among the toxigenic strains of M. aeruginosa, Dolichospermum, and Planktothrix (Christiansen et al. 2003; Dittmann et al. 2013), which predominantly occur in Sri Lankan freshwaters (Hettiarachchi et al. 2013). Interestingly, from all the toxins analysed during the study, MC was the most dominant toxin in the Sri Lankan water resources screened. MC was reported in 16 out of 43 locations, some of which are intensely used for drinking, irrigation, and recreation. At the time of the sampling, the level of MC detected in all drinking water reservoirs (Figure 3) was below the WHO provisional guidelines (12 μg/L, WHO 2020). However, long-term exposure (e.g., drinking, consumption of contaminated food) to the toxin could lead to vascular congestion, necrosis, nuclear pyknosis, intraluminal proteins, swelling of the epithelial cells, and interstitial inflammation in the cortex of the tubules (Chernoff et al. 2011). The ELISA results indicated that the concentration of the MC detected in drinking water reservoirs sits below the WHO provisional guidelines for lifetime exposure (1 μg/L). However, a fact that must be considered is the bioaccumulation of the toxin, as most of these water bodies are used for freshwater fishing and rice and vegetable farming. The bioaccumulation of MC in vegetables, edible leaves, and fish and the associated risks are well documented elsewhere (Bi et al. 2019; Xiang et al. 2020). The MC concentration detected from Beira Lake (water volume 2,903,000 m³ and surface area 0.654 km²) was the highest toxin concentration (3.31 μg/L) we enumerated within the samples. Water from this resource is not currently used for drinking purposes. However, the water body is intensively used for recreational activities since it is located in the most highly populated city, Colombo, in Sri Lanka. The MC values detected by ELISA are notably below the WHO provisional guidelines value for recreational water, which is 24 μg/L (WHO 2020). However, such levels of MC in water could still pose a risk for those exposed to this water during recreational activities such as rowing and swimming. Exposure during recreational activity can be associated with oral (accidental ingestion of contaminated water), dermal (direct contact), and inhalation (Backer et al. 2010), and MC intoxication via recreational activities is reported elsewhere (Vidal et al. 2017).

Our results revealed that the samples collected from the Anuradhapura district have a higher prevalence of MC-producing genes than other districts. More than 80% of the locations were positive for at least one toxin-producing gene, and 70% of sampling locations indicated the presence of the mcyA gene, a genetic marker for MC. This could be attributed to the area's intense rice and vegetable farming practices. Moreover, the recent flooding in the area may have led to widespread cyanobacteria contamination among the reservoirs and irrigational channels.

Another cyanotoxin of concern in Sri Lanka is NOD, a potent hepatotoxin and a carcinogen primarily produced by the strains of Nodularia spumigena. The hepatotoxicity and carcinogenicity of NOD are associated with the inhibition of protein phosphatase 1 and 2A (Wharton et al. 2019). Although the NOD-producing toxicogenic cyanobacteria are distributed worldwide, the occurrence of Nodularia sp. emerged in Sri Lankan fresh waters, raising the warning call, especially for the reservoirs dedicated to drinking and recreational activities (Jayatissa et al. 2006). Although, compared to MC, NOD was only found in four reservoirs, and the concentrations ranged from 0.05 to 0.08 μg/L (Figure 2, Table 1), the toxicity associated with NOD must be considered, given that the water from these sampling locations is used for drinking, irrigation, and fishing, and their ingestion causes a variety of adverse symptoms in humans and animals (Honkanen et al. 1991; Wharton et al. 2019).

CYN is a highly water-soluble polycyclic uracil derivative mainly produced by cyanobacteria such as R. raciborskii, Oscillatoria sp., Dolichospermum sp., A. flos-aquae, and Lyngbya wollei (Dittmann et al. 2013; Moreira et al. 2013). CYN has been reported to cause liver and kidney damage due to its ability to interfere with protein synthesis (Falconer & Humpage 2001). Given the tropical climate in Sri Lanka, CYN-producing strains such as R. raciborskii tend to occur all year round (Peduruarachchi et al. 2022). Interestingly, previous studies conducted in Sri Lanka's fresh waters reported that compared to the wide distribution of MC variants, CYN was restricted to specific geographical areas in the country (Piyathilaka & Manage 2017). However, using this two-step approach, PCR and ELISA, as a tool, we report that CYN occurrence and contamination are more significant than understood. We found that major reservoirs (e.g., Balalu, Padaviya, Nachchaduwa, Basawakkulama, Ambagaswewa, Parakrama Samudraya, Unnichchi, and Maussakele) used as a source for drinking water and farming contained the genes responsible for producing CYN. We appreciate that the prevalence of CYN-producing genes in the water sample collected does not prove that the water is unsafe for consumption. However, it must be noted that four out of nine water samples that tested positive for CYN genes also contained the toxin, as confirmed by the ELISA immunoassay. Recent reports suggest that the consumption of CYN-contaminated water might be one of the factors that caused the widespread kidney disease in the rice farming areas of Sri Lanka (Wijewickrama & Manage 2019). The gene cluster responsible for producing CYN is composed of 15 open reading frames (ORFs), which span 43 kb. In this study, we targeted the amplification of cyrA, cyrB, cyrC, and cyrJ genes responsible for amidinotransferase, NRPS/PKS peptide biosynthesis-related activities, and the tailoring functions of CYN during its biosynthesis (Moffitt & Neilan 2004).

Although there have been preliminary reports of cyanobacteria-producing neurotoxic alkaloids, SAX, and ANA-a in Sri Lankan freshwaters, to our knowledge, none of the reports had confirmed their occurrence of toxins via either analytical or molecular methods. SAX is primarily produced by the strains of R. raciborskii (Kellmann et al. 2008), which are also accountable for producing CYN. R. raciborskii is one of the main cyanobacterial strains found in several drinking and irrigational reservoirs in Sri Lanka (Peduruarachchi et al. 2022), which could have been responsible for SAX in the water samples collected. In this study, we used primers to amplify three major genes (stxA, stxG, and stxI) responsible for regulating and producing toxins. sxtA is the largest gene in the SAX-producing gene cluster (35 kb, 31 ORFs). However, some studies have reported that sxtG and sxtI are better candidates for identifying the presence of SAX than sxtA (Moreira et al. 2020). Interestingly, no PCR products related to sxtA were detected for the water samples collected in Nuwara, Upper Kothmale, and Rantabe reservoirs. In addition, they contain sxtG and/or sxtI. This could be attributed to the potential sequence dissimilarities between targets and primers, since the primers for this study were designed based on the cyanobacteria reported from European environments (Moreira et al. 2020). In addition, some evolutionary changes, such as insertions and deletions that occurred in the sxt gene cluster in the Sri Lankan cyanobacteria over time might have caused the negative amplification.

Moreover, the strains of Dolichospermum, Microcystis, Aphanizomenon, Oscillatoria, and Planktothrix, which produce ANA-a, are also widely spread over the freshwater bodies of Sri Lanka (Jayatissa et al. 2006). However, for the first time, we present the identification of genes responsible for ANA-a and their wide distribution (e.g., Nuwara, Padaviya, Parakrama Samudraya, Unnichchi) in Sri Lankan freshwater bodies, supported by the enumeration based on ELISA. The PCR test also showed that ANA-a-producing genes could be found even without the presence of ANA-a in the water sample at the sampling time. Interestingly, the MC and the mcy genes were also detected in the same reservoirs. Therefore, these reservoirs may contain cyanobacterial genera capable of producing toxins, such as Dolichospermum, Microcystis, Oscillatoria, and Planktothrix. At the time of sample collection, we did not observe (visually) any bloom conditions in water reservoirs, apart from Beira Lake. However, further genetic (e.g., 16s RNA) and morphological analysis, such as microscopy, is required to confirm the potential strains of cyanobacteria in the water samples.

The degree of correlation between positive amplicons and positive cyanotoxin enumerations was also assessed in the current study. There were no positive correlations between the two procedures in terms of toxins. However, it is interesting to note in most cases, whenever the water sample indicates the toxin's occurrence by the immunoassay, they also test positive for the genes responsible for the production of the toxin. In contrast, Moragahakanda, Sagama tank, and Randenigala water samples showed no PCR amplification for mcy genes; nevertheless, ELISA results showed they all contained MC ranging from 0.1 to 0.13 μg/L. However, there were no cyanobacterial blooms or cells (visual observation) during sample collection in any of the above reservoirs. Therefore, it may be the case that these reservoirs previously had MC-producing cyanobacteria and released the toxin that remained in the water when we collected the sample.

Nevertheless, due to their strong, stable cyclic structures, MCs could persist in water for several months after the blooms collapse (Chorus et al. 2000). Another factor that could contribute to the occurrence of toxins with no trace of toxin-producing genes in surface water samples collected is benthic cyanobacteria. Although there are no previous reports of benthic cyanobacteria in Sri Lankan water bodies, the toxins produced by benthic cyanobacteria (e.g., genera – Nostoc, Fischerella, Nodularia, Oscillatoria, Synechococcus, Lyngbya) may appear on surface water samples due to strong winds triggering mixing of the entire water body.

In contrast, the water samples collected from numerous locations contained no toxins, as seen by ELISA, although the PCR analysis indicated the prevalence of the toxin-producing genes. For instance, the water samples collected from Kala, Balalu, Nachchaduwa, Tissa, Nuwara, and Yan Oya from the district of Anuradhapura showed PCR products when they were assayed along the primers designed for the mcy gene cluster. Similarly, the water samples from Basawakkulama and Nuwara showed neither CYN nor STX, whereas the PCR indicated the existence of the genes responsible for producing these toxins. Thus, the toxins may exist in the water samples, perhaps below the limit of detection, or the cyanobacterial cells have not yet reached the period for producing these secondary metabolites.

We present the first report on the molecular screening of genetic markers of cyanobacteria in Sri Lankan reservoirs. The integrated PCR and ELISA approach is affordable and easy to implement without complex and expensive laboratory equipment. More importantly, it provides excellent insight into the early detection of toxigenic cyanobacteria and the level of cyanotoxin in the Sri Lankan water reservoirs. The presence of toxin-producing genes in waters used for drinking and recreation suggests the presence of a potential health threat even when toxins themselves are not detected. Drinking water treatment plants can use PCR testing as an early warning that source waters may be contaminated with cyanobacterial toxins. They will be forewarned of the potential need to adjust treatment to remove toxins. For a country like Sri Lanka, where cost-intensive and sophisticated analytical systems (e.g., LC-MS/MS) are scarce, having an established, affordable method is highly important.

Furthermore, waters from the selected water bodies are used intensively for drinking, farming, and serving as cattle and wildlife's primary drinking water source. Our data suggest that applying PCR and ELISA could be a potential approach to evaluate a cyanotoxin-related scenario of a given ecosystem, allowing one to redirect the chemical assays for cyanotoxin enumeration.

The level of occurrence of cyanotoxins and the genes that produce them we found here is concerning. The occurrence of cyanotoxin-producing genes in water samples may indicate the likelihood of toxic blooms shortly, threatening the drinking water industry and human health. Thus, authorities must consider policies and suitable safe drinking water management approaches. On a global reach, we believe this study could be used as a principal model for surveying water bodies located in geographically similar (e.g., near the equator) countries to estimate the potential contamination of cyanotoxins and use this as an early warning system (screening for the occurrence of toxin-producing genes) for a safe drinking water approach.

The authors thank the Center for the Laboratory Facilities for the Water Quality and Algae Research, Department of Zoology, Faculty of Applied Sciences, University of Sri Jayewardenepura.

This study was supported by Centre for Water Quality and Algae Research, Department of Zoology, Faculty of Applied Sciences, University of Sri Jayewardenepura, Sri Lanka.

PM contributed to conceptualisation, sample collection, methodology, and reviewing and editing the writing. GY contributed to sample analysis, data curation, formal analysis, methodology, writing and editing the manuscript. DS contributed to sample collection, sample analysis, molecular work, formal analysis, and writing the manuscript.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.