First report on the occurrence of a single species cyanobacterial bloom in a lake in Cyprus: Monitoring and treatment with hydrogen peroxide releasing granules

Background: Cyanobacteria are phytoplankton microorganisms, also known as blue-green algae, and an essential component of the food web in all aquatic ecosystems. Excess loads of nutrients into waterbodies can cause their rapid and excessive growth which leads to the formation of cyanobacterial harmful algal blooms (cyano-HABs). Toxic species of cyanobacteria genera excrete into the water a broad range of bioactive metabolites, some of which are known as cyanotoxins. These metabolites can negatively affect the ecosystem, and human and animal health in various ways, thus their presence needs to be closely monitored. This study aimed to monitor a lake at the Athalassa National Forest Park in Cyprus, in order to correlate its trophic condition with its water quality characteristics and identify the key environmental variables driving cyanobacteria blooming and their toxicity. In addition, surface water during the blooming period was collected and used in bench-scale experiments in order to test novel hydrogen peroxide releasing granules as mitigation processes for cyano-HABs. Results: The monitoring lasted throughout 2019 with ten sampling events taking place during this period. Samples were mainly analyzed for phytoplankton community, and various physicochemical parameters: pH, conductivity, salinity, total and dissolved nutrients. Obtained data indicated that cyanobacteria blooming lasted for four months (June – September), while microscopic observation of preserved samples showed that 99% of the phytoplankton biovolume was attributed to a single picocyanobacterial species, the Merismopedia sp. Select samples were analysed for the presence of toxins genes with positive results mainly for mcyB and mcyE genes. Further analysis with HPLC MS/MS, revealed that cyanotoxins’ concentration was lower than the method detection limit - MDL (<2-6 ng/L). Conclusion: The present study highlights the importance of monitoring several water characteristics to conclude on the main drivers of a bloom and its toxicity. The �ndings demonstrated that it is not enough to test cyanotoxin genes as indicator of their presence since, in case of mono-domination, cyanobacteria may not be active


Background
Cyanobacteria are phytoplankton microorganisms whose ability to oxygenate the atmosphere 3.5 billion years ago contributed to life formation [1].It is a group of bacteria ranging from 1 to 100 μm in diameter, while some of them are even smaller having a diameter less than 2 μm.Those are described as picoplankton and includes species such as Merismopedia sp, Aphanocapsa sp., and Synechococcus [2].Cyanobacteria gain energy from photosynthesis by capturing light through pigments which are chlorophyll-a and phycocyanin with excitation wavelength at λ=450 and λ=620 nm; respectively.The difference between the two excitation wavelengths of the pigments found in green algae and blue-green cyanobacteria, allows us to distinguish their presence in freshwater and marine environments [3].
Anthropogenic activities such as agricultural, urban, and industrial activities have intensely increased the load of nutrients in surface waters around the globe, making cyanobacterial blooming more persistent and prevalent [4].Both nutrients and cyanobacteria are essential to maintain the balance in an aquatic ecosystem.Nutrients support sh and shell sh production [5] while cyanobacteria are an essential component of the food web in all aquatic environments.Disruption of this balance by excess load of nutrients causes the rapid and excessive growth of cyanobacteria which leads to the formation of cyanobacterial harmful algal blooms (cyano-HABs).Blooming enhances water quality depletion by reduced light and oxygen penetration with serious consequences on biodiversity [6], while adding undesirable color, taste, and odor to the waterbodies.Toxic genera of cyanobacteria can excrete into the water a broad variety of bioactive metabolites, also known as cyanotoxins [7].These metabolites can negatively impact the ecosystem and human health, making it an important environmental issue of concern [8].Although their acute toxicity on humans is not extensively studied, mass mortalities of shes, birds, mammals and many other animal taxa have been reported [9].Recent studies have correlated liver-related deaths in U.S with several cyanotoxins [10].Cyano-HABs presence in freshwaters used as drinking water reservoirs is not only a health issue, but it also raises the overall treatment and monitoring costs which are in the range of millions of euros annually [11].Currently, there is no method for in-situ detection or a predictive model for the occurrence of these toxins since not all cyanobacterial species are active toxin producers under the same conditions.Therefore, it is imperative to nd both predictive models and monitoring tools as well as e cient treatment methods to mitigate the problem to safeguard water quality and reduce water treatment costs at source and in the waterworks.
The concentration (total and dissolved fraction) of the main nutrients -nitrogen (N) and phosphorus (P), is a strong indicator of the eutrophic state of waterbodies thus, several models and relationships have been developed over the years to form correlations [12].The most applied stoichiometric reference is the Red eld ratio which describes the nutrient limitation of planktonic production in coastal waters based on the TN:TP molar ratio [13,14].Despite the fact that Red eld proposed this ratio for its use on oceanic studies, it was well adopted as a universal nutrient limitation threshold with multiple citations in different types of aquatic systems.Red eld referred to an average N:P molar ratio that when exceeds 16, phosphorus becomes the limiting element for phytoplankton growth while when the ratio is below 7, nitrogen is the limiting element.Over the years, different ratios and approaches have been developed for better understanding of the limiting element based on the speci c conditions of each waterbody (e.g., nutrient sources, phytoplankton species present).Based on the sited literature, lakes with different characteristics showed different correlations of nutrients with their trophic status, and the e ciency of the Red eld ratio varied in each case [15].Recent studies indicated that higher TN:TP molar ratios (>22, Plimitation) are more suitable for surface waters while DIN:TP and NO 3 -:TP mass ratios have been used more often during the past years for determining the limiting elements in lakes [16].Studies on dissolved nutrients mass ratios (DIN:TP, NO 3 -:TP, NH 4 + :TP) suggested that in freshwater systems the DIN:TP mass ratio is a better indicator than the TN:TP molar ratio.Also, the NO 3 -:TP mass ratio performed even better than DIN:TP as DIN includes ammonium which in some studies it showed a weak correlation with N-limitation, resulting in a weaker model [17].
The ratios and thresholds used in the present study for evaluating the trophic status of a lake in Cyprus and the limiting elements during the monitoring period are presented in Table 1.Monitoring the level of nutrients in waterbodies and estimating their corresponding ratios are critical for improving the applied management strategies based on its speci c needs in order to better control the limiting factors and to prevent future blooming events.
The main input of nutrients into the waterbodies comes from agriculture due to the misuse of fertilizers.The European Commission of Environment of the European Union has composed the Water Framework Directive (2000/60/EC) which aims to protect surface waters from chemical pollution [20].Recently, it was proposed to include a group of cyanotoxins (microcystins) in a revised Drinking Water Directive on the Quality of Water Intended for Human Consumption [21].If this comes through, it will enforce public authorities to include cyanobacteria and cyanotoxins into their monitoring and mitigation strategies in order to comply with the new directives and legislations of EU.
Additionally, the Organization for Economic Co-operation and Development (OECD) has actively participated in the development of frameworks and guidelines by reforming the surface water quality regulations in EECCA countries [22].Based on those, OECD proposed a surface water classi cation (I -V class; excellent -bad) system based on different water quality characteristics.Its use has been widely adopted for monitoring water quality and apply restorative measures when needed.In the present study, water class was evaluated based on the limits set by OECD for acidi cation status and total and dissolved nutrients concentration as shown in Table 2. Surface water monitoring is essential for maintaining a healthy status and protect the biodiversity of aquatic biotopes around EU. Unfortunate events of high nutrient loads that lead to the formation of cyano-HABs are mostly unpredictable and thus highly e cient methods are required to be applied in-situ for restoration.There are several physical, chemical, and biological methods that have been developed and applied over the years with the chemical ones to be more cost effective, rapid, and e cient [23].The need to make chemical treatment "greener" has led to the application of hydrogen peroxide (H 2 O 2 ) to different cyanobacterial species as an alternative to copper algicides, resulting in selective reduction of cyanobacterial species among other taxa of phytoplankton [24,25,26].The hydroxyl radicals ( .OH) formed by the oxidant inhibit the electron transport of photosystem II, causing reduction of its photosynthetic activity leading to cellular death [27,28,29].There are studies suggesting that its e ciency varies and depends on the nutrient load of the matrix, the species composition and abundance, the boom density, and light intensity.Usually high doses of H 2 O 2 (>5 mg/L) are required for a complete destruction of cyanobacterial cells [30].An alternative to liquid hydrogen peroxide is its granular form found as metallic peroxide granules which decompose slowly and release H 2 O 2 [31].Those are calcium and magnesium peroxide (CaO 2 and MgO 2 respectively) granules.
The goal of the present study was to examine the e ciency of treatment in a dense single-species bloom that occurred in St. George Lake of the Athalassa National Forest Park, in Cyprus with metallic peroxide granules and compare it with the traditionally used liquid hydrogen peroxide.We hypothesized that peroxide granules would have the ability to destroy cyanobacterial cells by inhibiting photosystem II electron transfer in the same way as H 2 O 2 does, but in a more graduate and controlled manner, simulating multiple additions of hydrogen peroxide.
The rst objective was to determine the kinetic of the H 2 O 2 release of two types of granules (CaO 2 and MgO 2 ) into an on-going bloom by monitoring its residual concentration during the treatment.The second objective was to compare the e ciency of granules in reducing the photosynthetic activity during a 48-hours treatment.This study also aimed to determine the most appropriate dose of oxidant for successful mitigation of the Merismopedia sp.
bloom in order to propose an e cient treatment method for in-lake application that will upgrade its water quality class, as proposed by EU.

Study Area
Saint George Lake is located at the Athalassa National Forest Park (ANFP) in Nicosia, the capital city of Cyprus.It is an arti cial lake which covers an area of 68,000 m 3 with an average depth of approximately 2 m.The ANFP covers an area of 8.5 km 2 and it is found between Aglatzia, Strovolos, Latsia, and Geri municipalities; four of the most densely populated locations in Nicosia.Among the most interesting parts of the forest park is Athalassa Lake and Saint-Gorge Lake (Scheme 1) that serve as aquatic life and bird habitats, making them an extremely important biotope for the island.The present study focuses on monitoring of St. George Lake and its treatment during its blooming period in 2019.
Scheme 1. Lakes in the National Forest Park of Athalassa: (A) St. George Lake and (B) Athalassas Lake located at the Athalassa National Forest Park.

Sampling and monitoring
Sampling was performed at a central part of the lake and water was collected from a depth of 0.1-0.2m below the surface with the use of a 5 L bucket and a rope.Water samples were collected and stored in acid washed polyethylene (PE) bottles for the physicochemical water characterization and treatment purposes, and in glass containers for cyanobacterial genes and cyanotoxins analyses.All samples were stored at 4 -6 o C in the dark, brought to the laboratory, and processed within 6 hours after sampling to ensure high accuracy and prevent decomposition of the water characteristics.
The monitoring in St. George Lake occurred between February to December 2019 and 10 samples were collected overall (Table 3).The main physico-chemical parameters (pH, conductivity, salinity, total and dissolved nutrients), the content of cyanobacteria and green algae, the presence of genes for main cyanotoxins synthesis, and the cyanotoxins concentration were determined.Samples were also collected during the blooming period for treatment experiments with liquid hydrogen peroxide and hydrogen peroxide releasing granules.).Nutrients were determined by using Spectroquant® cell test kits (Merck Millipore) equivalent to EPA and APHA standard analytical methods and the Spectroquant® Pharo 300 spectrophotometer (Merck) with method standard deviations ± 0.15 mg/L-N, 0.027 mg/L PO 4 -P, 0.043 mg/L NH 4 -N, 0.13 mg/L NO 3 -N, 0.0027 mg/L NO 2 -N; respectively.Dissolved inorganic nitrogen was calculated as the sum of dissolved nitrogen ions (NH 4 + , NO 3 -, NO 2 -).Temperature, pH, conductivity, and salinity were measured at the sampling site using the ExStik® portable pH Meter (EXTECH, FLIR Systems).
Algal content and Instantaneous Chlorophyll Fluorescence (FT) and Quantum Yield (QY) Instantaneous Chlorophyll Fluorescence (FT) and Quantum Yield (QY) were determined using AquaPen AP 110/C (Photon Systems Instruments, Czech Republic) equipped with blue and red LED light emitters to monitor the growth of algae and cyanobacteria in St. George Lake and for evaluating the e ciency of the applied oxidants in the reduction of the photosynthetic activity of treated bloom.
For the characterization of cyanobacterial species in water samples, raw sample was placed directly or after ltration on a microscopy slide and tested under ECLIPSE Ci-L microscope (Nikon) equipped with OPTIKAM Wi-Fi camera (OPTIKA®, Italy).Phytoplankton samples were preserved with Lugol's iodine solution (2 % nal concentration), stored in 4-6 o C under dark conditions and used within 3 weeks.

DNA isolation and PCR ampli cation
DNA isolation from the biomass collected on cellulose nitrate lters was performed as described by Rogers and Bendich (1994) with minor modi cations [32].Brie y, lters were placed in 2 mL Eppendorf tubes, frozen in liquid nitrogen and grinded.Glass beads were added in ratio 1:1 and the content was dissolved in 700 µL of the extraction buffer I (100 mM Tris, 1.3 M NaCl, 20 mM EDTA, 4% cetrimonium bromide, 1% polyvinylpyrrolidone, 0.1% 2-mercaptoethanol).The mixture was beaten for 10 min using vortex shaker.After 45 min of incubation in 65 °C with 0.5% RNase A, 600 µL of the chloroform-isoamyl alcohol mixture (24:1) was added and the content was shaken and centrifuged at 14000 g for 10 min.The upper phase was transferred into a new tube and mixed with 50 µL of buffer II (10% cetrimonium bromide, 0.7 M NaCl).The chloroform washing step was repeated.After the addition of cold isopropanol in ratio 1:1 the mixture was centrifuged at 14000 g for 10 min.The pellet was washed in 500 µL of 70 % ethanol and the samples were centrifuged at 14000 g for 10 min.The supernatant was discarded, and the pellet was dried on air and resuspended in 50 µL of nuclease-free water.
PCRs for the identi cation of main genes of cyanotoxins were conducted using Dream Taq DNA polymerase (Thermo Fisher Scienti c).Approximately 80 ng of isolated DNA was added to the reaction mixture (20 µl total volume) with 0.2 µM of each primer.PCR was performed with the following parameters: initial denaturation for 3 min at 95 °C, 30 cycles at 95 °C for 30 s, a primer-pair speci c temperature for 30 s and 72 °C for 60 s; a nal extension at 72 °C for 10 min.The electrophoresis of PCR products was conducted on 1% agarose gels at 100 V for 25-40 min.Gels were stained with Midori Green Advance DNA Stain (ABO).

HPLC-HRMS
The high-performance liquid chromatography -high-resolution mass-spectrometry (HPLC-HRMS) method was used to check the presence of intra-cellular cyanotoxins in biomass stored on GF/C lters at -20 o C until extraction.
Sample preparation included extraction of cyanotoxins with 1 mL of 75% methanol in an ultrasonic bath [33].All chemicals used for analytical procedures were the analytical grades.Acetonitrile (HPLC-grade) and methanol (LiChrosolv hypergrade for LC-MS) were purchased from Merck (Darmstadt, Germany); formic acid (98-100%) was obtained from Fluka Chemika (Buchs, Switzerland).High quality water (18.2MΩ cm −1 ) was produced by the Millipore Direct-Q water puri cation system (Bedford, MA, USA).The MC-LR, MC-RR, MC-YR standards were purchased from Sigma Aldrich.Sample preparation procedures were run according to Chernova et al. (2016).
Analyses of extracts were performed using the LC-20 Prominence HPLC system (Shimadzu, Japan) coupled with LTQ Orbitrap XL Hybrid Ion Trap − Orbitrap Mass Spectrometer (Thermo Fisher Scienti c, San Jose, USA).Separation of the toxins was achieved on a Thermo Hypersil Gold RP C18 column (100 mm × 3 mm, 3 μm) with a Hypersil Gold drop-in guard column (Thermo Fisher Scienti c) by gradient elution (0.2 mL min −1 ) with a mixture of water and acetonitrile, both containing 0.05% formic acid.Mass-spectrometric analysis was carried out under conditions of electrospray ionization in the positive ion detection mode.The identi cation of target compounds was based on the accurate mass measurement of [М+Н] + or [М+2Н] 2+ ions (resolution of 30000, accuracy within 5 ppm), the collected fragmentation spectrum of the ions and the retention times.Limits of the detection for different microcystin congeners (2-6 ng L -1 ) were evaluated in model experiments using standard compounds, natural water and biomass as matrixes.

Experimental set-up for treating cyano-HABs
Experiments on the treatment of Mersmopedia sp.bloom in St. George Lake were performed in 250 mL borosilicate glass containers and the oxidants used for this purpose were liquid hydrogen peroxide and metallic peroxide granules.Hydrogen Peroxide (30%) was purchased from Sigma-Aldrich and diluted to 1000 mg/L for the stock solution.Calcium peroxide (CaO 2 ) and magnesium peroxide (MgO 2 ) granules were provided in the form of IXPER® 70CG and IXPER® Magnesium Peroxide Granules 30MG by Solvay Chimika S.A. (free samples).H 2 O 2 stock solution was added to 250 mL of raw sample from St. George Lake to reach a nal concentration of 1, 2, 3, 5 mg/L H 2 O 2 ; and a quantity of 1, 2, 3 g calcium peroxide and magnesium peroxide granules for treating cyano-HABs.The oxidant concentration was monitored by a colorimetric method as introduced by Sellers et.al (1980) [34].In brief, 5 mL of sample was ltered through a PVDF syringe lter and immediately reacted with 0.5 mL of titanium oxalate ([C]=50 g/L) and 0.5 mL sulfuric acid (1+17 v/v) (both reagents purchased from Sigma -Aldrich).The absorbance at 400 nm was measured by the Spectroquant® Pharo 300 spectrophotometer in a quartz cuvette and the concentration of H 2 O 2 was quanti ed based on a calibration curve ranged between 0.5 and 20 mg/L.For determining the e ciency of oxidants on mitigating naturally occurred cyanobacterial bloom (Merismopedia sp.); the FT and QY values in both wavelengths (450, 620 nm) were recorded at 1, 2, 3, 4, 6, 24, 48 h with AQUAPEN as described previously.Physicochemical characteristics such as pH, conductivity, TDS and salinity were measured before and after treatment with the use of ExStick probe (EXTECH).
Data processing and statistical analysis were performed with the use of PRISM®-GraphPad software.George Lake had a stable pH with small variations between 8.3 -8.9 as shown in Figure 1.Conductivity ranged between 1200 and 2000 μS/cm while salinity found to be from 700 to 1000 ppm.Conductivity and salinity followed the same trend showing a noticeable increase during the summer period, both having their peak in August when the bloom occurred.
Nutrient concentrations, depicted in Table 5, also varied with time.Phosphorus was in most of the samples higher than 0.1 mg/L making St. George a hypertrophic lake [22].The water class of St. George Lake determined based on OECD standards was found to be between II and III with respect to its nutrients' concentration.High nitrogen loads were detected in the early months of the year when the status of the lake was oligotrophic with low cyanobacterial content, while high phosphorus concentrations were recorded during the blooming period and declined afterwards (Figure 3).Soluble reactive phosphorus (SRP) was below the MDL before blooming period, stable during the blooming period and had a small drop which followed by a sharp increase during winter.This may be due to heavy rainfalls that caused nutrient run-offs.Dissolved inorganic nitrogen (DIN) remained high at the beginning of the year, radically decreased during summer and increased again in winter (Figure 4).The blooming in St. George Lake was a seasonal phenomenon that peaked during the summer period.More speci cally, July and August 2019, cyanobacterial density of the water was extremely high but after a light rainfall at the beginning of September the density of the bloom began dropping (Figure 5A).Until the end of September Merismopedia were present in the samples, but their density rapidly declined by the beginning of October.

Cyanobacteria species and cyanotoxins analyses
Microscopic observation of preserved samples (n o 5, 6) showed that 99% of the phytoplankton biovolume was attributed to a single picocyanobacterial species, Merismopedia sp.(Figure 6).These species are reported in the literature as microcystin and nodularin producers [35,36] which are both among the most detected cyanotoxin groups in surface waters.Therefore, cyanotoxins genes analyses and cyanotoxins concentrations analysis were performed to examine the toxicity of this bloom.
After isolating the DNA of biomass collected on lter samples, the targeted genes were ampli ed and injected into gel electrophoresis wells for identi cation.The presence of cyrB and cyrJ was recorded only in sample 1.The presence of MC genes was recorded in samples 1, 3, 8 (mcyB) and 1, 3-8 (mcyE).AnaC and sxtA were not found in any sample (Figure 7).
Cyanotoxins genes analyses showed positive results in several samples especially for microcystins genes (mcyB & mcyE), therefore the samples were analyzed for a variety of microcystins analogues concentrations.Microcystins were not detected above the MDL in any of the samples.However, matrix compounds with m/z very close (2-4 ppm) to the one of microcystins were detected in a very low concentration, but the fragmentation patterns of their parent ions differ from ones of microcystins.In Figure 8 the fragmentation pattern of an microcystin-LR standard with the one found in the extract is compared.Lack of characteristics fragments for MCs as the m/z = 599.42(Arg -Adda -Glu) con rmed the absence of microcystins in the Merismopedia sp.bloom.

Treatments
The oxidants used for cyano-HAB mitigation exhibited different e ciencies and impact on the Merismopedia sp.bloom.Hydrogen peroxide treatment was not effective for treating the dense bloom in concentrations of 1 -5 mg/L.The average initial instantaneous uorescence and quantum yield at λ=620 nm representing the cyanobacterial photosynthetic activity and wellness of the photosystem II, were 8500 and 0.37, respectively.
Treatments with lower H 2 O 2 doses (1 and 2 mg/L) were ine cient to treat cyanobacteria that continued to grow steadily.Treatment with H 2 O 2 concentration 3 and 5 mg/L showed only a minor drop of the corresponding FT values compared with the control (Figure 9A).All treated samples showed a stable average of QY around 0.37, meaning that the bloom remained unaffected during the treatment with H 2 O 2 (Figure 9B).
FT and QY at 450 nm excitation wavelength were also monitored during the 48 hour treatment, to determine the photosynthetic activity of green algae and plant suspensions, as illustrated in Figure 9 (C -D).There was a drop of photosynthetic activity in samples treated with 3 and 5 mg/L of H 2 O 2 , giving also visual changes in the color of the treated water (as depicted in graphical abstract.Even though photosynthetic activity dropped at high H 2 O 2 concentrations, QY was stable in all samples during the treatment meaning that the phytoplankton was not affected during the treatment. Treatment with CaO 2 granules (2 and 3 g/L) effectively decreased the photosynthetic activity of cyanobacteria (Figure 10Α).Even though 2 g/L of CaO 2 reduced the value of FT, QY was restored after 6 h of treatment, making it less e cient than 3 g/L which maintained a lower QY value for the duration of the 48 hour treatment (Figure 10B).Photosynthetic activity, measured at 450 nm of samples treated with 2 and 3 g/L CaO 2 , showed a drop of about 50% but the quantum yield of the same samples was not affected, meaning that FT could be restored after days of treatment, making it appropriate for in-situ applications (Figure 10C-D).
Magnesium peroxide treatment was ine cient for concentrations up to 3 g/L.Both FT and QY values at λ=620 and λ=450 nm were stable during the treatment period (Figure 11).The rst 4 h of treatment, FT values decreased but then it was recovered within 6 h of treatment.In general, magnesium peroxide was not able to in uence the bloom, and had no effect on phytoplankton.
It is apparent that MgO 2 had a much lower H 2 O 2 releasing capacity than CaO 2 , making CaO 2 a much more e cient treatment method.Release curves showed that maximum accumulative hydrogen peroxide concentration from 1, 2 and 3 g/L of CaO 2 was 3.5, 8.0 and 11 mg/L respectively; while for 1, 2 and 3 g/L of MgO 2 it was 0.7, 1.2, and 1.8 mg/L of H 2 O 2 ; respectively (Figure 12).
Physicochemical parameters (pH, conductivity, Salinity and TDS) variations during the treatments were monitored since the treatment with oxidants may negatively affect water matrixes.Hydrogen peroxide did not affect the water quality characteristics while magnesium peroxide granules slightly increased all the tested parameters.MgO 2 and CaO 2 granules made the solution more alkaline while H 2 O 2 had the least effect on the pH of the water matrix (Table 6).The initial water characteristics for comparison can be found on Figures 1-2, 6 th sample.Effects on the physicochemical characterization need to be accounted as well when deciding on a type oxidant and dosing to secure possible side-effects in the lake during treatment.Monitoring of St. George Lake showed that the blooming period lasted for 4 months during summer and early autumn period (June -September).The increase of the water temperature and the low turbidity during summer in combination with nutrient load and/or release from the sediments [37] may probably result in periodical blooming of cyanobacteria.In St. George Lake, high nutrient content recorded throughout the year, favored Merismopedia sp. to become the dominant species and to form a dense bloom.Annual average of nutrients classi es the Lake at class III meaning that water quality improvements are essential to be applied since EU Directive requested member countries to maintain surface waters at class I and II.
An almost linear correlation (R 2 = 0.80) between FT (at 620 nm, related to cyanobacteria) and the TP content was documented (Figure 13).Concentrations of TP higher than 0.2 mg/L favored Merismopedia sp.blooming as shown in Figure 13.Phosphorus concentration during the bloom was higher than 0.2 mg/L which means that in such a hypertrophic lake (TP > 0.1 mg/L) nitrogen became the limiting element.To support these ndings, different approaches on estimating nutrient limitation were tested in order to investigate which one ts better to the studied environment (Table 7).
Application of the Red eld ratio in our study indicated that P was the limiting factor for the whole season which does not re ect the actual trophic condition of St. George Lake.Guildford and Hecky (2000) proposed that lake systems tend to have higher than Red eld thresholds for P-limitation [16].Those thresholds showed a better t than TN:TP molar ratios, shifting from P-limitation to co-limitation during the bloom, but it does not still represent well the trophic status.The best adjustment was found when the ratio proposed by Levine (2001) and Symons (2012) applied.Both DIN:TP and NO 3 -:TP mass ratios showed relatively similar results, suggesting an N-Limitation during the bloom, co-limitation before and after the bloom and P-limitation in the remaining period.Overall, approaches that were based on ocean dynamics were found to have poor tting on the trophic status of the lake in contrast with more recent approaches that were intendent for fresh waterbodies.This stresses the need of better understanding the nutrient dynamics in lakes and the development of holistic approaches based on the different physicochemical characteristics of each waterbody, taking into consideration also other limiting factors that may affect trophic status such as light and temperature [38].Nutrients and trophic status of each waterbody are highly correlated, indicating that when a high nutrient load is documented, algal and cyanobacterial monitoring is essential for the early detection of (toxic) blooms.High FT values at 620 nm (>3000) recorded in St. George Lake early in summer indicated an on-going bloom which was expected based on its trophic condition.Since the photosynthetic activity was high in samples 5 -7, they were observed under the microscope.A mono-domination of a picocyanobacterial species, Merismopedia, a known MC and NOD producer, was found.However, even though genes involved in MC and NOD synthesis were present (mcyB and mcyE), no cyanotoxins were detected in any of the obtained samples.Several species have a potency to produce MC, but in given conditions the genes may not be expressed.Scienti c reports related to the regulation of mcy gene expression in response to external biotic and abiotic factors have been published recently.For example, in mixed cultures of M. aeruginosa and P. agardhii, both suppressed growth and downregulation of mcyE expression were observed [39] which suggests that the competition between two toxic strains may results in a lower MC production.It is also possible that the dominant species observed in the investigated lake lost their capability of MC synthesis.The presence of mcyB only in 3 samples and mcyE in 8 samples suggest that the genetic machinery for MC synthesis may be de cient.Furthermore, an indirect downregulation of MC synthesis was observed in response to iron limitation [40], as a result of a lower photosynthetic activity, therefore it was suggested that this parameter (iron level) may be helpful in predicting bloom toxicity.The light intensity is also known as an important abiotic factor in uencing mcy expression and MC production [41].
However, it should be underlined that any genetic method applied should be complemented and there accompanied with analytical con rmation.The level of mcy transcripts is often not correlated with the MC concentration.The toxicity based on the mcy levels might be both under-and overestimated [42,43,44].Therefore, these assays should not be considered as good indicators of bloom toxicity, but rather as a complementary tool in risk assessments.Similarly, it can be assumed that the detection of mcy does not ensure MC presence which should be proven through advanced analytical methods (LC-MS/MS).Attention should be paid on the obtained m/z and the corresponding fragmentation patterns so that cyanotoxins concentration is not overestimated.

Treatment
Dense blooms of Merismopedia sp. in a hypertrophic lake suggested the requirement of extremely high doses of H 2 O 2 for e cient treatment.Since in those cases, doses over 5 mg/L of H 2 O 2 may be harmful for the other components of microbial communities (bacteria, phytoplankton and zooplankton), alternative solutions are needed.CaO 2 and MgO 2 granules studied herein are an alternative to traditionally used liquid hydrogen peroxide [45].To determine the most e cient dose we took into consideration not only the ability of the oxidant to destroy cyanobacterial cells but also the wellness of remaining phytoplankton.The 2 g/L CaO 2 treatment was most effective in bloom elimination, but it caused a noticeable change in the remaining phytoplankton.Despite that, the wellness of the photosystem II, recorded as the quantum yield was not affected at all, which means that phytoplankton's photosynthetic activity may be restored shortly after treatment.CaO 2 outperformed MgO 2 because of its higher H 2 O 2 release from the granules.This may be due to the fact that the dissolution product of MgO 2, which is magnesium hydroxide, is less soluble than the dissolution product of CaO 2 , which is calcium hydroxide, at the same pH.This also affected the suspended solids content of the treated water.Calcium peroxide granules caused drastic changes of pH as while decomposing it releases highly basic Ca(OH) 2 (equation 1-2).
Both granules released H 2 O 2 with a reaction that follows a pseudo-zero-order kinetics pattern and their kinetics are greatly affected by temperature and the pH of the solution as explained by Wang et.Al (2016) [31].
Combining treatment e ciency and their releasing capacity, treatment with MgO

Conclusions
This is the rst report from Cyprus on the occurrence of a dense cyanobacterial bloom with the Merismopedia sp.being the dominant species.The blooming occurred in St. George Lake of ANFP, in Cyprus during the summer and early fall of 2019.Conventional monitoring tools such as microscopic enumeration of phytoplankton species and trophic condition determination were applied.However, limited information on the factors driving the cyanobacterial blooming is obtained through these tools.Therefore, additional characterization of the lake ecosystem including physicochemical characteristics; total and dissolved nutrients; temperature, air and light intensity; cyanobacterial and green algae content; cyanotoxins genes and cyanotoxins analyses were found to be essential.Monitoring is useful for building predictive models as early response tools to avoid cyanotoxins contamination of source waters used for recreational activities and drinking water.Correlations between nutrients and eutrophication have been developed recently [46] with the DIN:TP and NO 3 -:TP mass ratio to be the most promising ratios, as con rmed also in our case.While these ratios are proving to be promising for understanding the eutrophic status of surface waters, they should be applied with caution after careful examination of each waterbody's unique characteristics (depth, size, water temperature etc.).Customized monitoring strategies for each waterbody and treatment application at the early stages of a bloom, are essential for protecting water quality of surface waters.
With climate change being linked to global expansion and persistency of harmful cyanobacteria, it is imperative that we nd e cient methods to mitigate harmful cyanobacteria blooms at source.Treating cyanobacteria effectively, without harming the rest of the ecosystem is vital for restoring and safeguarding surface water quality.Currently, hydrogen peroxide is widely used for mitigating cyano -HABs as an alternative to algicides and an ecofriendly method.But treating dense blooms, such as the Merismopedia bloom occurred in St. George Lake, requires high oxidant doses (>5 mg/L) at once which is known that the remaining ecosystem (e.g., zooplankton; phytoplankton) is also affected.Peroxide granules that are H 2 O 2 slow releasing oxidants were tested herein as an alternative method for hydrogen peroxide treatment.Their hydrogen peroxide releasing properties and treatment e ciency varied.Calcium peroxide outperformed other peroxides which can be a potential treatment method worth to investigate further for its e ciency on different cyanobacterial species and matrixes.Despite that it releases high amount of H 2 O 2 , it is acting gradually on the species by rst reacting with matrix organic load and then by reaching the contaminant making it more e cient than liquid hydrogen peroxide.Further studies will provide a clearer view on its properties as a promising mitigation technique since pH and other physicochemical characteristics were found to affect the treatment. Cyano

Results 1 . Monitoring 1 . 1
Physicochemical water characteristics and nutrients Water characteristics of St. George Lake varied during the monitoring period (February -December 2019).The recorded values of physicochemical characteristics and nutrients concentrations during the period of study are illustrated in Figures 1 -4.During the summer months air temperature was as high as 34 o C while during Spring, Fall and Winter the temperature varied between 15 -25 o C. St.

Figure 3 Total
Figure 3

Table 1 .
Nutrient limitation approaches and thresholds based on TN:TP, DIN:TP and NO 3

Table 2 .
Proposed surface water quality standards by OECD based on the EU Directive (2000/60/EC).

Table 3 .
Sample number and date of sampling event in St. George Lake.
Physico-chemical water characteristics analysesRaw samples were analyzed for total nitrogen (TN) and total phosphorus (TP) while samples ltered through cellulose nitrate membrane lter were analyzed for the dissolved nutrients content (ammonium -NH 4 + , nitrates -

Table 4 .
Primers used in the detection of cyanotoxin producing genes in St. George samples and ampli cation parameters used at PCR.

Table 5 .
Total and dissolved nutrients concentration (mg/L) in St. George Lake during monitoring period.

Table 7 .
Calculation of different ratios proposed in the literature for nutrient limitation and only 2 g/L of granules can gradually release up to 8 mg/L H 2 O 2 which is su cient for treating blooms without affecting the rest phytoplankton species.
2 is ine cient and would require high amount of granular oxidant to mitigate a contaminated site.CaO 2 has the ability to release H 2 O 2 in a more effective way -HABs Cyanobacteria Harmful Algal Blooms