Cyanobacterial occurrence and detection of microcystins and saxitoxins in reservoirs of the Brazilian semi-arid

: Aim: The rapid spread of cyanobacteria in water sources and reservoirs has caused serious environmental damage and public health problems

In Rio Grande do Norte (RN) state, cyanobacteria are common in reservoirs used for public supply, but few studies report the presence of STXs and MCYs (Costa et al., 2006b(Costa et al., , 2009)), indicating the necessity for a systematic monitoring of the concentrations of these cyanotoxins.Thus, the purpose of this study was to determine the potential toxicity of cyanobacterial blooms in four reservoirs in the semi-arid of RN for a continuous period between 2009 and 2011.Besides this, we investigated seasonal (rainy and dry periods) differences of the distribution of microcystins and saxitoxins as well as the levels of cyanobacteria, and checked whether they were above the limit established by Brazilian law.

Introduction
In order to minimize the impact of long periods without rainfall, water reservoirs were built in the semi-arid region of northeastern Brazil to capture water from rainy days and make it available in periods of drought.Besides being used for domestic water supply, the reservoirs can be used for fishing, aquiculture and entertainment (Eskinazi-Sant 'Anna et al., 2006;Costa et al., 2006a).These reservoirs have typical characteristics of watersheds of semiarid regions, such as high temperatures and high turbidity throughout the year, associated with the constant state of eutrophication by nutrient input, render these environments naturally vulnerable to cyanobacteria (Eskinazi-Sant 'Anna et al., 2006, Costa et al., 2009, Sousa et al., 2008;Vasconcelos et al., 2011;Huszar, 2000;Silva et al., 2011;Bouvy et al., 1999;Molica et al., 2005;Panosso et al., 2007).

Sampling
Monthly water samples were taken at each point.Water samples were collected throughout the water column involving six depths, three of which were in the aphotic zone and the other three in the photic zone.For a single sample of each point, all the six samples were integrated.With the aid of a Van Dorn bottle, integrated aliquots of the water samples were collected for microcystin (MCYs) and saxitoxins (STXs) analysis and phytoplankton count (200 mL).The latter was fixed with 1% acetic Lugol.Water samples for MCs and STXs analysis were frozen until data processing.Water samples were obtained from a plankton net (20 µm) in vertical drags for identifying phytoplankton using living material (200 mL) and fixed with 4% formaldehyde (200 mL).
In each sampling point conductivity, turbidity, temperature, pH and dissolved oxygen parameters were measured using a multiparameter probe.To estimate the water transparency the depth extinction of the Secchi disk was used.The photic zone was obtained by calculating 2.7 times the estimated water transparency with the Secchi disk (Cole, 1975).

Identification and quantification of cyanobacteria
The identification of cyanobacterial populations and other groups of phytoplankton was performed by microscopy, which was made to the level of species, whenever it was possible, by analysing morphological and morphometric characteristics of the vegetative and reproductive stages.The classification system adopted was Komárek & Anagnostidis (1998) for the Chroococcales gender, Komárek & Anagnostidis (2005) for the Oscillatoriales and Komárek & Anagnostidis (1989) for Nostocales and specialized works for other phytoplankton.The population density was estimated (cel.mL -1 ) by the method of Utermöhl (1958) using the inverted microscope and quantification was done in random fields (Uhelinger, 1964), reaching 100 individuals of the most frequent species, with the error less than 20% at a confidence interval of 95% (Lund et al., 1958).In case of bloom, 400 individuals of the dominant species were quantified, resulting in an accepted error of 10% (Chorus & Bartram, 1999).

Microcystins and saxitoxins analysis
Water samples were frozen and defrosted three times, filtered with glass fiber filters (Whatman GF/C) and sonicated to lyse the cells, ensuring that the total amount of cianotoxins were analysed: both in the water or in the cytoplasm of the cells.After this process, the samples were analyzed by assay technique of enzyme-linked immunosorbent assay (ELISA) using kits (plate type) Commercial ELISA Beacon mark according to the manufacturer's instructions.

Statistical analyses
Statistical analyses were performed using SPSS Software.Data are expressed as mean values + standard error of mean (SEM).Significant differences between groups (Rainy season X Dry season) were defined as a p value less than 0.05, and they were determined by t-test.In order to identify potential relationships between both physico-chemical variables and microcystins, and physico-chemical variables and saxitoxins, we performed Pearson correlation test with p value less than 0.05.

Abiotic variables
The investigated reservoirs showed low transparency ranging between 0.2 m and 3.5 m; elevated temperatures between 25.2 °C and 33.5 °C and pH between neutral and alkaline usually ranging between 6.5 and 10 (Table 2).The volume of the reservoirs (Table 2 and Figure 2) varied over the three years and ranged between 42% and 100% of their total capacity, registering lower volumes in 2010.

Microcystins and saxitoxins detection
Saxitoxins and microcystins were detected in all samples of the four studied reservoirs (Figures 4  and 5), and microcystin was the most frequent with concentrations above those permitted for human consumption, according to Order No 2914/2011 of the Brazilian Ministry of Health (MH) (Brasil, 2011) which regulates the water potability standards.In 2010 the highest values of MCYs was detected, except for the maximum value of 24.1954 µg.L -1 in Garagalheiras in 2009, followed by 2009 and 2011.The minimum value was found in Itans (March 2010) corresponding to 0.00227 µg.L -1 .During the study, the mean values of MCYs in the rainy season were higher than in the dry season (Table 3) and these differences were statistically significant for the period of 2009-2011 (F=44.073; p<0.05) (Figure 6) and in each year (2009: F= 16.225; p<0.05; 2010: F= 46.643; p<0.05 and 2011: F= 7.989; p<0.05 ).
Regarding saxitoxins concentrations, the year that showed the highest values was 2010, followed by 2009 and 2011.The maximum value found was 0.766 µg.L -1 in São Rafael in August 2010 and the minimum was 0.003 µ.L -1 in Jucurutu in August 2009.
In this study, the comparisons showed no statistical difference in the levels of saxitoxins between dry and rainy season in the period of 2009-2010 (p>0.05).Also, the analyses of each year independently showed no difference in 2010 (Table 3).However, in 2009 the levels of saxitoxins were higher in the dry period compared to the rainy season (F= 14.771; p<0.05).

Composition and density of cyanobacteria
All samples were analyzed under the microscope and revealed the presence of cyanobacteria.It was possible to identify 21 species of cyanobacteria, 11 of them considered potentially producers of cyanotoxins, according Chorus & Bartram (1999).The average relative density of cyanobacteria to total phytoplankton was above 89% in all samples.
In relation to the density of cyanobacteria (Figure 7), the year that showed the highest values of each setting was 2009, except for the highest value in PT corresponding to 23.5 x10 5 cel.ml -1 in December 2010.The minimum value was    In all years, there were dominant microcystin producing species (Table 4) (Microcystis aeruginosa, Anabaena circinalis and Planktothrix agardhii) and dominant saxitoxins producing species (Cylindrospermopsis raciborskii, Planktothrix agardhii, Aphanizomenon gracile and Anabaena circinalis).Among them, the most frequent in all sampling points during the three years of study were Planktothrix agardhii, Microcystis spp and Cylindrospermopsis raciborskii.Microcystis spp, was represented by M. aeruginosa, M. panniformis and M. Protocystis .

Discussion
In Brazil and in the five continents of the world, the presence of microcystins and saxitoxins in water for human consumption has been reported over the years to the present day.In Brazil, the presence of MCYs in the northeast region has been reported by Piccin-Santos & Bittencourt-Oliveira ( 2012     The presence of saxitoxins was reported in sources for human water supply in the northeast region of Brazil (Costa et al., 2006b;Molica et al., 2005), Southeast (Ferrão-Filho et al., 2009;Anjos et al., 2006) and in the southern region (Yunes et al., 2003).
The occurrence of microcystins have also been reported in several countries as in Europe: in Spain in reservoirs with levels between 0.055 and 1.032 µg g -1 (Asencio, 2013) and Pawlik-Skowrońska et al. ( 2013) described maximum value of microcystin corresponding to 22.2 µg L −1 in a reservoir situated in Lublin (Poland) and, Bláha et al. (2010) described levels of MCYs in water of 36.9µgL−1 in the Czech Republic.In South America, in Rio de La Plata -Uruguay, Pírez et al. (2013) detected 65 µg L -1 .In Central America, Romero-Oliva et al. ( 2014) described values of MCYs (intra-and extracellular 1931 and 90 mg L -1 , respectively) in Lake Amatitlán (Guatemala).
In North America, in the San Francisco Estuary, USA (Lehman et al., 2013), between 0.79 ng L -1 and 29 ng L -1 .
In Africa, in the Nile River -Egypt, microcystins were reported between 1.6 and 4.1 mg g -1 in water for public supply (Mohamed et al., 2006).In Oceania, in the Murray River (Australia) levels between 0.028 and 0.036 mg L -1 were also reported in reservoirs for public use (Bowling et al., 2013).In Asia, in Saudi Arabia, Mohamed (2008) described values of MCYs in the water corresponding to 5.7 µg L −1 and both microcystins and saxitoxins were also reported in lakes and reservoirs in Greece (3.9 to 108 mg L -1 of MCs and 0.4 to 1.2 mg L -1 of STXs) (Spyros & Nikos, 2014).In France, Ledreux et al. (2010) described maximum levels of saxitoxin of 6.7 µg L −1 and maximum levels of MCYs of 89 µg L −1 .
The increasing eutrophication and dominance of cyanobacteria in the Brazilian semi-arid region is reported in the public water supply reservoirs in Pernambuco (Bouvy et al., 1999, Huszar et al., 2000); Paraíba (Vasconcelos et al., 2011) and in Rio Grande do Norte (Costa et al., 2006b;2009;Eskinazi-Sant'ana et al., 2006;Panosso et al., 2007;Silva et al., 2011;Sousa et al., 2008).Reservoirs located in semi-arid region of Rio Grande do Norte, show annual averages of total phosphorus and chlorophyll a (Costa et al., 2009), above the limit established by The Brazilian National Environment  (Brasil, 2005).The eutrophic condition in semiarid regions is established by values above 50-60 µg.L -1 of total phosphorus average and above 12-15 µg.L -1 of chlorophyll a average (Thornton & Rast, 1993).This permanent eutrophic condition, in semiarid region of Brazil, and the frequent events of intense blooms of potentially toxic cyanobacteria are associated with multiple uses of the river basin, low water transparency, warm waters, long daily sun exposure, long residence time and good availability of phosphorus and nitrogen (Costa et al., 2006b, 2009, Vasconcelos et al., 2011, Sousa et al., 2008;Silva et al., 2011).
In this study, between the years 2009 to 2011, we emphasized the continuity of permanent dominance and abundance of potentially toxic cyanobacteria, featuring events of toxic blooms, confirmed by the presence of microcystins and saxitoxins, at six investigated points in four reservoirs.The presence of MCYs with concentrations above 1µg.L -1 -limit allowed for human consumption according to of the Brazilian Ministry of Health (Ordinance 2914/2011) (Brasil, 2011) -was detected in 27% of samples.Moreover, despite the presence of saxitoxins (100%), the four investigated reservoirs showed values below the indicated value for drinking water (3 µg.L -1 ), as was also reported at the Armando Ribeiro Gonçalves reservoir (Costa et al., 2006b).Vasconcelos et al. (2011) reported the presence of microcystins above 1 µg.L -1 in 55% in the reservoirs in Paraiba (Brazil) during the dry season and 20% during the rainy season, 15% of them presented concentrations below 1µg.L -1 .Microcystin levels, by ELISA method, were also found at levels above 1µg.L -1 by Spyros & Nikos (2014) (between 3.9 and 108µg.L -1 ), by Pírez et al. (2013) (65 µg.L -1 ) and by Sotero-Santos et al. 2008 (between 28 and 45  µg.L -1 ) and all samples along with the co-occurrence of microcystins and saxitoxins Spyros & Nikos (2014) and Costa et al. (2006b).
The raise of the growth rate of cyanobacteria can be influenced by certain climatic conditions such as temperature increases, surface light and raise of nutrients due to greater precipitation (Paerl & Paul, 2012).Due to larger amount of cyanobacteria, the competition may be greater between species, which can promote the appearance of toxic strains and influence the increase in production of cianotoxins by toxic strains.Considering the influence of the environmental variables in the microcystin distribution, the temperature may represent a significant factor that may influence the production of cyanotoxins.Mohamed (2008) found values of MCYs, in the water above the limit established by WHO (World Health Organization) (1 µg L −1 ) (Chorus & Bartram, 1999) in temperatures ranging from 48° until 70°C.Regarding the physico-chemical variable, in the present study we identified a positive correlation only between temperature and microcystin levels, but no statistical difference between the other variables and both saxitoxin and microcystin.Similar results were described by Mohamed (2008) who found a positive correlation within intracellular microcystin content and temperature.However, Asencio (2013) did not find a correlation neither between environmental parameters (including water temperature, dissolved oxygen, conductivity and pH) with values of microcystin nor between values of elevated nutrients (phosphorus and nitrogen) with values of microcystin.
However, the synergistic effect of this variables and levels of cyanotoxins were not analyzed and according to Paerl & Paul (2012), there is a probable synergistic effect between global warming and eutrophication promoting the increase of toxinproducing cyanobacteria and in agreement, Ekvall at al. (2013) described, through an experiment, that both temperature and water color (humic content and nutrient) increase the production of mycrocystin, but when both water color and temperature were acting alone, there was no increase of microcystin.Davis et al. (2009) demonstrate that high temperature along with elevated levels of phosphorus (P) frequently influence in the production rate of toxic strains instead of non-toxic microcystis strains.
Another component which can influence the microcystin production rate is light intensities.Generally, low levels of cyanotoxins usually are found in low light intensities, and high light intensities seems to influence the transcription of genes associated with synthesis of MCYs (Kaebernick & Neilan, 2001;Kaebernick et al., 2000).Zilliges et al. (2011) described a function of MCYs as intracellular protein-binding peptides that act in the protection against high UVR levels, which can cause a negative effect of oxidative stress.
In our study, the highest values of microcystin occurred in the rainy season of 2010.However, Lehman et al. (2013) described as higher both values of microcystis and MCYs in the dry years instead of wet years.These authors defined that the concentration of MCYs raised when the abundance of microcystis increased in dry years.They also identified a seasonal threshold which was responsible for the presence of microcystis only when the temperature was higher than 19°C.In Cordoba (Argentina), Ruiz et al. (2013) found MCYs in 97% of 35 samples collected between 1998 and 2001, with levels ranging from not detectable to 119 µg L -1 .These authors also found seasonal differences in the distribution pattern of MCYs, with the highest values occurring in the summer and spring, when temperatures were higher.In spite of that, no statistical differences were detected.However, Asencio et al. (2013) did not find a significant difference between seasons for microcystin production.
The occurrence of higher MCYs levels in the rainy season (Feb-May) of 2010, corresponded to the periods of lower volume of water in reservoirs and in the rainy season of 2011 occurred the lowest levels of microcystin, corresponding to the longest and abundant period of rain during this study.The decrease of microcystin levels could have occurred because of the dilution caused by the abundant precipitation.However, another hypothesis to the decrease of MCYs (intracellular and extracellular) may have occurred because MCYs in the water column may have accumulated in the sediment due to senescence and possible subsequent sedimentation of intact cells (Wörmer et al., 2011).Although these cells are able to survive for long periods, they may suffer lysis or grazing, thus releasing the cyanotoxins in the water column (Boström et al., 1989).The sedimentation of toxic cells may represent a higher risk for water consumption without treatment or with inadequate treatment, because the removal of water occurs at the bottom of the supply reservoir, which can lead to resuspension of these cells and releasing of cyanotoxins.Besides this, when the blooms occur, toxic strains can replace non-toxic strains and vice-versa (Rinta-Kanto et al., 2009).Davis et al. (2009) described an elevated variability of toxic-strains in populations of Planktotrix and Microcystis ranging from 0.01 until 100% and this can also influence the decrease of levels of MCYs, causing low levels of toxic-strains in both population of Planktotrix and Microcystis (Krienitz et al., 2013).
The Brazilian National Environment Council (CONAMA) (357/05 -Class II waters: reservoirs) (Brasil, 2005), set the maximum values for cyanobacteria density in waters: 50,000 cel.mL -1 .Regarding the density of cyanobacteria, 76% of the values were above the limit.By analyzing the average of three years for each environment, only Jucurutu showed a mean value within the allowed (0.34 x10 5 cel.mL -1 ).The lower density of cyanobacteria at this point can be explained by the characteristic of a lotic environment, which is distinguished from other sampling points that are lentic environments.The low residence time favours the dispersion and discriminate the permanence of cyanobacteria.
Such high levels of cyanobacteria density indicate the continuity of the eutrophic state and represents an important parameter for evaluating the quality of water for human consumption.High density values of cyanobacteria were also found by Messineo et al. (2009), ranging between 10 6 and 200x10 9 cel.ml - , and other studies in the the Brazilian semi-arid region by Molica et al. (2002Molica et al. ( , 2005)), Bouvy et al. (1999Bouvy et al. ( , 2003) ) and Vasconcelos et al. (2011).
In four reservoirs, we identified five genera of potentially toxin-producing cyanobacteria, including four microcystin toxin-producing with high values of microcystin corresponding to 24.19 µg.L -1 .Asencio (2013) described in seven reservoirs 10 toxin-producing genera including seven microcystin toxin-producing and the highest value of microcystin was 1.032 µg.g -1 , which was above the limit established.Krienitz et al. (2013) described both Microcystis and Planktothrix as potential microcystin producers in Lake Naivasha (Kenya).He also found values of MCYs in all samples analyzed (2008 to 2013) in the range of 0.001-0.041µg L -1 .Mohamed (2008) found a positive correlation (r = 0.45-0.93)between values of intracellular microcystin with biovolume and abundance of many species (at least seven), yet only two of them, Oscillatoria limosa and Synechococcus lividus, were able to produce MCYs.Besides that, Oscillatoria limosa was present only in one of three studied sites.
The maximum density values did not correspond to the maxima found cyanotoxins.Other studies have also found no relationship between the density of cyanobacteria and cyanotoxins values (Messineo et al., 2009;Pírez et al., 2013) neither have found a linear relationship between cell density and MCYs, demonstrating that only 30% of the microcystin level variations could be explained by the density of cyanobacteria.Furthermore, they demonstrated that only 18% of microcystin levels have been explained by chlorophyll.Mohamed & Carmichael (2000) found no correlation between the chlorophyll-a content and microcystin concentration.This may be due to the fact that MCYs are tending to remain chemically stable for long periods in the environment (Chorus & Bartram, 1999).In addition, high levels of MCYs are most common during or after a large amount of cell lysis (Pírez et al., 2013;Messineo et al., 2009).Another study found that the highest MCYs concentrations were found when there was a predominance of M. aeruginosa bloom (potentially microcystin-producing), even a low density of cyanobacteria (Spyros & Nikos, 2014).This can also occur due to the possibility of non-producing, producing and potentially producing strains coexistence in the same population (Bittencourt-Oliveira et al., 2010;Bittencourt-Oliveira & Molica, 2003;Spyros & Nikos, 2014).
Water contamination by microcystin and saxitoxins occurs in several locations worldwide, commonly affecting human sources of water supply.This emphasizes the necessity for the long-term monitoring of water quality.However, it is still observed the lack of a standard research and a detection method in monitoring programs, hindering a more robust comparative results analysis and making it impossible to determine a uniform profile.Regarding the water quality monitoring, among the different methods used, the ELISA cyanotoxins quantification method appears to be more beneficial to carry out constant monitoring, as it is considered to be cheap and simple and able to detect some of the 80 variants of microcystin, including the principal one, which is the most toxic (MC-LR) (Pírez et al., 2013) and some variants of saxitoxins among the 27 already known (Ho et al., 2012).Thus, this method is considered satisfactory to meet the requirements of the Ordinance of the Ministry of Health 2914/2011 (Brasil, 2011).
In summary, the MCYs and cell density of potentially toxic cyanobacteria levels above allowed in drinking water, found in this and other studies, represent a potential health risk to the human population since the conventional water treatment is not able to remove cyanotoxins.This is alarming because in many semiarid places the proper treatment is not performed.In many places the water is treated in a simplified manner, or is not treated, increasing the risk of intoxication by direct ingestion of contaminated water, even in small doses for long periods.Besides the risk of contamination by direct water consumption, there is still the possibility of contamination by eating fish, as microcystin can accumulate in fish muscle (Magalhães et al., 2001;Pawlik-Skowrońska et al. (2013) and the consumption of fresh vegetables irrigated with contaminated water (Romero-Oliva et al., 2014).As the highest density values of cyanobacteria and cyanotoxins are not correlated, it is necessary and urgent to contemplate counting and quantification of cyanobacteria and cyanotoxins through a systematic monitoring specially in the water for drinking consumption.

Figure 3 .
Figure 3. Positive correlation between the level of microcystins and temperature.

Figure 6 .
Figure 6.Mean values of cyanotoxins during dry and rainy seasons.

Table 1 .
Hydrological and morphometric characteristics of the reservoirs.

Table 2 .
Limnological variables in reservoirs studied between 2009 and 2011 (Mean, minimum and maximum).

Table 3 .
Concentrations of microcystins and saxitoxins (Mean ± standard deviation) in reservoirs studied between 2009 and 2011.

Table 4 .
Potentially toxic species of cyanobacteria that were dominant in the studied sample points between 2009 and 2011.