OCCURRENCE OF TOXIN-PRODUCING CYANOBACTERIA BLOOMS IN A BRAZILIAN SEMIARID RESERVOIR

We report the occurrence of cyanobacterial blooms and the presence of cyanotoxins in water samples from the Armando Ribeiro Gonçalves reservoir (06° 08’ S and 37° 07’ W), located in the state of Rio Grande do Norte, in the semiarid region of northeastern Brazil. The cyanobacterial species were identified and quantified during the rainy and dry seasons in the year 2000. Cyanotoxins such as microcystins, saxitoxins and cylindrospermopsins were analyzed and quantified using hPLC and ELISA methods. The mixed toxic blooms of Cylindrospermopsis raciborskii, Microcystis spp (M. panniformis, M. protocystis, M. novacekii) and Aphanizomenon spp (Aphanizomenon gracile, A. cf. manguinii, A. cf. issastschenkoi) were persistent and represented 90-100% of the total phytoplankton species. Toxic cyanobacterial blooms from the Armando Ribeiro Gonçalves reservoir were analyzed and found to have three phases in relation to the annual cycle. During the rainy season, an intense toxic bloom of Cylindrospermopsis raciborskii was recorded along with saxitoxins (3.14 μg.L). During the transition period, between the rainy and dry seasons, different species of Microscytis occurred and microcystin as high as 8.8 μg.L was recorded. In the dry season, co-dominance of Cylindrospermopsis raciborskii, Microcystis spp and Aphanizomenon spp occurred and the concentrations of saxitoxin remained very low. Our results indicate the presence of microcystins (8.8 μg.L) and saxitoxins (3.14 μg.L) into the crude water, with increasing concentrations from the second fortnight of April to late May 2000. The occurrence of toxic blooms in this reservoir points to a permanent risk of cyanotoxins in supply waters, indicating the need for the implementation of bloom control measures to improve the water quality. Exposure of the local population to cyanotoxins through their potential accumulation in fish muscle must also be considered.

after undergoing renal dialysis (Jochimsen et al., 1998).Teixeira et al. (1993) showed a strong correlation between cyanobacterial blooms in the Itaparica reservoir (Bahia) and the death of 88 people, among 200 intoxicated after drinking water from the reservoir between March and April 1988.In the state of Rio Grande do Norte (northeastern Brazil), cyanobacterial dominance in eutrophic freshwater systems was studied extensively over the last decade (Chellappa, 1990;Chellappa et al., 1996;Costa et al., 1998;Costa et al., 2001) and the need for cyanotoxin analysis became apparent from the findings of fish mortality associated with the toxic blooms of Microcystis aeruginosa (Chellappa et al., 2000).
The present study was undertaken at the Armando Ribeiro Gonçalves reservoir, the second largest reservoir in northeastern Brazil (2.4 x 10 9 m 3 ), which supplies water to 400 thousand inhabitants.Intensive use of fertilizers, as well as shrimp and fish farming, take place in the hydrographic basin of this manmade reservoir, leading to high concentrations of phosphorus and nitrogen which result in water eutrophication.The low annual rainfall, characteristic of regions with high evaporation rates, also contributes to concentrate mineral salts and inorganic nutrients in the water (Naselli-Flores, 1999).Several other factors such as extended solar radiation with low amplitude of annual and daily variation, high evaporation rates and high water residence period, promotes increasing growth of cyanobacterial species (Costa,

INTRODUCTION
Eutrophication is the result of uncontrolled human population growth and the discharge of urban, industrial and agriculture effluents into the aquatic ecosystems of several countries, including Brazil (Tundisi & Matsumura-Tundisi, 1992).This represents a primary problem for water management all over the world, especially in dry regions, since one of the major consequences of eutrophication is the appearance of cyanobacterial blooms (Carmichael, 2001;Falconer, 2001;Azevedo et al., 2002).
Cyanobacterial blooms in freshwater usually comprise both toxin and non-toxin producing species (Baker and humpage, 1994).The main toxin producing cyanobacteria genera include Anabaena, Aphanizomenon, Microcystis, Planktothrix, Lyngbya and Cylindrospermopsis (Chorus and Bartran, 1999).Cyanotoxins (hepatotoxins and neurotoxins) are responsible for the intoxication of wild and domestic animals and the contamination of drinking water, also inducing fish mortality and eliminating other aquatic biota (Carmichael 2001;Falconer 2001;Sivonen & Jones, 1999).
The purpose of this work is to report the occurrence of cyanobacteria blooms and their toxins in the annual cycle.Potential health risks to humans drinking the reservoir's water are also discussed.

Sample collection and phytoplankton analysis
Samples were collected from the Armando Ribeiro Gonçalves reservoir (n = 9) and in the Pataxó Channel (PC) (n = 4), which carries water from the reservoir to the water treatment station (n = 4).The samples from these two sampling points are referred to as "raw water samples".A third sampling point was set up at the outlet of the treatment station, which is referred to as "treated water".During 2000, fortnightly samples were taken in the rainy season (April-May), in the period of transition from the rainy to the dry season (July) and in the dry season (November-December).For qualitative analyses of the phytoplankton, samples were collected by vertical towing with a 20 µm mesh size net and preserved in 4% formaldehyde.For phytoplankton counting, water samples were collected with a Van Dorn bottle along vertical profiles which were further pooled and preserved with Lugol solution.Phytoplankton analyses were performed using an inverted microscope, according to Lund et al. 1958.The classification system used was that of Komárek & Anagnostidis (1989, 1995, 1998).
Cyanotoxin analyses (microcystin, saxitoxin and cylindrospermopsin) were carried out on the seston samples, which were filtered through Whatman GF/C glass fiber filters and frozen at -20 °C.The cyanotoxins in these samples were analyzed by the high Efficiency Liquid Chromatography technique (hPLC-Shimadzu SPD-M10A).The samples of treated water were also analyzed for microcystin by immunoassay, using ELISA Microcystin Plate Kit (ENVIROLOGIX INC.).

Extraction and quantification of microcystin (MYCT)
Microcystin extraction of the seston samples was performed using a slightly modified version of Krishnamurthy et al. method (1986).The cell content was extracted three times with butanol: water (5:20:75 v/v) and cell debris was removed by centrifugation.The supernatant was evaporated to 30% of its initial volume and the remaining extract was filtered though a C-18 cartridge (Bond Elut C-18 Varian).The samples were eluted from the cartridge with 20 mL of a solution of deionized water, 20% methanol and 100% methanol, respectively.The 100% methanol fraction was evaporated to dryness, resolubilized in 1 mL methanol 50% and filtered through nylon filters (0.45 µm).Toxins were separated by hPLC.The hPLC system was equipped with a C-18 reversed phase column (250 mm x 10 mm, 5 µm).The analysis was carried out under isocratic conditions with a mobile phase of 20 mM ammonium acetate, ph 5, and acetonitrile (7:3) for 90 min, followed by a gradient of 30-50% acetonitrile in water for 50 min.The volume injected was 100 µL, with UV detection at 238 nm and a flow rate of 4.0 mL.min -1 .The peaks obtained in the chromatograms were analyzed and compared to the absorption spectrum with the Microcystin-LR standard.

Extraction and quantification of saxitoxin
The saxitoxins (saxitoxin-STX, Gonyautoxins-GTX and C-toxins) were analyzed by hPLC using the preoxidation technique described by Lawrence et al. (1991Lawrence et al. ( , 1995Lawrence et al. ( , 1996) ) and Lawrence & Ménard (1991), adapted to cyanobacterial cell extraction.Toxins were extracted from the seston for 1 h in a shaker containing 0.5 M acetic acid.The supernatant was evaporated and resuspended in 1 mL 5% acetic acid.The reaction with peroxide was done in 100 µL of the extract by adding 250 µL of NaOh 1 M followed by 25 µL of 5% h 2 O 2 .The solution was stirred and allowed to react for 10 min at room temperature.Then, 20 µL of acetic acid was added to the solution, which was analyzed by hPLC.The reaction with periodate was run with a solution prepared with 0.03 M periodic acid, 0.3 M ammonium phormiate and 0.3 M Na 2 hPO 4 (1:1:1 v/v).The ph was adjusted to 8.2 with NaOh 0.2 M. 500 µL of this solution were added to 100 µL of the extract, then stirred and allowed to react for 10 min at room temperature.Finally, 10 µL of acetic acid was added and the samples were analyzed by hPLC.
The analytical equipment included a pump (LC 10 AT vP), a control (SCL-10 A VP) and an injector with 20 µL Loop.A fluorescence detector (RT-10 A XL) with 330 nm excitation and 400 nm emission wavelengths was used to monitor the effluent.An LC-18 column (15 cm x 4.6 mm, 5 µm) with partition level was used.The mobile phase had an acetonitrile gradient: 0.1 M ammonium phormiate (v/v), adjusted to ph 6.0, varying from 0-1% in the first 15 minutes and 1-4% in the last seven minutes.The flow rate was 1 mL.min -1 .

Extraction and quantification of cylindrospermopsins (CYL)
The cylindrospermopsin analysis was done according to harada et al. (1994) and hawkins et al. (1997).The extraction was obtained by adding 5% acetic acid to the samples while stirring for 1 h, followed by centrifugation for 10 min at 4000 g, followed by chromatographic analysis using an Allsphere ODS-2 µm, 250 x 4.6 mm internal diameter column (Altech) and detection at 262 nm.The retention times and UV spectrum (200-300 nm) of the major peaks were compared with cylindrospermopsin standard kindly provided by Dr. Assaf Sukenik (Kinneret Institute, haifa, Israel).A linear gradient was used from 0 to 5% methanol in deionized water for 10 min, followed by isocratic conditions of 5% methanol for another 10 min.The samples displaying similar peaks were coeluted with the standard and reanalyzed to confirm the presence of the toxin.

RESULTS AND DISCUSSION
This paper work reports on the presence of high density potentially toxic cyanobacterial cells, and the presence of microcystins and saxitoxins in raw water seston samples from the Armando Ribeiro Gonçalves reservoir, the Pataxó channel, and treated water distributed through this system.
The phytoplankton community of the Armando Ribeiro Gonçalves reservoir presented a cyanobacterial predominance of 90-100% of the total phytoplankton density in the raw water in 2000.Mixed blooms of cyanobacteria comprised almost exclusively Microcystis spp (Microcystis sp, M. protocystis, M. panniformis and M. novacekii), Cylindrospermopsis raciborskii and Aphanizomenon spp (Aphanizomenon gracile, Aphanizomenon cf.manguinii, Aphanizomenon cf.issastschenkoi), whose predominance alternated during the period under study (Table 1).The total cyanobacterial density in the raw water collected from the Armando Ribeiro Gonçalves reservoir varied from 3 to 6 log.cel/mL, corresponding to 9.08 x 10 4 and 8.2 x 10 5 cells/mL during the period in question (Fig. 1).
Our results revealed the presence of microcystins and saxitoxin.however, it was not possible to confirm the occurrence of )

TABLE 1 Predominant and subdominant cyanobacteria in the plankton samples from the Armando Ribeiro Gonçalves reservoir.
Predominance was assigned arbitrarily to one or more species that were estimated to constitute > 50% of the total phytoplankton density.
( cylindrospermopsin in any of the analyzed samples.Maximum values of 3.14 µg.L -1 of saxitoxin and 8.8 µg.L -1 of microcystin were found in the seston samples of raw water from the Armando Ribeiro Gonçalves reservoir.The highest concentrations of saxitoxins were found when Cylindrospermopsis raciborskii occurred in the highest densities during the rainy season (April/May).Blooms of Microcystis spp and microcystins were also detected in this period, lasting until July (Fig. 1).Type C1 and C2 saxitoxins were the most common, corresponding to 30-50% of all the saxitoxins found in our analyses, followed by GTX, B1 and STX (data not shown).
Total cyanobacteria density averaged 4 log.cel/mL(between 2.2 x 10 4 and 6.2 x 10 4 cells/mL) in both the Pataxó Channel raw water and in the treated water.however, the microcystin and saxitoxin concentrations were lower than the recommended levels for drinking water (Fig. 2).
Cyanobacteria density during the period of this study exceeded the levels for drinking water (< 2 x 10 3 cells/mL) recommended by the WhO (Chorus & Bartram, 1999), and also the limit set by the Brazilian health Ministry (20 x 10 3 cell/mL).The presence of numerous C. raciborskii and Microcystis spp cells in the treated water indicates serious deficiencies in the filtering system of the water treatment station.
The present study reveals that the highest concentrations of saxitoxin and microcystins appeared during the same period as the blooms of Cylindrospermopsis raciborskii and Microcystis spp, indicating the predominance of potentially toxin-producing species in the reservoir.The production of saxitoxin, microcystin and cylindrospermopsin has been demonstrated from both natural and cultured populations of cyanobacteria in many regions of the world, including Brazil (Falconer, 2001;Jochimsen et al., 1998;Azevedo et al., 1994).
In this study we observed three distinct periods of predominance of cyanobacteria species, types of toxin and toxin concentrations.The rainy season was marked by the overwhelming predominance of Cylindrospermopsis raciborskii with high levels of saxitoxin, followed by Microcystis spp in the transition period between the rainy and dry seasons, and finally the coexistence of Aphanizomenon spp, Cylindrospermopsis raciborskii and Microcystis spp with equivalent densities in the dry season.The maximum microcystin concentration (8.8 µg.L -1 ) was detected in the second half of May 2000, when Microcystis spp predominated with 90% of the phytoplankton density, and a concomitant reduction of saxitoxins.In July 2000, which represents the transition period, blooming of Microcystis spp was less intense and the level of microcystin was lower.In November (dry season) the phytoplankton was dominated by three coexisting dominating species of cyanobacteria, Cylindrospermopsis raciborskii, Microcystis spp and Aphanizomenon spp, while microcystin and saxitoxin concentrations were very low.
C. raciborskii blooms, associated with the presence of high saxitoxin concentrations, suggest the production of saxitoxins by this species, as already demonstrated in other freshwater environments in Brazil (Lagos et al., 1999 andMolica et al., 2002).however, in future researches, it will be necessary to confirm the origin of this toxin by means of saxitoxin analyses on monospecific cultures isolated from the sites under study.
The toxicological studies of Fitzgerald et al. (1999) suggested 3 µg.L -1 as a limit of saxitoxins for human consumption, based on 60 Kg weight of an individual and a daily water consumption of 2 L. This limit has already been adopted as recommended in the Brazilian health Ministry's new resolution (M/S 1.469 of 29/12/2000) in terms of water quality for human consumption.The new resolution also establishes a maximum acceptable level of microcystin of 1 µg.L -1 in drinking water.
Our results revealed peaks as high as 8.8 µg.L -1 and 0.16 µg.L -1 of microcystins in raw and treated water, respectively.This demonstrates that the local water treatment is quite efficient in removing toxins from the water.however, the occurrence of 8.8 µg.L -1 in raw water exposes to high doses of these toxins at least part of the local population that uses untreated water directly from the Pataxó Channel.This situation should be considered a serious public health threat, since prolonged exposure to microcystins can lead to a higher incidence of hepatic cancer (Chorus & Bartran, 1999;Azevedo, 1998).Exposure of the local population through accumulation of microtoxins in fish musculature must also be considered (Magalhães et al., 2001).
The presence of saxitoxins in the reservoir water during this study, as in other reservoirs in the states of Rio Grande do Norte (Costa et al., 2003a(Costa et al., , 2003b) ) and Pernambuco (Molica et al., 2002), should be taken as a warning that saxitoxins may be more widely distributed in freshwater than we were aware of, particularly in northeastern Brazil.
Based on our results, we were unable to establish a correlation between cyanobacteria densities and cyanotoxin concentration, which would have enabled us to identify the seasons with the highest and lowest risk of toxin exposure.Nevertheless, we found fewer cyanobacterial cells in the running waters of the Pataxó Channel, suggesting that the use of an adequate management technique involving physical mixing processes at the water collection points of the treatment stations in the Armando Ribeiro Gonçalves reservoir could significantly reduce the number of cyanobacteria cells.
The occurrence of toxic cyanobacteria blooms recorded in this and other studies (Costa et al., 2003a(Costa et al., , 2003b) ) points to the permanent risk of cyanotoxins in supply waters, not only in the Armando Ribeiro Gonçalves reservoir but also in other reservoirs in the state of Rio Grande do Norte.Therefore, there is an urgent need to regularly monitor these systems in order to advise the population, and to prevent and control toxic cyanobacterial blooms.Moreover, further investigations are recommended into cyanobacteria and cyanotoxin density in Rio Grande do Norte's other rivers and reservoirs.