Cyanobacterial blooms in stratified and destratified eutrophic reservoirs in semi-arid region of Brazil

This study investigated the dynamics of cyanobacteria in two deep, eutrophic reservoirs in a semi-arid region of Brazil during periods of stratification and destratification. Four collections were carried out at each reservoir at two depths at three-month intervals. The following abiotic variables were analyzed: water temperature, dissolved oxygen, pH, turbidity, water transparency, total phosphorus, total dissolved phosphorus, orthophosphate and total nitrogen. Phytoplankton density was quantified for the determination of the biomass of cyanobacteria. The data were analyzed using CCA. Higher mean phytoplankton biomass values (29.8 mm3.L−1) occurred in the period of thermal stratification. A greater similarity in the phytoplankton communities also occurred in this period and was related to the development of cyanobacteria, mainly Cylindrospermopsis raciborskii (>3.9 mm3.L−1). During the period of thermal destratification, this species co-dominated the environment with Planktothrix agardhii, Geitlerinema amphibium, Microcystis aeruginosa and Merismopedia tenuissima, as well as with diatoms and phytoflagellates. Environmental instability and competition among algae hindered the establishment of blooms more during the mixture period than during the stratification period. Thermal changes in the water column caused by climatologic events altered other physiochemical conditions of the water, leading to changes in the composition and biomass of the cyanobacterial community in tropical reservoirs.


INTRODUCTION
The dominance of cyanobacteria in aquatic ecosystems is often associated with CO 2 availability (Shapiro 1990, Caraco andMiller 1998), high temperatures (Bouvy et al. 2000, Huszar et al. 2000, Figueredo and Giani 2009, Soares et al. 2009a), low luminosity (Niklisch and Kohl 1989), high pH (Reynolds andWalsby 1975, Briand et al. 2002), high concentrations of nutrients (especially phosphorus) (Watson et al. 1997), a low N:P ratio (Smith 1983) and buoyancy regulation (Walsby et al. 1997).However, a large number of papers have demonstrated that the dominance of these algae occurs due to the synergism of a combination of these factors, particularly the bottom-up influence and physical changes (Briand et al. 2002, Marinho and Huszar 2002, Figueredo and Giani 2009, Soares et al. 2009a).The generalization of cause-and-effect in the understanding of the success of cyanobacteria in aquatic ecosystems is difficult, as these algae belong to an ancient, diversified group of species with often contrasting eco-physio-logical strategies (Marinho and Azevedo 2007).Thus, generalizations lead to divergences with regard to the cause and consequence of particular factors in the development of cyanobacteria.
There is consensus among researchers regarding the factors that best explain the dominance and blooms of cyanobacteria, such as ample nutrient availability, an increase in temperature, stratification, an increase in residence time and salinization (Paerl and Huisman 2008).However, although nutrients are an important factor, physical variables better explain changes in the composition of species in systems dominated throughout the year by cyanobacteria (Soares et al. 2009a).
Temperature exhibits vertical behavior that affects the availability of light and nutrients, and is a determinant in the seasonal dynamics of phytoplankton (Becker et al. 2009).Meteorological phenomena, such as wind and rain, play an important part in phytoplankton ecology in reservoirs (Calijuri et al. 2002) by changing the thermal pattern.The effect of temperature on the establishment of cyanobacterial blooms has been related to global warming (Bouvy et al. 1999, Briand et al. 2004, Paerl and Huisman 2008).
Different studies on the occurrence of cyanobacterial blooms under conditions of stratification and destratification reveal seasonal aspects, the influence of aeration in artificial destratification, the succession of the phytoplankton community and algal behavior under different trophic conditions of the ecosystem (Heo and Kim 2004, Antenucci et al. 2005, Bormans et al. 2005, Burford and O'Donohue 2006, Fonseca and Bicudo 2008, Benhardt et al. 2008, Chen et al. 2009).These studies stress the dominance of cyanobacteria under conditions of stratification and during the warm periods of summer.
However, divergences are encountered with regard to these findings in tropical regions.Persistent cyanobacterial blooms throughout the year are reported in shallow polymictic ecosystems or in those with a constantly isothermal water column (Bouvy et al. 1999, 2001, Crossetti and Bicudo 2008, Figueredo and Giani 2009, Bittencourt-Oliveira et al. 2011), as well as in colder periods (Soares et al. 2009b).In deep ecosystems, thermal stratification during warm months has been reported as the main cause of formation of large biomasses of cyanobacteria in reservoirs in Australia and Brazil.However, the dominance of these algae has been found to prevail in periods with a deep mixture zone (Saker andGriffiths 2001, Soares et al. 2009a).
High density and biomass of cyanobacteria have been reported for reservoirs in a semi-arid region of Brazil (Bouvy et al. 1999, 2001, Huszar et al. 2000, Chellappa and Costa 2003, Moura et al. 2007a, b, Chellappa et al. 2008, Dantas et al. 2008, von Sperling et al. 2008).However, these studies did not investigate the occurrence of cyanobacteria in relation to stratification.The aim of the present study was to investigate the composition and dynamics of plankton cyanobacteria in two deep eutrophic reservoirs (Zmax > 10 m) in the semi-arid region of Brazil during periods of thermal stratification and destratification.

STUDY AREA
The Arcoverde reservoir (8 • 33 28 S and 36 • 59 15 W) and Pedra reservoir (8 • 29 35 S and 36 • 56 39 W) are located in the municipality of Pedra in the state of Pernambuco, Brazil (Fig. 1).The climate in this region is semi-arid, with two well-defined seasons: rainy (March to August) and dry (September to February) (Melo-Júnior et al. 2007).
The Arcoverde reservoir has an area of 2.0 × 10 6 m 2 , extension of 4.2 km and accumulation capacity of 18.8 × 10 6 m 3 , with a maximal and mean depth of 24.0 m and 9.4 m, respectively.The Pedra reservoir is located upstream and has an area of 0.4 × 10 6 m 2 , extension of 1.6 km and a volume of 2.9 × 10 6 m 3 , with a maximal and mean depth of 15.0 m and 7.2 m, respectively.These reservoirs are used for the water supply of approximately 100 thousand inhabitants (SRH 2000).

MATERIALS AND METHODS
Sampling was carried out at three-month intervals between November 2007 and August 2008, with four collections per reservoir.The collection sites were established at 200 meters from the border in the central body of each reservoir.For the determination of abiotic variables and the quantitative analysis of the cyanobacteria, water samples were collected with a Van Dorn bottle at two depths: subsurface (0.1 m) and bottom (1 m above the sediment).The depth of the bottom collection ranged from 12 to 20 m in the Arcoverde reservoir, and from 9 to 13 m in the Pedra reservoir.
In situ determinations were made of water temperature ( • C) and dissolved oxygen (mg.L −1 ) using an oximeter (Schott, Handylab OX1), water transparency with a Secchi disk, maximal depth using an echosounder, turbidty (NTU) with a turbidimeter (Hanna Instruments, HI 93703) and pH with a potenciometer (Digimed, DMPH-2).The euphotic zone (Zeu) was calculated as three times the value of the Secchi disk (Cole 1994).The mixture zone (Zmix) was determined from the profile of the temperature in the water column and was equal to maximal depth (Zmax) when there was no thermal gradient of a minimal difference of 0.5 The determination of total nitrogen (μmol) and total phosphorus (μmol) was based on the procedure described by Valderrama (1981), whereas total dissolved phosphorus (μmol) and orthophosphate (μmol) were determined based on the procedure described by Strickland and Parsons (1965).Carlson's Trophic Index was used for the characterization of the ecosystems (Carlson 1977).The molar TN:TP ratio was determined based on Downing and McCauley (1992).For the estimation of limitation by nutrients, TN:TP < 20 was con-sidered for limitation by nitrogen; TN:TP > 38 was considered limitation by phosphorus (Kosten et al. 2009); [TN] < 7.0 μmol was considered nitrogen limitation; and [TP] < 0.1 μmol was considered phosphorus limitation (Reynolds 1999).
For the taxonomic and quantitative study, aliquots of 100 mL were collected in duplicate with a Van Dorn bottle and preserved in acetic Lugol's solution (Bicudo and Menezes 2006).The identification of cyanobacteria was performed to the lowest possible hierarchical level with the use of specialized literature (Komárek and Anagnostidis 1989, 1999, 2005, Komárek and Cronberg 2001).Other algal divisions were identified only on the group level.Density (ind.mL−1 ) was determined with the Utermöhl method (Utermöhl 1958).Biomass (mm 3 .L −1 ) was calculated from cell biovolume values (n = 30) based on Hillebrand et al. (1999).
Analysis of variance (ANOVA) was used to test differences in abiotic and biotic variables ( p < 0.05) between months of thermal stratification and destratification using the Statistica 2004 software (StatSoft, Inc., Tulsa, OK, USA).Canonical correspondence analysis (CCA) was used to assess associations between cyanobacteria and environmental variables.The matrix with biotic data was constructed only with species of cyanobacteria with biomass greater than 1 mm 3 .L −1 in at least one sampling month.The abiotic variables were log transformed (x) and progressively reduced using the "forward selection" routine on the Canoco 4.5 program.Significance ( p < 0.05) was tested using the Monte Carlo test, with 999 unrestricted permutations.

RESULTS
The Arcoverde and Pedra reservoirs were stratified in the dry season, and destratified in the rainy season (Fig. 2).
Nitrogen limitation (N:P ratio < 3 and [TN] < 7 μmol) occurred throughout the study.During thermal stratification, the reservoirs were anoxic in the hypolimnion, with neutral to alkaline pH and total phosphorus and orthophosphate concentrations below 8.7 μmol and 2.3 μmol, respectively.During destratification, hypoxia occurred at the bottom, with a slightly acid to alkaline pH, the reservoirs were deeper (F = 10.79,p < 0.05) and had greater concentrations of total phosphorus (F = 5.83, p < 0.05) and orthophosphate (F = 10.41,p < 0.01).No variations were found between the periods of stratification and destratification with regard to turbidity (mean: 47.8 ± 29.4 NTU) or euphotic zone (mean: 1.5 ± 0.7 m) (Table I).
There was a significant difference in phytoplankton biomass between the periods of stratification and destratification (F = 7.94, p < 0.05).Mean biomass values during stratification were 29.3 ± 30.3 mm 3 .L −1 at Arcoverde and 36.0 ± 22.5 mm 3 .L −1 at Pedra reservoirs, which were respectively 5.1-fold and 4.0-fold higher than the values found in the period of destratification.
The canonical correspondence analysis confirmed the association between the abiotic variables and cyanobacterial biomass ( p < 0.01).Axes 1 and 2 explained 72.3% of the variance in biomass.The correlation between the species and abiotic variables explained 83.7% of the canonical ordination on the first two axes (Table III).
Stratification and destratification were related to Axis 1. Sampling units related to stratification (November and February) were on the negative side of the axis, whereas those related to destratification (May and August) were on the positive side.C. raciborskii was inversely related to Axis 1 and was the most abundant taxon during thermal stratification.This organism was also associated to higher pH, a greater availability of light in the epilimnion, lesser depths and lower phosphorus values.P. agardhii was directly related to the axis and was abundant under conditions that were contrary to those that favored C. raciborskii (Table III, Fig. 4).
The second axis revealed an evident vertical variation in the period of thermal stratification and demonstrated heterogeneity between sampling months in this An Acad Bras Cienc (2011) 83 (4)   period in the Arcoverde reservoir.M. aeruginosa was associated to sampling units from the subsurface, higher pH values and lower phosphorus content.M. tenuissima was related positively to this axis and associated to sampling units from the bottom at both depths in May and higher phosphorus values (Table III, Fig. 4).

DISCUSSION
Changes in the phytoplankton structure in tropical and subtropical eutrophic ecosystems are predominantly related to thermal variations (Marinho and Huszar 2002, Heo and Kim 2004, Becker et al. 2006, Pennard et   ).The studies cited report that cyanobacteria have higher biomasses in periods of thermal stratification and are replaced by diatoms under conditions of thermal circulation.However, this is not a rule, as there are reports of cases in which cyanobacteria dominate the environment throughout the year (Crossetti and Bicudo 2008, Soares et al. 2009a).In such cases, the seasonal substitution of species of cyanobacteria occurs, with high biomass values of colonial organisms during periods of thermal stratification and filamentous cyanobacteria during the destratification phase.Nonetheless, colonial taxa form blooms under mixture conditions, and filamentous taxa occur during stratification.In ecosystems in northeastern Brazil, the dominance of a single species throughout the year has been recorded, with oscillating biomass values that may be related to variations in temperature or thermal pattern (Bouvy et al. 1999, Huszar et al. 2000, Dantas et al. 2008).
In the present study, cyanobacteria were predominant in the two eutrophic reservoirs in the semi-arid region of northeastern Brazil, regardless of the thermal condition.Greater biomasses were found under conditions of stratification, with the dominance of a single species.During periods of destratification, the structure of the phytoplankton community was formed by different species.Other algal divisions coexisted with the cyanobacteria, especially diatoms and phytoflagellates, particularly during periods of the thermal instability.
Destratification affects the availability of light and input of nutrients from the sediment (Mischke 2003).Unfavorable light conditions are disadvantageous for organisms that are obligatorily phototrophic and advantageous for mixotrophic organisms (O'Farrell et al. 2003).During the period of destratification (rainy season) in the ecosystems studied, there was an increase in the euphotic zone and a reduction in the availability of light in the water column.This facilitated the development of phytoflagellates and diatoms, which certainly competed for the niche with the cyanobacteria.
The biomass of species and composition of the community vary under conditions of stratification and destratification (Crossetti and Bicudo 2008, Dantas et al. 2008, Soares et al. 2009a, Bittencourt-Oliveira et al. 2011).In the reservoirs studied, C. raciborskii dominated during stratification, with biomasses above 3.9 mm 3 .L −1 .During destratification, this organism was replaced by P. agardhii, the biomass of which did not surpass 5.5 mm 3 .L −1 .G. amphibium and M. tenuissima exhibited greater relative abundance during thermal destratification, coexisting with P. agardhii and C. raciborskii.
The dominance of C. raciborskii occurs in strongly stratified ecosystems with a high degree of intermittence in the availability of nutrients (Bouvy et al. 2000, Marinho and Huszar 2002, Bormans et al. 2005).Although stratification was not found in other reservoirs in the semi-arid region of northeastern Brazil (Bouvy et al. 1999, Huszar et al. 2000, Chellappa and Costa 2003, von Sperling et al. 2008), the authors cited report the dominance of this species in periods with lower concentrations of nutrients, especially nitrogen.This occurs because C. raciborskii is able to store phosphorus and synthesize nitrogen, regardless of the thermal conditions.Borics et al. (2000) report the replacement of C. raciborskii by P. agardhii, with the disappearance of C. raciborskii accompanied by reductions in water temperature, biomass of the community and abundance of the cyanobacteria P. agardhii and Pseudanabaena limnetica (Lemmermann) Komárek.Fonseca and Bicudo (2008) report the dominance of Raphidiopsis/Cylindrospermopis under conditions of stratification, with the biomass of these taxa gradually reduced during destratification and the subsequent coexistence with other cyanobacteria, diatoms and chlorophytes.
Thermal stratification leads to the establishment of high biomasses of cyanobacteria (Marinho andHuszar 2002, Soares et al. 2009a), whereas destratification is related to a change in the algal structure, with the replacement of cyanobacteria by other groups of algae (Tundisi 1990, Naselli-Flores and Barone 1998, Heo and Kim 2004, Becker et al. 2006) or other species of cyanobacteria (Crossetti and Bicudo 2008, Soares et al. 2009a).However, this is not a rule, as high biomasses of cyanobacteria also occur during the mixture period (Bouvy et al. 1999, 2001, Figueredo and Giani 2009, Soares et al. 2009b).
In conclusion, the dynamics of these deep, eutrophic reservoirs in the semi-arid region of northeastern Brazil were influenced by the thermal pattern, with the structure of the algal community, behavior of the euphotic zone and availability of nutrients being similar during the period of thermal stratification.While stratification determined the dominance and high biomass of C. raciborskii, destratification hindered the formation of blooms and increased the competition among species.Despite the reduction in algal biomass during the period of destratification, the cyanobacteria were not replaced by other divisions and remained dominant throughout the year.Thermal changes in the water column caused by climatologic events alter other physiochemical conditions of the water, leading to changes in the composition and biomass of the cyanobacterial community in tropical reservoirs.

An
Fig. 1 -Geographic location of Arcoverde and Pedra reservoirs, northeastern Brazil.

TABLE I Values of variables determined at the subsurface (S) and bottom (B) of two reservoirs in the state of Pernambuco, Brazil, from November 2007 to August 2008.
An Acad BrasCienc (2011) 83 (4)