OUP user menu

Seasonal dynamics and toxicity of Cylindrospermopsis raciborskii in Lake Guiers (Senegal, West Africa)

Céline Berger , Ngansoumana Ba , Muriel Gugger , Marc Bouvy , Filippo Rusconi , Alain Couté , Marc Troussellier , Cécile Bernard
DOI: http://dx.doi.org/10.1111/j.1574-6941.2006.00141.x 355-366 First published online: 1 September 2006


Cylindrospermopsis raciborskii is a toxic bloom-forming cyanobacterium that occurs at tropical and temperate latitudes. Despite several reports from Africa, no data were previously available about its dynamics or toxic potential there. We therefore carried out a 1-year survey of the dynamics of C. raciborskii in the main water reservoir in Senegal, Lake Guiers. Cylindrospermopsis raciborskii never formed a bloom in this lake during the period studied, but was dominant during the dry season. The only observed bloom-forming species was a diatom, Fragilaria sp., which displayed a seasonal pattern contrary to that exhibited by C. raciborskii. Principal component analysis applied to environmental and phytoplankton data showed that high C. raciborskii biomasses were mainly related to high temperature and water column stability. Tests for C. raciborskii species-related toxicity and/or toxin synthesis were performed on 21 isolated clones. All the strains isolated tested negative in mouse toxicity bioassays, toxin analysis (MS/MS) and tests for known cylindrospermopsin genes (ps, pks). The limited number of isolates studied, and the occurrence of toxic and nontoxic clones in natural cyanobacterial populations, mean that we cannot conclude that there is no C. raciborskii-associated health risk in this drinking water reservoir.

  • Cylindrospermopsis raciborskii
  • cyanobacteria
  • dynamics
  • toxins
  • reservoir
  • Africa


Over the past decade, a growing number of studies have investigated the ecological and toxicological characteristics of a cyanobacterial species Cylindrospermopsis raciborskii. This was for two reasons: (i) this species is increasingly spreading into different aquatic ecosystems (Padisák, 1997; Briand et al., 2004), and (ii) it is able to produce a wide range of toxins (Bernard et al., 2003). The ecological success of this species appears to be related to its ability to tolerate a rather wide range of climatic conditions. Increasing colonization of mid-latitudes by C. raciborskii may result from the global warming phenomenon, which is providing this species with more favorable environmental conditions for its growth (Briand et al., 2004). However, perennial blooms of Cylindrospermopsis have only been reported in tropical areas (e.g. in Brazil, Bouvy et al., 2001), whereas in sub-tropical and temperate areas, this species displays seasonal dynamics (Mc Gregor & Fabbro, 2000). Cylindrospermopsis raciborskii is able to produce a wide range of toxins, such as cylindrospermopsin (CYN, a hepatotoxic cyclic peptide) in Australia (Hawkins et al., 1985; Mc Gregor & Fabbro, 2000; Saker & Griffiths, 2000), USA (Carmichael, 2002) and Asia (Li et al., 2001; Chonudomkul et al., 2004), saxitoxins (STXs, neurotoxic alkaloids) in South America (Lagos et al., 1999; Pomati et al., 2004), and an unidentified hepatotoxin (CYN-like) in Europe (Bernard et al., 2003; Fastner et al., 2003).

Cylindrospermopsis is distributed worldwide, but few studies of this genus in Africa have been reported (Padisák, 1997; Cogels et al., 2001; Komárek & Komárková, 2003; Bouaïcha & Nasri, 2004; Berger et al., 2005; Gugger et al., 2005b). Morphological evidence and biogeographic comparisons have suggested that Cylindrospermopsis may have originated in this continent (Padisák, 1997), and recent genetic data reveal a close relationship between some Senegalese Cylindrospermopsis raciborskii isolates and toxic Australian strains (Gugger et al., 2005b). However, the dynamics and toxicity of Cylindrospermopsis in African freshwater bodies are of particular importance, because freshwater is a scarce resource, and toxic cyanobacterial development could restrict its use.

Lake Guiers constitutes the largest freshwater reservoir in Senegal, and supplies drinking water for the capital, Dakar (>2 300 000 inhabitants in 2000), as well as for numerous local villages. Recent hydraulic managements, in particular the construction of the Diama dam on the Senegal River, have stopped the incursions of seawater, and stabilized the water level in Lake Guiers (Cogels et al., 2001). The consequences of such changes have been a shift in the composition of the phytoplankton community, and the development of cyanobacteria such as C. raciborskii (Cogels et al., 2001). This phenomenon was therefore studied during a multidisciplinary program conducted in the central zone of the Lake Guiers from March 2002 to March 2003.

In this article, we report the seasonal dynamics of C. raciborskii in Lake Guiers taking into account changes in the phytoplankton community and in the environmental context of this sub-Saharan region. Toxicological properties, toxins, and ps and pks genes associated with CYN production in Australian strains, were investigated for 21 C. raciborskii strains isolated from Lake Guiers during the surveyed period.

Materials and methods

Study area and field sampling

The Senegal River arises in Guinea and flows via Mali, Mauritania and Senegal to the Atlantic Ocean, which corresponds to a large watershed (≈290 000 km2). The Northern Senegal region has a Sahelian climate, with very irregular rainfall during the two to three months from June to August (200–250 mm per year). The rainy season combined with the large watershed results in a substantial flood period lasting till October.

Lake Guiers is the main lake in Senegal (between 16°39′ N and 15°N) with a surface area of 300 km2, a volume of 600 million m3, and an average depth of 2.0 m (Fig. 1). This shallow depth prevents the periodic thermal stratification of the water column except for a few hours in the middle of the day (data not shown). The lake constitutes the main freshwater reservoir in Senegal, and is considered to be a vital national resource, especially as the source of drinking water for the city of Dakar (Varis & Fraboulet-Jussila, 2002). Water is pumped from the central part of the lake by the N'Gnith water treatment plant, and then piped to Dakar through a 160 km long underground pipe. The mean daily throughput is around 2000 m3 h−1.


Map of Lake Guiers showing the location of the sampling site at N'Gnith.

The lake is connected via the Taoué channel to the Senegal River 150 km from the estuary mouth. The water level in Lake Guiers is controlled by two dams on the Senegal River: the Diama dam built in 1985, which is also intended to prevent incursions of seawater, and the Manantali dam built in 1987 (Albergel et al., 1993). Two annual periods can be distinguished in terms of hydrology: (1) a filling period between August and October, during which water flows from the Senegal River into the lake, and (2) a 9-month isolation period when little water is transferred between the river and the lake (Cogels & Gac, 1993).

Between March 2002 and March 2003, water samples (n=35) were collected on average three times per month from 0.5 m under the surface using a 2 L vertical Niskin bottle. The samples were taken from the central part of the lake (N'Gnith village), at a station located near the pumping station of the water treatment plant (16°19′ N, 15°90′ W) (Fig. 1).

Environmental variables

Water temperature and conductivity were measured in situ using a thermistor probe (Optic StowAway temp; ONSET, Bourne, MA) and a Tacussel CD78 conductimeter (Imlab Sarl, Lille, France), respectively. Water samples for determinations of the nutrient content, chlorophyll-a concentration, and phytoplankton community analyses were collected using a vertical Niskin bottle. Samples for the dissolved inorganic nutrient determinations (NO3-N, NH4-N, PO4-P) were filtered through Whatman GF/F fiberglass filters, stored at −20°C and further analyzed according to Strickland and Parsons (1972). Water transparency (light penetration) was estimated using a black-white Secchi disk (20 cm diameter). Irradiance and wind parameters were measured using a DAVIS meteorological station (Hayward, CA) attached to the jetty of the N'Gnith water treatment plant. The monthly mean solar radiation received through the water column was calculated after taking into account the irradiance, the depth of the water column and the light extinction. The mean wind direction was calculated monthly, including a distinction between the four cardinal wind directions (north–south and east–west components). Positive values indicated wind from the northerly and easterly directions, whereas negative values indicated wind from the southerly and westerly directions.

Phytoplankton variables

The chlorophyll-a concentration was determined fluorometrically after methanol extraction (Yentsch & Menzel, 1963). For phytoplankton determination, 50 mL of water were fixed immediately with Lugol's iodine. Phytoplankton species were identified using an inverted microscope (Olympus CK 40) on the basis of descriptions in the literature (Lauterborn, 1915; Skuja, 1948, 1956; Grönblad et al., 1958; Couté & Rousselin, 1975; Compère, 1991; Komárek & Anagnostidis, 1998). Taxon abundances were determined by the Utermöhl (1958) method (5–10 mL volume, 48 h sedimentation, at least 400 phytoplankton cells counted). Specific biovolumes were estimated according to formulae reported by Dia & Reynaud (1982) and Sun & Liu (2003). The wet weight biomass was computed from the abundance and specific biovolume estimates, assuming that the phytoplankton cells had a density of 1 g cm−3.

Species richness was measured as the number of taxa per sample. Phytoplankton diversity was expressed by the Shannon index (Shannon & Weaver, 1963), based on the abundance of each species.

Relationships between environmental and phytoplankton variables were studied by principal component analysis (PCA) using the R software version 4 (Casgrain & Legendre, 2001). PCA was performed on the correlation matrix instead of the covariance matrix in order to standardize the descriptors. Cluster classification of observation scores from the two first factorial planes of the PCA analysis (Ward's criterion aggregation) was performed to characterize the samplings. Linear regressions were performed using the StatView statistical package (SAS Institute, NC).

Isolation of strains and culture conditions

In May and December 2002, specific samples were collected with a 20 μm mesh size net in order to isolate clones of Cylindrospermopsis. Trichomes of Cylindrospermopsis were isolated under an inverted microscope and incubated at 25°C on a Z8 minus nitrogen (Kotai, 1972) solid medium (7 g L−1 of agar), with 16 : 8 h light : dark cycle under a constant solar radiation of 20 μE cm−2 s−1 during the light period. 21 isolates were obtained and maintained in solid and liquid Z8 medium minus nitrogen under the same conditions in the Paris Museum Collection (PMC) (Table 1).

View this table:

Cylindrospermopsis strains used in this study

SpeciesStrainMorphotypeGeographic originToxicity/Toxin content
C. curvisporaPMC 144.02CoiledGuiers (Sn) May 2002Not toxic/NTThis study
PMC 184.03CoiledGuiers (Sn) May 2002nd/NTThis study
C. raciborskiiPMC 115.02StraightGuiers (Sn) May 2002Not toxic/NTThis study
PMC 116.02FlexuousGuiers (Sn) May 2002Not toxic/NTThis study
PMC 117.02FlexuousGuiers (Sn) May 2002Not toxic/NTThis study
PMC 118.02StraightGuiers (Sn) May 2002Not toxic/NTThis study
PMC 139.02StraightGuiers (Sn) May 2002nd/NTThis study
PMC 140.02StraightGuiers (Sn) May 2002nd/NTThis study
PMC 141.02StraightGuiers (Sn) May 2002nd/NTThis study
PMC 142.02StraightGuiers (Sn) May 2002nd/NTThis study
PMC 143.02FlexuousGuiers (Sn) May 2002nd/NTThis study
PMC 145.02StraightGuiers (Sn) May 2002nd/NTThis study
PMC 146.02FlexuousGuiers (Sn) May 2002nd/NTThis study
PMC 217.03StraightGuiers (Sn) Dec 2002nd/NTThis study
PMC 218.03StraightGuiers (Sn) Dec 2002nd/NTThis study
PMC 219.03StraightGuiers (Sn) Dec 2002nd/NTThis study
PMC 220.03StraightGuiers (Sn) Dec 2002nd/NTThis study
PMC 221.03StraightGuiers (Sn) Dec 2002nd/NTThis study
PMC 222.03FlexuousGuiers (Sn) Dec 2002nd/NTThis study
PMC 223.03FlexuousGuiers (Sn) Dec 2002nd/NTThis study
PMC 224.03FlexuousGuiers (Sn) Dec 2002nd/NTThis study
PMC 00.01StraightJucazinho (Brazil)Neurotoxic/STXBernard (2003)
AWQC CYP-023JStraightBourke (Australia)Hepatotoxic/CYNTerao et al. (1994) Falconer et al. (1999) Seawright et al. (1999) Bernard (2003)
PMC 99.06StraightEpazote (Mexico)Not toxic/NTBernard (2003)
  • C.: Cylindrospermopsis; STXs: saxitoxins; CYN: cylindrospermopsin.

  • * Culture collections: PMC, Paris Museum Collection, C. Bernard of the Muséum National d'Histoire Naturelle, Paris, France; CYP, P. Baker of Australian Water Quality Center, Adelaide, Australia.

  • nd, not determined; NT, no toxin detected.

  • named Cylindrospermopsis africana in Gugger (2005)

Toxin content and toxicity

All the Cylindrospermopsis isolates (n=21) were screened for the production of CYN, microcystins, STXs and anatoxin-a by mass spectrometry (Q-Star Pulsar i Applied Biosystems). 25 mL of culture was centrifuged. The pellets were extracted in 5 mL of water acidified with 1% formic acid (pH 2) for 1 h with orbital stirring, sonicated for 10 min, and then centrifuged for 10 min at 3220 g. The supernatant was filtered on 0.22 μm filters (Analypore, Labosi). Toxins were then enriched on C18 solid phase extraction cartridges (3 mL, 500 mg, Waters). The cartridge was moistened with 3 mL of methanol, and equilibrated in 3 mL of milli-Q water. The following step gradient was applied: 3 mL of acidic water (1% formic acid), 3 mL of acidic methanol 20% (1% formic acid), 3 mL of acidic 80% methanol (1% formic acid). The flow-through was screened for STXs, the first and second fractions for CYN and anatoxin-a, and the last fraction for microcystins. Reference strains including C. raciborskii PMC 00.01 and AWQC CYP-023J (Table 1), which produce STXs and CYN respectively (Bernard et al., 2003), as well as Phormidium favosum PMC 240.04, which produces anatoxin-a (Gugger et al., 2005a), and Planktothrix agardhii PMC 75.02, which produces microcystins, were used to confirm that the methodology described here resulted in appropriate extraction of the toxins and their characterization by tandem mass spectrometry. Mass spectra analyses were performed as described in Gugger (2005)a.

To check for the presence of unknown toxic compounds, five Cylindrospermopsis isolates were tested by mouse bioassays (Table 1). The assay was performed as described by Bernard et al. (Bernard et al., 2003) with an intraperitoneal injection (i.p.) of extract containing 5, 10 or 20 mg of lyophilized cells per mL. Intraperitoneal injection with a 0.9% aqueous solution of NaCl was used as the negative control. The mice were observed for a period of 48 h. Necropsy was performed immediately after death or after spinal dislocation of the survivors at the end of the 48 h observation period. Specified organs (liver, spleen, kidneys and thymus) were collected, and then fixed for further histopathology studies as described by Bernard (2003).

Specific amplification of the genes implicated in toxin production

We looked for the 16S rRNA gene, ps and pks genes in each isolate. Their amplification was combined in two multiplex PCRs amplifying the 16S rRNA-ps genes and 16S rRNA-pks genes, respectively. The multiplex PCRs were performed using cyanobacteria-specific primers: cya 106, cya 781, M13, M14, M4 and K18 (Nübel et al., 1997; Schembri et al., 2001; Fergusson & Saint, 2003). Reactions were performed in a volume of 25 μL containing 1 μL of clonal culture, 200 μM of deoxynucleoside triphosphate, 20 μM of each primer (cya 106, cya 781 with M13, M14 or M4, and K18), 2.5 μL 10X PCR buffer and 2.5 mM of magnesium chloride. The template was frozen at −20°C, and then boiled for three 1 min periods before adding 1 U Taq polymerase (Eppendorf, Germany). The thermal-cycling conditions were 5 min denaturing at 94°C; 35 cycles of 30 s denaturing at 94°C, 30 s annealing at 57°C, and 30 s extension at 72°C, followed by 5 min elongation at 72°C. The amplified products were checked on 1.5% agarose gel stained with ethidium bromide.


Phytoplankton dynamics in Lake Guiers

Changes in the chlorophyll-a concentration, and of the total abundances and biomasses of the phytoplankton, were observed at the N'Gnith station from March 2002 to March 2003 (Fig. 2). Phytoplankton biomass reached a minimum (4.75 mg L−1) in May 2002 and a maximum (41.63 mg L−1) in January 2003 (Fig. 2). The chlorophyll-a concentration was less closely correlated to the two other phytoplankton parameters (r≤0.549, P<0.0006); the mean chlorophyll-a concentration was 23.1 μg L−1 (CV%=46.5), the lowest value (<10 μg L−1) was observed in March and April 2002, and the highest value (42.8 μg L−1) in December 2002.


Changes in chlorophyll-a levels, and in the abundance and the total biomass of phytoplankton from March 2002 to March 2003 in Lake Guiers.

The phytoplankton community was composed of 78 taxa belonging to 34 genera within six taxonomic groups: the Bacillariophyceae, Chlorophyceae, Cyanobacteria, Dinophyceae, Euglenophyceae and Xanthophyceae. The Chlorophyceae were the most diverse group, with 43 species (55% of the total species richness), followed by the Cyanobacteria with 25 species (32%), and the Bacillariophyceae with 4 species (5%). Xanthophyceae, Euglenophyceae and Dinophyceae species were not detected all year round, and the six species belonging to these genera (8%) were pooled for counting.

The Chlorophyceae comprised the most diverse group, but they only accounted for 13% of the total biomass. Cyanobacteria and Bacillariophyceae dominated the phytoplankton community, averaging 53.5% and 30% of the total biomass, respectively.

Among the Bacillariophyceae, Fragilaria sp. was dominant from March to mid April 2002 and from November to the end of February 2003 averaging 29.8% of the total phytoplankton biomass (CV%=92.2) (Fig. 3a). The lowest values were observed between June and September 2002 (ranging from 0% to 6%). The highest values were recorded between October and February 2003 with 82.5% in December (mean=56.9% of total biomass over this period). The most abundant cyanobacterial species was Cylindrospermopsis raciborskii (mean=27.1% of total phytoplankton biomass, CV%=60.1) and Lyngbya versicolor (mean=5.7%, CV%=55.6). The biomass of Cylindrospermopsis remained relatively constant during the period studied, but this species dominated from mid April to October, when Fragilaria sp. was not detected. During this period of dominance, the mean C. raciborskii biomass constituted 41.3% of the total biomass, reaching a peak level of 58.7% in April.


Changes in (a) the biomasses of C. raciborskii and Fragilaria sp., (b) the water temperature and the solar radiation, (c) the wind direction and velocity from March 2002 to March 2003 in Lake Guiers.

Thus, during the period studied, the phytoplankton community was successively dominated by Fragilaria sp. and C. raciborskii. Only the biomass reached by the diatom in winter can be considered to have been a bloom (Fig. 3a). This did not affect the species richness of the community (mean=38.3; CV%=15.7), but was matched by a consistent decrease in the Shannon diversity index (mean=2 bits per individual; bit ind−1; CV%=20.2), which reached its lowest value (0.96 bit ind−1) in January 2003 (Fig. 4).


Variations of the Shannon index and the species richness of the phytoplankton community from March 2002 to March 2003 in Lake Guiers.

Physico-chemical variables

Temperatures ranged between 19.7°C (end of January 2003) and 30.3°C (October 2002) (Fig. 3b). The monthly mean solar radiation intensity (Fig. 3b) indicated that there were two contrasting periods: (i) a high solar radiation period extending from March to October 2002 (≥7 mol quanta m−2 day−1) which corresponded to a steady increase in temperature and (ii) a period of lower solar radiation extending from November 2002 to January 2003 (7 mol quanta m−2 day−1), during which the temperature started to decrease.

The wind also displayed two contrasting periods during the survey: (i) predominantly north-east continental trade winds from November to April and (ii) predominantly north-west marine trade winds frequently alternating with southerly winds from the end of April to the end of October (Fig. 3c). As the lake is oriented from north-east to south-west, and has a relatively high fetch, continental trade winds produce thorough mixing of the entire water column, whereas the milder maritime trade winds produce less wind-mixing.

Conductivity values (Fig. 5a) showed very little year-round change, ranging from 171 to 205 μS cm−1 (mean=181 μS cm−1, CV %=5.2%). Secchi disk values (Fig. 5a) were low, averaging 46 cm (CV %=21.5) with the highest values (>65 cm) observed between June and July, which corresponded to the flooding period of the Senegal River. The lowest values (32 and 33 cm) were observed just before and just after this period, coinciding with the highest values of phytoplankton biomass (Fig. 2).


Variations in physicochemical variables: (a) conductivity and Secchi depth, (b) ammonium and phosphate concentrations from March 2002 to 2003 in Lake Guiers.

Ammonium concentrations averaged 0.65 μM, with considerable variation during the period studied (CV%=133.2). The highest value (around 4 μM) was detected during the flood period (on 25 June 2002, Fig. 5b). Phosphate concentrations were very low (Fig. 5b), and always less than 0.3 μM (mean=0.1 μM, CV%=105.4). Minor phosphate peaks were observed from April to July, in September and in December. However, the concentrations were generally below the quantification threshold.

Relationships between the phytoplankton and environmental variables

A PCA of the environmental and phytoplankton variables was performed on the whole data sets (35 samplings, nine environmental and eight phytoplankton variables). The first three axes of the PCA accounted for 72% of the total variance. Projection of the environmental and phytoplankton variables in the reduced space formed by the first two axes (axis 1/axis 2) (Fig. 6a) revealed a positive correlation between the total abundance and Fragilaria sp. biomass on the first axis. This axis also displayed an inverse relationship between the biomasses of C. raciborskii and Fragilaria sp. The Shannon index was inversely related to the dominance of Fragilaria sp. exhibiting a decrease in phytoplankton diversity while this species was dominant. Temperature was positively correlated to levels of the two main cyanobacteria, especially to the Cylindrospermopsis biomass, and was inversely correlated to the Fragilaria sp. biomass. This axis also revealed that the biomass of Fragilaria sp. was inversely related to the solar radiation values, and clearly associated with the easterly trade winds. Conversely, the presence of C. raciborskii was explained by the predominantly westerly trade winds.


PCA analysis. (a) Positions of the nine environmental and the eight biological variables in the plane of the first two principal axes. (b) Positions of the 35 samples in the reduced space of the first two principal components, with identification of two groups of samples (Ward's criterion aggregation). The two first axes accounted for 72% of the total variability. Legend: letters indicate the months, and numbers indicate the different samplings within the each month.

The second axis described the nutrient context (Fig. 6a). The dissolved nutrients (NO3, NH4 and PO4) could not account for the dynamics of the dominant phytoplankton species. Secchi disk and conductivity measurements also contributed little to explain the presence or absence of any of the dominant algal species.

Analysis of the reduced space formed by axes 1 and 3 confirmed the results found during the first analysis, with a close correlation between temperature, irradiance, wind direction and the presence of dominant phytoplankton species (data not shown). Axis 3 revealed the inverse relationship between the concentrations of NH4 and PO4 on the one hand, and the concentrations of NO3 and conductivity levels on the other. Ward's aggregation, based on the cluster analysis of the PCA scores for the samples, clearly identified two groups (Fig 6b). The first group comprised the samples taken between mid October (O2) and April (A1), when phytoplankton biomass concentrations were high, and Fragilaria sp. was especially dominant. The second group, from April (A2) to October (O1), was characterized by the dominance of C. raciborskii, especially during June and July 2003. These data indicated a close parallel between the two dominant phytoplankton groups and the environmental conditions, the phytoplankton biomass being directly related to temperature, solar radiation intensity and wind direction rather than to nutrient availability. To further analyze the relationships between the biomasses of C. raciborskii, Fragilaria sp. and these three variables, linear regression models were fitted to these data (Table 2). The C. raciborskii biomass was significantly directly correlated to temperature, and inversely correlated to easterly and northerly wind velocities. In contrast, Fragilaria sp. biomass was significantly inversely correlated to temperature and solar radiation intensity, and positively correlated to easterly wind velocity.

View this table:

Pearson's correlation values (r, associated probabilities) between Cylindrospermopsis raciborskii, Fragilaria sp. biomasses and some environmental variables

Water temperatureSolar radiation intensityEast wind velocityNorth wind velocity
C. raciborskii+0.578 (0.0002)ns−0.500 (0.0019)−0.4409 (0.0076)
Fragilaria sp.−0.7500 (<0.0001)−0.542 (0.006)+0.754 (<0.0001)+0.380 (0.0236)
  • * ns: not significant

Screening for toxins and toxicity of Cylindrospermopsis strains

Twenty-one Cylindrospermopsis clones were isolated from Lake Guiers at times when this genus was dominant (May 2002, n=13) and at others when it was not (December 2002, n=8) (Table 1). Three different morphotypes of the genus Cylindrospermopsis were observed: straight and flexuous trichomes belonging to the species C. raciborskii, and coiled trichomes belonging to the species C. curvispora. The detailed taxonomic description is available in Berger (2005). All Cylindrospermopsis strains were checked for the presence of anatoxin-a, CYN, microcystins and STXs by mass spectrometry. No known toxin has been detected.

Genes known to be involved in CYN production were also searched in each Cylindrospermopsis isolate, because of their close relationship with Australian CYN-producing Cylindrospermopsis strains, and the possibility that our laboratory conditions could be unfavorable for the production of this toxin. Thus, the 16S RNA gene, which acts as a control of the amplification reaction, was amplified in each isolate, but no amplicon was obtained for ps or pks genes.

Five of the 21 Cylindrospermopsis strains, representing the various morphotypes (C. raciborskii PMC 115.02 and PMC 118.02 (straight morphotype), C. raciborskii PMC 116.02 and PMC 117.02 (flexuous morphotype) and C. curvispora PMC 144.02 (coiled morphotype), were tested for global toxicity using a mouse bioassay (Table 1). During the 48 h observation period, no death, and no symptom of any toxicity was observed vs. hepatotoxic and neurotoxic controls. Histopathology studies revealed no macroscopic or microscopic lesions when compared to control (mice injected with 0.9% NaCl).


Lake Guiers plays a crucial role in Senegal, where it is the main freshwater resource of the country, and is used for diverse purposes, notably to provide drinking water for the city of Dakar. Hydrodynamic management of the Senegal River, based on dam constructions, has made it possible to stabilize the water level in the lake, and to avoid the incursion of seawater (Cogels et al., 2001). These hydrodynamic and hydrological changes have improved the quantitative water control, but have also induced marked changes in the abiotic and biotic components of the ecosystem (Cogels et al., 2001), as had already been observed in other impacted reservoirs (Naselli-Flores & Barone, 2005). Microorganisms, and especially phytoplankton communities, appeared to be the organisms most sensitive to various changes in environmental conditions (Paerl et al., 2003). At least two noxious responses of phytoplankton communities have now been well documented: (i) dystrophic crisis and oxygen depletion due to a large increase in biomass and (ii) the occurrence of toxic species, especially cyanobacterial species. The worst possible outcome in terms of health is the development of blooms of toxic species, which precludes most of the normal uses of the water.

Before and just after the construction of the Diama dam, the maximum chlorophyll-a concentration was about 25 μg L−1 (Dia & Reynaud, 1982; Cogels et al., 2001). Our study shows that there has been a subsequent increase in the phytoplankton biomass, with higher values of chlorophyll-a (mean=23.1 μg L−1, max=42.8 μg L−1). This increase confirms an eutrophication tendency in Lake Guiers due to several factors, such as expanding agriculture (and particularly of irrigation) and other pressures linked to a rapidly increasing population (Varis & Lahtela, 2002).

Since the Diama dam has been built, there have been notable changes in the composition of the phytoplankton community composition: (i) a severe decrease in the species number of Bacillariophyceae [66 sp. (Compère, 1991) vs. 4 sp. in this study], (ii) a more limited change in the number of cyanobacteria (11 sp. vs. 25 sp.) and Chlorophyceae (39 sp. vs 43 sp.), (iii) a change in taxon composition, in particular in the cyanobacterial group (Compère, 1991; Berger et al., 2005), and (iv) the emergence and development of the potentially toxic cyanobacterial species Cylindrospermopsis raciborskii (Cogels et al., 2001; Berger et al., 2005).

These changes may be linked to the hydrological modifications induced by the dam, such as water level stability or the elimination of the intrusion of saline water, leading to greater diversity of ecosystem types. Previously, brackish water may have allowed brackish phytoplankton species to become established (Compère, 1991), but may also have precluded the occurrence of some other salt-sensitive species, such as C. raciborskii (Cogels et al., 2001; Berger et al., 2005).

The dynamics of C. raciborskii appear to be seasonal (maximum biomass during the spring – summer period, minimum biomass during the winter period). Although it may be a dominant species, constituting up to 58.7% of the total phytoplankton biomass from April to October, it never formed a bloom during the period studied. The only blooming species observed during the survey was a diatom, Fragilaria sp. This species displayed an opposite temporal pattern that contrasted with that exhibited by C. raciborskii, with its greatest biomass occurring in winter and its lowest biomass in summer. Contrary to other studies, we did not find that the dominance of Cylindrospermopsis in Lake Guiers was linked to any decrease in diversity (Bouvy et al., 2000; Dokulil & Teubner, 2000).

Similarly to other Cylindrospermopsis surveys conducted in tropical climates (Padisák, 1997; Bouvy et al., 2000; Mc Gregor & Fabbro, 2000), the increase of C. raciborskii biomass was mainly linked to high temperature. In vitro studies have shown that the optimum temperature for C. raciborskii growth is around 30°C, and that net growth increases between 20 and 35°C (Saker & Griffiths, 2000; Briand et al., 2004). In contrast, Fragilaria sp. can grow at lower water temperatures (Butterwick et al., 2005). The coefficient of determination (r2, Table 2), fitted to the C. raciborskii biomass vs. temperature relationship, showed that 33.5% of the change in the C. raciborskii biomass could be explained by changes in temperature. In comparison, temperature changes accounted for 56% of the variation in Fragilaria sp. biomass. High light values also appeared to have an adverse effect on the latter species, as illustrated by the significant negative correlation between biomass and solar radiation intensity. The absence of correlation between C. raciborskii biomass and solar radiation intensity may indicate that this species can tolerate (growth or/and survival) various irradiance conditions well, as recorded by both in vitro and in situ studies (Dokulil & Mayer, 1996; Fabbro & Duivenvoorden, 1996; Briand et al., 2004).

The period when the temperature was above 25°C also corresponded to a drastic change in wind direction, which may have resulted in reduced turbulence (i.e. greater stability) of the water column. Indeed, from May to November, the marine trade winds blowing from the east dominate, and induce greater stability of the water column. This environmental condition, combined with high temperatures, favors the maintenance of C. raciborskii, as already suggested by Paerl (1988) and Padisák (1997). In a subtropical reservoir, Havens (1998) concluded that the driving force in phytoplankton succession was the underwater irradiance, which is ultimately determined by wind, solar radiation and the thermal stability of the water column. In two Brazilian reservoirs, Bouvy (2003, 1999) have clearly demonstrated that climate changes (such as droughts) can alter the physical conditions, thus creating ideal conditions of temperature and irradiance for the dominance of Cylindrospermopsis sp.

In view of the probably multifactorial nature of the seasonal dynamics of C. raciborskii, the more intriguing question is: why did C. raciborskii not form a bloom during the period studied, whereas other species were able to bloom? The only hypothesis we can formulate is based on the fact that the occurrence of a bloom requires the correct timing of several environmental conditions, and that this did not occur for C. raciborskii in Lake Guiers during the period studied. When physical conditions (i.e. temperature, solar radiation, wind direction, water column stability) were the most favorable for the growth of C. raciborskii, the concentration of inorganic nutrients, especially of phosphorus, was very low, and thus may have prevented C. raciborskii from forming a large biomass. In contrast, favorable physical conditions for Fragilaria sp., including a lower temperature and a well-mixed water column, were caused by the winter wind regime. From November to April, continental trade winds (north-westerly winds) strongly mixed the entire water column due to the orientation of the lake (north-west to south-west), which results in a high level of fetch. This may have allowed diatoms to be exposed both to enough light (at the surface) and enough nutrients (at the bottom) to ensure the formation of large biomass during winter. This survey is the first study of the seasonal dynamics of the genus Cylindrospermopsis in sub-Saharan Africa, and reveals that the main controlling factors are linked to the seasonal meteorological changes.

In a previous study based on genetic data, Gugger (2005b) showed a close relationship between three Senegalese strains and toxic Australian strains. None of the Senegalese strains tested displayed any toxicity or known toxins. To confirm this nontoxicity, or at least the absence of any known cyanotoxins, 21 Cylindrospermopsis clones were isolated during the periods of dominance and non-dominance of this cyanobacterium in Lake Guiers. None of these strains produced any detectable toxin.

However, other studies have shown that cyanobacterial blooms are composed of both toxic and nontoxic clones (Hisbergues et al., 2003; Kurmayer & Kutzenberger, 2003; Kurmayer et al., 2004, 2005). A possible hypothesis to explain the absence of toxic clones in our samples may be that toxicity may confer an adaptive advantage under adverse conditions. When such adverse conditions are not present, the proportion of toxic clones can remain low. In Lake Guiers, Cylindrospermopsis is not exposed to high grazing pressure, due to the absence of potential grazers such as zooplanktonic species (Ka et al., 2006). If toxicity is a response to limit grazing pressure, this means that toxic clones will not be selected in Lake Guiers, and so may constitute only a small proportion of the Cylindrospermopsis population.

The limited number of clones studied may not be representative of the diversity of Cylindrospermopsis populations from Lake Guiers. The other potential weakness of our sampling strategy is that we only sampled the C. raciborskii population during two different periods and at one station. The distribution of toxic clones may display both temporal and spatial heterogeneity, as has been reported for the genus Microcystis (Kurmayer & Kutzenberger, 2003), and consequently a more intensive temporal and spatial sampling design may be required to detect their presence. An interesting way of avoiding the difficulty of culturing rare strains, and the time-consuming isolation-based approach, would be to use molecular tools to monitor bloom-forming cyanobacteria. This would involve screening total DNA from in situ populations in order to identify specific genes linked to CYN-production (Fergusson & Saint, 2003).

The persistent presence and/or dominance of C. raciborskii in Lake Guiers is worrying in the light of the use of the Lake by local populations. Due to its ability to produce toxins, this species has been blamed for human poisoning (Hawkins et al., 1985) and mortality amongst cattle (Saker et al., 1999). The present study did not identify any C. raciborskii bloom, nor were any toxic strains isolated, but we cannot exclude the possibility of either an increasing adaptation of this species to the environment of Lake Guiers or the development of a toxic ecotype. A cyanobacterial-health-risk survey of the Lake Guiers must include at least a conventional count of this species, but could also include regular screening of toxin genes using simple molecular methods.


This study was funded by the Institut de Recherche pour le Développement, Research unit 098, the Délégation Générale pour l'Armement (No 026065094), as was Céline Berger (grant N°026069094); both are gratefully acknowledged. The authors would like to thank S. Krys, R. Biré (AFSSA, Paris, France) and J.J. Fontaine (ENVA, Paris, France) for performing mouse toxicity experiments and analysis. We also thank S. Leibe-Magnin and C. Duval for technical and culture assistance. We are grateful to Dr Robert Arfi for the meteorological data and valuable discussion of the multivariate analysis. We also thank both reviewers for their comments.


View Abstract