496 MONITORING A TOXIC CYANOBACTERIA BLOOM IN LAKE BOURGET (FRANCE) AND ITS CONSEQUENCES FOR WATER QUALITY Jean-F. Humbert1*, Gérard Paolini2 and Brigitte Le Berre1 1. INRA, Station d’Hydrobiologie, BP 511, 74203 Thonon-les-Bains cedex, France. 2. Station d'Hydrobiologie du lac du Bourget, 73103 Aix-les-Bains, France. ABSTRACT This work describes the appearance of a toxic cyanobacterial bloom in Lake Bourget (France) in the winter 98/99 and its effect on the quality of the water that supplies two population centers. The horizontal distributions of the Planktothrix rubescens filaments were very similar in the three sampling stations and the vertical distributions were homogeneous from 0 to 140 m depth after complete mixing of the lake (January 1999). Before that, P. rubescens filaments were mainly located in the epilimnion and metalimnion. A microcystin-RR was identified by HPLC analysis. The RRmicrocystin concentration in the lake was higher than 4 µg l-1 at the maximum of the bloom, while the concentrations in drinking water were always below 1 µg l-1. This event shows that a large decrease in phosphorus concentration in the lake over the past 10 years, and pumping deep water (30 m) for water supply are not sufficient to avoid problems with cyanobacteria. INTRODUCTION
liters) were taken at depths of 3, 10, 15, 30, 45 and 65 m. 300 ml of each of them were preserved in Lugol’s iodine solution for microscopic enumeration. 200 µm units of P. rubescens filaments were counted using the Utermöhl inverted microscope technique after sedimentation of 25 or 50 ml water. Estimation of the number of cells was obtained knowing that the mean length of a P. rubescens cell (estimated on 100 measurements) was 5 µm.
B *
*
M
* W ater treatm en t u n its T* N
Planktothrix (Oscillatoria) rubescens is very common in sub-alpine and other lakes in central and northern Europe [1]. This species can form blooms and produce toxins such as microcystins and anatoxin-a [2]. Thus, a P. rubescens bloom in a lake whose water is used for human consumption or for recreation must be monitored. There have been problems with this organism in Lake Bourget over the past two years. A P. rubescens bloom in November 1998 sharply increased the turbidity of the water in treatment units providing drinking water. These central supplies obtain their lake water from a depth of 30 m. We have surveyed this new phenomenon by studying the spatial and temporal distributions of P. rubescens in the lake and estimated the microcystin concentrations before and after water treatment. MATERIAL AND METHODS Lake Bourget is a large (45 km²), deep (140 m at the point B) lake in the French Alps (Fig. 1). Water samples were taken at 3 points (B : northern lake, M : middle of the lake and T : southern lake) twice a month, to evaluate the horizontal distribution of P. rubescens. Samples (2 Harmful Algal Blooms 2000 Hallegraeff, G.M., Blackburn, S.I., Bolch, C.J. and Lewis, R.J. (eds) Intergovernmental Oceanographic Commission of UNESCO 2001
4 km
Fig. 1. Location of Lake Bourget in France. B, M (Mémard) and T (Tresserve) are the three sampling stations for this study Microcystins were extracted from 1 liter of water by the protocol of Feuillade et al. [3]. The samples were taken in the water treatment units, before (water intake in treatment unit) and at the end of the treatment steps (in water reaching the consumers). Intracellular microcystins were identified and quantified by high-performance liquid chromatography (HPLC). The eluting compounds were detected by photodiode-array UV spectroscopy (Waters). Identification of toxins was realized on their retention time and on their UV spectra (absorbance maximum at 238 nm). The amount of microcystin was estimated by measuring the area of the UV absorption peaks at 238 nm and quantitation (Millennium Software from Waters) was made using a mcyst-RR and a mcyst-LR calibration curve. These two microcystins were purchased from SIGMA.
497 RESULTS AND DISCUSSION The vertical distributions of P. rubescens in the three sample areas (M, T & B) were very similar (Fig. 2). In December 98, the filaments were located above the thermocline with a maximum density (> 20000 cells ml-1) at 30 m depth. There was a small difference in the depth of the thermocline at the beginning of the month, deeper at points M and T (50 m) than at point B (35 m). This was due to the strong dominant north wind at this time of year (data not shown), which deepens the thermocline in the southern lake. From January to the end of the bloom in March 99, the cell concentration, similar throughout the water column, slowly decreased. At point B, the density was the same from top to bottom (140 m) from January to March as a result of complete mixing of the lake. Depth (m) 10
B 18000
6000
20
12000
30 40 50 60
M 00
40
12000
30
180
20
6000
10
50 60
T
10
Cell concentration (cell number ml-1)
MCYS-RR concentration (µg l-1) 6
18 00 0
Before water treatment
16 00 0
20
5
6000
14 00 0
30 40
The cell concentration of P. rubescens in the untreated water of the two central water supplies (Fig. 3) was significantly (p < 0.05) correlated with the one estimated in the lake at 30 m (Pearson correlation coefficient r = 0.94 for Mémard and r = 0.98 for Tresserve supplies respectively). There was also a significant correlation (p < 0.05) between the numbers of cells before and after water treatment in the Mémard central supply (r = 0.9), but not in Tresserve (r = 0.6) (Fig. 3). This phenomenon was probably due to differences in the performances of water treatment protocols used in the two treatment units. Differences in the cell concentrations were especially significant from January to March, after the treatment protocol in the Tresserve unit was changed (Fig. 3). These changes concerned the washing of the sand filters (every day instead three times a week) and the quantity of ozone used (4 g/m3 O3 instead of 3 g/m3). This change was followed by a strong decrease of the cell concentration in the water at the 01/13/1999. Most of the microcystin (> 90 %) found in the cellular extracts was mcyst-RR. There was a good linear correlation (Mémard, r = 0.8; Tresserve, r = 0.9; p < 0.05) between the number of P. rubescens cells and the microcystin-RR concentration in water before treatment (Fig. 3). There were several intracellular microcystin-RR concentrations above 0.5 µg l-1 in treated water during three weeks in December 1998 and January 1999 but no extracellular toxin was detected by HPLC, in treated water throughout the study period (data not shown).
4
12 00 0 10 00 0
3
18000
8 00 0 2
6 00 0
50
4 00 0 1 2 00 0
12000
Fig. 2. Spatial distributions of Planktothrix rubescens at the three Lake Bourget stations during the bloom (98-99). The cyanobacterial concentration is expressed in number of cells per milliliter of water.
0
/9 /2 8 9/ 9 /0 8 5 0 1 /9 9 /1 3 0 1 /9 9 /2 0 0 1 /9 9 /2 8 0 2 /9 9 /0 2 0 2 /9 9 /1 1 0 2 /9 9 /1 9 0 2 /9 9 /2 5 0 3 /9 9 /0 5 0 3 /9 9 /1 0 0 3 /9 9 /1 8 0 3 /9 9 /2 5 0 4 /9 9 /0 1/ 99
March 99
01
February 99
/2 2
January 99
12
December 98
0
12
60
Cell concentration (cell number ml-1) 7000 6000
Mémard Tresserve After water treatment
5000 4000 3000 2000 1000
22 12 /98 /2 9 01 /98 /0 5 01 /99 /1 3 01 /99 /2 0 01 /99 /2 8 02 /99 /0 2 02 /99 /1 1 02 /99 /1 9 02 /99 /2 5 03 /99 /0 5 03 /99 /1 0 03 /99 /1 8 03 /99 /2 5 04 /99 /0 1/ 99
0
12 /
The spatial distribution of P. rubescens in the water column is very interesting as there are no published reports of a homogeneous distribution of this cyanobacterium from the surface to the bottom in a deep lake. Micheletti et al. [4] showed that P. rubescens occupies the metalimnion (10-15 m) in lake Zürich in summer and gradually spreads to the epilimnion in autumn. Filaments were found in the epilimnion throughout the year in lake Nantua [5].
Fig. 3. Planktothrix rubescens cell concentrations (bars) and microcystin-RR concentrations (lines) in water before treatment, and cell concentrations after water treatment in the two water supplies (Mémard and Tresserve).
498 A drop in the level of the ground water has resulted in surface water being increasingly used as a source of drinking water. Falconer et al. [6] have recommended that the water body used for domestic consumption should be selected on the basis of its history (cyanobacterial blooms), nutrient concentration (total phosphorus < 10 µg P l-1) and hydrodynamic regime. In Lake Bourget, the total phosphorus is over 10 µg P l-1 after complete mixing of the water in winter. But phosphorus concentration was six times higher in the Eighties and there were never any cyanobacterial blooms. So, the reduction of phosphorus concentration is not always sufficient to prevent cyanobacterial bloom formation. In Lake Nantua, the decline in phosphorus inputs has resulted in P. rubescens population sinking to the metalimnion rather than causing a major reduction in the cyanobacteria biomass [5. In the same order, Mez et al. [7] have observed cyanobacterial blooms in oligotrophic lakes from Switzerland. Hrudey et al. [8] pointed out that the vertical movement of cyanobacterial populations must be known in order to select the best depth for withdrawing drinking water to avoid contamination. Taking water from depth generally avoids problems of cyanobacteria. The bloom in Lake Bourget shows that this generality does not always overcome problems with cyanobacteria in deep lakes because complete mixing of the water in winter distributes the filaments throughout the water column. The concentrations of intracellular microcystin-RR in drinking water were always below the maximum dose (1 µg l-1) of microcystinLR recommended by the WHO [9], which is considered to be more toxic than microcystin-RR [10]. However, the human health risk resulting from chronic exposure to this toxin is very difficult to estimate. We are therefore preparing an Alert Levels Framework (monitoring and management action sequence) for Lake Bourget and the pumping of ground water has been re-instituted for use during the blooms.
Acknowledgements : The authors thank J.P. Bosse and E. Menthon for laboratory assistance and the anonymous reviewers for their useful commentaries. This work was supported by the MISE (Mission Inter-Services de l’Eau, Chambéry). The English text was checked by Dr. Owen Parkes. REFERENCES 1. J. Feuillade, Arch. Hydrobiol. 41, 77-93 (1994) 2. K. Sivonen and G. Jones, in: Toxic cyanobacteria in water, I. Chorus & J. Bartram, eds (E & FN Spon, London ), pp. 41-111 (1999) 3. M. Feuillade, G. Jann-Para and J. Feuillade, Arch. Hydrobiol. 138, 175-186 (1996) 4. S. Micheletti, F. Schanz and A.E. Walsby, New Phytol. 139, 233-246 (1998) 5. M. Feuillade and J.C. Druart, Arch. Hydrobiol. 41, 55-76 (1994) 6. I. Falconer, J. Bartram, I. Chorus, T. KuiperGoodman, H. Utkilen, M. Burch and G. Codd, in: Toxic Cyanobacteria in Water, I. Chorus & J. Bartram, eds (E & FN Spon, London ), pp. 155-178 (1999) 7. K. Mez, K.A. Beattie, G. Codd, K. Hanselmann, B. Hauser, H. Naegeli and H.R. Preisig, Europ. J. Phycol. 32, 111-117 (1997) 8. S. Hrudey, M. Burch, M. Drikas and R. Gregory, in: Toxic Cyanobacteria in Water, I. Chorus & J. Bartram, eds (E & FN Spon, London ), pp. 275-312 (1999) 9. WHO, in: Guidelines for Drinking-water Quality., Second Edition. Addendum to Volume 2, Health Criteria and other Supporting Information, World Health organization, Geneva (1998) 10. T. Kuiper-Goodman, I. Falconer and J. Fitzgerald, in: Toxic Cyanobacteria in Water, I. Chorus & J. Bartram, eds (E & FN Spon, London ), pp. 113-153 (1999)
