Abstract:

Cyanobacteria (blue-green algae) occur in freshwater all over the world and frequently
dominate the phytoplankton community. Beside ecophysiological adaptations such as
buoyancy due to the synthesis of gas vesicles or the efficient uptake of bicarbonate ions at
high pH it is generally believed, that the production of toxins and other bioactive substances
contributes to their ecological success. The toxins produced by cyanobacteria include cyclic
peptides, most prominently the hepatotoxic microcystins and alkaloids, such as the
hepatotoxic cylindrospermopsin or the neurotoxic saxitoxins, anatoxin-a or anatoxin-a(S).
While it is known that cyanobacteria in general play an important role in the saline and
freshwater lakes in East Africa, the knowledge about their ability to produce toxins in
freshwater in this region is very scarce. This lack of knowledge is of particular relevance, as a
significant part of the population in Uganda daily uses the water as drinking water and
consumes phytoplanktivorous fish harvested from the freshwater lakes. In order to estimate a
potential health risk due to the toxin production by cyanobacteria in those countries, not only
a sound estimate of the phytoplankton biovolume in the lakes is needed, but also the
understanding, when toxins are produced and which are the factors that favour toxin
production in those systems.
It was the aim of the study, to estimate the production of the most widely distributed
toxin microcystin seasonally in several large freshwater lakes in Uganda as well as to identify
not only the producers of microcystin but also the environmental factors favouring
microcystin production. During one year five lakes in Uganda were sampled monthly (Lake
Victoria: Murchison Bay, Napoleon Gulf; three shallow lakes within the catchment area of
the Ruwenzori: Lake George, Lake Edward, Lake Mburo; one Crater lake: Lake Saka) and
limnologically characterized using physical and chemical parameters. In parallel the
composition of phytoplankton was determined qualitatively and quantitatively by counting
from sedimentation chambers using the inverted microscope technique. The concentration of
microcystin was analysed by means of high-performance liquid chromatography (HPLCDAD).
In addition clonal strains of the genus Microcystis were isolated from field samples
and characterized both genetically and morphologically according to their ability to produce
microcystin. PCR based diagnostic approaches were used to identify microcystin-producing
genotypes both in field samples and in the laboratory.
vii
I
In the first publication (chapter 2, Okello et al. 2009, Environmental Toxicology, 16 July
2009) it could be shown, that Microcystis is the only genus producing microcystin in
Ugandan freshwaters. All PCR based tests revealed gene regions that only could be assigned
to Microcystis and never to co-occurring other taxa such as Anabaena or Planktolyngbia.
Further it could be shown that eutrophication of freshwater water bodies -either natural or
human-induced - always result in dominance of Microcystis and therefore ultimately in the
production of microcystin. Two structural microcystin variants have been determined that
seem to be unique and dominant and have not been described previously.
In the second publication (chapter 3, Okello und Kurmayer, Water Research, submitted)
the production of microcystin was analysed in the five freshwater lakes in Uganda during one
year (April 2007-April 2008). In all samples microcystins were detected. Corresponding to
publication Nol it could be shown that the cell number of Microcystis as determined in the
microscope correlated significantly with the concentration of microcystin. Other potential
microcystin-producing genera such as Anabaena or Planktothrix also showed partly
significant correlations with the microcystin concentration in water. However since no
evidence for the presence of microcystin genes that could be assigned to either Anabaena or
Planktothrix was obtained during publication Nol and the correlation coefficients were
always much lower when compared with those observed for Microcystis it is concluded that
the significant correlations observed for Anabaena or Planktothrix are due to their cooccurrence
with Microcystis.
Surprisingly the lakes showed a relatively narrow range in microcystin concentrations
during the season, while differed significantly (on average up to 28-fold) in microcystin
production. Samples from Lake Saka had the maximum MC concentration (10 pg L'1) in July
2007. The minimum concentration (0.02 pg L'1) was recorded in Lake George (Kahendero)
in the months of May 07, June 07, January 08 and April 08. The difference in microcystin
concentrations between lakes only partly could be explained by the variation in Microcystis
cell numbers. For example, while on average the maximum Microcystis cell concentration
was observed in L. George (1.2 ± 0.3 (1 standard error) Mio cells mF1), on average the
microcystin concentrations were at the minimum (0.2 ± 0.1 pg L’1). In contrast, in L. Mburo
on average a rather low cell concentration was recorded (0.17 ± 0.02 Mio cells ml'1) while the
microcystin concentration on average was significantly higher (1.6 ± 0.2 pg L1). In order to
test the hypothesis that populations of Microcystis differ in the proportion of genotypes
containing the genes involved in microcystin synthesis (wcy) resulting in the significant
variation. in MC cellular contents between the study lakes, the absolute and relative
viii
abundance of nicy genotypes was determined. The average proportion of nicy genotypes in L.
George (1,14±0.13 (SE)% was 17-fold lower when compared with the average nicy genotype
proportion in L. Mburo (19.6±4.8%). It is concluded that the variation in average MC content
of Microcystis that is observed between lakes could be explained by the divergence in nicy
genotype proportion between populations.
In the third publication (chapter 4, Okello and Kurmayer, in preparation) the
morphological characteristics of the genus Microcystis in field samples were compared with
those observed under culture conditions in the laboratory. While in field samples, the
morphotypes M. aeruginosa, M. flos-aquae and M. wesenbergii occurred most frequently,
only M. aeruginosa, M. flos-aquae could be found under culture conditions. However,
altogether the morphological characters that were observed under field conditions were found
reproducible. In total only 13 of 96 strains (14%) contained the zncyE/B gene indicative of
microcystin production. In addition twelve isolates were sequenced for the intergenic spacer
region of the phycocyanin region (PC-IGS) and assigned to the existing classification system
of Microcystis by phylogenetic methods. To construct a phylogenetic tree 128 additional
entries of Microcystis from the Genbank database (602 bp) were compared with the
sequences obtained during this study. In general Microcystis strains showed a non-random
distribution among the phylogenetic lineages, for example while the strains of this study
occurred in two clades (cluster A, D) only but were absent from another clade (cluster F)
containing the largest number of strains. It is concluded that the identification of Microcystis
according to morphological criteria is reliable, and even in the presence of morphologically
similar genera such as Gomphospaeria, Synechocystis, Chroococcus can be used to estimate
Microcystis cell numbers under the microscope. Altogether from publications No 1-3 it is
concluded, that microscopical counting of Microcystis is a reliable tool, in order to provide a
relatively quick and simple estimate of microcystin concentrations by multiplying the average
cell numbers by the microcystin cell quotas as determined in this study.