There is growing concern about observations of harmful algae blooms promoted by nutrient enrichment both in freshwater and estuarine systems. The dominant species frequently include bloom-forming cyanobacteria such as the genera Anabaena, Microcystis and Planktothrix, whose growth may also be favored by elevated temperature . However, efforts of lake restoration, such as the reduction of nutrient input, can also lead to the proliferation of stratifying cyanobacteria such as P. rubescens because of an increased underwater light regime .
It is widely agreed that the production of microcystin (MC), which is the most abundant toxin in freshwater, is directly related to the cell division rate of a particular isolate grown under controlled laboratory conditions . In nature, blooms of cyanobacteria are typically composed of toxic and nontoxic genotypes, the latter resulting from the loss  or the inactivation of the MC synthetase (mcy) gene cluster [5, 6]. Some indices show that the production of MC has a selective advantage for the producers . However, besides the fact that MC is a potent inhibitor of eukaryotic protein phosphatases 1 and 2A, the cellular function of MC is not known. To elucidate parameters that influence the competitive ability of toxic and nontoxic genotypes, several growth experiments with toxic and nontoxic strains of Microcystis or Planktothrix have been performed under controlled laboratory conditions. Vézie et al.  reported that under high nutrient levels (nitrogen, phosphorus) toxic strains of Microcystis grew faster than nontoxic strains, while Briand et al.  found an advantage of toxic over nontoxic Planktothrix strains when environmental conditions limited growth (through dim light, low temperature or nitrogen-limiting conditions). Furthermore, toxic Microcystis strains were shown to be the better competitor at high irradiances  when compared with nontoxic strains. The rather contrasting conclusions obtained from different toxic and nontoxic strains possibly result from physiological adaptations of the individual genotypes to specific environmental conditions, which are not related to MC production. Indeed, it has been shown that, within the genus Planktothrix, the origin of nontoxic strains is rather ancient, and that toxic and nontoxic strains evolved independently and differentiated physiologically in response to environmental factors not directly related to toxin production . Only one monophyletic lineage of green-pigmented strains of the genus Planktothrix that lost the mcy gene cluster (henceforth referred to as lineage 1) could be found, which invaded shallow, polymictic lakes throughout Europe . A second lineage containing both red- and green-pigmented strains (henceforth referred to as lineage 2) retained the mcy gene cluster. This phylogenetic evidence can explain why red-pigmented (phycoerythrin-rich) populations of Planktothrix, which typically occur in deep, stratified lakes and reservoirs, are commonly composed solely of the genotype containing the mcy gene cluster [5, 11]. By contrast, green-pigmented populations that dominate in shallow, eutrophic and polymictic water bodies have a much higher proportion of the nontoxic genotype. A recent survey on toxic genotype abundance in European lakes revealed that red-pigmented populations of Planktothrix show a significantly higher proportion of the toxic genotype when compared with green-pigmented populations .
Only few studies investigated the selective advantage of red- versus green-pigmented Planktothrix ecotypes under field conditions. Davis et al.  investigated a mixed-pigmented Planktothrix population in Blelham Tarn, Lake District, England and analyzed the vertical distribution of the biomass of the two ecotypes. For both ecotypes, the biovolume was increasing under stratified conditions of the water column. However, the red-pigmented ecotype could be shown to grow at greater depths under stratifying and mixed conditions, as its compensation light intensity for growth was lower compared with the green-pigmented ecotype. It is known that P. rubescens is adapted to low light conditions whereas P. agardhii is more tolerant to high light intensities [14, 15]. Oberhaus et al.  suggested further that the combined effects of temperature and light quality and quantity influence the proliferation of P. rubescens and P. agardhii. They found the red-pigmented strain to be more competitive at lower temperatures (15°C) and low intensities of green light, resembling the conditions present in the metalimnion, whereas the green-pigmented strain was more competitive at higher temperatures (25°C) and generally less specialized to light quality. Similarly, Stomp and colleagues  showed that the underwater light regime was an important factor for niche differentiation of red- and green-pigmented picocyanobacteria and reported their coexistence in waters of intermediate turbidity, whereas red-pigmented picocyanobacteria dominated in clear waters and green-pigmented picocyanobacteria were dominant in turbid waters. Additionally, Walsby and co-workers  suggested that the resistance of gas vesicles against hydrostatic pressure is of major importance during lake mixing, especially in deep lakes when filaments become entrained in the hypolimnion. A selective difference between red- and green-pigmented strains producing different types of gas vesicles has been suggested , which could further influence the dominance of the red-pigmented ecotype in deep habitats and the common abundance of the green-pigmented ecotype in more shallow water bodies. However, still not much is known about the temporal stability of those contrasting ecotype strategies in ecosystems subject to severe shifts in local environmental conditions.
Here, we report the detailed analysis of the genotypic population structure of Planktothrix spp. in Lake Zürich, Switzerland, covering a time span of almost 30 years, which was facilitated by the isolation of DNA from phytoplankton preserved on filters. Because Lake Zürich represents an important drinking water source for about 900,000 inhabitants, its planktonic phytoplankton composition has been monitored intensively. Lake Zürich underwent a well-documented history of eutrophication that reached its maximum around 1965. The re-oligotrophication process was initiated by reducing the input of phosphorus . Except for the period of maximum eutrophy (1965 to 1975), Planktothrix occurred in Lake Zürich during the whole century. The first Planktothrix bloom was recorded in 1897. During the eutrophic period, eukaryotic algae frequently formed surface blooms, which subsequently disappeared as a result of re-oligotrophication measures, while Planktothrix consistently increased .
The aim of the study was to find out whether (i) the abundance of the toxic Planktothrix genotype changed during the observation period, that is from the period with minimum population density and almost complete disappearance to a stable dominance of the phytoplankton community; (ii) nontoxic genotypes (including inactive mutants) increased in proportion during the observation period: following the hypothesis of Briand et al.  we would expect an increase of nontoxic genotypes parallel to the increase of the total population density; (iii) the green-pigmented ecotype was present and was of selective advantage under certain environmental conditions, for example at the beginning of the 1980s when the euphotic zone was rather shallow and the red-pigmented ecotype was disfavored because of the high absorption coefficient in the water column.