Buoyancy changes of the cyanobacterium Planktothrix rubescens- the Burgundy-blood alga - were modelled from its buoyancy response to light and irradiance changes in Lake Zürich during autumnal mixing. The daily insolation received by filaments at fixed depths and circulating to different depths was calculated from the measured light attenuation and surface irradiance. The active mixing depth, za5, was determined from the vertical turbulent diffusion coefficient, Kz, calculated from the wind speed, heat flux and temperature gradients. The fixed depth resulting in filament buoyancy, zn, decreased from 13 to 2 m between August and December 1998; the critical depth for buoyancy, zq, to which filaments must be circulated to become buoyant, decreased from >60 m in the summer to <10 m in winter. When za5 first exceeded zn, in September, P. rubescens was mixed into the epilimnion. In October, zq > za5: circulating filaments would have lost buoyancy in the high insolation. Often in November and December, after deeper mixing and lower insolation, za5 > zq: filaments would have become buoyant but would have floated to the lake surface (the Burgundy-blood phenomenon) only under subsequent calm conditions, when Kz was low. The model explains the Burgundy-blood phenomenon in deeper lakes; waterblooms near shallow leeward shores arise from populations floating up in deeper regions of the lake.

In cyanobacteria the protein on the outside of the gas vesicle, GvpC, is characterised by the presence of a 33 amino acid residue repeat (33RR), which in some genera is highly conserved. The number of 33RRs correlates with the diameter of the gas vesicle and inversely with its strength. Gas vesicles isolated from Microcystis aeruginosa strain PCC 7806 were found to be wider and have a lower critical collapse pressure than those from Microcystis sp. strain BC 8401. The entire gas-vesicle gene cluster of the latter strain was sequenced and compared with the published sequence of the former: the sequences of nine of the ten gvp genes differed by only 1-5% between the two strains; the only substantial difference was in gvpC which in strain BC 8401 lacked a 99-nucleotide section encoding a 33RR. This observation further narrows the correlation of gas vesicle width to the number of 33RRs and suggests how Microcystis strains might be used in experimental manipulation of gas vesicle width and strength.

The ability of the Planktothrix rubescens to stratify in Lake Zürich is related to the size and shape of the cyanobacterial filaments. Detailed measurements made in the lake are used in a dynamic computer model of buoyancy regulation to investigate the vertical movements of filaments tracking the depth at which the irradiance would support neutral buoyancy. The movement of the filament lags behind the constantly changing target depth owing to (a) the time taken for the filament to respond to the irradiance by changing its density and (b) the time it takes to move by sinking down or floating up through the water column. The model simulates the stratification depth over a 5-month period of the summer from the continuous measurements of irradiance and weekly measurements of light attenuation and temperature, without any further adjustment over the period. Models using filaments of the size observed in Lake Zürich explain several details of the observed depth changes: smaller planktonic cyanobacteria (e.g. Limnothrix sp.) are unable to migrate fast enough and larger ones (e.g. Anabaena spp.) will overshoot and become entrained in the epilimnion. The model can be used to simulate recruitment of Planktothrix filaments from different depths after vernal stratification. Recruitment of filaments from depths down to 45 m will contribute to the metalimnetic population increase in early July.

Transcripts of the gas vesicle genes gvpA and gvpC were detected in single filaments of the cyanobacterium Planktothrix rubescens using reverse transcription and quantitative real-time PCR. Primers were designed to amplify short sequences within gvpA and three length variants of gvpC. With genomic template DNA, and using Sybr Green to monitor product accumulation, similar amplification efficiencies were observed for each of these genes. The relative copy numbers of gvpC length variants in genomic DNA from five Planktothrix gas vesicle genotypes determined by real-time PCR were similar to those indicated by sequencing the gas vesicle gene clusters. The precipitation of gvp cDNA reverse-transcribed from cellular RNA from single filaments was required before amplification of the gene fragments; without this step it was not possible to detect the accumulation of the expected amplicons by dissociation analysis. Precipitation was also necessary to ensure the generation of product curves that allowed linear regression in an early stage of PCR, a prerequisite for the quantification of low-input cDNA amounts without the need for standard curves. This report shows that different gvpC length variants are transcribed within single Planktothrix filaments, both from laboratory cultures and from natural samples taken from Lake Zurich. This has implications for the efficiency of buoyancy provision by the possible production of gas vesicles of different strengths within individual cyanobacterial filaments. The hypothesis that post-transcriptional regulation may influence the type of protein (GvpC) present in gas vesicles is presented.

Summary • Comparisons were made of the depths where the cyanobacterium Planktothrix rubescens stratified in lakes and the depths where the daily insolation supported neutral buoyancy of the organism. • The percentage of filaments floating and sinking were measured in cultures grown under light : dark cycles at different irradiances. Qn, the daily insolation for neutral buoyancy, was determined as that at which 50% of the filaments were floating, averaged over the 24-h period. • For P. rubescens 9316 from Lake Zurich, Qn was 0.28 mol m−2; during the summer period of stratification in Lake Zurich, the neutral buoyancy depth (zn) at which this insolation occurred varied between 5 and 14 m, depending on cloud cover and light attenuation. The weekly mean depth of the Planktothrix population maximum (zp) was usually within 1 m of the mean zn. For P. rubescens 9972 from Blelham Tarn, Qn was 0.51 mol m−2; during the stratified period in Blelham Tarn, zn varied between 2.3 and 6.2 m, also similar to zp; the shallower stratification depth is explained by the steeper light attenuation. • The depth at which Planktothrix stratifies in lakes is therefore explained by buoyancy regulation in relation to the irradiance. Metalimnetic stratification occurs only when zn exceeds the mixed depth, zm; when zm > zn, the organism becomes entrained in the surface mixed layer.

SUMMARY 1. The changes in the vertical distributions of red coloured Planktothrix rubescens and green P. agardhii filaments in Blelham Tarn, English Lake District, were related to vertical profiles of temperature and light attenuation and to continuous records of the surface irradiance and windspeed, from August 1999 to October 2000.2. The potential growth rate of each organism was calculated from the irradiance and temperature at 0.5 m depths and hourly intervals throughout the year, using algorithms determined from growth rates in culture. The analyses indicated that there was sufficient irradiance for growth, integrated over the 24‐h cycle, at depths down to the metalimnion where the Planktothrix populations stratified in summer. The compensation depth for growth by P. rubescens reached a maximum of 9.3 m in spring and midsummer, and fell to a minimum of 1.6 m in midwinter; the corresponding values for P. agardhii were 7.9 and 0.5 m.3. The mixed depth (zm) exceeded the critical depth for growth (zb) by P. rubescens (the condition preventing population increase) on only 3 days of the year; for P. agardhii, however, zm exceeded zb on 31 days, contributing to its faster decline. The stratified population of P. rubescens was the major cause of light attenuation during the summer of 2000, and resulted in competitive exclusion of P. agardhii.4. The calculated growth rates integrated over the depth of the water column in Blelham Tarn equalled, or exceeded, the measured changes of the populations during periods when they were increasing, during summer and autumn. Close agreement between the two values was found for much of the year when allowance was made for dilution of the lake population by rainfall over the watershed. During periods of rapid decline, of P. agardhii in September 1999, P. rubescens in December 1999 and both in July–August 2000, additional losses (e.g. by chytrid parasitism and grazing) are invoked.

The blooms of cyanobacteria that develop each summer in the Baltic Sea are composed of two functional groups, namely the small-sized picocyanobacteria (Synechococcus sp.) and the larger, colony-forming, filamentous N2-fixing cyanobacteria. The former encompassed both red (phycoerythrin-rich) and blue-green (phycocyanin-rich) species. The majority of the picocyanobacteria measured less than 1μm and this size fraction comprised as much as 80% of the total cyanobacterial biomass and contributed as much as 50% of the total primary production of a cyanobacterial bloom. The picocyanobacteria are incapable of fixing N2, do not possess gas vesicles and are not toxic. However, a small filamentous Pseudanabaena sp. that could potentially fix N2 was isolated from the picocyanobacteria fraction. The larger cyanobacteria may form surface scums because they possess gas vesicles that make them buoyant. Although their biomass was less than the picocyanobacteria, they therefore form the more conspicuous and nuisance-forming part of the bloom. The larger cyanobacteria were composed mainly of three different species: Nodularia spumigena, Aphanizomenon flos-aquae and Anabaena sp. These all belong to the heterocystous, N2-fixing cyanobacteria. N. spumigena and A. flos-aquae were the dominant species; only N. spumigena was toxic. Although individual Nodularia filaments showed a range of different phenotypes, they all belong to one species as judged from 16S rDNA sequencing. Through determination of the genotypes of many individual Nodularia filaments, it was shown that this population was not clonal and that horizontal exchange of genetic information occurs. N. spumigena and A. flos-aquae were different with respect to their photosynthetic and N2-fixing potentials. Depending on prevailing environmental conditions, these differences would promote the proliferation of one species over the other and hence would determine overall the toxicity of a bloom. Daily integrals of photon irradiance rather than temperature determined the onset of bloom formation. During a bloom, the diazotrophic cyanobacteria fixed N2 at a rate that was 10–20% in excess of their own demand for N. Picocyanobacteria assimilated most of this excess N as shown by 15N incorporation. During bloom conditions, the diazotrophic cyanobacteria met about 50% of the N demand of the total cyanobacterial community. The picocyanobacteria were predominantly N-limited while the diazotrophic cyanobacteria were probably iron limited. These findings allow us to understand the formation of toxic cyanobacterial blooms and also to develop tools to predict bloom formation.

A limnological database for Lake Kinneret was used to determine the period each year when the interaction of mixed depth and light availability would permit growth of the planktonic cyanobacterium Aphanizomenon ovalisporum. The organism, recorded for 56-211 days in each of the 7-years, 1994-2000, was usually distributed through the mixed epilimnion and reached a maximum in late October. The mixed depth (z(m)) was calculated from the temperature profile, wind speed and wind direction. Changes in the total population were related to changes in the daily integrals of photosynthesis, calculated from vertical profiles of water temperature (Theta) and light attenuation (K-d), and hourly changes in photon irradiance (E-0) each day, using temperature-compensated algorithms based on measurements of photosynthesis at different irradiances. Net photosynthesis was calculated for different ratios of respiration (R) and maximum photosynthesis (P-m). In a modelling analysis, the daily integral of net photosynthesis fill below zero on the day of the population maximum, and remained negative during the period of population decline, when R was set at -0.08 P-m. Similar modelling, based on the 7 year mean values of Theta, K-d and E-0 for each day of the year, showed the critical depth for photosynthesis by Aphanizomenon would exceed the mixed depth in Lake Kinneret from mid-March to late October. Direct measurements of -R/P-m ratios could be used with information in the database to analyse the periods of photosynthetic production by different phytoplankton species in Lake Kinneret.

We have developed a technique for determining the genetic structure of populations of filamentous cyanobacteria. The sequence diversity at specific gene loci is first characterised in a range of clonal cultures; subsequent analysis involves individual trichomes collected directly from natural populations. This technique has been used to examine the population genetic structure of Nodularia in the Baltic Sea and Planktothrix in Lake Zürich. For Nodularia, studies utilising four polymorphic loci reveal that even though there is a degree of linkage disequilibrium, horizontal transfer of genetic information has been sufficient to generate many of the possible allelic combinations. Analyses reveal both spatial and temporal variation in population genetic structure. Other studies of both Nodularia and Planktothrix have shown a correlation between particular alleles at the gvpC locus and the critical pressure of the gas vesicles that accumulate within the cell. We are now investigating how the natural selection of different gas vesicle phenotypes, imposed by changes in the depth of the upper mixed layer of the water column, affects the relative success of individual cyanobacteria possessing different gvpC alleles.

Wild-type strains of the cyanobacterium Planktothrix rubescens have a cluster of gas vesicle (gvp) genes with repeats of alternating gvpA and gvpC. The gvpC occurs in three length variants, all with the same 3'-sequence, OmegaC. Spontaneous non-buoyant mutants had lost some of the alternating gvpAC copies and their gvpC genes had a novel 3'-end sequence, PsiC; additional gvpC genes terminating in this sequence were also found in the wild-type and representatives of other GV genotypes. Alleles of gvpC terminating in PsiC occurred only at the downstream ends of the gvpAC clusters investigated; all other gvpCs terminated in OmegaC. Mutants of strains with the GV3 genotype produced only 30-50% of the gas vesicles present in the wild-type; their gas vesicles had lower mean critical pressures (0.70-0.78 MPa) than those in the wild-type (1.05-1.10 MPa).

Transformation experiments with Haloferax volcanii show that the amino acid sequence of the gas vesicle protein GvpA influences the morphology and strength of gas vesicles produced by halophilic archaea. A modified expression vector containing p-gvpA was used to complement a Vac(-) strain of Hfx. volcanii that harboured the entire p-vac region (from Halobacterium salinarum PHH1) except for p-gvpA. Replacement of p-gvpA with mc-gvpA (from Haloferax mediterranei) led to the synthesis of gas vesicles that were narrower and stronger. Other gene replacements (using c-gvpA from Hbt. salinarum or mutated p-gvpA sequences) led to a significant but smaller increase in gas vesicle strength, and less marked effects on gas vesicle morphology.

Summary • Analyses were made to determine which changes in a Lake Zürich population of Planktothrix rubescens were dependent on light‐ and temperature‐dependent growth rates, and when growth was limited by the mixing depth. • Changes in vertical distribution of the cyanobacterium, determined weekly from August 1998 to September 1999, were related to growth increments calculated at 1‐h time and 1‐m depth intervals from values of irradiance, attenuance, temperature and biomass in the lake, using algorithms based on growth rates in culture. • Population biovolume varied annually from 1.2 to 120 cm 3 m −2 . During summer, modelled growth in the metalimnion matched the 50‐fold population increase. Modelled growth exceeded the observed increase when Planktothrix was mixed into the nutrient‐depleted epilimnion, suggesting nutrient limitation. The measured increase ceased when the mixed depth exceeded the critical depth for growth in autumn (Sverdrup’s principle). Light limitation explained the gradual decrease of the population in winter. The steep decline in spring had other causes. • Population changes were largely determined by interactions of light and depth distribution; decreases in nutrient loading have had little impact on Planktothrix growth in Lake Zürich.

• Differences in photosynthetic production and conversion to biomass of the red-coloured cyanobacterium Planktothrix rubescens and the green-coloured Planktothrix agardhii , were investigated in relation to their growth in Blelham Tarn, UK, using clonal isolates from the lake. • Growth rates (µ) were measured in cultures under 12 h : 12 h light : dark cycles at 15 irradiances ( E ) in temperatures (Θ) of 10-25°C. Photosynthetic rates ( P ) were measured under the same conditions. • For P. rubescens , µ reached a maximum of 0.33 d -1 at 25°C in photon irradiances > 40 µmol m -2 s -1 and exceeded µ for P. agardhii over the range of temperatures in Blelham Tarn (< 21°C), although not at temperatures > 25°C. In P. rubescens , the dif ference (Δµ) between the growth rate of cell carbon (µ C ), calculated from P , and µ was only 3% at 10°C but increased with temperature to 30% at 25°C; in P. agardhii , Δµ values were higher at low temperatures and lower at the higher temperatures. • Using algorithms describing the irradiance- and temperature-dependent growth rates and measured values of E and Θ at different depths in Blelham Tarn, it was demonstrated that P. rubescens would outgrow P. agardhii , though the latter might grow better in warmer and shallower lakes. We discuss the problems of modelling phytoplankton growth from measurements of in situ photosynthesis.

In late summer and autumn, before the vertical circulation reaches the thermocline, the phytoplankton population of Lake Zürich is dominated by the red-coloured filamentous cyanobacterium Planktothrix rubescens, which stratifies in the metalimnion at depths close to the photosynthetic compensation point. The filament volume concentration reached a maximum of 12 cm3 m-3; the depth of the maximum varied from 10.5 to 12.5 m. Changes in the depth distribution were attributed to a combination of (1) seiche movements, which raised or lowered the thermocline by up to 2 m over 36 h, and (2) flotation by the buoyant filaments relative to the isotherms, by up 0.4 m d-1. These changes caused a 2-fold change in insolation at the Planktothrix peak. Estimates were made of the daily integral of photosynthetic O2-production, ΣΣ(NP), by the population of P. rubescens over a period of four cloudless days. The estimates were calculated from measurements of surface irradiance (at 5-min intervals), vertical light attenuation, temperature, filament volume concentration and the photosynthesis/irradiance (P/I) curves of filaments concentrated from the metalimnion. Despite the similar, high insolation on each of the four days, the calculated values of ΣΣ(NP) varied from 9 to 53 mmol m-2 d-1, owing to the changing depth distribution of the filaments. Measurements of P/I curves of lakewater samples incubated at a depth of 11 m showed changes in the photosynthetic coefficients during the day. These also generated large changes in calculated values of ΣΣ(NP). The computer spreadsheet used to calculate ΣΣ(NP) was modified to incorporate time-based changes in the photosynthetic coefficients and vertical distribution of the organism. These refinements provide a more accurate description of photosynthesis by the deep-living P. rubescens, which adjusts its position by buoyancy regulation to exploit the light field in the metalimnion, where it outcompetes other phytoplankton.

The relationship between photosynthesis and irradiance was investigated in a strain of the cyanobacterium Planktothrix rubescens isolated from Lake Zürich, using cultures grown on a 12:12-h light-dark cycle. From the photosynthesis-irradiance (P/I) curves, values of the light-affinity coefficient, α, and the maximum rate of photosynthesis, P m, were determined: in different cultures α ranged from 0.02-0.08 μmol mg-1 h-1 (μmol m-2 s-1)-1 and P m from 0.6-2.6 μmol mg-1 h-1 (rates of O2 production related to dry biomass). Comparisons made at different phases of the light-dark cycle in individual cultures showed that in the first hour of the light phase α rose by an average of 21 % and P m by 7 %. In cultures that had been grown in many light-dark cycles, however, the rise occurred only if light was given at the beginning of the notional light phase; cultures kept in darkness showed no rise. It is concluded that the observed increases in α and P m after dawn were in response to the exogenous irradiance rather than due to endogenous rhythms. These findings are discussed in relation to diel variations in the photosynthetic coefficients of natural populations of P. rubescens in Lake Zürich.

Daily integrals of photosynthesis by a cyanobacterial bloom in the Baltic Sea, during the summer of 1993, were calculated from the vertical distributions of light, temperature and the organisms in the water column and from photosynthesis/irradiance curves of picoplanktonic and diazotrophic cyanobacteria isolated from the community. The distribution of chlorophyll a in size-classes <20 µm and >20 µm was monitored over 9 days that included a deep mixing event followed by calm. Picocyanobacteria formed 70% of the cyanobacterial biomass and contributed 56% of the total primary production. Of the filamentous diazotrophs that formed the other 30%, Aphanizomenon contributed 28% and a Nodularia-containing fraction 16% of the primary production. For the whole population there was little change in standardized photosynthetic O2 production, which remained at about 31 mmol m−2 before and after the mixing event. There were differences, however, between the classes of cyanobacteria: in picocyanobacteria primary production hardly changed, while in Aphanizomenon it increased by 2.6 and in Nodularia it fell below zero. Total phytoplankton photosynthesis was strongly dependent on total daily insolation with the compensation point at a photon insolation of 22.7 mol m−2 d−1. Similar analyses of N2 fixation showed much less dependence on depth distribution of light and biomass: Aphanizomenon fixed about twice as much N2 as Nodularia their; their fixation exceeded their own N demand by about 12%. Together, these species contributed 49% of the total N demand of the phytoplankton population. Computer models based on the measured light attenuation and photosynthetic coefficients indicate that growth of the cyanobacterial population could occur only in the summer months when the critical depth of the cyanobacteria exceeds the depth of mixing.

In cyanobacteria of the genus Planktothrix:, there are three length variants of gvpC, the gene that encodes the outer protein of the gas vesicle. Sequence analyses indicated that the three allelic variants of gvpC differ principally in the presence or absence of a 99 nt and a 213 nt section. Strains with the new variant, gvpC(28), which encodes a 28 kDa form of GvpC, produce gas vesicles that collapse at the relatively low critical pressure (p(c)) of 0.61-0.75 MPa. The authors have identified 12 classes of gvp genotypes that differ in the number and arrangement of alternating gvpA-gvpC genes and in the presence of OmegaC, a fragment of gvpC. The gvpC(28) gene was found to be the most common variant of gvpC amongst 71 strains of Planktothrix: isolated from Nordic lakes: 34 strains contained only gvpC(28); 22 strains, which possessed only the shorter gvpC(20) gene, produced gas vesicles with a higher p(c) of 0.76-0.91 MPa; and 15 strains, which possessed both gvpC(20) and gvpC(28), also produced the stronger gas vesicles. Genotypes with only the gvpC(28) genes were more common amongst green Planktothrix: strains (33 out of 38) than red strains (one out of 33). It is suggested that there is competition between the strains producing the two types of gas vesicles, with the stronger forms favoured in lakes deeper than 60 m, in which the combination of cell turgor pressure and hydrostatic pressure can collapse the weaker gas vesicles. The fact that none of the Nordic lakes are deeper than 67 m would explain the absence of the gvpC(16)-containing strains that produce even narrower gas vesicles of p(c) 1.0-1.2 MPa, which are common in the much deeper Lake Zürich.