Interactive comment on “ Influence of elevated CO 2 concentrations on cell division and nitrogen fixation rates in the bloom-forming cyanobacterium Nodularia spumigena ”

Thank you very much for the many constructive remarks on our manuscript that helped us to improve it substantially. 1. The statement that the surface ocean absorbs 1/4 of current CO2 emissions is not referenced (naturally, since references aren’t typically included in an abstract), but to my knowledge this is still a relatively uncertain number and somewhat controversial. Maybe it would be better to start out by simply saying “As CO2 emitted to the atmo-


Introduction
Massive anthropogenic emissions caused atmospheric CO 2 concentrations to rise from an interglacial level of 280 ppm in preindustrial times, (Indermuehle et al., 1999) to presently 385 ppm (Keeling et al., 2008).In the case of unabated CO 2 emissions this value is expected to double until the end of the century (IPCC, 2007).A combination of dissolution and mixing combined with biological processes makes the ocean absorb about one third of the anthropogenic CO 2 emitted to the atmosphere (Sabine et al., 2004).Due to the reaction of CO 2 with water, the surface oceans' pH has already decreased by ∼0.1 units and will continue to drop by additional 0.3 to 0.4 units until 2100 under a business-as-usual CO 2 emission scenario (IPCC IS92a, see Meehl et al., 2007).
As photosynthetic CO 2 fixation is substrate-limited under current atmospheric CO 2 /O 2 ratios all photoautotrophic organisms evolved active carbon concentrating mechanisms (CCM), providing elevated [CO 2 ] at the site of carboxylation.In seawater, CO 2 concentration represents less than 1% of the inorganic carbon species and does indeed limit photosynthetic carbon fixation rates (Giordano et al., 2005).Elevated CO 2 concentrations are suggested to reduce the energetic costs for CCM (Fridlyand et al., 1996) and the regeneration of oxidised carbon acceptors and should thereby facilitate other energy consuming processes (Raven and Johnston, 1991;Riebesell, 2004).Indeed, rising CO 2 concentrations have been shown to enhance carbon fixation in several single species experiments (Hinga, 2002;Riebesell, 2004;Rost et al., 2008) and in natural plankton communities (Hein and Sand-Jensen, 1997;Tortell et al., 2002b;Riebesell et al., 2007).
Published by Copernicus Publications on behalf of the European Geosciences Union.
Elevated nitrogen and carbon fixation rates mostly accompanied by enhanced cell division under projected future CO 2 conditions were measured in the filamentous oceanic cyanobacterium Trichodesmium (Barcelos e Ramos et al., 2007;Hutchins et al., 2007;Levitan et al., 2007;Kranz et al., 2009), as well as in the unicellular picocyanobacterium Crocosphaera (Fu et al., 2008).A combined positive effect of CO 2 enrichment and temperature on cell division rates was observed for the non-nitrogen fixing Synechococcus.But, under the same conditions Prochlorococcus, another picocyanobacterium, did not show this response (Fu et al., 2007).
Cyanobacteria can be found in a wide range of environments and are successful competitors even under conditions where inorganic carbon becomes ultimately limiting to primary production.This commonly occurs under high growth densities as prevailing in microbial mats and surface scums (Shapiro and Wright, 1990;Oliver and Ganf, 2000).Depending on the conditions in the natural habitats, CCM expression and activity can strongly differ between different species of cyanobacteria (Badger et al., 2005).While the expression of some components of the CCM appears to be constitutive, others were shown to be regulated in response to environmental factors (Shibata et al., 2001;Beardall and Giordano, 2002).
The stoichiometric composition of a mono-specific culture can show highly dynamic changes due to its ability to store nutrients in internal pools (Arrigo, 2005).A strong tendency to store nutrients is typical for cyanobacteria, especially if they are exposed to high fluctuations in nutrient supply (Allen, 1984).CO 2 related shifts in elemental ratios were also found in the studies on Trichodesmium, Crocosphera and on Synechococcus (Barcelos e Ramos et al., 2007;Fu et al., 2007Fu et al., , 2008;;Hutchins et al., 2007).The biomass composition of cyanobacteria in these experiments as well as of seven microalgal species examined by Burkhardt et al. (1999) reacted differently to rising CO 2 .Observed reactions in elemental ratios included increasing, decreasing and constant C/N and C/P ratios.These results imply that phytoplankton responses to future CO 2 concentrations will likely not follow a common pattern but may depend on the physiology of single species.
Exclusively in the Baltic Sea and in the Peel-Harvey estuary in Australia, the filamentous heterocystous cyanobacterium Nodularia spumigena, frequently forms extensive blooms that play a major role in the annual productivity of these regions (Sellner, 1997).Aside from Nodularia, heterocystic filamentous Anabaena and Aphanizomenon often contribute considerably to the biomass of cyanobacterial blooms in the Baltic Sea.Under calm conditions Nodularia accumulates at the surface forming big aggregates and even dense scums.Estimates of annual nitrogen fixation in the Baltic Sea due to cyanobacterial blooms are roughly equal to the total nitrogen input from river runoff and atmospheric deposition (Larsson et al., 2001;Schneider et al., 2003).Therefore, Nodularia is of high biogeochemical importance for this re-gion.In the Baltic Sea dissolved inorganic carbon (DIC), salinity and total alkalinity are lower compared to ocean values due to a strong riverine influence in this marginal sea.As a result of that and of a high biological productivity, the carbonate system shows a much stronger diurnal and seasonal variability than that of the open ocean, with strong temporal changes in pH, CO 2 and CO 2− 3 concentrations (Thomas and Schneider, 1999).
In this study, aggregation is avoided as microclimatic effects on CO 2 and O 2 levels inside and outside the aggregations might induce a different physiological response to elevated pCO 2 than in dispersed filaments (Ploug, 2008).The experimental setup simulates a pre-bloom situation were single filaments are dispersed within the upper water column and aggregations have not yet formed.Carbon and nitrogen fixation of Nodularia spumigena were investigated under five different CO 2 concentrations.Simulated CO 2 conditions correspond to concentrations between values in glacial periods and future values projected for 2100.Other environmental conditions were kept constant simulating a pre-bloom situation where single filaments are suspended within the surface layer.
We aim to determine whether the stimulating effects of elevated [CO 2 ] on carbon and nitrogen fixation found in Trichodesmium represent a general phenomenon among diazotrophic cyanobacteria and hence also apply to heterocystic Nodularia.

Setup
Semi continuous batch cultures of non-axenic Nodularia spumigena (IOW-2000/1) were grown at 16 • C.This temperature was chosen because Nodularia blooms frequently develop in the southern Baltic Sea at this value (Kononen, 1992).The cultures were illuminated at an average irradiance of 85 µmol photons m −2 s −1 under a 14/10 h light/dark cycle, simulating the depth integrated irradiance of the mixed layer.To ensure identical light conditions for all bottles, their positions were shifted daily.Aggregation of cell filaments and the development of microenvironments, in which growth conditions can deviate from those in the bulk medium, was avoided by keeping the cultures homogenously mixed at all times.This was achieved through a rotating device (Planktongravistat) that slowly rotated the incubation bottles orthogonally to their axis at a constant velocity of 1 rpm.
For acclimation of the cultures to the CO 2 treatments, precultures were grown for 13 days (replicates 1-9), and 20 days for the two high CO 2 -concentration treatments (replicates 11-15).Pre-cultures were grown in the same manner as the cultures in the experimental units.Different incubation periods were used in order to reach similar cell densities at the start of the experiment, despite the different division rates.Pre-culture incubation corresponded to 7 to 8 cell generations.The experiment was started with an initial Chl a concentration of 1 µg l −1 in three replicate bottles per treatment.Start values of dissolved inorganic carbon (DIC), total alkalinity (TAlk) and cell counts were measured.All treatments were sampled after an incubation time of seven days, during which CO 2 was drawn down by photosynthesis, as occures during the development of a bloom.DIC consumption in all bottles was less than 3%.The denoted [CO 2 ] and pH values were calculated for the point where half of the consumed DIC was taken up by the bacteria.TAlk was measured in the end of the experiment.

Growth medium
An artificial seawater based medium (Kester et al., 1967) was prepared using modified YBC II nutrients (Chen et al., 1996) without inorganic nitrogen and with reduced phosphate (5.4 µmol l −1 ).A salinity of 8 was chosen in correspondence with the origin of the culture (southern Gotland Sea) and because this is the salinity where intensive blooms of Nodularia are commonly observed (Kononen, 1992).
DIC concentration was adjusted to a value typical for the open Baltic Sea (1981 µmol kg −1 ).After preparation, the medium was 0.2 µm sterile-filtered directly into the one litre glass culture bottles.To minimise gas exchange with air, filtration was performed by means of a peristaltic pump.The media for the pre-culture and experiment were prepared from one batch.

CO 2 manipulation
Both TAlk and DIC samples were taken from several culture bottles after filtration of the media.Based on these measurements, manipulation was carried out by adding HCl or NaOH to obtain an experimental pCO 2 range between ∼150 and ∼700 ppm and a corresponding pH range between 8.6 and 7.9 (on the free scale).The acid/base manipulation technique was chosen here because it is the most simple and precise way for CO 2 manipulation in closed bottles without headspace.As the pCO 2 range applied in this experiment is relatively small, the carbon ion speciation compared to a method where DIC is changed while TAlk is kept constant, shows equal trends of similar magnitude (Schulz et al., 2009).Therefore, we do not expect significant differences in the measured results that are due to the manipulation approach.The low end pCO 2 value of 150 ppm was chosen because at the time of Nodularia bloom development, i.e. early to mid summer, pCO 2 levels in the Gotland Sea are typically between 100 and 200 ppm (Thomas and Schneider, 1999).These comparatively low CO 2 levels result from intense biological activity earlier in the season combined with a low buffer capacity of the Baltic Sea brackish waters.

Seawater carbonate system
DIC was measured after Stoll et al. (2001) in a QUAA-TRO analyzer (Bran & Lübbe GmbH, Norderstedt, Germany) equipped with a XY-2 sampling unit.The precision and accuracy of this method are ∼2-3 µmol kg −1 .TAlk was determined in duplicate samples via potentiometric titration (Dickson, 1981) with 0.005 M HCl at 20 • C in a Metrohm Tiamo automatic titration device (Metrohm GmbH & Co. KG, Filderstadt, Germany) with a precision of ∼2 µmol kg −1 .TAlk measurements were calibrated against Dickson seawater standard for CO 2 measurement (Marine Physical Laboratory, University of California, Prof. A. G. Dickson).pCO 2 and pH (on the free scale) were calculated from DIC and TAlk measurements with the program CO2SYS version 01.05 by E. Lewis and D. Wallace (for distinct treatment values see Table 1).

Chlorophyll a
Chl a concentrations were determined fluorometrically according to Welschmeyer (1994).Triplicates of 50 ml per bottle were filtered under a low vacuum of ∼200 mbar on glass fibre filters (Whatman GF/F 25 mm ∅) and stored frozen at −18 • C. Filters were homogenised in acetone and the extract measured in a Turner fluorometer 10-AU (Turner BioSystems, CA, USA).

Particulate organic matter
Quantification of POC and PON was carried out via an elemental analyzer with a heat conductivity detector Eu-roVektor EA (EuroVektor S.p.A., Milan, Italy) according to Sharp (1974).200 ml of sample were filtered at a pressure of ∼200 mbar on a combusted filter (Whatman GF/F 25 mm ∅) and subsequently stored at −18 • C. Before measurement, filters were dried at 60 • C for at least 5 h and packed into tin boats.Samples were calibrated against acetanilide C/N=10.36/71.09kg/kg).
POP was determined following Hansen and Koroleff (in Grasshoff et al., 1983) adapted to the measurement of samples on glass fibre filters.200 ml of sample were filtered at a pressure of ∼200 mbar on combusted filters (Whatman GF/F 25 mm ∅) and subsequently stored at −18 • C. The biomass was completely oxidised by heating the filters in 50 ml glass bottles with 35 ml of alkaline peroxodisulphate solution in a pressure cooker.Solutions were measured colorimetrically in a spectrophotometer U 2000 (Hitachi-Europe GmBH, Krefeld, Germany) with a precision of about ±0.2 µmol l −1 .

Cell counts
For determination of cell numbers, heterocyst frequency and cell dimensions, samples were filtered on white celluloseacetate filters (25 mm ∅ 1.2 µm pore size AE95 Schleicher & Schüll, Dassel, Germany) under low vacuum (200 mbar).Photographs were taken with an Observer A1 microscope and an AxioCam MRc (Carl Zeiss, Jena, Germany).Width and length of vegetative cells and heterocysts were measured using the free computer program Image-J (Wayne Rasband, wayne@codon.nih.gov,NIH, Bethesda, MD, USA).For cell counts, duplicate samples of 30 ml were filtered and stored dry.Twenty two to 28 photographs were taken systematically in a transect covering the diameter of the filter and additional 15 photos were taken randomly.The contrast of the photographs was altered by a macro in Photoshop software version CS3 (Adobe systems, San José, CA, USA) in order to achieve the spreading value based on a colour histogram that could then be correlated linearly with the surface covered by the cells.This spreading value was calibrated against exemplary cell counts that were estimated by using the com-puter program Image-J.Corrections were made considering the heterocyst frequency and dimensions.

Statistics
Cell division rates (µ) were calculated according to: (n 1 =cell number at t 1 , n 2 =cell number at t 2 , δt=t 2 − t 1 ).Scatter plots were constructed using the program Statistica 6.0 (StatSoft Inc., Tulsa, USA).Each replicate bottle is represented by one data point.Regression lines represent a Pearson correlation with regression bands depicting the 95% confidence limits and determination coefficient r 2 for the fitted line.The p value is calculated from an F-test, testing the null hypothesis that the overall slope is zero and that there is no linear relationship between x and y.

Results
Cell division rates differed significantly among treatments, reaching the maximum values of ∼0.52 d −1 at the lowest CO 2 level and the minimum values of ∼0.33 d −1 at elevated CO 2 levels (Fig. 1a).This resulted in a total decrease in cell division rate of 36% over the experimental CO 2 range, while cellular Chlorophyll a did not change with CO 2 treatment (Fig. 1b).A slight decrease in cellular nitrogen fixation in relation to rising pCO 2 was tested to be barely significant (p=0.012)(Fig. 1c).A significant increase in cellular carbon and phosphorus content with rising [CO 2 ] was detected (Fig. 2a and c).Taking the regression line as a mean, carbon and phosphorus cell quota increased from low to high CO 2 treatments by 32% and 30%, respectively.In contrast to carbon and phosphorus cell quota, the cellular nitrogen content did not show a clear trend with CO 2 (Fig. 2b).Rates of cellular carbon and phosphorus production (calculated from cell quota and cell division rate) did not show a significant trend over the experimental CO 2 range (Fig. 3a and c).A decreasing trend with CO 2 was obtained for nitrogen production derived from cell quota and division rate, comparable to and of similar statistical significance as measured nitrogen fixation rates (Fig. 1c), but with a steeper slope (Fig. 3b).Despite distinct differences in cell quota, no change in cell dimensions or heterocyst frequency could be detected in response to the CO 2 treatment.The mean length of vegetative cells was 3.8±0.46µm (n=601), at a filament width of 10.7±0.91 µm (n=125).Heterocyst length was 8.8±1.2 µm (n=63).In all samples, one of 12 cells±1 was a heterocyst.Carbon to nitrogen ratios exhibited a highly significant (p<0.001)increase (26%) in response to elevated [CO 2 ] and lowered pH.At low [CO 2 ], C/N was about 5.5 and thus below the Redfield ratio while a maximum value (7.0) at high [CO 2 ] slightly exceeded the Redfield ratio (Fig. 4a).In contrast to this trend in C/N, the carbon to phosphorus ratio was not affected by changes in [CO 2 ], remaining constant at about 170, which is 58% above the Redfield value (Fig. 4b).
Consistent with the different responses of nitrogen and phosphorus cell quotas, the N/P ratio showed a declining trend with increasing CO 2 , complementary to C/N ratio (Fig. 4c).

Discussion
Various studies in recent years have demonstrated direct effects of rising CO 2 concentrations on cell division rates and/or carbon fixation in mono-specific cultures of eukaryotic (Burkhardt et al., 1999;Hinga, 2002;Yang and Gao, 2003;Riebesell, 2004) and prokaryotic (Barcelos e Ramos et al., 2007;Fu et al., 2007Fu et al., , 2008;;Hutchins et al., 2007;Levitan et al., 2007;Kranz et al., 2009) marine phytoplankton.Results from culture experiments are corroborated by studies on natural marine phytoplankton assemblages (Hein and Sand-Jensen, 1997;Riebesell et al., 2007), demonstrating that, in cases where a CO 2 effect has been detected, it resulted in the stimulation of cell division, carbon and nitrogen fixation.In contrast, in the present study a mono-specific culture of Nodularia spumigena revealed a decrease in division rates in response to increasing pCO 2 .Aside from the unexpected slope of this trend, a surprising observation also was that the inverse relationship of cell division rate and pCO 2 extended to a CO 2 partial pressure of 150 ppm, i.e. well below the preindustrial level of 280 ppm.The fact that the trend did not level off towards the low CO 2 concentrations suggests that maximum cell division rate may occur at even higher pH and lower [CO 2 ] than tested here, values quite untypical for seawater.
The decrease in cell division rate with decreasing pH may be explained in the context of the natural growth conditions of Nodularia in seasonally or locally alkaline environments.These alkaline conditions are frequently caused when primary production results in a strong CO 2 drawdown in poorly buffered brackish water (Thomas and Schneider, 1999).Cyanobacteria in terrestrial habitats (i.e.lichens and microbial mats; Hallingbaeck, 1991) and in lakes are known to react in a similar way to acidification as Nodularia did in this experiment (Shapiro and Wright, 1990;Whitton and Potts, 2000).Acidification of different ranges (between pH 7.7 and 4.4) caused by anthropogenic atmospheric deposition of strong acids (H 2 SO 4 , HNO 3 ) substantially changed the phytoplankton communities of many lakes.Cyanobacteria were the only phytoplanktonic group that became nearly extinct in these acidified lakes (Findlay, 2003).Especially the genus Nostoc, a close relative of Nodularia, is described by Mollenhauer et al. (1999) as an "endangered constituent of European inland aquatic biodiversity".The explanation for the particular success of cyanobacteria under alkaline conditions is still unclear (Whitton and Potts, 2000).
With rising [CO 2 ], an increase of organic carbon and phosphorus within the cells could be detected while cell size remained constant.Therefore, a substantial increase in cell density at elevated [CO 2 ] can be assumed (Allan, 1984; Sellner, 1997).As cellular nitrogen content was not influenced by the treatment a highly significant increase of the elemental C/N ratio was measured in the formed biomass.As cellular Chla content as well as the production rates of carbon and phosphorus in organic matter did not seem to be negatively affected by the treatment, it is clear that cell division was neither limited by carbon fixation nor by energy supply.In fact, the accumulation of phosphorus and carbon in cellular reservoirs is the result of the reduced division rate.The mechanism responsible for the observed pH/CO 2 sensitivity of cell division rate is still unknown.
If the reduced division rates alone resulted in the storage of nutrients that would have otherwise been distributed among daughter cells, storage of carbon, phosphorus and nitrogen proportional to the cell division rate would be expected.However, in this experiment, cellular carbon and phosphorus content showed a much stronger increase with rising experimental [CO 2 ] than cellular nitrogen content.As the decrease in cellular nitrogen fixation rates was not large enough to account for the strong shifts in N/P ratios, impeded nitrogen transfer from heterocysts to vegetative cells seems to be the most reasonable explanation.
While in the non-heterocystic cyanobacterium Trichodesmium increased nitrogen fixation rates with rising CO 2 levels were interpreted to benefit of surplus energy from photosynthesis, in the present study nitrogen fixation in the heterocystous Nodularia was not enhanced by the treatment (Barcelos e Ramos et al., 2007;Hutchins et al., 2007;Levitan et al., 2007;Kranz et al., 2009).In heterocystous cyanobacteria, nitrogen fixation is spatially separated from the bulk of photosynthetically derived energy.The energy (ATP) and the reductive power (NADPH, ferredoxin) for nitrogen fixation in heterocysts are partly derived from cyclic electron transfer in photosystem I inside the heterocysts and partly from the pentose phosphate cycle that is supplied with carbohydrates from adjacent vegetative cells (Wolk et al., 1994).Heterocysts are probably not directly affected by [CO 2 ] as they do not fix CO 2 themselves because they lack photosystem II and RUBISCO (reviewed in Böhme, 1998).However, a pH effect on exchange processes between heterocysts and vegetative cells seems possible.A defective transport between heterocysts and vegetative cells under low pH could provide an explanation for the relatively weaker accumulation of particulate organic nitrogen compared to carbon and phosphorus.Nitrogen fixed by heterocysts is transferred to vegetative cells by high affinity (active) and low affinity (passive) transport of amino acids (Montesinos et al., 1995).Among others, the acidic amino acid glutamic acid and the basic amino acid arginine play major roles as vehicles of fixed nitrogen out of heterocysts and into the vegetative cells (Böhme, 1998).This exchange mainly occurs by diffusion between the end membranes of two adjacent cells (Wolk et al., 1994), therefore amino acids have to pass through the external media.Due to the ion charge, weak acids can pass transporters only in the protonated form and weak bases can pass only in the deprotonated form.Thus, the transport of weak acids and bases exhibits a high sensitivity to pH differences across the cell membrane (Decoursey, 2003).Based on the finding that cell division rate is pH dependent, one may speculate that Nodularia is, also concerning its intracellular pH, adapted to the temporal occurrence of basic microenvironments.Assuming a rather constant internal pH of the cells, the uptake www.biogeosciences.net/6/1865/2009/Biogeosciences, 6, 1865Biogeosciences, 6, -1875Biogeosciences, 6, , 2009 of basic substances like arginine, in more acidic environment, would be hindered in vegetative cells, resulting in an increased loss rate of fixed nitrogen.Simultaneously, an accumulation of metabolites in heterocysts due to an impeded release of acidic compounds like glutamic acid because of an adverse proton gradient across the cell membrane is possible.This could cause an unbalance in the metabolism of heterocysts that could provide an explanation to the slight decrease in nitrogen fixation rates under low pH.The regulation of heterocyst differentiation by the availability of fixed nitrogen has been demonstrated in several studies (see references in Wolk et al., 1994).The constant heterocyst frequency observed in the present study could indicate that there was no limitation by the supply of fixed nitrogen to vegetative cells in all treatments.As the cultures were grown in N-free growth media, the observed frequency may also be the maximum number of heterocysts per vegetative cells that can be differentiated.When calm weather leads to an accumulation of Nodularia at the surface, high irradiance and associated high photosynthetic activities are likely to promote high nitrogen fixation rates (Paerl et al., 1985).A study of Surosz et al. (2006) showed that Anabeana flos-aquae, a relative of Nodularia with a similar autecology, reacts to nitrogen starvation with enhanced aggregation due to increased production of transparent exopolymer particles (TEP).In another study, Anabaena flos-aquae agglomerated in layers several centimetres thick was shown to exhibit higher nitrogenase activity than in dispersed filaments, despite high [O 2 ] caused by photosynthesis (Kangatharalingam et al., 1991).Hereby the ratio between nitrogen fixation and carbon fixation at the outer areas of the aggregation was lowest and increased towards the centre.Additionally, it seems reasonable that nitrogen storage is enhanced within surface scums also as the microenvironment is enriched with amino acids excreted by the heterocysts of neighbouring filaments.In summary, there are many hints supporting the hypothesis that aggregation of heterocystic cyanobacteria is a strategy to improve nitrogen fixation and storage.In contrast, in the present study C and P accumulated in cellular reservoirs was found in the constantly dispersed Nodularia filaments.In nature, phosphorus and carbon are in short supply within the surface scum, while short mixing events may provide the possibility for bacteria to store phosphorus and carbon.This ecological scenario could give a possible explanation for the observed pH preference and the strong accumulation of phosphorus and carbon relative to nitrogen in a homogeneous non-agglomerated culture.A cross check of this study using a setup were filaments are allowed to aggregate would be the logical follow-up.Unfortunately, aggregates formed by cultured Nodularia bear little similarity to natural aggregations and can not be sampled adequately using quantitative methods as applied in this study.Therefore, an experiment of this kind is likely to deliver ambiguous results that may prove difficult to extrapolate to in situ conditions.
Aggregation in clusters and microbial mats is a phenomenon observed for many planktonic, benthic and terrestrial cyanobacteria.In the Baltic Sea, Nodularia is wellknown for forming dense toxic surface scums that cause considerable nuisance along the coastlines every summer.It often dominates the cyanobacterial community under relatively calm weather conditions, when aggregate formation is most prominent.When turbid conditions or storms interrupt calm weather, picocyanobacteria and other filamentous species that are usually more dispersed in the water column become dominant (Kononen, 1992;Sellner, 1997;Stal et al., 2003).Hence, it appears that Nodularia benefits from the physical or chemical microenvironment prevailing in surface aggregations but the mechanisms involved are poorly known and there are only a few studies showing that aggregation can be a purposeful process in cyanobacteria (Ohmori et al., 1992, Koblížek et al., 2000).
In surface scums, especially in poorly buffered brackish waters like the Baltic Sea (Thomas and Schneider, 1999), pH can rise several units and DIC can be significantly lowered due to high photosynthetic demand for CO 2 .Alternating conditions of pH 9 at daytime and pH 7 in darkness were measured inside Nodularia aggregations (Ploug, 2008).According to several authors, cyanobacteria outcompete eukaryotes under high pH and low CO 2 conditions of freshwater blooms (Shapiro and Wright, 1990;Oliver and Ganf, 2000).Observations that dispersed Nodularia filaments were not fertilised by elevated [CO 2 ], as seen for Trichodesmium, could indicate that Nodularia possesses a similar ecological strategy as their freshwater relatives.A high affinity CCM apparatus that would allow Nodularia to outcompete other phytoplankton in a CO 2 limited microenvironment of a dense cyanobacterial bloom could be an ecological specialisation that can not be down regulated sufficiently to benefit from [CO 2 ] as high as applied in the present study.
An explanation for the reduced division rates at pH-values commonly found in seawater can not be given.This emphasises that there is an urgent need to investigate pH dependent mechanisms that could be responsible for the observed effects.It can be speculated that the pH preference found in Nodularia is an adaptation to the chemical microenvironment caused by photosynthesis in aggregations.But how and whether aggregated Nodularia, adapted to diurnal pH variations ranging from 7 to 9 (Ploug, 2008), react to comparably small changes caused by an atmospheric [CO 2 ] increase of a few hundreds of ppm is questionable.Considering a pH of 9 inside the aggregations, it is conceivable that a more acidic seawater projected for the future could cause a relief to problems in carbon acquisition.Aggregation is the crucial factor controlling the chemical environment of many important aquatic diazotrophs that must be considered when assessing future oceanic carbon and nitrogen fluxes.
The present study clearly shows that under pre-bloom conditions, with single filaments dispersed in the upper water column, cell division is negatively influenced by ocean acidification.This could lead to a progressive delay in the formation of aggregates and thus in the initiation of Nodularia blooms.As the development of Nodularia is delayed, it could be outcompeted by other phytoplankton species that are either less or positively affected by rising [CO 2 ].Anabaena, a cyanobacterium co-occurring with Nodularia, is also not fertilised by elevated [CO 2 ] and reacts with shifts in elemental composition different than those reported here for Nodularia (unpublished data).Since cyanobacterial blooms in the Baltic Sea are always composed of different species, it is probable that there will be a gradual change in species composition (Tortell et al., 2002a).Therefore, a CO 2 related decrease of the Baltic Sea nitrogen budget cannot be postulated.Carbon export is more likely to be enhanced if N. spumigena is replaced by other species since it is known that N. spumigena, due to the buoyancy of persisting gas vacuoles and living filaments, decomposes largely in the upper water column (Hoppe, 1981;Sellner, 1997).

Fig. 1 .
Fig. 1.Cellular division rate (A), cellular chlorophyll a content (B) and nitrogen fixation rate (C) as a function of CO 2 partial pressure and corresponding pH (on the free scale).Each data point represents one bottle.The correlation coefficient r 2 and p value of the regression (solid line) are given.The dashed line represents the 95% confident intervals.

Fig. 2 .
Fig. 2. Variations in cell quota of carbon (A), nitrogen (B) and phosphorus (C) as a function of CO 2 partial pressure and corresponding pH (on the free scale).Each data point represents one bottle.The correlation coefficient r 2 and p value of the regression (solid line) are given.The dashed line represents the 95% confident intervals.

Fig. 3 .
Fig. 3. Cellular net production rates of POC (A), PON (B) and POP (C) as a function of CO 2 partial pressure and corresponding pH (on the free scale).Each data point represents one bottle.The correlation coefficient r 2 and p value of the regression (solid line) are given.The dashed line represents the 95% confident intervals.

Fig. 4 .
Fig. 4. Variation in C/N (A), C/P (B) and N/P (C) molar ratios as a function of CO 2 partial pressure and corresponding pH (on the free scale).Each data point represents results from one bottle.The dotted line indicates the Redfield ratio.The correlation coefficient r 2 and p value of the regression (solid line) are given.The dashed line represents the 95% confident intervals.

Table 1 .
Values of the carbonate system in the 14 experimental units.Alkalinity was measured in the end of the experiment while DIC was determined at the beginning and in the end.pH values (on the free scale) and CO 2 partial pressures are calculated for the point at which half of the DIC consumed during the experiment has been taken up.