Effects of increased atmospheric CO 2 on small and intermediate sized osmotrophs during a nutrient induced phytoplankton bloom

Abstract. We report the transient population dynamic response of the osmotrophic community initiated by a nutrient pulse in mesocosms exposed to different pCO2 levels. Differences in phytoplankton and heterotrophic bacteria abundances associated with the CO2 treatment are also described. Coastal seawater was enclosed in floating mesocosms (27 m3) and nutrients were supplied initially in order to stimulate growth of microbial organisms, including the coccolitophorid Emiliania huxleyi. The mesocosms were modified to achieve 350 μatm (1×CO2), 700 μatm (2×CO2) and 1050 μatm (3×CO2) CO2 pressure. The temporal dynamics was related to nutrient conditions in the enclosures. Numerically small osmotrophs (picoeukaryotes and Synechoccocus sp.) dominated initially and towards the end of the experiment, whereas intermediate sized osmotrophs bloomed as the initial bloom of small sized osmotrophs ceased. Maximum concentrations of E. huxleyi were approximately 4.6×103 cells ml−1 whereas other intermediate sized osmotrophs reached approximately twice as high concentrations. The osmotrophic succession pattern did not change, and neither were we able to detect differences with regard to presence or absence of specific osmotrophic taxa as a consequence of altered pCO2. Towards the end of the experiment we did, however, record significantly higher picoeukaryotic- and lower Synechococcus-abundances in the higher CO2 treatments. Slightly increased cell concentrations of E. huxleyi and other nanoeukaryotes were also recorded at elevated pCO2 on certain days.

initially and towards the end of the experiment, whereas intermediate sized osmotrophs bloomed as the initial bloom of small sized osmotrophs ceased. Maximum concentrations of E. huxleyi were approximately 4.6×10 3 cells ml −1 whereas other intermediate sized osmotrophs reached approximately twice as high concentrations. Osmotrophic succession pattern did not change, and we were not able to detect differences with re-1 Introduction 20 The pelagic food web is a complex and dynamic system where production is based largely on regenerated rather than new nutrients (Thingstad, 1998). In the pelagic zone nutrient limitation is believed to be a fundamental controlling factor for the community composition of osmotrophic microorganisms (organisms that feed on dissolved substrates) (Thingstad et al., 2005). Consequently, a change in inorganic nutrient avail-EGU phytoplankton community composition and succession, Pinhassi et al. (2006). Such amendments can in turn change the bacterioplankton community structure as a response to the growth and decay of various phytoplankton species or groups, indicating that dissolved organic matter from different algae select for different bacteria (Pinhassi et al., 2004;Grossart et al., 2005). Not only nutrients affect the osmotrophic commu-5 nity, however. Predation and lytic viruses are important mechanisms creating diversity and allowing for coexisting size classes of osmotrophs (Thingstad, 1998;Thingstad, 2000).
Phytoplankton and bacteria are key components of energy fluxes and nutrient cycling in the sea (Grossart et al., 2005). The major function of heterotrophic bacteria in 10 interactions with phytoplankton is organic matter degradation (Cole et al., 1988;Smith et al., 1995;Grossart and Simon, 1998). Because heterotrophic bacteria are the major consumers of dissolved organic matter in the aquatic environment, limitation of bacterial growth by organic or inorganic nutrients can have important consequences in terms of biogeochemical C cycling (Pinhassi et al., 2006). Also, an important mechanism for 15 the regulation of atmospheric CO 2 concentration is the fixation of CO 2 by marine phytoplankton and the subsequent export of the organically bound carbon to the deeper ocean (Engel et al., 2004).
The atmospheric CO 2 has increased from a pre-industrial level of 280 µatm to the present level of 370 µatm. Further increased atmospheric CO 2 concentration will lead 20 to a rise in the CO 2 concentration in the surface ocean and consequently a shift in its chemical equilibrium (Brewer et al., 1997). Some phytoplankton species (diatoms and the haptophyte Phaeocystis globosa) seem to get their CO 2 requirement fulfilled at the present day levels, whereas others (like the haptophyte Emiliania huxleyi) may benefit, in terms of increased primary production, from an increase in atmospheric CO 2 (Riebe- EGU eukaryotes were not affected by altered CO 2 environments (Engel et al., 2005). Seawater mesocosms allow studies of pCO 2 related impact on dynamics at a community level (Delille et al., 2005). Although not identical to the natural system they offer a good alternative that allow manipulation of complex ecosystems. We report result from the third mesocosms experiment carried out by the project Pelagic Ecosystem 5 CO 2 Enrichment Studies (PeECE). The two fist experiments had a maximum CO 2 concentrations corresponding to the atmospheric level expected in 2100 (710 µatm). We here go a step further with maximum level of 1050 µatm. The population dynamic in the osmotrophic community initiated by an initial nutrient pulse in mesocosms exposed to different pCO 2 levels as well as quantitative and qualitative variations in phytoplankton 10 and heterotrophic bacteria created by the difference in CO 2 exposure were monitored by flow cytometry and are currently described.

Experimental design and sampling
A mesocosm experiment was carried out at Marine Biological Station, University of 15 Bergen, Norway between 11 May and 10 June 2005. Nine polyethylene enclosures (2 m diameter and 9.5 m deep, volume 27 m 3 ) were mounted on floating frames, in a West-East line, and secured to a raft located in a small enclosed bay (Raunefjorden). The enclosures were filled on 11 May with 27 m 3 unfiltered, nutrient-poor, post-bloom fjord water. The atmospheric and seawater pCO 2 were manipulated to achieve levels 20 of 1050 µatm simulating 2150 conditions (3×CO 2 ), to 700 µatm in a year 2100 scenario (2×CO 2 ) and to 350 µatm CO 2 as the present scenario (1×CO 2 ). To initiate the development of a bloom of the coccolithophore Emiliania huxleyi (Haptophyta) nitrate and phosphate were added on day 0 (16 May) of the experiment, in a ratio of 25:1 yielding initial concentrations of approximately 15 µmol L −1 NO 3 and 0.6 µmol L −1 PO 4 EGU Samples for flow cytometric investigations were collected every second day for the first 6 days of the experiment and thereafter every day until the end of the investigation. For a full description of the experimental setup and sampling procedures, see Schulz et al. (2007) 1 .
2.2 Flow cytometry (FCM) 5 All FCM analyses were performed with a FACSCalibur flow cytometer (Becton Dickinson) equipped with an air-cooled laser providing 15 mW at 488 nm and with standard filter set-up. The phytoplankton counts were obtained from fresh samples at high flow rate (average 104 µl min −1 ). The trigger was set on red fluorescence and the samples were analysed for 300 s. Discrimination between populations was based on dot plots 10 of side scatter signal (SSC) and pigment autofluorescence (chlorophyll and phycoerythrin). We followed the dynamics of five different autotrophic phytoplankton populations (Synechococcus sp., Emiliania huxleyi, two unknown groups of nanoeukaryotes (differing in FL3 signal and hence in chlorophyll content) and picoeukaryotes ( Fig. 1a and b). Samples for enumeration of heterotrophic bacteria samples were fixed with glu-15 taraldehyde at a final concentration of 0.1% for 30 min at 4 • C, frozen in liquid nitrogen and stored at −70 • C until further analysis (Marie et al., 1999). Enumeration was performed for 60 s at an event rate between 100 and 1000 s −1 . Each sample was diluted at minimum two different dilutions from 10-to 200-fold in 0.2 µm filtered seawater and stained with SYBR Green I (Molecular Probes Inc., Eugene, OR) for 10 min at 80 • C 20 in the dark (Marie et al., 1999). The flow cytometer instrumentation and the remaining methodology followed the recommendations of Marie et al. (1999). Detection and enumeration of bacteria was based on scatter plots of SSC signal versus green DNA dye (SYBR Green) fluorescence, and we followed the development of total bacteria (Fig. 1c)

EGU
All concentrations were calculated from measured instrument flow rate, based on volumetric measurements, and all data files analyzed using EcoFlow (version 1.0.5, available from the authors).

5
The nutrients added at day 0 caused an increase in algal biomass (chlorophyll-a concentration) from approximately 2 µg chl-a l −1 to maximum values between 16 and 20 µg chl-a l −1 on day 9- 10 (Fig. 2, Schulz et al., 2007 1 ). Towards the end of the experiment a second, and much smaller, peak (3-4 µg chl-a l −1 ) was observed. The major part of the two chl-a peaks consisted of diatoms and dinoflagellates, respectively (large osmotrophs) (Schulz et al., 2007 1 ;Riebesell et al., 2007).

CO 2 effects
Chl-a concentrations did not vary greatly between the different treatments, but at the peak of the bloom (day 9-10) there was a tendency of higher chl-a concentrations in 15 the 2× and 3× compared to the 1×CO 2 mesocosm, and from day 13 onwards higher in the 3×CO 2 than in the rest of the mesocosms ( Fig. 2; Schulz et al., 2007 1 ). From day 0 to day 8 we did not observe any effect of the CO 2 treatment in any of the six groups of small and intermediate sized omotrophs (Fig. 3). As the bloom of E. huxleyi proceeded (day 9), however, somewhat higher E. huxleyi concentrations were registered 20 in the 3×CO 2 (≈4.6 x 10 3 cells ml −1 ) compared to the 1x (≈2.9 x 10 3 cells ml −1 ) and 2×CO 2 mesocosms (≈3.9 x 10 3 cells ml −1 ; Fig. 3c). A similar trend was also detected in nanoeukaryotes 1 and nanoeukaryote 2 from day 8 onwards (Fig. 3d, e). Towards the end of the experiment a more conspicuous CO 2 effect was observed within the small autotrophic osmotrophs (Fig. 3a,  EGU centrations (3×CO 2 ). The heterotrophic bacteria were not affected much by changes in CO 2 concentrations but a minute tendency of higher bacteria numbers in 3×CO 2 compared to the 1× and 2×CO 2 mesocosms was registered the last few days of the experiment (Fig. 3f). 5 4.1 Osmotrophic population dynamic

Discussion
As described in Tanaka et al. (2007) the inorganic nutrient environment that succeeded the initial nutrient manipulation can be divided into five different phases. Phase 1 (days 0-6) was characterized by no nutrient depletion and during phase 2 (days 7-11) the silicate (Si) got exhausted (phosphate (P) and nitrate (N) still being replete). In phase ses revealed that diatoms accounted for most of the chlorophyll during the main bloom (Riebesell et al., 2007;Schulz et al., 2007 1 ). The flow cytometry results presented here revealed a much more varied dynamic among the various osmotrophic groups: The initial nutrient pulse resulted in a community shift from small sized (picoplankton: heterotrophic bacteria, Synechococcus and picoeukaryotes) to intermediate (E. 20 huxleyi and other eukaryotic nanoflagellates) in addition to the big sized (diatoms) osmotrophs. On a competition to defence specialist axis (Thingstad et al., 2005) intermediate/big osmotrophs represent the latter characterized by features (e.g. size, silicate scale) making them less vulnerable for grazing (Thingstad, 1998;Hamm, 2000;Hamm et al., 2003) and/or infection (Raven and Waite, 2004), whereas the small osmotrophs 25 are thought to out-compete bigger ones when nutrients are low (Kuenen et al., 1977; 4180 ate/big) be explained? By looking more closely into the defence group (intermediate and big osmotrophs) it is evident that when silicate was exhausted (phase II) and thus limiting for further diatom growth, this gave room for the nanoeukaryotes (including E. huxleyi). Emiliania huxleyi has a high P-affinity (Riegmann et al., 2000) and ability to produce enzymes for utilization of phosphorus from organic substrates (Kuenzler and Perras, 1965). It could therefore potentially have a competitive advantage to other nanoeukaryotes as phosphate became depleted in phase III. The coccolithophorid experienced a viral attack, however (Larsen et al., 2007) giving room for Nanoeukaryotes 1 and 2 which retained with oscillations until phase V. Our analyses did not allow for species designation of Nanoeukaryotes 1 and 2, but several Chrysochromulina (Prym-15 nesiophyceae) and Pyramimonas (Prasinophycea) species are common nanoeukaryotes in our coastal waters (Throndsen et al., 2003), and species within these genera have proven susceptible to virus within the Phycodnaviridae familiy (Suttle and Chan, 1995;Sandaa et al., 2001). Studies of the viral community showed that CeV and two other closely related to viruses within the Phycodnavirideae were present (Larsen et 20 al., 2007). It may therefore well be that the different peaks contains different species with one species taking over when others are infected and killed. The observed oscillating development within the intermediate osmotrophs thus demonstrate how the "killing the winner mechanism" also apply for algae and algal viruses (Thingstad and Lignell, 1997;Thingstad, 2000). 25 We observed a simultaneous decrease of all small osmotrophs (heterophic bacteria, Synechococcus and picoeukaryotes) in phase I and IV (and towards the end of phase V). Such within-community similarities suggest a common size-selective predator (heterotrophic flagellates) as the major loss mechanism for the competition group (Fenchel,  1980;Fenchel, 1987;Thingstad et al., 2005). The coexistence within the group needs further explanations though and two theoretical ones come to mind: 1) growth rate limitation of heterotrophic bacteria by bioavailable organic carbon (Thingstad et al., 2007) and 2) differences in the ability to use organic nitrogen sources. Tanaka et al. (2007) concludes that bacterial growth was not limited by the availability of labile 5 DOC whereas mineral nutrients were depleted from phase four. The latter explanation thus seem more plausible and can also explain why the picoeukaryotes dominated the small sized autotrophic community in the beginning of the experiment (phase I) whereas Synechococcus took on the lead role in phase V. The bacterio-, cyanophages and algal virus dynamic demonstrated in Larsen et al. (2007) suggests viruses played 10 an essential role in the population dynamics within each of the three groups of small osmotrophs (Thingstad, 2000). It has already been mentioned that the initial nutrient addition was followed by a noticeably decrease in abundance of competition specialists (small sized osmotrophs: heterophic bacteria, Synechococcus and picoeukaryotes). However, when comparing 15 the concentration of these three groups with what we observed in the reference seawater it is evident that some mechanism prior to nutrient addition caused them to increase substantially. One possible explanation is that filling the mecosoms and/or bubbling the water to achieve the desired CO 2 levels killed off possible predators and/or released DOM, which they could have benefited from if they were nutrient limited in the fjord wa-20 ter prior to the experiment. The plankton community contains species that are fragile and therefore may be sensitive to the filling/bubbling procedure, but as neither DOM nor predator abundances were measured before and after onset of filling/bubbling the mesocosms we can no more than speculate that these were the mechanisms leading to the high initial concentration of small osmotrophs.

CO 2 effects on the osmotrophic community
The current study did not reveal omotrophic successional shifts that can be traced back to the altered CO 2 concentrations. Nor were we able to detect introduction or 4182 with lowered Synechococcus and elevated picoeukaryotes abundances, at the highest CO 2 level. Similar effects of varying CO 2 concentrations was not as evident for the remaining osmotrophs, but the trend of higher cell numbers with increasing CO 2 for all groups, except for Synechococcus, emerged more clearly when calculating total cell numbers for the entire experimental period for the autotrophic osmotrophs (Fig. 4).
Higher abundances of primary producers at the highest CO 2 level as the experiment progressed is in agreement with a somewhat higher total primary production in the second half of the experiment (Egge et al., 2007 2 ).
It has previously been documented that some phytoplankton species (E. huxleyi, G. oceanica) benefit, in terms of increased photosynthetic carbon fixation rates, from an 15 increase in CO 2 concentrations compared to the present day level (Riebesell et al., 2000;Rost et al., 2003) whereas others do not (P. pouchetii, several diatom species; Burkhardt et al., 1999Burkhardt et al., , 2001Rost et al., 2003). Riebesell (2004) conclude from this that the current increase in atmpospheric CO 2 will promote growth of calcifying primary producers. Our results do not necessarily support this conclusion as all intermediate 20 autotrophic osmotrophs (including the non calcifyers) seemed to experience a similar (and small) increase in abundance as CO 2 increased. One aspect that could interfere with our interpretation of possible CO 2 effect on the osmotrophs is the phytoplanktonvirus interactions that have a profound influence on the marine microbial systems (reviewed by Brussaard, 2004). Larsen et al. (2007) showed that one virus which infect E. 25 huxelyi and one that assumingly infect some other nanoeukaryote, occurred in higher numbers in mesocosms with the lowest CO 2 level. This is obviously an additional reason for lower E. huxleyi-and nanoeukaryotes 1 and 2 concentrations in these very same enclosures.

EGU
The only group with higher biomass (this study) and production (Egge et al., 2007 2 ) at 1×CO 2 than at 2× and 3×CO 2 was Synechococcus. Engel et al., 2005, report that average abundances of Synechococcus in a similar mesocosm experiment in 2001 was not affected by the CO 2 concentrations, but a closer inspection of the osmotrophic dynamic (presented by Rochelle-Newall, 2004, Fig. 2) reveal that also in that case the 5 densest Synechococcus population occurred within the enclosure exposed to the lowest CO 2 concentration. In both experiments this is a result only visible towards the end when inorganic N and P are depleted and osmotrophic production depends on remineralised nutrients. Direct competition experiments have demonstrated that low CO 2 concentrations favour the growth of cyanobacteria over other phytoplankton species in 10 freshwater systems (Shapiro, 1973), and that freshwater Synechococcus compete well for dissolved inorganic carbon (Williams and Turpin, 1987). Cyanobacteria in general (Badger and Price, 2003), and more specifically marine Synechococcus (Hassidim et al., 1997), have demonstrated effective photosynthetic CO 2 concentrating mechanisms (CCMs). The observed Synechococcus dominance in phase V could thus be a com-15 bined effect of its superiority over picoeukaryotes in competition for dissolved organic nitrogen (as discussed above) and for dissolved inorganic carbon (DIC). In order for the latter to be the case, however, DIC must have been limiting. The fact that picoeukaryotic abundance increased considerably when CO 2 concentration was raised to 1050 µatm (Fig. 3) indicates that this could have been true. Prasinophytes (the ma-20 rine counterpart to green algae, frequently represented by Micromonas pusilla) are often dominating the picoeukaryotic communities in coastal and nutrient rich environments (Not et al., 2005). Our results may thus illustrate that comparable to fresh water green algae (Shapiro, 1973), this group increase on behalf of cyanobacteria when CO 2 increases. 2×CO 2 equals the highest CO 2 level tested in 2001, and in neither exper-

Printer-friendly Version
Interactive Discussion EGU this study, and Fig. 2 in Rochelle-Newall, 2004). Grossart et al. (2006) were not able to detect significant changes in heterotrophic bacterial abundance as a result of a variable CO 2 environment and link the indirect effect of changes in pCO2 on bacterial activities to phytoplankton dynamics. In the current experiment the effect, if any, was a slight tendency of higher concentration in 5 3×CO 2 mesocosms than in 1× and 2×CO 2 , and only detectable towards the end of the experiment. This might have been a secondary effect of more nanoeukaryotic cells being terminated, releasing higher amounts of DOM in phase IV, in these enclosures.

Concluding remarks
The osmotrophic community within our mesocosms may have experienced three per-10 turbing events: Nutrient addition, a potentially effect of the filling and/or bubbling procedures, and CO 2 manipulations. By contributing significantly to the early success of the small sized osmotrophs, the bubbling/filling did perhaps influence the onset of the observed community composition shifts. However, the bloom of defence specialists/intermediate sized phytoplankton that can be foreseen as a consequence of ele-15 vated nutrient concentrations (Thingstad et al., 2005) was apparently not disturbed by this. A series of community composition shifts succeeded the initial nutrient amendment and as such this seemed, not surprisingly to be the single one parameter affecting the microbial community most profoundly. The effect of the CO 2 manipulations was not quite as obvious, probably because short time experiments like the current 20 do not provide sufficient time to create differences detectable as successional shifts and introduction or removal of certain taxonomic groups. Nevertheless, our results do substantiate previous works suggesting that CO 2 variations influence the relative taxonomic composition of marine phytoplankton (Tortell et al., 2002;Grossart et al., 2006;Engel et al., 2007). These differences were most noticeable towards the end of the EGU diate sized osmotrophs had increased their importance relatively to the diatoms (this study; Riebesell et al., 2007;Schulz et al., 2007 1 ). A number of CCM variants, differing in manner of operation and efficiency, are found among different phytoplankton groups, and nutrient availability is also known to play a significant role in modulating CCMs (reviewed by Giordano et al., 2005). It is therefore difficult to judge whether 5 our observations suggest that increase in atmospheric CO 2 will have a greater effect when production is based on regenerated nutrients, or whether they rather reflect that small and intermediate sized osmotrophs are not equipped with carbon concentration mechanisms as efficient as the diatoms and therefore benefit more from increased CO 2 levels than the latter (John et al., 2007). The experiment do, however, illustrate as previously suggested (Tortell, 2000), that the competitive balances between microbial taxa may be altered when atmospheric CO 2 changes. 4,2007 Effects of CO2 increase on osmotrophs  J. Phycol., 18, 275-284, 1982. Smith, D. C., Steward, G. F., Long, R. A., and Azam, F.: Bacterial mediation of carbon fluxes during a diatom bloom in a mesocosm, Deep-Sea Res. Pt Ii, 42, 75-97, 1995. Suttle, C. and Chan, A.: Viruses infecting the marine prymnesiophyte Chrysochromulina spp.: