Micro-and mesozooplankton community response to increasing CO 2 levels in the Baltic Sea : insights from a large-scale mesocosm experiment

Introduction Conclusions References


Introduction
Since the industrial revolution, anthropogenic CO 2 emissions have increased at an unprecedented rate and cause a concomitant increase of CO 2 concentration in the oceans.Thereby, ocean carbonate chemistry is altered with the main changes being reduced carbonate ion concentrations CO 2− 3 and increased proton concentrations [H + ] leading to a pH decrease.This phenomenon is nowadays well recognized as ocean acidification (OA).Since the beginning of the industrial revolution, ocean pH has decreased by approx.0.1 units and projections suggest a further decrease of 0.14-0.43units by the end of the century (IPCC, 2013).The Baltic Sea, one of the largest brackish water systems, is especially sensitive to CO 2 changes because it naturally has low alkalinity and thus carbonate buffer capacity.Models project a drop of 0.5 pH units for the Baltic Sea by the year 2100 (Hjalmarsson et al., 2008;Havenhand, 2012;Omstedt et al., 2012).Eutrophication specifically affects coastal areas and can add to the fCO 2 fluctuations by provoking low oxygen partial pressure due to increased degradation processes, respectively respiration.Therefore, diel and seasonal variations of carbonate chemistry parameters particularly of coastal areas of the Baltic Sea are already huge today and the amplitude of fluctuations has even increased since the beginning of the industrialization and concomitant eutrophication (Omstedt et al., 2009;Melzner et al., 2013;Jansson et al., 2013).Consequently, zooplankton in the coastal Baltic naturally experiences large pH fluctuations on a daily and seasonal basis and possibly are at least to some extent adapted to these highly variable abiotic conditions (Melzner et al., 2013;Almén et al., 2014).
Ocean acidification is suspected to have severe consequences for marine organisms and acts synergistically with the concurrent temperature increase due to greenhouse Introduction

Conclusions References
Tables Figures

Back Close
Full gas emissions (Riebesell et al., 2009).Until now, most attempts to test for sensitivities of marine organisms to OA were conducted as single species experiments under controlled (optimal) laboratory conditions.Such an approach can not account for community interactions in natural environments, and thus application of results to natural environments is limited.Laboratory experiments suggest calcifying organisms to be most vulnerable to OA because the formation and preservation of calcareous structures is hindered (Riebesell et al., 2000;Hoegh-Guldberg et al., 2007;Lischka et al., 2011).Non-calcareous micro-and mesozooplankton is generally considered quite robust to elevated CO 2 concentrations.Effects on the microzooplankton level seem to be of more indirect nature through changes in primary production, phytoplankton community composition and stoichiometry (Suffrian et al., 2008;Feng et al., 2009;Rossoll et al., 2012).Mesozooplankton is often dominated by copepods (Longhurst, 1985) which are relatively insensitive to fCO 2 /pH changes expected for this century and direct negative effects usually do not occur unless exposed to much higher fCO 2 levels projected only much later (Kurihara et al., 2004;IPCC, 2013).More recent evidence suggests, however, that nauplii stages may be the weak point in copepod's life cycles (Cripps et al., 2014).As for the microzooplankton, studies on copepods and cladocerans suggest CO 2 effects may be more indirectly mediated to the zooplankton level through CO 2 induced changes in the biochemical and/or stoichiometric composition of their food (Urabe et al., 2003;Rossoll et al., 2012).
Holistic approaches studying CO 2 effects on entire natural plankton communities including zooplankton are still rare.In a preceding similar mesocosm experiment, Aberle et al. (2013) and Niehoff et al. (2013) found no effects on Arctic micro-and mesozooplankton communities, neither with respect to abundance of single species or total numbers nor with respects change in community diversity.In terms of ciliates, these communities were dominated by large-sized forms (> 30 µm), in terms of mesozooplankton by copepods and cirripedia larvae.Among microzooplankton, ciliates and heterotrophic dinoflagellates dominate in summer in Tvärminne/Storfärden, among mesozooplankton rotifers, copepods and cladocera (Kivi, 1986;Viitasalo, 1992;Koski et al., 1999).Introduction

Conclusions References
Tables Figures

Back Close
Full The amount of carbon transferred to higher trophic levels depends on the mesozooplankton species composition (Hansen et al., 1994).In Storfärden, during late summer and autumn, the microbial loop is of particular importance when filter-feeding cladocerans mediate carbon transfer to higher trophic levels including fish (Koski et al., 1999, and references therein).
As part of the KOSMOS Tvärminne mesocosm experiment, we examined CO 2 effects on the enclosed micro-and mesozooplankton community.Between June and August 2012, an fCO 2 gradient was set up in approximately six 55 m 3 mesocosms covering fCO 2 projections for this century or beyond (IPCC, 2013).Abundance and community composition was followed through enumeration of regularly taken waterand net samples.Per definition, micro-and mesozooplankton include heteroptrophic proto-and/or metazoa ranging between 0.02-0.2mm (20-200 µm) and 0.2-20 mm (200-20,000 µm) in size, respectively.In this study, we do not follow this classification strictly.The category "microzooplankton" (MiZP) comprises ciliates only, including some species that can be facultative autotrophs or obligate phototrophs (for instance Myrionecta rubra), whereas all metazoa independent of their body size were assigned to the category "mesozooplankton" (MZP).Temperature can have a general effect on MiZP abundance and community composition and governs the dynamics of crustacean species (for instance affects productivity of cladocerans) in late summer in our study area (Nanazato and Yasuno, 1985;Koski et al., 1999;Rose et al., 2009;Aberle et al., 2013).To consider possible impact of temperature variation and/or CO 2 driven chlorophyll a differences (Schulz et al., 2013), in addition to fCO 2 , we also included temperature and chlorophyll a as explanatory variables in our statistical analyses.

Methods
To study the effect of elevated fCO 2 on a natural plankton community in the Baltic Sea, nine KOSMOS offshore pelagic mesocosms (Kiel Off-Shore Mesocosms for future Ocean Simulation) were deployed and moored on 12 June 2012 until the middle Introduction

Conclusions References
Tables Figures

Back Close
Full of August in the Tvärminne/Storfjärden archipelago area at the south-west coast of Finland at 59 • 51.5 N and 23 • 15.5 E. The water depth at the mooring site was approximately 30 m.The mesocosm bags extended down to 17 m and were closed with 2 m long sediment traps at the bottom of the bags to enclose an isolated water body with its natural plankton community.After deployment, the mesocosm bags were initially kept open and submerged ∼ 0.5 m below the surface to allow for a free exchange of the water and plankton community in the bags with the surrounding water masses.
Organisms > 3 mm such as fish and cnidaria were excluded by 3 mm nets at the top and bottom openings of the bags during the first five days.These nets were removed on t −7 (i.e.seven days before the first CO 2 addition on t 0 ), the sediment traps were attached to the bottom, and the top ends of the mesocosm bags pulled up to 1.5 m above the surface to isolate the enclosed pelagic community from the Baltic Sea.The final volumes of the mesocosms ranged between 53.1 and 55.1 m 3 (Paul et al., 2015).
The nine mesocosms were enriched with different amounts of CO 2 saturated seawater to set up an initial gradient of fCO 2 from 240 µatm (ambient, control mesocosms) up to ∼ 1650 µatm.Three mesocosms (M2, M4, M9) were lost during the course of the experiment due to leakage.fCO 2 values in the six remaining mesocosms averaged over the sampling period (t 1 -t 43 ) were 365 µatm (M1 control), 368 µatm (M5, control), 497 µatm (M7), 821 µatm (M6), 1007 µatm (M3) and 1231 µatm (M8).CTD profiles and samples for dissolved inorganic nutrients (silicate, phosphate, nitrate, nitrite, ammonium) and carbonate chemistry system parameters (DIC, TA, pH T ) were either taken daily or every second day.For more technical details about the experimental set-up, the CO 2 manipulations, and sampling procedures for various analyses see Paul et al. (2015).Sampling days were enumerated consecutively with t −3 indicating three days before CO 2 manipulation, t 0 as the day of the first CO 2 manipulation, and t 1+X as the days following the first CO 2 manipulation.Introduction

Conclusions References
Tables Figures

Back Close
Full

Microzooplankton sampling
Water samples for the enumeration of ciliates were taken every second day with a depth-integrating sampler (0-17 m), IWS (HYDRO-BIOS, Kiel, Germany), between 09:00 and 12:00 a.m. from six mesocosms.After careful mixing, 250 ml of seawater were filled into brown-glass bottles and preserved in acidic Lugol's iodine (1 % final concentration).50 ml of the sample were transferred to Utermöhl sedimentation chambers.After 24 h settling time, ciliates were counted with a Zeiss Axiovert 100 inverted microscope at 200× magnification Utermöhl (1958).At high cell numbers (> 400 cells), half the bottom plate area was counted.If less than 400 cells were found in the first half of the bottom plate area, the entire chamber was counted.Rare species were counted on the whole bottom plate.Ciliates were identified to the lowest possible taxonomic level (genus/species) according to Setälä et al. (2009) andTelesh et al. (2009).138 samples were analyzed in total.Abundances were calculated as cells L −1 .

Mesozooplankton sampling
Mesozooplankton samples from six mesocosms were taken with an Apstein net of 17 cm diameter and 100 µm mesh size.Zooplankton were sampled between 08:00 and 11:00 am by towing the net vertically from 17 m depth to the mesocosm surface.
In total, at eleven sampling days, vertical net hauls were done from the mesocosms: prior to the CO 2 addition (t −3 , t −2 , t −1 ), at the day of the first CO 2 addition (t 0 ), and after the first CO 2 addition (t 3 , t 10 , t 17 , t 24 , t 31 , t 38 , t 45 ).After collection, the samples were brought back to the lab in the Tvärminne zoological station (University of Helsinki) and preserved in 70 % ethanol.Zooplankton abundance was calculated assuming 100 % filtering efficiency of the net.The samples were divided with a Folsom plankton splitter (1 : 2, 1 : 4, 1 : 8, 1 : 16, 1 : 32) and the aliquots of the samples were counted.Organisms were counted and determined under a stereo microscope (WILD M3B) to the lowest taxonomical level possible.Abundant species/taxa (> 30 individuals in an aliquot) were only counted from subsamples, while less abundant species/taxa Introduction

Conclusions References
Tables Figures

Back Close
Full were counted from the whole sample.Juvenile bivalves did not distribute equally in the Folsom splitter due to their relatively large mass and were therefore counted from the whole sample.Copepods (Acartia spp., Eurytemora spp., Temora spp.) were identified according to different stages (adult females, adult males, copepodite stages CI-CV).
Copepod nauplii were counted but not determined to species level.The counting of the cladoceran species (Bosmina spp., Evadne spp., Podon spp.) was distinguished according to organisms with empty or filled brood chambers, respectively (i.e.organisms that had empty brood chambers or bore embryos/resting eggs, respectively, in their brood chambers) and categorized as 'empty' or 'filled'.For data analyses, the ratio between the number of organisms with 'empty' to 'filled' individuals was calculated for each mesocosm and sampling day, i.e. a small ratio stands for a higher proportion of reproducing organisms in the population in a particular mesocosm at a particular sampling day.A total of 66 samples were analyzed.Abundances were calculated as individuals m −3 .

Data analysis and statistics
To assure equally spaced data, some sampling days were excluded from statistical analyses.For the microzooplankton data this applied to t −3 , t 0 , t 2 and t 4 , and for the mesozoopankton this applied to t −3 , t −2 , t −1 and t 0 .However, for demonstration purpose only, the data of these sampling days were included in the figures.As explanatory variables, fCO 2 , temperature and chlorophyll a were used to test for effects on different response variables (see below).Collinearity was checked prior to analyses.To account for the change in fCO 2 over time due to ingassing/outgassing as well as temperature and chlorophyll a changes over time, all explanatory variables were used as continuous variable for each t day included in the analyses.All analyses were carried out with R using the package nlme, mgcv, Hmisc and MASS.All plots were done in ggplot (R Development Core Team, 2012).The Shannon index (H) was calculated as a measure of diversity in each of the mesocosms and to estimate changes in the relative contribution of single species/groups 20032 Introduction

Conclusions References
Tables Figures

Back Close
Full in the whole micro-/mesozooplankton community over time and in response to different abiotic parameters such as the fCO 2 levels.When all considered species/groups contribute equally to the community in terms of their abundances, H calculated on the natural logarithm becomes 2. For the meosozooplankton, 17 species or taxonomic groups were included in the calculation of H: copepodite stages and larval stages of Balanus sp.(nauplii and cypris larvae) were summarized on the genus level (Copepoda: Acartia sp., Eurytemora sp., Temora sp., Harpacticoida sp., copepod nauplii; Cladocera: Bosmina sp., Daphnia sp., Evadne sp., Podon sp.; Rotifera: Asplanchna sp., Keratella sp., Synchaeta sp., Rotifera sp.; larvae of Balanus sp., juvenile bivalves, juvenile gastropods, and larvae of polychaets).

Microzooplankton
Statistical analyses were done on total cell numbers, the Shannon index H as well as the abundance of particular groups that showed distinct differences such as small size-class Myrionecta rubra, Balanion comatum, Strombidium cf.epidemum, and small Strobilidium sp.Linear mixed effects modelling (LME) was applied on a Gaussian distribution to determine the effect of CO 2 , temperature and chlorophyll a. Actually, count data should be modelled on a Poisson distribution, but model selection (s.b.) yielded in convergence problems in R for Poisson distribution.Therefore, we used a Gaussian distribution, which can also be applied on count data (Zuur et al., 2009) Full ing data exploration suggested interactions between the factors, respective interaction terms were included in the model.Model selection was based on the Akaike information criterion (AIC) by removing non-significant terms to find the simplest adequate model.However, missing values for chlorophyll a occurred for M3/t 25 and for M5/t 23 , these values were estimated as means of the preceding and following day.Chlorophyll a values were also missing for t 41 and t 43 .A polynomial fit curve applied on phase III (according to temperature variations, three experimental phases were defined: phase I, II, II.Phase III lasted from t 31 until t 43 ; Paul et al., 2015) resulted in no meaningful values, therefore these values were estimated as phase III means.
The different response variables were modelled as a function of the daily change in fCO 2 , temperature and chlorophyll a and if suggested with interaction terms as mentioned above.To account for the time dependency and the nested nature of the data, GLM models (generalized mixed effects) were applied on a Gaussian distribution using fCO 2 (values on a continuous scale for each sampling day) and sampling day nested in mesocosm as random intercept.In case of violation of the assumptions for linear models yielding to non trustworthy p values, the GLM model was re-applied as a GA(M)M (generalized additive (mixed) model) and a smoother for sampling day included to prove the validity of the GLM outcome.In some cases, some residual patterns mostly due to sampling day still remained even after applying the GAMM.But GAMM is as much as can be done with current hard-and software, and therefore, for highly significant p values, our results should still be reasonably robust, and p values that are not highly significant should be seen with some caution (Zuur et al., 2009).

Mesozoplankton
The statistical approach with respect to MZP corresponded with description in section 2.2.1.Total abundance, the Shannon index H as well as total abundance of species that suggested distinct differences such as Bosmina and the ratio of Bosmina with empty to individuals with full brood chambers (i.e.either bearing embryos or resting eggs in their brood chambers) were analyzed statistically.Missing values for fCO 2 occurred on t 24 , 20034 Introduction

Conclusions References
Tables Figures

Microzooplankton total abundance
Total abundance of microzooplankton at experiment start (t 0 ) varied between 78 120 cells L −1 (M5) and 52 360 cells L −1 (M3) and more or less continually decreased from the beginning over time until t 17 when a plateau was reached with low cell numbers between 7080 (M8) and 10 940 (M3) until t 33 .During the last five sampling days (t 35t 43 ), total cell numbers were more variable again with some small ups and downs and reached minimum values between 900 cells L −1 (M6) and 3580 cells L −1 (M8) on the last sampling day (Fig. 1).

Abundance of Myrionecta rubra
Myrionecta rubra was (by far) the most dominant ciliate species during the entire period (Fig. 2a).M. rubra occurred in three different size classes ( 10 µm, 11-20 µm, > 20 µm) of which organisms of the smallest size range made up the highest num-Introduction

Conclusions References
Tables Figures

Back Close
Full and further decreased with some fluctuations until the end of the experiment to reach final values of between 130 cells L −1 and 1740 cells L −1 among all mesocosms.Some striking difference, however, occurred between t 25 -t 35 when abundance in the three highest CO 2 mesocosms was higher compared to the two controls and the lowest CO 2 enriched mesocosm (mean: 4518 cells L −1 (SD 1082) and mean: 3459 cells L −1 (SD 383), respectively).M. rubra of the medium size class also had maximum numbers on t 0 ranging from 17 600 cells L −1 to 25 680 cells L −1 .From the experiment start, numbers more or less continually decreased and reached minimum values of between 480 cells L −1 and 0 cells L −1 from t 19 on.The largest M. rubra occurred only rarely but as in the other two size classes, highest numbers were found during the first few sampling days varying between 2680-5800 cells L −1 on t 0 and reaching very low numbers already on t 7 /t 9 (1080-280 cells L −1 ).After t 19 , M. rubra > 20 µm occurred only exceptionally.

Abundance of other species/genera/groups
Other dominant groups/species that contributed to the total cell numbers of microzooplankton were Mesodinium sp., Balanion comatum, Strombidium cf.epidemum, Rimostrombidium sp., Strobilidium sp.(< 20 µm and > 20 µm), Strombidium sp., Tintinnids, Spathidium sp., cysts, and ciliates that could not be identified (Fig. 2b and  c).Among those, Strombidium cf.epidemum was most dominant and showed three peaks, around t 9 /t 11 , t 23 , and t 37 .Peak values ranged between 1160 cells L −1 and 4000 cells L −1 on t 9 /t 11 and showed some distinct difference between control and CO  Full The Shannon diversity index H ranged from 0.58-1.66over the whole period of time (Fig. 4).In general, it showed a slightly increasing trend varying between 1.04 and 1.23 on t −3 and, respectively 1.30 and 1.66 on t 43 .Overall, H showed a non-monotonic relationship with a slightly increasing trend at lower fCO 2 and a decreasing trend the more the fCO 2 increased, and as well as a decreasing trend with temperature.

Statistical analyses microzooplankton
GAMM's determined significant effects for total abundance, small size class Myrionecta rubra, Balanion comatum, Strombidium cf.epidemum, Strobilidium sp., and the Shannon index H in response to one or more of the included explanatory variables.Detailed statistical results are shown in Table 1.Model validation showed some residual pattern in all cases, but most of the obtained p values are highly significant and are therefore reasonably trustworthy (Zuur et al., 2009).Only with respect to Balanion comatum, p values should be seen with some caution as they are not highly significant.

Mesozooplankton total abundance
Total abundance of mesozooplankton caught in the net samples on t −2 varied between 4841 ind.m −3 in M1 and 31471 ind.m −3 in M8 (Fig. 5).During the course of the experiment abundances increased in all mesocosms continuously until peak abundances were reached between t 24 and t 31 .
Low abundances of cladocerans occurred only at the beginning (t −2 , t −1 ) and on t 17 when they only had a share of between 8 and 28 % of the whole community.Copepods had a relatively low share early in the experiment and at the end (∼10-20 %), but half way through the experiment copepods constituted 74-84 % (t 17 ) of the whole community.Rotifera were a major part of the zooplankton only during the first days of the experiment with about 11 % to 42 % between t −1 and t 3 , later on they almost disappeared in all mesocosms (<1 %) except for M8 where on t 45 2 % of the whole community were rotifera.Among the group mollusca, gastropods always had a smaller share than bivalves with usually below 2 % contribution to the total abundance of this group, only in a few cases (M1 and M5 on t 10 and t 24 ) gastropods had a share of 5 % and at the last day in M8 of 14 % of the mollusca group.Juvenile bivalves mainly occurred from the start until day t 10 and had maximum contributions of 17-45 % to the total zooplankton community between t −2 and t 0 .By day t 10 their abundances decreased to 7-0.6 % of the total community.The group 'crustacea' comprises mainly larvae of Balanus sp.
(nauplii and cyprids).Only very rarely a mysid was found and specimen of this order were also included in the group crustacea.The main occurrence of 'crustacea' was from t −1 until t 10 contributing between 10 and 2 % to the total zooplankton community

Conclusions References
Tables Figures

Back Close
Full during this time with highest numbers at the beginning of the experiment.The group 'others' always contributed less than 0.5 % to the total abundance.In all mesocosms, the Shannon diversity index was highest at the beginning of the experiment (T 3 : 1.78-1.89)and decreased continuously with time reaching lowest values on the last sampling day (T 45 : 0.23-0.5)indicating that towards the second half of the experiment and at the end, the dominance of single species/groups increased.

Copepoda
Eurytemora sp. was the dominant copepod species in the zooplankton community over the entire period.Acartia sp.occurred regularly but in much lower abundances.Temora sp.occurred only in very low numbers mainly during the first part of the experiment (Fig. 7a).The abundances of Eurytemora sp. were relatively low at the beginning until day t 10 ranging between 82 ind.m −3 and 2496 ind.m −3 .Peak abundances were reached around day t 17 and t 24 with between 19 192 ind.m The three copepod species were determined to copepodite stages (CI-CV) and adult females and males (Fig. 7b).Eurytemora sp.copepodites CI-CV were present in high proportions almost during the whole period of time with up to > 90 %, except on t 3 and t 10 they had a lower share in all mesocosms ( 50 %).Adult females and males had Introduction

Conclusions References
Tables Figures

Back Close
Full cosms.At the beginning and towards the end of the study, most of Acartia sp. were in the copepodite stage CI-CV.Adult females and males occurred during the whole period of time and had maximum proportions half way through the experiment (t 17 , t 24 ) when the share of copepodite stages was negligible.During this time, reproduction took place indicated by the following increase in copepodite stages during the second half of the study.The stage distribution of Temora sp. was similar to Acartia sp. with a peak of copepodite stages CI-CV during the first and the last sampling days.Most of the time, however, adult females and males dominated and overall Temora sp. was not of great importance in the copepod community.

Cladocera
Four species of cladocera were found in the mesocosms: Bosmina sp.Full The counting of the two dominant cladoceran species Podon sp. and Bosmina sp. was divided into organisms with empty brood chambers and organisms bearing embryos/resting eggs in their brood chambers to inspect for a possible direct or indirect effect of CO 2 on asexual/sexual reproduction and subsequently a ratio was calculated, s.a.(Fig. 9a and b).Mostly, the percent contribution of organisms with filled brood chambers varied between 40 and 10 % in all mesocosms among the study period.Only during the very first days, Bosmina sp. with filled chambers had contributions of up to 67 % (not shown).The ratio of Bosmina brood chambers varied during peak occurrence (t 24 -t 31 ) between 3.47 and 17.18.During the remaining days values fall between 0.15 and 8.44, except for M1 on t −3 when no organisms with filled brood chambers were counted.Podon sp.occurred mainly during the first part of the experiment until t 17 .During that time, the share of organisms with full brood chambers varied roughly between about 25 and 50 %.Later, the importance of Podon sp. was negligible, but among the few organisms that still occurred were both animals with empty and with full brood chambers.Podon actively reproduced during the first days of the experiment indicated by a low ratio of organisms with empty/full brood chambers (0.79-2.77), whereas lowest reproductive activity occurred on t 17 /t 24 (5.09-33.10).

Statistical analyses mesozooplankton
For total abundance of mesozooplankton we determined no significant relationship with fCO 2 or any of the other explanatory variables (temperature, chlorophyll a) (Table 1).
The cladocera Bosmina sp.showed distinct abundance peaks in M7, M6, and M3 with approx.110-130 ind. 10 3 m −3 higher numbers between t 24 and t 31 compared to the two control mesocosms and M8.The GLM model revealed neither a significant relation of the total abundance of Bosmina sp. with fCO 2 nor temperature.Chlorophyll a concentration was determined to significantly affect the Bosmina occurrence but model validation showed heterogeneity of the residuals mostly due to experiment day.Running the GAMM model with a smoother on experiment day did not confirm this result.Introduction

Conclusions References
Tables Figures

Back Close
Full GAMM analysis on the ratio between Bosmina with empty brood chambers to organisms with full brood chambers yielded in significance of all three main terms as well as significant interaction term between fCO 2 and chlorophyll a.Some minor residual structure remained after GAMM on the Bosmina ratio that should be kept in mind with respect to resulting p values (Zuur et al., 2009).
According to a GAMM applied on the Shannon diversity index H, neither of the factors significantly affected MZP species diversity.

Microzooplankton
The MiZP abundance and species succession in our experiment corresponded well with description by Kivi (1986) on annual succession of protozooplankton in Tvärminne/Storfjärden.In May, shortly after the chlorophyll maximum, this author observed the highest protozoan biomass whereas a minimum was found in June/July two weeks after the spring bloom (mostly ciliates and heterotrophic dinoflagellates).Dominant ciliates during the summer month were Lohmaniella spp. or small Strombidium spp.
(35 µm).Myrionecta rubra was always present with maximum abundance in late spring.Lohmaniella spp.also occurred in the present study but was classified with Strobilidium spp.( 20 µm) due to difficulties with clear identification.In our study, the MiZP community was dominated by the primarily photoautotrophic ciliate M. rubra (=Mesodinium rubrum) (Lohmann 1908, Jankowski 1976) (Mesodiniidae, Litostomatea) most of the time (Lindholm, 1985).Only towards the end of our experiment, heterotrophic ciliates became more important in the MiZP community when small Strombidiids such as Strombidium cf.epidemum occurred with similar abundances as M. rubra.M. rubra is also a common species in the Baltic Sea with maximum reported densities of 26 600 cells L −1 in the Arkona Basin usually above the thermocline and associated with the euphotic layer (Setälä and Kivi, 2003).Maximum total ciliate densities in the entrance Introduction

Conclusions References
Tables Figures

Back Close
Full of the Gulf of Finland varied between 10-50 000 cells L −1 in 1988 and 1990, respectively, and hence are in the same range as in our study, and also consisted of the same typical species/groups (Setälä and Kivi, 2003).

Changes in microzooplankton species diversity
Previous studies on sensitivities of MiZP communities towards ocean acidification are inconsistent.For example Rose et al. (2009) report on significant changes in MiZP abundance and community composition in the open North Atlantic Ocean between their single factor (only temperature) and two factor (temperature and CO 2 ) experiments and conclude that a combination of direct and indirect (bottom-up) effects were responsible for observed changes.Mesocosm studies off the coast of Norway and in the Arctic revealed no effect of different CO 2 concentrations on the MiZP community neither with respect to abundance nor community composition (Suffrian et al., 2008;Nielsen et al., 2010;Aberle et al., 2013).In the latter study, positive effects on the autotrophic biomass with higher and lower CO 2 concentrations were found for dinoflagellates and respectively prasinophytes and haptophytes but these effects did not translate to the MiZP level (Schulz et al., 2013).While we found no significant relation between microzooplankton total abundance and fCO 2 concentration, total abundance was significantly affected by temperature and the microzooplankton community seemed to change with respect to species diversity H towards a higher dominance of single species with increasing temperature and fCO 2 , respectively.Most likely, small species/genus are responsible for this change in diversity.During the first days of the experiment (t 5 , t 5 -t 9 , and t 7 -t 13 , respectively) small species such as Balanion comatum, Strombidium cf.epidemum, and Strobilidium sp.(< 20 µm) show some distinct differences in abundance between the three higher and lower fCO 2 mesocosms.While B. comatum occurs at higher abundance in the control mesocosms and the lowest CO 2 enrichment level (M7, 497 µatm), S. cf.epidemum and Strobilidium sp. have higher abundances in the three high CO 2 mesocosms.Later in the experiment, between t 19 and t 31 , the small size class Introduction

Conclusions References
Tables Figures

Back Close
Full Myrionecta rubra for example occurred in much higher numbers in the mesocosms with the three highest fCO 2 concentrations.For the mentioned species, significant relations were determined for all factors included in our analyses, except for Balanion comatum that showed no significant response to chlorophyll a and Strombidium cf.epidemum that only showed a significant relation with chlorophyll a. Rose et al. (2009) also report on increased dominance of smaller taxa (mostly Lohmaniella sp.among ciliates) during the course of their experiment, but dependent on a combination of different factors, i.e. temperature, CO 2 and changes in the top-down control.Finally, they conclude on a more general effect of temperature on MiZP abundance and community composition.
A relationship between temperature and Shannon diversity H on ciliate communities and on heterotrophic ciliates, respectively, was also shown by Setälä and Kivi (2003) and Aberle et al. (2007).In contrast to our present study, Aberle et al. found H to increase with higher temperature and it was larger ciliates (mostly Strobilidium species) that caused the community shift.

May Myrionecta rubra benefit from OA?
For the present study, a positive CO 2 effect on community chlorophyll a is described with significantly higher concentrations between day t 21 and t 39 (phase II and III) attributed for up to 90 % to picophytoplankton ( 2 µm).The relative contribution of the 2-20 µm size fraction to total chlorophyll a was estimated as about 20 % (Paul et al., 2015).This period of increased chlorophyll a concentrations under high CO 2 also coincides well with increased abundances of the probably predominantly photoautotrophic ciliate Myrionecta rubra ( 10 µm) in the high CO 2 mesocosms.Blooms of M. rubra can cause red tides and are characterized by high uptake rates of inorganic nutrients and they can contribute significantly to chlorophyll a values and primary production in estuaries, fjords and upwelling areas.M. rubra robs plastids from cryptophytes.In the absence of cryptophytes, they sustain a larger cell volume but exposure to cryptophytes stimulates incorporation and cell devision of M. rubra resulting in a decreased average cell but increased population size (hence biomass) (Lindholm, 1985;Gustafson Jr 20045 Introduction

Conclusions References
Tables Figures

Back Close
Full  , 2000, and references therein).Cryptophytes were among the main contributors to total chlorophyll a in particular during phase I and showed a significant negative effect of CO 2 during the first phase of the experiment (Paul et al., 2015).Moreover, small picophytoplankton of approx.2.9 µm cell diameter most likely representing cryptophytes had highest abundances during phases II and III and showed a distinct negative correlation with fCO 2 (Crawfurd et al., 2015).Cryptophyte biomass decreased from t 3 to t 17 as did the total abundance of M. rubra, but whereas the 11-20 µm size class of M. rubra almost disappeared by that time or even earlier for the 20 µm size class, the small size-class cells remained and developed a distinct difference in abundance between the higher and lower CO 2 mesocosms.Growth and photosynthetic performance of M. rubra is ultimately dependent on the availability of cryptophytes, but the ciliate can sustain long periods without feeding by functioning as a phototroph and has the ability to control cryptophyte plastids' division and synthesize chlorophyll (Johnson and Stoecker, 2005;Johnson et al., 2006).Photosynthetic performance of M. rubra may have been stimulated by elevated CO 2 concentrations and thus this ciliate may be "co-responsible" for the CO 2 driven total chlorophyll a differences observed during phases II and III.Consequently, higher cell numbers of small sized M. rubra may be a combination of indirect and direct CO 2 effects through 1) availability of cryptophytes in phase I, II and III maybe being partly responsible for the negative CO 2 effect on cryptophyte biomass and abundances reported by Crawfurd et al. (2015) and Paul et al. (2015), and 2) through a CO 2 -mediated higher photosynthetic rate of M. rubra supporting its own growth.We have no (strong) support, however, that the negative CO 2 effect on cryptophyte abundance during phase I is related to higher grazing pressure of M. rubra resulting in an abundance increase.A certain affinity of Mesodinium rubrum (synonymous Myrionecta rubra) to low pH is supported by a microcosm study with a coastal plankton community that revealed clearly higher abundance at the lowest pH level (6.0) Nielsen et al. (2010).During this phase, M. rubra abundances showed much variability between sampling days.Some higher abundances in the small and medium size class M. rubra especially in M6 and M8 between t 2 and t 11 , however, may point

Conclusions References
Tables Figures

Back Close
Full in this direction.During phase II and III, however, M. rubra may have benefitted from CO 2 stimulated photosynthetic activities and controlled cryptophyte abundances, and hence, decreased cryptophyte numbers would represent an indirect CO 2 effect through M. rubra (Crawfurd et al., 2015).

Mesozooplankton
The MZP community enclosed in the mesocosms reflected fairly well the natural succession of MZP in Tvärminne/Storfjärden where rotifers, cladocerans and calanoid copepods comprise the major zooplankton taxa.Usually rotifers numerically dominate in spring/early summer (Synachaeta sp.) and reach a second peak in midsummer/autumn (Keratella sp.).The calanoid copepods Acartia bifilosa and Eurytemora affinis show two abundance peaks, in mid-June and mid-September, respectively, and Temora longicornis occurs only at low numbers year-round.Cladocerans peak in summer (August/September) with Bosmina longispina maritima clearly dominating among Podon spp.and Evadne nordmanni.Highest MZP biomass is build up in summer (August/September) (Kivi, 1986;Viitasalo, 1992;Koski et al., 1999).The species composition in the mesocosms resembled well natural conditions and were dominated by the most common and successful genus/species known for the Gulf of Finland and the Tvärminne region such as Acartia bifilosa, Eurytemora affinis, Bosmina longispina maritima.Due to the rather late start of our mesocosm experiment after the spring phytoplankton bloom, the usual peak of Synchaeta sp. in spring/early summer -also one of the most successful species (i.e.Synchaeta baltica) -was barely visible during the first days, later rotifers still occurred until termination but were not of great importance anymore (Viitasalo, 1992).Total population densities known for mesozooplankton in the Tvärminne area more or less coincide with abundances found in the mesocosms and range from median values between ∼ 22 000-∼ 40 000 ind. m −3 with occasional peak abundance for Acartia bifilosa and Bosmina sp. of up to 45 000 and 82 000 ind. m −3 , respectively.Average peak abundance of Acartia bifilosa and Bosmina sp., respectively, during a period from 20047 Introduction

Conclusions References
Tables Figures

Back Close
Full 1967-1984 was ∼ 10 000 ind. m −3 and respectively ∼ 20 000 ind. m −3 (Viitasalo et al., 1995;Viitasalo, 1992).Between t 24 and t 31 , however, some exceptional high numbers (> 150 000 ind. m −3 ) occurred in the mesocosms mainly attributed to extremely high occurrence of Bosmina sp..Even higher densities exceeding 1 000 000 ind. m −3 during bloms of blue-green algae are known for B. fatalis in an eutrophic lake in Japan (Hanazato and Yasuno, 1987).The MZP community in the surrounding water did not entirely correspond with the mescosms over the course of the experiment.Whereas the dominance of particular species corresponded quite well until t 3 , it diverged progressively after t 10 when in the surrounding water the occurrence of colonies of blue-green algae (Aphanizomenon) and rotifera where higher than in the mesocosms, and the abundance of copepods and cladocerans comparatively lower (S.Lischka, pers. obs., 2012).Most likely, this is a result of isolation of the mesocosm bags from surrounding water mass exchange and incoming plankton communities and selective advantage of single species in the mesocosms.

Copepods
Up to date, there are only very few large CO 2 enrichment mesocom studies with an as far as possible holistic plankton community approach (Riebesell et al., 2008(Riebesell et al., , 2013b)).Therefore, this study is still one of the first to follow MZP community development subjected to ocean acidification scenarios projected for this century in a close-to natural whole plankton community.Previous study using the same mesocosm set-up investigated effects on an Arctic MZP community and found no significant difference neither in total abundance or abundance of single taxa nor in species diversity.This Arctic MZP community was dominated by meroplanktonic larvae and copepods played a minor role, and thus differed in species type composition compared to the Baltic community enclosed in our mesocosms (Niehoff et al., 2013;Riebesell et al., 2013a).In general, on the mesozooplankton level, calcifiers seem to be more sensitive to CO 2 increases than crustaceans (Kurihara, 2008;Kroeker et al., 2013), i.e. copepods which dominate

Conclusions References
Tables Figures

Back Close
Full zooplankton communities in boreal and higher latitude regions.While copepods are thought to be rather robust against ocean acidification with negative effects occurring usually not until pCO 2 levels far beyond projections for end of this century (Kurihara et al., 2004;Mayor et al., 2007;Weydmann et al., 2012;McConville et al., 2013;Almén et al., 2015), more recent studies give evidence that copepods' sensitivity may be highly stage dependent and thus so far mostly underestimated due to the fact that most studies done to-date considered only adult stage copepods.Mortality of the nauplii stage Acartia tonsa for example increased threefold already at CO 2 concentrations expected for the end of this century (Cripps et al., 2014).These authors highlight the importance of a holistic life-stage approach in order to provide meaningful data for climate change projections.
Copepods comprised one of the two dominant taxonomic groups in the present study and the mesocosm approach allowed to investigate CO 2 effects on the succession of all different life stages from eggs to reproducing adults.The CO 2 scenario chosen in the present study covered the range projected for this century (IPCC, 2013).Over this range, we found no distinct abundance differences for neither of the species.The permanent occurrence of adult males and females together with copepodite stages and nauplii suggest more or less continuous reproduction.Concurrent lab experiments investigating the effect of CO 2 on reproductive success of Eurytemora affinis are in agreement with the observations from the mesocosms (Almén et al., 2015, this issue).
Incubated Acartia bifilosa showed fCO 2 unaffected egg production, but slight negative effects on egg hatching and development were found and adult females were smaller in the two highest CO 2 mesocosms (Vehmaa et al., 2015, this issue).Our results are also in line with Niehoff et al. (2013) who do not describe any apparent CO 2 effect on an Arctic MZP community including copepods.Therefore, results from our study are completely in line with earlier studies describing copepods rather robust against CO 2 changes at least in the range projected for the end of this century.Copepods in the study region naturally experience fCO 2 , pH and also temperature fluctuations of more than 0.5 pH units and 5 • C temperature during daily vertical migrations which is more Introduction

Conclusions References
Tables Figures

Back Close
Full than the predicted climate change for the year 2100.I.e.these copepods are probably well adapted to short-term physico-chemical changes (Lewis et al., 2013;Almén et al., 2014).

Mollusks
Mollusks enclosed in the mesocosms comprised for more than 90 % of juvenile bivalves of the species Macoma balthica and occurred during the first ten days of the experiment.Calcifiers are among the most vulnerable organisms to ocean acidification, and within this investigation bivalve larvae were therefore subjected to a more detailed study on their occurrence and length distribution over time.The main findings from this study suggest reduced stettling rates and a developmental delay with increasing fCO 2 .For more details see Jansson et al. (2015, this issue).

Cladocera -OA effect on Bosmina spp. through increased food availability?
Most conspiciuous differences found in mesozooplankton abundance are due to the cladoceran Bosmina sp. between t 24 and t 31 .In three of the four CO 2 enriched mesocosms (497 µatm, 821 µatm, 1007 µatm) peak numbers were twice or even more than twice as high compared to the control and the highest CO 2 mesocosms, though a significant relation with fCO 2 could not be proved.Nevertheless, this striking difference may possibly point to an indirect CO 2 effect through higher food availability under high CO 2 .
In the inner parts of the Baltic proper, the endemic Bosmina longispina maritima is most abundant among the genus Bosmina.Its life cycle includes many parthenogenetic generations in the summer, sexual reproduction occurs only during late summer and autumn from which fertilized resting eggs result that overwinter in the sediment and develop in spring (Kankaala and Wulff, 1981, and  of Bosmina spp.varies between 20-25 days, age of first reproduction is between 4-7 days (food dependent), they can bear several developmental stages in the brood pouch of the mother and at favourable conditions, populations can increase twofold within 5-10 days (Purasjoki, 1958;Kankaala and Wulff, 1981;Hanazato and Yasuno, 1987;Biswas et al., 2014).Population dynamics of Bosmina longirostris are highly foodsensitive with food concentrations having a significant effect on growth, net reproductive rate and rate of population increase.For example at high food concentration more broods occur and more eggs are produced (Kankaala and Wulff, 1981;Urabe, 1991), whereat not only food quantity but also quality influences life time and together were shown to shorten it to up to 10 days.(Hanazato and Yasuno, 1987).Bosmina is also very quick in taking advantage of favourable environmental conditions and has a high plasticity to customize its life cycle and growth patterns.For example, to avoid predation they can allocate energy to somatic growth instead of early reproduction to attain larger size and escape prey size spectrum of copepods (Kankaala, 1983;Jankowski, 2004).
The above mentioned population increase of Bosmina in the mesocosms coincides with significant CO 2 mediated differences during phase III in chlorophyll a, chlorophytes and particulate organic matter during the respective days and probably represented favourable food conditions for this species enhancing asexual reproduction in particular in the elevated CO 2 mesocosms (Paul et al., 2015).Only M8, the mesocosm with the highest CO 2 concentration, diverged from this trend.The stoichiometry of food organisms can affect growth in cladocerans, for example, growth of Daphnia pulicaria was reduced due to reduced C : P ratios of high CO 2 cultivated food algae (Urabe et al., 2003).In the present study the average C : P ratio of particulate organic matter did not vary with fCO 2 , however this can only be seen as an estimate for the food that was Introduction

Conclusions References
Tables Figures

Back Close
Full effectively ingested by Bosmina (Paul et al., 2015).Peak abundance in all mesocosms occurred only on one sampling day, i.e. did not stay high for a longer period but was low at the preceding sampling day and had dropped already at the following sampling day.Most likely, the drop in population size that occurred earlier than to be expected from Bosmina's lifespan of around 20 days was due to high mortality and/or change to sexual reproduction producing resting eggs.Therefore, a possible explanation why Bosmina in M8 did not follow the trend observed in the other CO 2 -elevated mesocosms may be that due to the rather low possible sampling frequency (every seven days) the actual abundance peak was missed (Riebesell et al., 2013a).Reason for mortality could be in response to the overall drop in available food during phases II and III and/or stress response due to extreme densities or reproductive rates of Bosmina itself.It is known, that Bosmina sp. can die earlier when they have higher reproductive rates and switch to sexual reproduction producing resting eggs, respectively, at too high population densities (so called "crowding phenomenon") (Purasjoki, 1958;Acharya et al., 2005).In Kankaala (1983), Bosmina started sexual reproduction at around 4500 ind.m −3 which is about 1-2 orders of magnitude less than observed peak numbers in the mesocosms.The significant results we found for the ratio of Bosmina with empty and full brood chambers strongly suggest that organisms in the high CO 2 mesocosms had higher reproductive activities during the time of actual peak abundance.In particular, Bosmina in M8 and M3 (two highest CO 2 levels) had continuously low brood chamber ratios (i.e.large proportion of actively reproducing organisms in the population) from t 10 onwards (with the ratio in M8 mostly even lower than in M3).This supports our assumption that we missed to sample the abundance peak of Bosmina in M8 possibly obstructing to prove a significant indirect fCO 2 effect on Bosmina abundance through increased food availability.Introduction

Conclusions References
Tables Figures

Back Close
Full

Conclusions
Our study gives evidence for direct and/or indirect CO 2 effects on the micro-and mesozooplankton community.For the MiZP community composition we determined significant changes with a shift towards smaller species/genus with increasing CO 2 levels.The phototrophic ciliate Myrionecta rubra, as well as Balanion comatum, Strombidium cf.epidemum, Strobilidium sp. and the short-lived cladocera Bosmina seemed to benefit from increased CO 2 concentrations, the first one probably directly, the others rather indirectly.Although our results show no direct significant relation with abundance, we assume Bosmina growth and reproduction was stimulated from increased food availability at elevated CO 2 mostly during phases II and III (higher post-bloom chlorophyll a and particulate organic matter).This may point to an indirect CO 2 effect that was masked as a consequence of too low sampling frequency not allowing to adequately capture the population dynamics of this short-lived and highly adjustable genus.For the study region, microbial loop has been shown to be of particular importance during late summer and autumn when most of the secondary production including fish is fueled by carbon channeled from the microbial loop to crustacean zooplankton.Filterfeeding cladocerans directly feed on bacteria and flagellates and effectively transfer carbon from the microbial loop to higher trophic levels.Contrary, in copepod dominated communities, the carbon transfer from microbial loop is comparatively low because an intermediate trophic level is needed (heterotrophic flagellates, ciliates) (Koski et al., 1999, and references therein).Therefore, we conclude, under increasing ocean acidification in cladoceran dominated MZP communities, the importance of trophic transfer from the microbial loop to higher trophic levels becomes more efficient.Introduction

Conclusions References
Tables Figures

Conclusions References
Tables Figures

Back Close
Full q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q 0 10 20 30 40 q q q q q q q q q q q q q q q q q q q q 0 10 20 30 40 Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | 2 enriched mesocosm (mean: 1250 cells L −1 (SD 180) and mean: 2205 cells L −1 (SD 851), respectively).On t 23 peak values ranged between 2300 cells L −1 and 3840 cells L −1 , and between 1980 cells L −1 and 6,740 cells L −1 on t 37 .Balanion comatum, Rimostrombidium sp., Strobilidium sp.(< 20 µm), Spathidium sp., and tintinnids were of some importance during the first days of the experiment showing peaks in cell numbers of 1760 cells L −1 on t 7 , 1680 cells L −1 on t 0 , 3640 cells L −1 on t 11 , 1760 cells L −1 on t 0 , and 1080 BGD Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | M7, M6, and M3 (497-1007 µatm) had highest peak values ranging between 130 276 ind.m −3 and 162 082 ind.m −3 , whereas abundance in M1 and M8 were somewhat lower with 111 980 ind.m −3 and 90 975 ind.m −3 , respectively.In M5, no abundance peak occurred but zooplankton developed a plateau between t 24 until t 38 of around 70-74 000 ind. m −3 .Towards the end of the experiment, zooplankton total abundance returned to about the initial values (29 325-44 824 ind.m −3 in M8 and M1, respectively).Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | numbers of Eurytemora sp.sharply declined.During the course of the experiment, Acartia sp.varied in numbers between 117 ind.m −3 and 4624 ind.m −3 and did not show clear abundance peaks in most of the mesocosms.Temora sp. was present during the whole time but always in low abundances ranging between 330 ind.m −3 and 3 ind.m −3 among all mesocosms.In M1, M5, and M7, Temora sp. was absent on day t 38 and t 45 , respectively.Copepod nauplii occurred during the entire experiment duration with peak abundance between t 10 and t 24 (9003-33 555 ind.m −3 ).Outside this period, nauplii abundance ranged between 17 ind.m −3 and 2700 ind.m −3 .
Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | et al.
Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | references therein).Cladocerans are highly reproductive at times of favourable environmental conditions.The lifespan Figures Back Close Full Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | the crew of R/V Alkor (AL394, AL397) for transportation, deployment and recovery of the mesocosms.The Tvärminne Zoological Station is greatfully acknowledged for kind hospitality, logistic and facility support.This collaborative study has received funding from the German BMBF (Federal Ministry of Education and Research) projects BIOACID II (FKZ 03F06550) and SOPRAN Phase II (FKZ 03F0611)Discussion Paper | Discussion Paper | Discussion Paper | Riebesell, U., Czerny, J., von Bröckel, K., Boxhammer, T., Büdenbender, J., Deckelnick, M., Fischer, M., Hoffmann, D., Krug, S. A., Lentz, U., Ludwig, A., Muche, R., and Schulz, K. G.: Technical Note: A mobile sea-going mesocosm system -new opportunities for ocean change research, Biogeosciences, 10, 1835-1847, doi:10.5194/bg-10-1835-2013,2013a.20048Discussion Paper | Discussion Paper | Discussion Paper | Urabe, J., Togari, J., and Elser, J. J.: Stoichiometric impacts of increased carbon dioxide on a planktonic herbivore Authors, Glob.Change Biol.

Figure 1 .Figure 3 .
Figure 1.Total abundance of microzooplankton during the course of the experiment.Note there is one missing value in M1 on t 13 .
t 38 and t 45 , and for temperature, and chlorophyll a on t 38 and t 45 .Missing observations for t 24 and t 38 were estimated by building the mean of values measured at t 23 /t 25 and respectively t 37 /t 39 .t 45 was the last sampling day and hence it was not possible to estimate a mean from the preceding and following day.Therefore missing values for t 45 were estimated from a polynomial fit curve applied on phase III values cells L−1 on t 0 , respectively.Peak abundance of Balanion comatum diverged with CO 2 concentration with higher mean cell numbers in the control and lowest enriched mesocosm compared to the three high CO 2 mesocosms (mean: 1680 cells L −1 (SD 139) and mean: 880 cells L −1 (SD 223), respectively).Likewise, small Strobilidium sp.developed some CO 2 related difference with mean abundance of 1360 cells L −1 (SD 170) and 2400 cells L −1 (SD 872) in the two controls and the CO 2 enriched mesocosms, respectively.Later in the experiment, these species/groups were not of importance anymore.Mesodinium sp. and Strobilidium sp.> 20 µm occurred always in relatively low cell numbers (< 550 cells L 3.1.4Percentcontributionofnumerically dominant species/genera/groups to total cell numbersFigure3band c show the percent contribution of dominant species/genera/groups to the total cell numbers over time for each of the mesocosms.For better clarity, Myrionecta rubra size classes, Strobilidium sp.size classes together with Rimostrombidium sp., Strombidium spp.and cysts together with ciliates sp. were combined.M. rubra dominated the microzooplankton community in all mesocosms most of the time.During the first days of the experiment, M. rubra contributed ∼ 90 % to the total cell numbers in all mesocosms and stayed above 50 % until t 21 .Minimum contributions occurred on t 37 when M. rubra had a share of only 6-24 %.After t 37 , M. rubra proportions ranged between 18 and 67 %.The second most important group was Strombidium sp. and among this Strombidium cf.epidemum.Strombidium sp. had highest shares during the second half of the experiment.It started with relatively low contributions during the first days of the experiment and increased depending on the mesocosm from t 19 /t 21 on to proportions of 17-36 %.Maximum contributions varied between 58 and 69 % during t 35 -t 39 .All remaining groups usually had contributions below 15 %.