High tolerance of microzooplankton to ocean acidification in an Arctic coastal plankton community

Impacts of ocean acidification (OA) on marine biota have been observed in a wide range of marine systems. We used a mesocosm approach to study the response of a high Arctic coastal microzooplankton community during the post-bloom period in Kongsfjorden (Svalbard) to direct and indirect effects of high pCO2/low pH. We found almost no direct effects of OA on microzooplankton composition and diversity. Both the relative shares of ciliates and heterotrophic dinoflagellates as well as the taxonomic composition of microzooplankton remained unaffected by changes in pCO2/pH. Although the different pCO2 treatments affected food availability and phytoplankton composition, no indirect effects (e.g. on the total carrying capacity and phenology of microzooplankton) could be observed. Our data point to a high tolerance of this Arctic microzooplankton community to changes inpCO2/pH. Future studies on the impact of OA on plankton communities should include microzooplankton in order to test whether the observed low sensitivity to OA is typical for coastal communities where changes in seawater pH occur frequently.


Introduction
Since the beginning of the industrial revolution, the oceans have absorbed ca.30 % of anthropogenic CO 2 (Sabine et al., 2004), and oceans thus serve as one of the largest sinks for anthropogenic CO 2 , which in turn affects the marine carbonate system.The on-going increase in atmospheric pCO 2 re-sults in decreasing seawater pH and carbonate ion (CO 2− 3 ) and increasing bicarbonate (HCO − 3 ) and CO 2 concentrations.Especially the predicted drop in pH, in the following referred to as "ocean acidification (OA)", is considered to affect a variety of biological and biogeochemical processes in the oceans with potentially far-reaching consequences on the community and ecosystem level (Riebesell et al., 2007).In this study, we wanted to test whether Arctic coastal plankton communities will be in any way affected by high pCO 2 /low pH and thus susceptible to ocean acidification.
Microzooplankton (MZP in the following) include heterotrophic protozoa and metazoa in the plankton within a size range between 20 and 200 µm.In this study we focussed on heterotrophic dinoflagellates and ciliates since these were numerically the most important MZP components and they are considered as major phytoplankton consumers worldwide (Calbet and Landry, 2004;Sherr and Sherr, 2007).Bacterivorous protozoa (e.g.heterotrophic nanoflagellates) were not included in our analysis but are presented by Brussaard et al. (2013).Today, there is strong evidence that MZP play a pivotal role in suppressing phytoplankton blooms in temperate (Johansson et al., 2004;Aberle et al., 2007) and cold waters (Levinsen and Nielsen, 2002;Seuthe et al., 2011).MZP usually show a rapid numerical response to changes in food availability, and the occurrence of specific species in the plankton can be directly linked to specific prey organisms (Loeder et al., 2011).MZP can thus be strongly bottomup controlled since their development can be directly linked to food availability.But at times when mesozooplankton abundance is high, top-down control (e.g. by copepods or meroplanktonic larvae) plays a crucial role in suppressing MZP abundance considerably (Levinsen and Nielsen, 2002).
Changes in pCO 2 are known to affect consumers indirectly via changes, e.g. in food availability, phytoplankton community structure, size classes and stoichiometry (Suffrian et al., 2008;Feng et al., 2009;Rossoll et al., 2012).On the other hand MZP are known for a direct pH sensitivity (Hinga, 2002;Pedersen, 2003), and a drop in seawater pH as a result of increasing pCO 2 could directly affect the physiology of both autotrophic and heterotrophic protists by changing, e.g.intracellular pH, membrane potentials and enzyme activities (Nielsen et al., 2010).
To investigate the impact of OA on a natural Arctic plankton community, a mesocosm experiment was conducted in Kongsfjorden, Svalbard, over a period of about one month in June/July 2010.The initiation of the phytoplankton spring bloom in the fjord starts already under the ice cover, culminating between April and early June after ice break-up (Eilertsen et al., 1989;Seuthe et al., 2011).After the spring bloom, phytoplankton remain moderately high during late spring and summer (Hop et al., 2002).MZP are underinvestigated in the Kongsfjorden so far, but there is profound evidence that ciliates and heterotrophic dinoflagellates play an important role as trophic intermediaries in Arctic plankton communities (Seuthe et al., 2011).
In the present study we hypothesized the following: 1. direct effects of high pCO 2 will alter MZP community composition and diversity, and 2. indirect effects of high pCO 2 by changes in food availability and phytoplankton composition will alter the carrying capacity and phenology of MZP.

Methods
In summer 2010 nine polyethylene mesocosms (∼ 50 m 3 , 17 m long) were deployed at 78 A detailed description of the mesocosm design, the deployment logistics, the methodology of CO 2 enrichment and the maintenance of the mesocosms throughout the duration of the experiment is given in Riebesell et al. (2012).

Microzooplankton sampling and identification
Seawater samples for MZP enumeration were taken once a week by a depth-integrating water sampler (depth integration: 0-12 m water depth).A total of 250 mL of seawater were transferred to brown-glass bottles and fixed with acidic Lugol's iodine (2 % final concentration).A total of 100 mL of each sample were transferred to sedimentation chambers, and MZP were counted by the inverted microscope method (Utermöhl, 1958) at a 200 × magnification with a Zeiss Axiovert 135.The whole area of the bottom plate was counted for each sample in order to guarantee comparability of the counting method both at periods of high and low MZP abundance.Most dinoflagellates were considered as heterotrophic since most species found in the samples are characterized by obligate or optional heterotrophic feeding modes.Only some small, chloroplast-bearing dinoflagellates were considered as predominantly autotrophic and included into the phytoplankton.MZP were identified to the lowest possible taxonomic level (species or genus level) according to Kahl (1932), Foissner et al. (1995), Strüder-Kypke et al. (2002), Tomas (1996) and Scott (2005).For biovolume calculations geometric proxies were used according to Hillebrand et al. (1999), and carbon biomass was calculated using the conversion factors given in Putt and Stoecker (1989).Diversity was measured by the Shannon-Wiener function (H ; log e ) (Shannon and Weaver, 1963).

Phytoplankton sampling and identification
A total of 100 mL water sample from a depth-integrating water sampler (depth integration: 0-12 m water depth) were filled in brown-glass bottles and fixed with alkaline Lugol's iodine (1 % final concentration).The counting was performed after Utermöhl (1958) with an inverted microscope (Zeiss Axiovert 100).Cells bigger than 12 µm were counted on the half bottom area at 200-fold magnification and cells less than 12 µm (∼ 5-12 µm) on two to four stripes at 400-fold magnification.The settling volume was 25 mL.For identification Tomas (1996), Hoppenrath et al. (2009) and Kraberg et al. (2010) were used.The biovolumes were calculated after Olenina et al. (2006), and for transformation to carbon the equations of Menden-Deuer and Lessard (2000) were used.

Chlorophyll measurements
A total of 250-500 mL of seawater were sampled and filtered onto GF/F for chlorophyll a analysis.Filters were stored frozen for at least 24 h and homogenized in 90 % acetone   with glass beads (2 and 4 mm) in a cell mill thereafter.After centrifugation at 5000 rpm, chlorophyll a concentrations were determined in the supernatant on a fluorometer (TURNER, 10-AU) according to Welschmeyer (1994).

Statistical analysis
To test for significant effects of pCO 2 on MZP diversity, a regression analysis was conducted.As an independent factor, pCO 2 and, as a dependent factor, H were chosen using STATISTICA 6.0.Diversity was calculated using PRIMER 5.2 (© 2001 Primer-E Ltd.).

Chlorophyll a, phytoplankton biomass and composition
The  chlorophyll a concentrations between 0.26 (pCO 2 1085) and 0.36 µg L −1 (pCO 2 180) (Fig. 1).In general, a three-phase bloom development occurred at all pCO 2 levels: the first phase before nutrient addition (before day 13), the second phase after nutrient addition until the second chlorophyll minimum (day 13-21) and the third phase until the end of the experiment (after day 21).Each phase was characterized by an increase in phytoplankton biomass and a subsequent decline in chlorophyll a right after the bloom.Chlorophyll a reached maximum peak heights during the third phase of bloom development while highest chlorophyll a concentrations occurred at low (pCO 2 : 175-250; Fig. 1a-c), and intermediate pCO 2 levels (pCO 2 : 340-600; Fig. 1d-f) and reduced chlorophyll a peaks were observed at high pCO 2 levels (pCO 2 : 675-1085; Fig. 1g-i).
During the first phase of bloom development, the phytoplankton (PP) community was dominated by chrysophytes and nanoflagellates (3-8 µm) at all pCO 2 levels (Fig. 2).During the second phase, a moderate bloom of prasinophytes (∼ 5 µm) occurred showing higher peak heights at intermediate and high pCO 2 levels (pCO 2 : 425-1085; Fig. 2d-i) (see Brussaard et al., 2013 for details).The third phase was dominated by autotrophic dinoflagellates (mainly Heterocapsa rotundata), which showed a higher biomass at intermediate and high pCO 2 levels (pCO 2 : 425-1085; Fig. 2e-i).Diatoms were of minor importance occurring only at low biomass at the end of the experiment.

Microzooplankton biomass
The experiment started with fairly high MZP biomass at all pCO 2 levels ranging between a minimum of 15 µg C L −1 (pCO 2 600) and a maximum of 54 µg C L −1 (pCO 2 675) (Fig. 1a-i).While the MZP biomass in the low (pCO 2 : 175-250; Fig. 1a-c) and the high pCO 2 levels (pCO 2 : 675-1085; Fig. 1g-i) decreased during the first phase of the experiment until day 13, biomass in the intermediate pCO 2 level (pCO 2 : 340-600; Fig. 1d-f) increased from day 0 to day 7, followed by a decline until day 13.During the second and the third phase, an increase in MZP biomass was observed at all pCO 2 levels reaching highest biomass peaks of 46 µg C L −1 (pCO 2 175, Fig. 1a) and 50 µg C L −1 (pCO 2 250, Fig. 1c) at low pCO 2 levels.However, no correlation was found between the MZP biomass maxima and the different pCO 2 treatments (Table 1).In addition, no trend was observed for the cumulative biomass (total MZP, ciliates and dinoflagellates) in relation to the different pCO 2 treatments (Table 1).

Microzooplankton composition and diversity
Heterotrophic dinoflagellates dominated the MZP community throughout the experiment, independent of pCO 2 level, while ciliates contributed to lower proportions to the MZP biomass (Fig. 3a-i).The taxonomic composition of the MZP community showed almost identical patterns at all pCO 2 levels.Table 2 shows the taxonomic composition of MZP (biomass in µg C L −1 ) over the course of the mesocosm  1).

Microzooplankton succession
During the first phase of bloom development, biomass response of MZP at the different pCO 2 levels was quite heterogeneous.The strongest positive biomass response to increasing chlorophyll a concentrations during the first phase was observed for the dinoflagellate Gymnodinium cf.arcticum  (Fig. 4a).Most MZP species showed a distinct biomass increase during the second or the third phase of bloom development at all pCO 2 levels (Fig. 4).While the biomass of the ciliates Lohmaniella oviformis, Rimostrombidium sp. and Strombidium cf.conicum peaked already during the second phase of bloom development at all pCO 2 levels (Fig. 4ce), others, e.g.Katodinium cf.glaucum, showed increasing biomass until the third phase.Overall responses of single species to increasing phytoplankton availability showed similar patterns at all pCO 2 levels, and no trend in biomass maxima of specific MZP species in relation to the different pCO 2 levels was observed (Table 1).

Discussion
The initial MZP biomass of 15-54 µg C L −1 in our mesocosm study in Kongsfjorden in late May 2012 was high compared to studies from the same season and location where MZP biomass ranged between 2 and 13 µg C L −1 (Seuthe et al., 2011;Hodal et al., 2012).However, our data are in line with post-bloom MZP biomass reported for other coastal Arctic regions, e.g.Disko Bay, West Greenland (Levinsen et al., 2000;Hansen et al., 2003).In good agreement to Hodal et al. (2012), small-sized ciliates and dinoflagellates were of minor importance while large-sized ciliates and dinoflagellates dominated the MZP.Furthermore, the taxonomic composition of the MZP community we found in 2012 was simi-lar to that of the post-bloom period in Kongsfjorden in 2006 (Seuthe et al., 2011).

Direct effects of high pCO 2 will alter microzooplankton community composition and diversity
Previous studies on the impact of OA on MZP communities showed no clear trend (Suffrian et al., 2008;Rose et al., 2009).In order to understand the causes and consequences of future CO 2 and pH conditions, it is essential to consider all components of the plankton, from protists to metazoans, and to compare among sites with different degrees of natural pH fluctuations (Nielsen et al., 2012).MZP play a major role in the global carbon cycle by fixing inorganic carbon, and the interplay between auto-and heterotrophic protists is crucial since up to 60-75 % of phytoplankton production in coastal and open oceans is consumed by MZP (Landry and Calbet, 2004).MZP thus act as a trophic link between phytoplankton and mesozooplankton and contribute substantially to the cycling of bulk organic matter and nutrients (Irigoien et al., 2005;Calbet and Saiz 2005).However, only few studies have addressed the impact of OA on MZP communities so far, although there is indication for pH sensitivity of MZP especially at elevated pH (pH of ∼ 8.0 to 9.5) (Hinga, 1992;Pedersen and Hansen, 2003).Experimental studies off the coast of Norway (Suffrian et al., 2008) and in the open North Atlantic Ocean (Rose   et al., 2009) found no direct effects of a high pCO 2 /low pH on MZP.This is in line with our observation, since no direct effects on MZP composition and diversity was observed; neither the relative shares of ciliates and heterotrophic dinoflagellates were affected nor did a low pH induce changes in taxonomic composition.However, Nielsen et al. (2012) showed that a reduced pH (∼ 6.3) can alter the performance of autotrophs (reduced abundance and photosynthetic rate) and MZP (reduced abundance).For coccolithophores such direct effects are assumed to be related to changes in the intra-cellular pH, which in turn alters enzymatic reactions and growth rates (Nimer et al., 1994;Suffrian et al., 2011).
In our study, the changes in pCO 2 and pH were within the range expected for the 21st century and thus not as extreme as in the scenario simulated by Nielsen et al. (2012).But similar to their study, we found a high tolerance of MZP towards moderate changes in pH.In general, open ocean plankton communities are considered more vulnerable to OA since the pH in coastal environments fluctuates more strongly with frequent variations by up to 1 or more pH units (Hansen, 2002;Hinga, 2002).So, even today coastal MZP temporarily experience high pCO 2 /low pH as predicted for the next century, but the question remains whether communities that are permanently subject to a high CO 2 world will display the same level of pH tolerance.

Indirect effects of high pCO 2 by changes in food availability and phytoplankton composition will alter the carrying capacity and phenology of MZP
Elevated pCO 2 is known to affect autotrophic processes directly (Riebesell and Tortell, 2011), while MZP are considered to be predominantly indirectly affected by high pCO2 from changes, e.g. in food availability, phytoplankton community and size structure (Suffrian et al., 2008;Rose et al., 2009).
Our initial expectation was that the different pCO 2 levels would induce general shifts in phytoplankton composition, which in turn could affect the carrying capacity and phenology of MZP.Chlorophyll a concentrations in the mesocosms did not show a pronounced response to CO 2 enrichment, but an increase with time in all mesocosms, which was not reflected in MZP biomass.Phytoplankton composition in our mesocosms showed some clear CO 2 -dependent trends, e.g. a positive CO 2 effect on dinoflagellates, while prasinophytes and haptophytes profited at low CO 2 levels during phase 3 of the experiment (Schulz et al., 2013;Brussaard et al., 2013).However, these changes in phytoplankton community composition altered neither the carrying capacity nor the phenol-ogy of MZP.This might be related to the fact that changes in community composition happened only on short time scales and responses were usually not maintained over a longer period of bloom development (Brussaard et al., 2013).Phytoplankton usually comprise a multitude of species, and thus short-term, pCO 2 -induced alterations in algal diets of consumers can often be mitigated (Urabe and Waki, 2009).Further, as pointed out by Suffrian et al. (2008), the distinct effects of elevated CO 2 on single plankton species observed in laboratory studies are not comparable to those obtained in experiments simulating close-to natural conditions since such complex systems seem to have a higher buffering capacity to changes in pCO 2 .This is in line with observations of Rose et al. (2009) where no relationship between the MZP community composition and elevated CO 2 could be observed.
In the mesocosms, nanophytoplankton (e.g.nanoflagellates and chrysophytes) played a major role during the first phase of bloom development and, apart from MZP grazing, viral lysis enforced the nanophytoplankton bloom to decline (Brussaard et al., 2013).The rapid decline in nanophytoplankton induced by viruses corresponds well with the steep biomass decline of ciliates and dinoflagellates during the first phase of bloom development, which points to a food shortage of MZP.This is also confirmed by the moderate grazing rates observed for MZP during the nanophytoplankton bloom at all pCO 2 levels (Brussaard et al., 2013).0.22 0.04 0.06 0.00 0.00 0.00 0.73 0.04 0.04 0.00 0.00 0.00 0.24 0.00 0.00 0.00 0.06 0.00 Strobilidium spiralis 0.10 0.10 0.10 0.36 0.18 0.08 0.01 0.10 0.04 0.28 0.22 0.05 0.06 0.04 0.10 0.12 0.58 0.05 Laboea strobila 0.14 0.06 0.00 0.28 0.12 0.00 0.01 0.14 0.04 0.18 0.00 0.12 0.22 0.08 0.04 0.04 0.10 0.00 During the second and the third phase of bloom development, increases in MZP biomass were observed at all pCO 2 levels.This corresponds well to the high nanophytoplankton biomass (mainly prasinophytes, autotrophic dinoflagellates and haptophytes; see Schulz et al., 2013 for details) during the second half of the experiment (phase 2 and 3).
In addition, high MZP grazing rates on pico-and nanophytoplankton were observed during the phases after nutrient addition from day 13 on (Brussaard et al., 2013).But despite a continuously high pico-and nanophytoplankton availabil-ity and an increasing biomass of autotrophic dinoflagellates (mainly H. rotundata), MZP showed a rapid decline during the third phase of bloom development.This phenomenon is most likely not bottom-up regulated but related to an enhanced top-down control of MZP by mesozooplankton.In our study the mesozooplankton were comprised of Cirripedia and copepod nauplii initially and shifted towards a community dominated by polychaete larvae and copepodites (mainly Calanus spp.) from day 18 onwards at all pCO 2 levels (Niehoff et al., 2012) the mesozooplankton clearly during the first phase of bloom development, are characterized by an herbivorous feeding mode in contrast to other meroplanktonic larvae such as copepod nauplii (Turner et al., 2001).This might explain the high initial MZP biomass in the mesocosms when predation pressure by omnivorous mesozooplankton was still low.Subsequently, polychaete larvae and Calanus copepodites, effective grazers of MZP (Turner et al., 2001), became highly abundant, thus pointing to a suppression of MZP by mesozooplankton.
In conclusion, our hypotheses that a high pCO 2 will alter MZP community structure, carrying capacity or phenology must be rejected on the basis of the present mesocosm experiment and it points to a relatively high robustness of MZP towards elevated CO 2 in coastal waters.

Table 1 .
Regressions of microzooplankton response variables on pCO 2 treatments.

Table 2 .
Taxonomic composition of microzooplankton (biomass in µg C L −1 ) over the course of the mesocosm experiment (white rows: dinoflagellates; light grey rows: ciliates) in the different pCO 2 treatments.

Table A1 .
Exact pH and pCO 2 values in the different pCO 2 treatments.

Table A2 .
Taxonomic composition of microzooplankton (abundance in individuals mL −1 ) over the course of the mesocosm experiment (white rows: dinoflagellates; light grey rows: ciliates) in the different pCO 2 treatments.