Microzooplankton grazing and phytoplankton growth in marine mesocosms with increased CO2 levels

Microzooplankton grazing and algae growth responses to increasing pCO


Introduction
Atmospheric CO 2 levels have increased from about 280 to 380 µatm since the beginning of the industrial revolution, and are projected to reach values as high as 700 µatm by the end of the 21st century (IPCC, 2001).This increase in atmospheric CO 2 (and other gases) is predicted to result in e.g.increasing global temperatures, rising sea EGU level and accelerating extreme weather incidences (IPCC, 2007).Increased atmospheric CO 2 levels have lead to increased ocean acidity with a pH drop of 0.1 since the beginning of the industrial revolution and with a predicted drop of another 0.4 units already before the end of this century (Caldeira and Wicket, 2003).As a consequence, the carbonate saturation in the ocean is decreasing, likely effecting a number of organisms, especially those with calcareous skeletons such as coccolithophorids, corals and molluscs (see discussion and references in Schulz et al., 2007).Auto-and mixotrophic protists play a key role in the global carbon cycle since they fix inorganic carbon that is either transferred to the higher trophic levels through grazing or exported to deeper ocean layers through the biological pump and sedimentation.But it is still unclear, how and to what extent the alteration in the ocean chemistry affects and is affected by the phytoplankton growth and grazing interaction.As shown in previous experiments, the decreasing pH and hence decreasing carbonate saturation in the ocean may have a negative effect on the calcite (CaCO 3 ) production by coccolithophores and foraminifera (Riebesell et al., 2000;Russell et al., 2004), while other algal species which rely on dissolved CO 2 concentration for photosynthesis, might benefit from an increase in the surface ocean CO 2 concentration.Thus, CO 2 perturbations at an ecosystem level may provoke very complex responses in phytoplankton species composition and succession, and thereby affect the structure and functioning of the marine food web by cascading effects on elemental recycling by virus and bacteria as well as carbon fluxes through the grazing food web and export through sedimentation.While such complicated effects can not be studied in laboratory, mesocosm experiments provide a powerful tool to better understand complex responses of marine systems to increasing CO 2 levels and its feedback effects on carbon cycle and global climate.Thus, to investigate how increased CO 2 levels in the atmosphere could affect the phytoplankton-grazer interactions, we conducted a series of dilution experiments to quantify microzooplankton grazing during the 2005 Pelagic Ecosystem CO 2 Enrichment study (Schulz et al., 2007).• 14 ′ E).To initiate phytoplankton blooms all the mesocoms were fertilized with NO 3 and PO 4 to initial concentrations of 15 and 0.6 µmol l −1 , respectively.The mesocosms were manipulated (in triplicates) to three pCO 2 levels (ca 350, 700 and 1050 µatm) by aerating with normal or CO 2 -enriched air.These CO 2 concentrations represented one (1×), two (2×) and three (3×) times the present atmospheric CO 2 conditions, respectively.

Setup and sampling of dilution experiments and nutrient analysis
Phytoplankton growth and microzooplankton grazing rates were assessed by a total of 12 dilution experiments (Landry, 1993;Landry and Hassett, 1982), listed in Table 1.
The experiments were performed using water from one of each of the three CO 2 treatments at 4 occasions corresponding to pre algal bloom (day 1-3), bloom (day 7-9) and post bloom conditions (day 13-15 and 20-22) (Schulz et al., 2007).Water for the dilution experiments was collected by submerging 25 l polycarbonate bottles with the main opening covered by a 200 µm nylon mesh to exclude mesozooplankton, and with the spigot open to let air out of the bottle in order to sample with minimal turbulence and sheer-stress of the delicate protists.An aliquot was filtered trough 0.2 µm cellulose acetate filter (Whatman, 142 mm) using tissue culture hoses and low pressure (<50 hPa).
Filtration was conducted in a cold room at in situ temperature immediately before the To assure that the experiments were not biased by nutrient limitation, nutrients were measured in the 100% sea water bottles at the start and the end of the incubations (Table 1).Nutrient samples were frozen and stored at −20 • C until analysis according to Grasshoff and Kremling (1999) as described in detail by Schulz et al. (2007).In order to avoid unnecessary changes in the experimental nutrient conditions (e.g.Landry, 1993), nutrients were added to the experimental bottles only when nutrient levels were below 2 µmol l −1 of nitrate or 0.2 µmol l −1 of phosphate (i.e. from day 13 and on).Final concentrations of nutrients added were 1 µM (NO − 3 , NH + 4 ), 0.1 µM (PO 3− 4 ) and trace metals corresponding to f /40 medium according to Guillard and Ryther (1962).Nutrients were never depleted in the experiments (Table 1).The 2 l bottles were tightly capped and incubated in situ outside the mesocosms for 24 h hanging horizontally on strings from a floating ring at 6 m depth.This setup reproduced light conditions comparable to the average conditions inside the mesocosm (measured with a horizontally mounted underwater LI-192 underwater quantum sensor).The incubation setup also created a gentle irregular tipping movement which prevented sedimentation in the flasks.Samples for microzooplankton counts and HPLC analysis were taken from the 10 l bottles at start (t 0 ) and from the 2 l incubation bottles at end (t 24 ), by gently siphoning off while slowly stirring with the hose.

Algal pigment analysis
Phytoplankton pigments were analysed with high performance liquid chromatography (HPLC) to obtain growth and grazing rates for the entire community and for selected algal groups based on their marker pigments (Table 2).Aliquots for HPLC analysis (400-500 ml) were filtered under low vacuum   Pigments were extracted in 1 ml of 100% acetone.Additionally 100 µl of an internal standard (canthaxanthin) and glass beads were added before sonication (4 • C, 5 min).
Chl a was used as a proxy for the whole phytoplankton community while taxon specific marker pigments were analysed to obtain specific growth and grazing coefficients for different algal groups (Table 2).19 '-hexanoyloxyfucoxanthin (19-hex) could not be used as a marker for prymnesiophytes during the bloom phase, as it could not be well separated from prasinoxanthin in the HPLC measurements of the samples.To get genuine values for this important group pure E. huxleyi samples from Bergen (provided by M. N. M üller, IFM-GEOMAR) were screened by HPLC to find an alternative marker.A 19'-hexanoyloxyfucoxanthin-like peak, which was regarded typical for prymnesiophytes or even specific for coccolithophorids (Zapata et al., 2004), was found in the samples, corresponding to 4-keto-19'-hexanoyloxyfucoxanthin (4-keto-hex) recently reported by Airs and Llewellyn (2006).Both markers, 19-hex and 4-keto-hex, were found at stable ratios to each other and to Chl a in the pure E. huxleyi samples from Bergen.Thus, one or both of these pigments were used to identify the prymnesiophytes in each experiment.Introduction

Conclusions References
Tables Figures Printer-friendly Version Interactive Discussion EGU

Microzooplankton abundance estimates
Subsamples (100-300 ml) for microzooplankton analyses were fixed with Lugol's iodine (1-2% final concentration) and stored in brown glass bottles at ambient temperature (ca 15 • C).Samples were settled for 24 h in 50 ml sedimentation chambers (Uterm öhl, 1958).One to two transects of each sample was counted with a Zeiss Axiovert 100 inverted microscope at 200x magnification.Additional transects at 400x magnification were used to determine smaller cells.A total of ca 120-1000 cells were enumerated in each sample.Cell sizes were measured with an ocular scale and used to calculate biovolume, using formulas for spherical (1) and prolate spheroid shapes (2), with diameter (d ) and height (h).
Plankton biovolume (except for ciliates) was converted to carbon biomass (3) according to Menden-Deuer (2000): Ciliate biovolume was converted to carbon biomass using a conversion factor of 0.19 pg C/µm 3 (Putt and Stoecker, 1989) The microplankton was differentiated into autotrophic plankton and microzooplankton (including both heterotrophic and mixotropic organisms) by comparison of morphological features to literature (Kuylenstierna andKarlson, 1996-2006;Str üder-Kypke et al., 2000-2001;Throndsen and Eikrem, 2005;Throndsen et al., 2003).The microzooplankton was grouped into dinoflagellates, ciliates and "other".All ciliates were regarded as heterotrophic by morphological features (ciliates only apical, no visible chloroplasts etc.).The group named "other" consisted mainly of microflagellates that were both scarse and of very low biomass (Fig. 1), thus for simplicity all microflagellates were considered heterotrophic.Changes in phytoplankton pigment concentrations over the incubation period were used to calculate the apparent phytoplankton growth rate (µ) and the mortality losses due to microzooplankton grazing (g).Assuming exponential growth: P 0 and P t are the initial and final pigment concentrations respectively; t is the incubation time (t=t 24 -t 0 ), k is the instantaneous coefficient of phytoplankton growth, g the coefficient of grazing mortality and c is the dilution factor expressed as percentage of ambient seawater.It can be inferred that µ is linearly related to the dilution factor c, that the negative slope is the grazing coefficient g and that the Y-intercept is the phytoplankton growth rate k (Landry 1993).Changes in grazer density were monitored in the 100% bottle at start (t 0 ) and end (t 24 ) of the experiment.Since such changes accounted always for less than 10% (±) of the community (not shown), no correction for grazer density was applied to the calculations (cf.Landry, 1993).Regressions were tested with ANOVA (Sigmaplot version 9, Systat Software Inc.).The percentage of initial pigment SS daily grazed by microzooplankton (% d −1 ) was calculated according to: SS = 1 − e −g•100 (4)

Microzooplankton community composition and development
There was no clear difference in microzooplankton community composition between the three different CO 2 treatments (Fig. 1).EGU whereof all were considered heterotrophic, for simplicity.Although the biomass of the heterotrophs thus was overestimated, it has no practical quantitative effect since the total biomass of "Other" was only 0-6.5% of the total "microzooplankton" biomass (Fig. 1).The total heterotrophic biomass reached its maximum (90-130 µg C l −1 ) during the experiment starting days 13-15, while it decreased again at the time of the last experiments (65±5 µg C l −1 ).Although dinoflagellates increased in abundance during the first 8-15 days, ciliates did not show any clear trend of development through the experiment.

Development of the overall phytoplankton community, growth and grazing
Overall phytoplankton community biomass, growth and grazing estimates based on Chl a showed similar patterns in the three CO 2 treatments during the incubation experiment (Tables 3a-c).Two phases can be observed: the first from day one to nine was characterized by the highest algal growth rates (0.12 to 0.99 d −1 ).Although the microzooplankton community grazing rates also were the highest (0.28-0.49d −1 ) with a daily Chl a SS removal of 25-39% during this first period, the algal community reached the maximum SS (1.99-12.23 µg Chl a l −1 , Tables 3a-c).Thus, microzooplankton only appeared to have a minor effect on the overall phytoplankton development when nutrients were abundant (Table 1) during the first 9 days, this is also apparent from the general distribution of the data points below the 1:1-lines in Fig. 2.
Between days 9 and 13 there was a significant decrease in instantaneous Chl a growth rates (k) in all CO 2 treatments.While the 2× and 3× CO 2 treatments showed a marked decrease in Chl aSS (Tables 3b and c), the 1×CO 2 treatment Chl aSS was relatively stable in that period (Table 3a).However, after day 13 the algal Chl a SS declined in all three mesocosms to 2.1-2.5 µg l −1 at day 20-22 (Tables 3a-c).During this latter period phytoplankton growth rates decreased (0.02-0.37 d −1 ) and overall microzooplankton grazing pressure stayed relatively low (5-24% SS d −1 , Tables 3a- c).Thus, the microzooplankton grazing impact on the overall phytoplankton community was limited.Of the seven analysed specific algal pigments (Table 2) only the pigments assumed to characterize some of the most dominant groups; Prymnesiophytes (4-keto-19'hexanoyloxyfucoxanthin and 19'-hexanoyloxyfucoxanthin), Diatoms (fucoxanthin), Dinoflagellates (peridinin) and Cyanobacteria (zeaxanthin) yielded significant growth or grazing rates in most of the experiments (Tables 3a-c).Thus data on the other pigments are not further discussed.
It is not surprising that the general pattern observed for the total phytoplankton community (Chl a) was mirrored in the effect of the microzooplankton grazing on the pigments assumed to reflect the dominant diatoms and prymnesiophytes.Grazing on diatoms and prymnesiophytes also showed similar patterns.During the first ten days the growth rates of these algae were generally higher than the feeding rates indicating that microzooplankton feeding was not a factor significantly limiting their blooming.Grazing rates overcame the growth rates during the days 13-15 while they were comparable in the end of the experiment (20-22).
The grazing pressure on cyanobacteria SS was higher compared with the ones on the larger autotrophs, ranging between 19% and 65% (Tables 3a-c).This intense feeding activity was balanced by higher instantaneous growth rates (0.19-2.25 d −1 ) and did not seem to limit the increase of the SS during the last two experiments.The apparent patterns of growth and grazing on dinoflagellates were more inconsistent, and few conclusions may be drawn from these data.
As observed for Chl a, the microzooplankton grazing on the specific pigments did not seem to be influenced by the different CO 2 treatments, and neither did the grazing pressure seem to have any major effect on the development of the bloom of the different groups, except perhaps in the very beginning of the experiment when the standing stocks of the phytoplankton were generally low.The highest percentages of SS removed by microzooplankton were 42% for diatoms, 43% for prymnesiophytes and 65% for cyanobacteria.When using specific pigments as markers for individual taxa it is of great importance to know what species are physically present and dominant (Antajan et al., 2004;Irigoien et al., 2004).Due to the initial high silicate concentrations in all mesocoms (Table 1) the phytoplankton community biomass rapidly became dominated by diatoms while the silicate became significantly reduced (Schulz et al., 2007;Egge et al., 2007).Thus, the development of the fucoxanthin showed a similar development as the draw down of the silicate and observations of phytoplankton samples from the mesocosms, and this pigment should thus be considered to closely mirror the development of the diatoms.
Although the calcifying prymnesiophyte Emiliania huxleyi only reached moderate numbers, other prymnesiophytes were abundant in the mesocosms corroborating our use of 4-keto-hex and 19-hex as indicators of prymnesiophytes in all the CO 2 treatments (Engel et al., 2007;Paulino et al., 2007;Schulz et al., 2007, J. K. Egge and A. Larsen, personal communication).Also the development of the dominating cyanobacteria Synechococcus sp.(Paulino et al., 2007) appeared to follow the same pattern as the development of the zeaxanthin measured here.Although we observed autotrophic dinoflagellates (not shown) while analysing the heterotrophic dinoflagellates the development of the peridinin concentration and rates (Tables 3a-c) is less clear, and may be obscured by the problem of defining mixotrophy in this group.The dinoflagellates will therefore not be further discussed here.
In conclusion, as the presence of the major phytoplanktonic groups -diatoms, prymnesiophytes and cyanobacteria (Synechococcus sp.) -was verified against flow cytometry and microscopy we consider the HPLC data as trustworthy for these three groups.The major aim with this investigation was to compare microzooplankton grazing and algae growth interactions in different CO 2 environments.However, we found no clear effects on microzooplankton grazing or phytoplankton growth when comparing the three CO 2 -treatments over a three week period.Despite that previous laboratory studies have shown a number of acute effects on single planktonic organisms (even if sometimes conflicting and contradictory, as discussed in Schulz et al., 2007), we suggest from our results that either; 1) Complex, close to natural systems such as investigated here may show such a complex response patterns that it needs more detailed studies (including e.g.biogeochemical studies of the material transport between the trophic compartments) to be disclosed, or 2) Such complex systems may simply have large "buffering capacities" making them able to absorb increased CO 2 , at least under certain conditions, such as described in Riebesell et al. (2007).However, as described in Riebesell et al. (2007) such CO 2 over-consumption would lead to offset Red field ratios, and possibly significant deterioration of the content of essential constituents in the prey of the microzooplankton.This has not been investigated here.If the observed CO 2 over-consumption observed by Riebesell et al. (2007) in this system leads to a deterioration of the food quality this may not be readily visible on the first trophic level, because at least some microzooplankton may have the capacity to upgrade low quality prey (Veloza et al., 2006) such as the carbon rich algae in the 3×CO 2 -treatment, and if this is true, the trophic cascade response may thus not be visible until higher levels in the marine food web, such as e.g. for copepods.But effects on higher trophic levels may need longer experimental duration than a few weeks to be clearly manifested.It is also interesting to notice that while the ciliates did not change substantially in biomass, the heterotrophic dinoflagellates did so (Fig. 1).This may be explained by that many dinoflagellates feed on diatoms (compare e.g.feeding guilds discussed in Nejstgaard et al., 1997 and2001), the phytoplankton group showing the highest growth and grazing rates here.It has also been hypothesized that at least some dinoflagellates may Introduction

Conclusions References
Tables Figures Printer-friendly Version

Interactive Discussion
EGU trophically upgrade the food for higher trophic levels such as copepods (see discussion and references in Veloza et al., 2006).These are potentially interesting aspects that need to be investigated in future studies.However, to our knowledge, this is the first study in such marine systems and more data is needed before such conclusions can be drawn.It should especially be focused on the possible effects on food quality vs.
quantity for higher trophic levels, such as copepods, and perhaps fish.
The general temporal dynamic of the phytoplankton community, observed in our bottle incubation experiments, mirrored the dynamic observed inside the mesocosms by other studies (Egge et al., 2007;Paulino et al., 2007;Schulz et al., 2007).They grew during the first ten days of the experiment as a consequence of the nutrients addition (Schulz et al., 2007).During this period microzooplankton was grazing actively (0.28-0.49d −1 ) on the autotrophic compartment but without limiting the development of the bloom.The effect of this trophic activity was evident from the microzooplankton biomass increase during the same period.The decline of the bloom after day 10, is therefore due to the nutrient depletion as reported by Schulz et al. (2007), or perhaps viral activity (Larsen et al., 2007) more than a result of grazing.It was only during the initial phase of the experiments when phytoplankton biomass was low, and possibly during the post bloom phase, when the instantaneous growth rates were close to zero or negative, that the grazing became more significant and the microzooplankton biomass reached its maximum.Neither did the cyanobacteria appear to be significantly limited by the microzooplankton.Despite of a high daily removal (19%-65% of the SS), the microzooplankton did not control the biomass increase registered during the last two incubation experiments.The lack of microzooplankton grazing may also explain the increase in the cyanobacteria community observed in the mesocosms by Paulino et al. (2007).The general relatively low levels of microzooplankton grazing activity may explain the observed lack of a net heterotrophic phase in this PeECE III mesocosm experiment (Egge et al., 2007), and support the hypothesis by Riebesell et al. (2007) that such a system may favour a high export of organic material through the pycnocline.
Target concentrations for the dilution of 25, 50, 75 and 100% undiluted sea water were carefully mixed in 10 l polycarbonate bottles and distributed to triplicate 2 l polycarbonate incubation bottles by siphoning.2 l bottles were filled alternating the flow into each bottle until they were all topped off at about the same time.Absolute dilutions were checked by Chl a concentrations at start in the 10 l bottles.
Development of SS, growth and grazing of specific algae groups Use of specific marker pigments as a proxy for different algae.

Table 1 .
EGUthe experiment; especially project leader U. Riebesell for coordinating the project.The staff at the Marine Biological Station, University of Bergen, in particular T. Sørlie and A. Aadnesen, and the Bergen Marine Research Infrastructure (RI) are gratefully acknowledged for support in mesocosm logistics.We further thank K. Nachtigall for technical assistance with pigment measurements, and P. Fritsche for assistance with nutrient and pigment data.IFM-GEOMAR library west bank is acknowledged for help with literature acquisition.Special thanks for the support with microplankton identification go to J.Egge, A. Sazhin.P. Simonelli was funded by the University of Bergen.J. C. Nejstgaard was supported by the Norwegian Research Council (NRC) project 152714/120 30.Y. Carotenuto was funded by the European Marine Research Station Network (MARS) Travel Award for Young Scientist 2004.Introduction Nutrient data measured at beginning and end of the dilution experiments.Data are shown for 1×, 2× and 3× CO 2 treatments at t 0 and t 24 .Introduction

Table 2 .
Name and abbreviation of the pigments used as algae taxon-specific markers."Taxon" denotes the major taxon the pigment was considered to reflect here, while the "Additional taxon" denotes other groups that potentially could contribute to the pool of the specific pigment (based on the reference given); the pigment was not used to characterize the additional taxon in this study.

Table 3a .
Compilation of pigment key data for dilution experiment based run with water from

Table 3b .
Compilation of pigment key data for dilution experiment based run with water from mesocosm 2×CO 2 otherwise as Table3a.

Table 3c .
Compilation of pigment key data for dilution experiment based run with water from mesocosm 3×CO 2 otherwise as Table3a.