Biogeosciences Effects of long-term high CO 2 exposure on two species of coccolithophores

The physiological performance of two coccolithophore species,Emiliania huxleyi and Coccolithus braarudii, was investigated during long-term exposure to elevatedpCO2 levels. Mono-specific cultures were grown over 152 ( E. huxleyi ) and 65 ( C. braarudii) generations while pCO2 was gradually increased to maximum levels of 1150 μatm ( E. huxleyi ) and 930 μatm ( C. braarudii) and kept constant thereafter. Rates of cell growth and cell quotas of particulate organic carbon (POC), particulate inorganic carbon (PIC) and total particulate nitrogen (TPN) were determined repeatedly throughout the incubation period. Increasing pCO2 caused a decrease in cell growth rate of 9% and 29% in E. huxleyiand C. braarudii, respectively. In both species cellular PIC:TPN and PIC:POC ratios decreased in response to rising pCO2, whereas no change was observed in the POC:TPN ratios of E. huxleyiandC. braarudii. These results are consistent with those obtained in shorter-term high CO2 exposure experiments following abrupt pertubations of the seawater carbonate system and indicate that for the strains tested here a gradual CO 2 increase does not alleviate CO2/pH sensitivity.


Introduction
Emissions of anthropogenic CO 2 since the beginning of the industrial revolution have lead to an increase in atmospheric carbon dioxide concentration, resulting in a fraction of the anthropogenic CO 2 being absorbed by the ocean.The ongoing oceanic uptake of anthropogenic CO 2 is steadily altering the seawater carbonate chemistry and has lead to a Correspondence to: M. N. Müller (mueller@obs-vlfr.fr)reduction in surface ocean pH by 0.12 units over the past 200 years (Alley et al., 2007).This process, termed ocean acidification, seriously alters the physiological performance of pelagic and benthic organisms (Fabry et al., 2008).Due to their ecological and biogeochemical importance and because they are easily grown in cultures, coccolithophores are among the best studied organisms with respect to their response to ocean acidification (reviewed by Rost et al., 2008 andZondervan, 2007).Results from these studies indicate considerable species-and even strain-specific differences in CO 2 sensitivity.
When exposed to elevated pCO 2 levels as projected for the course of this century, Emiliania huxleyi, Gephyrocapsa oceanica and Calcidiscus quadriperforatus respond by decreasing calcification to varying degrees, corresponding to a decrease in the particulate inorganic to organic carbon ratio, PIC:POC (Riebesell et al., 2000;Zondervan et al., 2002;Langer et al., 2006;Feng et al., 2008;Langer et al., 2009).Coccolithus braarudii, on the other hand, seems to be insensitive to elevated pCO 2 (Langer et al., 2006) with respect to calcification.Differences in pCO 2 sensitivity between species were also observed with respect to organic carbon production.While E. huxleyi and G. oceanica increased POC production by up to 50% under elevated pCO 2 (Riebesell et al., 2000;Zondervan et al., 2002;Feng et al., 2008;Shi et al., 2009), no change with increasing pCO 2 was observed in C. braarudii and C. quadriperforatus (Langer et al., 2006).Recently, Iglesias-Rodriguez et al. (2008) reported an increase of up to 100% in both calcification and organic carbon production per cell in response to elevated pCO 2 .This trend reverses when accounting for the large difference in cell size between pCO 2 treatments as observed in previous experiments on E. huxleyi, yielding a slight decrease in calcification with increasing pCO 2 when normalizing the data to cellular carbon content (Riebesell et al., 2008b).
Most acidification experiments with coccolithophores were done applying an abrupt pertubation of seawater carbonate chemistry, either by bubbling with CO 2 enriched/depleted air (Feng et al., 2008;Iglesias-Rodriguez et al., 2008) or by adding acid/base (Riebesell et al., 2000;Zondervan et al., 2002;Langer et al., 2006), and preacclimating the cultures before starting the experiment (commonly 7-10 generations).After the acclimation time, experimental high CO 2 exposure time ranged between 1-2 (Iglesias-Rodriguez et al., 2008) and 7-10 generations (Riebesell et al., 2000;Zondervan et al., 2002;Langer et al., 2006;Feng et al., 2008).This results in a maximum of 20 cell divisions under high pCO 2 conditions.The physiological response to high pCO 2 exposure of E. huxleyi within one cell generation was recently investigated by Barcelos e Ramos et al. (2010) and was found to be consistent with the observations gained from longer-term experiments.However, it is presently unknown whether the observed CO 2 sensitivities of coccolithophores are persistent over long-term high pCO 2 exposure.Here, we report on semi-continuous batch culture experiments with two coccolithophore species (E.huxleyi and C. braarudii), which were grown over multiple generations (152 and 65 generations, respectively) under gradually increasing pCO 2 levels.

Experimental setup
Coccolithophores were grown in duplicates under semicontinuous batch culture conditions (as described above) in 280 ml autoclaved polycarbonate Erlenmeyer flasks.pCO 2 was adjusted by additions of HCl or NaOH to the media.Cultures were allowed to grow for about 5-11 generations corresponding to a dissolved inorganic carbon (DIC) and nitrate consumption of maximal 10% and 23%, respectively.At this stage exponentially growing cultures were sampled for DIC, pH, cell number, total particulate and particulate organic carbon (TPC and POC), and total particulate nitrogen (TPN) before being transferred into fresh medium (f/20 nutrient conditions and the carbonate system already adjusted) to a concentration of 100 and 50 cells ml −1 , corresponding to a minimum population size of 28 000 and 14 000 cells (E. huxleyi and C. braarudii, respectively).This culture dilution and sampling protocol was continuously repeated throughout the course of the experiment.The media in the control flasks (low pCO 2 ) was adjusted to a pH total of 8.21±0.05(±1SD), resulting in a pCO 2 of 260 ± 20 µatm with a corresponding calcite saturation state ( ) of 5.6 ± 0.4.At the beginning of the experiment all flasks started under conditions described above and then the treatment flasks were slowly acidified over several generations (79 and 28 generations for E. huxleyi and C. braarudii, respectively) to the target pCO 2 values (1150 ± 140 µatm and 930 ± 180 µatm for E. huxleyi and C. braarudii, respectively) and kept at these levels until the end of the experiment.

TPC, POC, PIC, TPN
Two sub-samples from each flask were filtrated onto precombusted (525 • C for 7 h) GF/F filters and frozen at −20 • C. For POC analysis filters were fumed over HCl for 24 h to remove all inorganic carbon and afterwards all filters were measured on a Euro EA Elemental Analyser (Ehrhardt and Koeve, 1999).PIC (particulate inorganic carbon) was calculated from the difference of TPC and POC.
Cell quotas of carbon and nitrogen increase during the light phase whereas the cell density stays constant because of synchronized cell division in the dark phase (Müller et al., 2008).Because samples were taken at different times of the light phase, normalizing the data on a per cell basis generates a bias in the data due to sampling time.We therefore present the data as cellular ratios of PIC:POC, PIC:TPN and POC:TPN, which show no temporal trend over the course of the light phase (Fig. 1).

Cell counts
One sub-sample from each flask was taken and the cell number was immediatly determined with a Coulter Counter (Z Series).Samples were measured three times and the mean was used to calculate the growth rate "µ" (d −1 ) as where c 0 and c 1 are the cell concentrations at the beginning (t 0 ) and end (t 1 ) of the incubation period (expressed in days).

Carbonate system
The carbonate system was monitored by DIC and pH measurements.DIC was analyzed after Stoll et al. (2001) using an automated segmented-flow analyzer (Quaatro) equipped with an auto-sampler (±10 µmol kg −1 accuracy and five µmol kg −1 precision) and pH was measured using a "Metrohm 713 pH-Meter", equipped with pH and reference electrodes and temperature sensor.Sensor and electrodes were stored in filtrated seawater at 16 • C to match the ionic strength of the sampled water.pH measurements were periodically checked by calculating pH from measurements of total alkalinity (Dickson, 1981) and DIC of filtrated seawater using the program CO2sys (version 1.05 by E. Lewis and D. W. R. Wallace) with dissociation constants for carbonic acid after Roy et al. (1993).Calculated pH values closely agreed with pH measurements with a maximum deviation of ±0.02.Here we present pH values on the total scale.

Scanning electron microscopy
5 ml samples were taken periodically from the control and high pCO 2 treatment and fixed with formaldehyde (1% end concentration).Subsequently, the samples were filtered onto polycarbonate filters (0.2 µm pore size), dried at 60 • C for 24 h and then sputter-coated with gold-palladium.Pictures were taken with a CamScan-CS-44 scanning electron microscope at the Institute of Geosciences of the Christian-Albrecht-University in Kiel.

Statistical analysis
As the data set does not allow for parametric statitics, it was analyzed by means of the binomial test.To test the null hypothesis that two categories are observed equally often, mean values of the duplicate samples under continuously high pCO 2 were compared to the corresponding mean values of the duplicates under low pCO 2 sampled at similar generations.A mean value under the continuously high pCO 2 treatment that was greater than the corresponding mean value of the low pCO 2 treatment constituted the first category (coded as "+"); otherwise it was coded as "−" for the second category.Note that under the null-hypothesis both categories should be observed equally often.Results are given with the p-value of the binomial test plus the corresponding distribution ratio (dr) of "+" to "−".A note of caution might be required here: the sample size for the binomial test was rather small throughout the experiment (varying between 3 and 8).Thus, non-significant results may be due to a lack of statistical power.

Emiliania huxleyi
Cells were cultured for 98 days, corresponding to 152 generations in the control treatment (low pCO 2 ) and 144 generations in the high pCO 2 treatment.Cellular division in the control treatment stabilized at a rate of µ=1.10 ± 0.  no change in growth rate was detectable.After reaching the maximum pCO 2 level of 1150 µatm the growth rate decreased to µ=1.00 ± 0.06 d −1 (p=0.07,dr=1:7) and remained at this value until the end of the experiment (Fig. 2a and b).The PIC:TPN ratio was relatively constant at 4.9 ± 1.0 molC molN −1 under low pCO 2 but with the onset of high pCO 2 the ratio decreased and was consistently lower compared to the control by an average value of 1.8 ± 0.7 molC molN −1 (Fig. 2c; p=0.031, dr=0:6).No consistent difference between pCO 2 treatments was observed in POC:TPN with a mean ratio of 10.0 ± 1.4 molC molN −1 (Fig. 2d; p=0.453, dr=2:5).The PIC:POC ratio was considerably lower under constant high pCO 2 with a mean value of 0.33 ± 0.13 molC molC −1 compared to 0.56 ± 0.13 molC molC −1 at low pCO 2 (Fig. 2e; p=0.031, dr=0:6).Coccolith morphology of E. huxleyi did not display a visible difference between pCO 2 treatments (Fig. 4a and b  870 µatm, which was followed by an immediate increase in cell growth.As soon as pCO 2 was elevated above 1000 µatm again, the cell growth rate decreased back to the previous level.

Growth rate
An average decrease of 9% was observed in E. huxleyi's growth rate in response to increasing pCO 2 from 260 to 1150 µatm.This reduction in growth rate under continuous high pCO 2 was tested to be different from the rate at low pCO 2 with a 93% probability (p=0.07); even though it is marginally non-significant, we consider this decrease as biological relevant.Previous studies detected no change in growth rates after exposure to elevated pCO 2 (Riebesell et al., 2000;Zondervan et al., 2002;Feng et al., 2008).While this difference may be due to the fact that the maximum pCO 2 of 900 µatm in these earlier studies was somewhat lower than in the present experiment, the small difference in growth rate of 9% may also have been below the detection limit in the single end-point sampling approach used in previous studies.Recent results by Barcelos e Ramos et al. ( 2010) and Langer et al. (2009) indicate a reduced growth rate at a pCO 2 >1000 µatm in short-term experiments.Interestingly, between day 73 and day 80, when pCO 2 accidently dropped to 870 µatm, the growth rate increased to control values (Fig. 2b) and immediately returned to the lower value after pCO 2 was raised again to 1150 µatm.A similar instantaneous effect of pCO 2 on the growth rate of E. huxleyi was reported to occur within one cell generation (Barcelos e Ramos et al., 2010).
Under constant high pCO 2 the growth rate of C. braarudii was reduced by 29%.At comparable pCO 2 values, Langer et al. (2006) observed no significant reduction in the growth rate of C. braarudii.This difference might be induced by the long-term culturing under constant high pCO 2 , but other potential factors such as differences in the experimental temperature and the light intensity cannot be excluded as being responsible.Recent results by S. Krug (personal cummunication, 2009) indicate a reduced growth rate of C. braarudii when exposed to pCO 2 >1400 µatm for 15 generations.

PIC:TPN and PIC:POC
Under constant high pCO 2 the ratios of PIC:TPN (and equally PIC:POC) in E. huxleyi and C. braarudii were reduced by ≈42% and ≈70%, respectively.The reduction of the PIC:POC ratio in E. huxleyi is a commonly observed response under high pCO 2 which is driven by the decrease of the cellular PIC quota and increase in POC quota and ranges between 10 and 60% depending on pCO 2 level, temperature and light intensity (Riebesell et al., 2000;Zondervan et al., 2002;Feng et al., 2008;Langer et al., 2009;Barcelos e Ramos et al., 2010).Since the POC:TPN ratio remains constant under pCO 2 values (as discussed in the  next paragraph) the PIC:TPN ratio should display a similar behaviour as the PIC:POC ratio.Indeed, the PIC:TPN ratios are decreasing in response to elevated pCO 2 values in both species under long-term high pCO 2 cultivation (Figs.2c  and 3c), which is also seen in short-term experiments under different light intensities for E. huxleyi (Fig. 5a, data from Zondervan et al., 2002).Coccolithus braarudii, on the other hand, is reported to maintain constant PIC:POC and PIC:TPN ratios within pCO 2 values ranging from 345 to 915 µatm (Langer et al., 2006).However, more recently S. Krug (personal communication, 2009) measured in shortterm experiments a significant decrease in both the PIC:POC and the PIC:TPN ratio when C. braarudii was exposed to pCO 2 values above 1000 µatm.

POC:TPN
Particulate organic carbon production was observed to slightly increase under high pCO 2 and nutrient replete conditions in E. huxleyi (Zondervan et al., 2002;Feng et al., 2008;Barcelos e Ramos et al., 2010), whereas C. braarudii maintains a constant rate (Langer et al., 2006).The POC:TPN ratio, however, remains unchanged under short-term high pCO 2 exposure in E. huxleyi (Feng et al., 2008) andC. braarudii (G. Langer, personal communication, 2009).These observations from short-term studies are well in agreement with those revealed in the current long-term study.Zondervan et al., 2002) in comparison to the present longterm data.The measured ratio of 10.0 ± 1.4 molC molN −1 under long-term cultivation is higher than ratios reported from short-term and mesocosms experiments, which vary between 6 and 7 molC molN −1 (Engel et al., 2005;Feng et al., 2008).Previous studies reported POC:TPN ratios of ≈ 10 in E. huxleyi only under nitrogen limitation (Engel et al., 2004;Sciandra et al., 2003).However, we can exclude nitrogen limitation of E. huxleyi in the present study since i) an initial nitrate concentration of 88 µmol l −1 would be sufficient to support exponential growth up to a cell density of 5×10 8 l −1 (calculated with a TPN content of 2.6 pgN cell −1 ) which was never reached during this study, and ii) the measured growth rate of 1.10 ± 0.06 d −1 corresponds to maximal growth rates under nutrient replete conditions for the temperature and light levels applied in our study (Buitenhuis et al., 2008).Therefore, we can rule out nitrogen limitation to be responsible for the higher POC:TPN ratio in the present study.
In general, we can confirm the observed trend in the POC:TPN ratios of short-term experiments and conclude that E. huxleyi increases the POC production per cell under longterm high pCO 2 exposure within the tested range.However, since the POC:TPN ratio stays constant the bulk organic matter per available nitrogen of an exponentially growing E. huxleyi or C. braarudii population will be equal under high and low pCO 2 .

From short-to long-term pCO 2 response of coccolithophores
Results from short-term pCO 2 perturbation experiments, typically involving 7-10 cell generations, were found to be consistent with those under long-term incubation (up to 150 generations).Additionally, Barcelos e Ramos et al. ( 2010) demonstrated that the pCO 2 response observed in short-term studies is even detectable within one generation (≈14 h after perturbation of the seawater carbonate chemistry).Considering the short "generation time" of coccolithophores (the average time between two cell divisions; ≈1-2 days for coccolithophores), evolutionary change and adaptation may occur during long-term incubations.However, the experimental setup of the current long-term study was not suited for the detection of genetic changes.In comparison to natural populations of E. huxleyi, where a high genetic variation can be found and sexual reproduction by a haploid stage occurs (Medlin et al., 1996;Iglesias-Rodriguez et al., 2006;Morin, 2008), the strains used in this study were grown asexual as vegetative diploids, what reduces the mutational supply rate and imposes a speed limit on adaptive evolution (de Visser et al., 1999).
Results from cultures of Chlamydomonas reinhardtii, grown over 1000 generations under elevated pCO 2 , indicated no change in growth rate and cells failed to evolve specific adaptation to high pCO 2 even though the phenotype of the evolved population was attributed to genetic change (Collins and Bell, 2004).Both investigated coccolithophorid species decreased growth rates in response to elevated pCO 2 and did not recover over the time frame of the experiment.The potential to re-establish previous growth rates through adaptation to a new environment might be increased by long-term experiments using a higher mutational supply rate (by larger populations and multiple strains culturing) and induced sexual reproduction (Colegrave, 2002).

Conclusions
Since the studies of Riebesell et al. (2000) and Langer et al. (2006Langer et al. ( , 2009) ) species-and strain-specific performance of coccolithophores under elevated pCO 2 levels is known from short-term experiments, typically involving 7-10 cell generations.Here, we discussed data from a multiple-generation experiment using two coccolithophore species which generally confirm the observed CO 2 sensitivities obtained in shortterm experiments, though based on our data it can not be distinguished whether this is the result of a sustained acclimation response alone, or involved genetic change.A gradual CO 2 increase did not alleviate the CO 2 /pH sensitivity under the experimental conditions.In contrast to earlier studies we observed reduced growth rates in response to elevated pCO 2 . Fig.1

Fig. 3 .
Fig. 3. Physiological responses of Coccolithus braarudii to elevated pCO 2 over the course of the experiment.Labels and symbols as in Fig. 2.
Fig. 4Fig.4. Representative SEM photographs of the two coccolithophore species.Cells of E. huxleyi grown in the control treatment (A) and under high pCO 2 at day 73 (B).Cells of C. braarudii grown in the control treatment (C) and under high pCO 2 at day 66 (D).

Fig. 4 .
Fig. 4Fig.4. Representative SEM photographs of the two coccolithophore species.Cells of E. huxleyi grown in the control treatment (A) and under high pCO 2 at day 73 (B).Cells of C. braarudii grown in the control treatment (C) and under high pCO 2 at day 66 (D).
Figure 5b displays the POC:TPN ratios of E. huxleyi (data from www.biogeosciences.net/7