Introduction

Anthropogenic activities are causing a rapid increase in atmospheric carbon dioxide (CO2), about 30 % of which is absorbed by the oceans (Sabine et al. 2004; Caldeira and Wickett 2003). Atmospheric CO2 levels are predicted to increase to 900 parts per million (ppm) or more by the year 2100 (Meehl et al. 2007). The resulting changes in ocean water chemistry are collectively termed ocean acidification (OA) and include increased pCO2, decreased pH, and a lowered saturation state for major calcium carbonate mineral forms, including aragonite and calcite, that are formed in biogenic calcification (Caldeira and Wickett 2003). Responses of calcifying organisms to OA have been intensively studied during the past decade (Orr et al. 2005; Hofmann et al. 2010; Kroeker et al. 2010), and diverse responses have been observed (Kroeker et al. 2010).

Coccolithophores are unicellular marine phytoplankton that produce intricate calcite plates called coccoliths (McIntyre and McIntyre 1970) and fix carbon via photosynthesis, therefore playing an important role in ocean–atmosphere CO2 exchange and the marine carbon cycle. Effects of the predicted changes on coccolithophore calcification and photosynthesis, however, have primarily been studied in only a few species, including Emiliania huxleyi and the closely related Gephyrocapsa oceanica.

The cosmopolitan coccolithophore genus Calcidiscus thrives across a large geographic range under diverse environmental conditions (Knappertsbusch et al. 1997; Renaud et al. 2002; Silva et al. 2009). Calcidiscus coccolithophores play an important role in air–sea CO2 exchange, because (1) they contribute substantially to global calcite production and (2) their relatively large coccolith mass and the resistance of their robust coccoliths to dissolution (Schneidermann 1977; Baumann et al. 2005) may affect the rate of organic carbon export to the deep ocean. While prolific species such as E. huxleyi may be numerically dominant in phytoplankton assemblages, large and heavily calcified species such as Calcidiscus leptoporus and Calcidiscus quadriperforatus often dominate calcite production (Baumann et al. 2004; Ziveri et al. 2007). For example, in regions of the South Atlantic Ocean, Calcidiscus coccolithophores can account for more than 50 % of carbonate production, while E. huxleyi rarely accounts for more than 10 % (Baumann et al. 2004). Furthermore, the majority of the transport of organic carbon to the deep ocean has been attributed to coccolithophore calcite ballasting (Klaas and Archer 2002), although this hypothesis has been challenged (Passow 2004; Passow and De La Rocha 2006; Lee et al. 2009). Thus, predictions of how coccolithophores will respond to OA, and how these responses will in turn affect ocean–atmosphere CO2 exchange and the marine and global carbon cycles, require the study of biogeochemically important coccolithophores such as Calcidiscus.

Studies focusing on E. huxleyi suggest OA may have negative impacts on morphology and calcification (Ridgwell et al. 2007; Kroeker et al. 2010), although responses appear to vary among strains (Langer et al. 2009, 2011). Studies that have examined the response of Calcidiscus spp. focus on C. leptoporus and have generally found negative impacts of OA on growth, calcification, and morphology (Langer et al. 2006; Langer and Bode 2011). However, studies examining the response of a different C. leptoporus strain to future pCO2 levels reported no physiological changes besides an increase in growth rate and decrease in production of transparent exopolymer particles (TEP; Fiorini et al. 2011; Pedrotti et al. 2012).

Coccolithophores partition carbon into three particulate pools: the cellular organic carbon pool, the calcite fixed to form coccoliths, and the extracellular organic carbon pool existing largely as free TEP or cell surface coatings. Varying responses to OA in terms of carbon partitioning are known among strains of the genetically diverse species E. huxleyi (Saez et al. 2003; Langer et al. 2009), and we hypothesized that responses of Calcidiscus strains and species to OA are similarly non-uniform. As outlined above, such differences could potentially have important consequences for carbon sequestration. In order to determine the variation in carbon partitioning, morphology, and growth response among Calcidiscus spp, we exposed two strains of C. leptoporus and one strain of C. quadriperforatus to a range of pCO2 levels ranging from pre-industrial (~200 ppm) to over 100 years into the future (~1350 ppm). Phenotypes assessed in each experiment included growth, coccolith morphology, organic and inorganic carbon allocation, and extracellular particulate (alcian blue-stainable) material.

Materials and methods

Responses to OA in the diploid life phase were determined for two strains of Calcidiscus leptoporus and one strain of Calcidiscus quadriperforatus after acclimation to 4–5 different pCO2 treatments (Table 1). Cells were grown in triplicate dilute batch cultures, and cell concentrations and carbonate chemistry were measured at the beginning and end of each experiment. After a minimum acclimation period of seven generations, samples were collected to determine cellular and extracellular carbon allocation, and coccolith morphology.

Table 1 Carbonate chemistry parameters of Calcidiscus cultures
Full size table

Culturing conditions

Monospecific cultures of C. leptoporus strains RCC1130 and RCC1141, and C. quadriperforatus strain RCC1168 were obtained from the Roscoff Culture Collection (http://www.sb-roscoff.fr/Phyto/RCC). Culture media consisted of nutrient-poor aged seawater (collected in October 2008 offshore in Half Moon Bay, CA, USA; 37°29′31′N, 122°30′02′W) that was filter-sterilized (0.2 μm) and supplemented with vitamins and trace metals as in f/2 medium (Guillard and Ryther 1962), phosphorus as Na3PO4 (final concentration 14 μM), and nitrogen as NaNO3 (final concentration 100 μM).

Cells were grown in dilute batch cultures in 2-L glass bottles (Pyrex) with Teflon-lined screw caps (Corning) to limit gas exchange with the atmosphere (Langer et al. 2006). Samples were collected during exponential growth phase after ≥7 generations. Cultures were sampled at a low final cell density (3000–8000 cells mL−1) to minimize changes in seawater chemistry throughout the batch experiment (Rost and Riebesell 2004; Langer et al. 2006), with less than 8 % change in total alkalinity (TA) and dissolved inorganic carbon (DIC; Table 1).

Experiments were conducted at 20 °C with a light intensity of 350 μmol photons m−2 s−1 (Quantum scalar irradiance meter, Bispherical Instruments QSL-101) supplied by cool-white fluorescent lamps (Vita-Lite 5500 K, DUROTEST) and a 16-/8-h light/dark cycle. Sampling occurred at the same time each day, at 6 h following light onset. Salinity of the culture media was 33.65 ± 0.05, n = 3, measured using a YSI 30 salinity/conductivity/temperature meter (YSI, Inc., Yellow Springs, Ohio).

Manipulation and determination of carbonate chemistry

Coccolithophore culturing was performed at pCO2 levels encompassing a global-change relevant range (Table 1). Seawater carbonate chemistry was altered by the addition of calculated amounts of 1 M HCl and 1 M Na2CO3, using carbonate chemistry from unaltered “ambient” media as a baseline. The 200 ppm pCO2 level was achieved by bubbling media overnight with air passed through soda lime CO2 absorbent (Sodasorb) to remove CO2, and then, appropriate amounts of HCl and Na2CO3 were added. Seawater carbonate chemistry parameters were calculated from measured TA, DIC, salinity, and temperature, using CO2calc (Robbins et al. 2010) with constants from Mehrbach et al. (1973) refit by Dickson and Millero (1987). The total hydrogen scale was used to assess changes in pH, and total phosphate and total silicate were set to 1 and 10 μmol kg−1, respectively, based on prior seawater analysis (data not shown).

TA samples of 100 mL were filtered through pre-combusted (5 h at 550 °C) GF/F filters (Whatman) and stored in borosilicate glass bottles (Pyrex) in the dark at 4 °C. Fifty microliters of the sample was used to measure TA by potentiometric titration (Bradshaw et al. 1981) using Gran plots for endpoint determination (Gran 1952). Prior to determination, samples were brought to room temperature (~22 °C), and a pH electrode (Thermo Scientific, Orion 8102bnuwp) was calibrated with NBS buffers of pH 4, 7, and 10 (Fisher, SB101, SB107, SB115). Fifty microliters of sample was weighed to the nearest 0.1 mg in a pre-weighed beaker containing a magnetic stir bar. Samples were titrated using a Metrohm dosimat 765 pH Meter (Metrohm Herisau Switzerland) and 0.05 M HCl (Merck Titrisol) prepared in a 3.5 % NaCl (Fisher Scientific) solution with ionic strength similar to the samples. Samples were measured only when analysis of certified reference materials (obtained from Andrew Dickson at the Oceanic Carbon Dioxide Quality Control, Marine Physical Laboratory at Scripps Institution of Oceanography, University of California San Diego (SIO), http://cdiac.ornl.gov/newsletr/fall98/reference.htm) yielded a value of less than ±0.8 % from the certified value.

DIC samples were sterile-filtered into 24-mL borosilicate glass vials using 0.2-µm syringe filters (Thermo Scientific) and stored without headspace at 4 °C in the dark. DIC was measured using a Monterey Bay Research Institute-clone DIC analyzer with acid-sparging and non-dispersive infrared (NDIR) analysis (Friederich et al. 2002) with pure O2 as carrier gas. Each sample was analyzed in triplicate (1.5 mL per injection) after addition of 5 % H3PO4. Certified reference materials for CO2 measurements were obtained from the Marine Physical Laboratory at SIO as stated above for TA quality control. Standards were measured at the beginning and end of each run, and also between every 4–5 samples. Based on the standard run and the slope of the drift during samples runs, corrected DIC values were determined. Values for pH and pCO2 were calculated from TA and DIC.

Determination of specific growth rate

At least 200 cells were counted using a Sedgewick-Rafter counting chamber (SPI Supplies) under a Zeiss Axioscop microscope. Specific growth rates (µ) were calculated using Eq. [1]:

$$\mu = \frac{{\ln (n_{\text{f}} ) - \ln (n_{\text{i}} )}}{d}$$
(1)

where n i is the first experimental cell concentration in cells mL−1, n f is the final cell concentration, and d is the difference in time (in days) between n f and n i. Cell counts were conducted at the same time each day, 6 h after light cycle onset.

Determination of carbon content and POC production

Total particulate carbon (TPC) and particulate organic carbon (POC) samples were collected by filtering 150 and 200 mL of culture, respectively, onto pre-combusted (550 °C for 5 h) GF/F filters (Whatman) and storing at −20 °C. For experiments with very low final cell concentrations (<2000 cells mL−1), sample volumes were increased to 200 mL for TPC and 400 mL for POC. Prior to processing, samples were dried at 60 °C overnight after which POC samples were acidified with 235 µL of 1 M HCl to remove inorganic carbon and then dried again. TPC and POC were measured on a PerkinElmer 2400 CHN analyzer at the University of Georgia Chemical Analysis Laboratory (http://cais.uga.edu/analytical_services/chemical_analysis/), using acetanilide as a carbon standard. Values were normalized to the number of cells filtered. Particulate inorganic carbon (PIC) was calculated as the difference between TPC and POC. POC production and calcification rates were estimated using growth rates.

Determination of coccosphere volume and coccolith morphology

Coccosphere volume and morphology were determined from scanning electron microscopy (SEM) micrographs. Samples were prepared for SEM by gently filtering 15–20 mL of the sample onto 0.2-µm pore polycarbonate filters (Millipore). Samples were dried overnight at 60 °C and stored in a desiccator. For imaging, filter pieces were cut and mounted onto aluminum specimen mounts (Ted Pella, Inc.) using liquid colloidal silver paint (Pelco, Ted Pella, Inc.), allowed to dry for at least 24 h, sputter-coated with iridium, and imaged with a ZEISS Ultra 55 Field Emission scanning electron microscope (Carl Zeiss Microscopy). All images were taken at the electron microscopy facility at San Francisco State University (http://emfimage.sfsu.edu/).

Determination of coccolith morphology was based on a categorization system modified from prior coccolithophore morphology studies (Bach et al. 2011; Langer et al. 2011; Langer and Bode 2011). These prior studies distinguished between “complete” and “incomplete” coccoliths; however, in the present study, this distinction could not be achieved objectively, and individual coccoliths were categorized as one of the following: normal, slightly malformed, malformed, and extremely malformed (Fig. 1). A minimum of n = 250 coccoliths were counted per treatment, with the exception of RCC1130 at 800 ppm, where n = 144 because fewer images were captured and fewer coccoliths were greater than 50 % visible.

Fig. 1
figure 1

Representative examples of coccoliths in each morphology category, including: a normal, b slightly malformed, c malformed, and d extremely malformed

Full size image

Coccosphere volume was determined using the open-source program Image J (Schneider et al. 2012). Scale bars on each SEM image were used as a reference to determine cell diameter. Only cells that were fully intact (i.e., clearly spherical and with no gaps in coccolith coverage) were used in the cell size analysis. We noted that at pCO2 levels higher than ambient, cell shape was not consistently spherical, which could be due to cells collapsing or other variations in cell shape. We were not able to determine and rectify the cause of these inconsistencies; thus, we did not include cell size data based on SEM images for these samples. A minimum of n = 30 cells were counted per treatment.

Determination of particulate extracellular acidic polysaccharides

Particulate extracellular acidic polysaccharides were determined colorimetrically from formalin-fixed samples using the dye alcian blue method (Passow and Alldredge 1995). Alcian blue is specific to acidic polysaccharides and stains free transparent exopolymer particles (TEP) as well as the mucopolysaccharide coating of Emiliania huxleyi (Engel et al. 2004). Alcian blue also stains the organic coating of Calcidiscus spp. (Pedrotti et al. 2012), and thus, the amount of particulate alcian blue-stainable material (PABM) includes free TEP and the extracellular surface coating of the cells. Samples (310 mL) were fixed immediately after collection using buffered formalin (end-concentration of 4 %) and stored at 4 ºC until all experiments were completed. Triplicate subsamples (100 mL each) per sample were filtered onto 0.4-µm PC filters (Poretics), stained and analyzed as described in Passow and Alldredge (1995). All samples were processed using the same alcian blue solution, which was standardized using gum xanthan. If one of the triplicates per sample was a clear outlier, that value was dropped. Results are expressed as averages of gum xanthan equivalents ± standard deviation or converted carbon equivalents using Eq. [2] following Engel et al. (2004).

$${\text{Carbon equivalent}} \left[ {\mu{\text{mol C L}}^{ - 1} } \right] = 0.033 \pm 0.007 \times {\text{gum xanthan equivalent}} \left[ {\mu{\text{g xanthan equivalents L}}^{ - 1} } \right] + 1.8$$
(2)

Statistical analyses

In order to directly compare responses to pCO2 across strains, pCO2 levels were recoded for statistical analyses as “low,” “ambient,” “medium,” or “high” for pCO2 levels of approximately 200, 500, 900, and 1200 µatm, respectively. For each response variable, linear regression models were generated for the interactive and independent factors strain and pCO2. Normality of the residual errors was tested using QQ normality plots (sensu Miller and Stillman 2013). Heteroscedasticity was tested using a Breusch–Pagan test (ncvTest, package car). When model residuals did not meet the normality and heteroscedasticity assumptions, the data were transformed. When model residuals were determined to be normal and without significant heteroscedasticity, a two-way analysis of variance (ANOVA) was performed to examine interactive effects of strain and pCO2 level. To examine the effects of pCO2 within each strain for each response variable, one-way ANOVAs were performed followed by Tukey HSD post hoc tests. Statistical significance was assumed at p ≤ 0.05, and all statistical analyses were conducted using R (version 2.14.2). Unless otherwise stated, results are reported as the mean ± SE and represent measurements of n = 3 biological replicate cultures.

Results

Growth rate

Growth rates of Calcidiscus leptoporus strain RCC1141 and Calcidiscus quadriperforatus strain RCC1168 decreased with increasing pCO2, while C. leptoporus RCC1130 displayed an optimal-curve-type response with a peak growth rate at 957 ppm pCO2 (Fig. 2a). Specific growth rates (µ) ranged from 0.15 ± 0.04 to 0.61 ± 0.03, with the highest growth rate observed in C. leptoporus RCC1141 in ambient conditions. Maximum growth rate in C. quadriperforatus occurred at 229 ppm, decreasing significantly (ANOVA, F = 4.7, p = 0.025, Table 2) by more than half at 1338 ppm (0.15 ± 0.04, n = 2). At 1110 ppm pCO2, RCC1141 cell counts and growth rates could not be determined, because large aggregates formed in all three replicate bottles (>40 cells, Fig. 2a). Aggregation was not observed in any other strain or in RCC1141 at any other pCO2.

Fig. 2
figure 2

Particulate carbon partitioning of Calcidiscus spp. strains (C. quadriperforatus RCC1168, C. leptoporus RCC1141, and C. leptoporus RCC1130) cultured under different pCO2 levels. From left to right and top to bottom: a specific growth rates (μ), b cellular concentration of total particulate carbon (TPC), c particulate inorganic carbon (PIC), d particulate organic carbon (POC), e calcification rate, f primary production rate, g PIC:POC ratio, and h cellular concentration of particulate alcian blue-stainable material (PABM). Each point represents the mean ± standard deviation for n = 3 cultures, except for PIC and PIC production rate in RCC1141 at 1100 ppm, in which n = 2. Statistically significant differences are given in Table 3. Blue vertical bars represent pre-industrial pCO2 (~280 ppm), and red vertical bars represent projected year 2100 pCO2 (~900 ppm)

Full size image
Table 2 Statistical significance within and among strains of differences in growth rate, cellular TPC, PIC, and POC as well as cell number-normalized calcification and POC production rate, and PIC:POC ratios
Full size table

Total particulate carbon (TPC)

Cellular TPC content increased in C. leptoporus strain RCC1141 and C. quadriperforatus RCC1168 with increasing pCO2 and showed no change in C. leptoporus strain RCC1130 (Fig. 2b). In C. leptoporus RCC1141, the lowest TPC content of 122 ± 10 pg C cell−1 was observed at ambient pCO2 and nearly doubled at pCO2 levels of 643 and 863 ppm. In ambient pCO2, the two C. leptoporus strains had roughly the same TPC content (Fig. 2b; Table 2). C. quadriperforatus RCC1168 had the highest TPC content of all strains at all pCO2 levels, ranging from 377 to 525 pg C cell−1 and showed a slight, though nonsignificant increase from low to high pCO2 levels (Fig. 2b; Table 2).

Particulate inorganic carbon (PIC) and particulate organic carbon (POC)

Cellular PIC content in all strains showed either a slight increase or no change (Fig. 2c) with increasing pCO2. In C. quadriperforatus, PIC content ranged from 251 to 382 pg C cell−1 and was highest at ambient but showed no overall trend or significant differences (Fig. 2c; Table 2). PIC content in C. leptoporus strain RCC1141 increased slightly with increasing pCO2, with the lowest mean PIC of 93 ± 11 pg C cell−1 observed at ambient. Cellular PIC content could not be measured in strain RCC1141 at the highest pCO2 level (1100 ppm), as aggregation occurred (Fig. 3). However, based on PIC:POC determinations and light microscopy observations, cellular PIC content may have decreased at this level (see Discussion). In C. leptoporus strain RCC1130, PIC content generally increased with increasing pCO2; highest PIC content was observed at 1255 ppm, following a slight decrease at 924 ppm.

Fig. 3
figure 3

Light microscopy images of aggregated cells from a culture of C. leptoporus strain RCC1141. Images were observed under a 20X objective. Cells grew for >2 weeks at 1100 ppm pCO2

Full size image

With increasing pCO2, the cellular POC content in all strains either increased or remained constant (Fig. 2d). POC increased significantly in C. leptoporus strain RCC1141 (ANOVA, F = 18.9, p = 0.0005, Fig. 2d); the lowest POC of 29 ± 2 pg C cell−1 was observed at ambient pCO2 and POC more than doubled to 80 ± 24 pg C cell−1 at 862 ppm. In C. quadriperforatus, the overall effect of pCO2 was not statistically significant (Table 2); however, there was a clear trend of increasing POC with increasing pCO2 (Fig. 2d), and POC content at 1338 ppm was 76 % higher than at ambient. In C. leptoporus strain RCC1130, POC content did not change with increasing pCO2 (Table 2; Fig. 2d).

Rates of calcification and POC production

Increasing pCO2 had a different effect on calcification rates (Fig. 2e) and POC production (Fig. 2f) in each strain examined. In C. leptoporus strain RCC1130, neither calcification nor POC production rate showed significant change with increasing pCO2, though there was a slight increase in POC rate at 924 ppm that coincided with a peak in growth rate (Fig. 2f). Mean POC production ranged from 22 to 30 pg C cell−1day−1, while calcification rate ranged from 15 to 21 pg C cell−1day−1. In C. leptoporus strain RCC1141, calcification rate, which ranged from 55 to 61 pg C cell−1day−1, did not change significantly with increasing pCO2. However, POC production, which ranged from 17 to 30 pg C cell−1 day−1, was lowest at ambient and increased by more than 70 % at higher pCO2 levels (ANOVA, F = 12.3, p = 0.0023, Table 2; Fig. 2f). In C. quadriperforatus strain RCC1168, both POC production and calcification rate decreased significantly with increasing pCO2 (ANOVA, F = 4.49, p = 0.029 and ANOVA, F = 8.28, p = 0.0044, respectively). Calcification rate ranged from 43 to 156 pg C cell−1day−1, while POC production ranged from 32 to 68 pg C cell−1day−1. The lowest calcification rate was observed at 1338 ppm, more than three times lower than at ambient conditions. Highest POC production was observed at 229 ppm pCO2, and production was less than half of that at 1338 ppm.

Ratio of particulate inorganic carbon to particulate organic carbon (PIC:POC)

At the pCO2 levels examined, PIC:POC was highest at ambient pCO2 in all strains and subsequently decreased with increasing pCO2 (Fig. 2g). In C. leptoporus strain RCC1141, PIC:POC decreased significantly with increasing pCO2 (ANOVA, F = 13.5, p = 0.00048, Table 2; Fig. 2g), and PIC:POC at 1110 ppm pCO2 was more than three times lower than that of ambient. PIC:POC also decreased slightly in C. leptoporus strain RCC1130, but the change was not significant. In C. quadriperforatus, the decrease in PIC:POC was not significant, but large; mean PIC:POC at 1338 ppm was less than half that of ambient (Fig. 2g).

Particulate alcian blue-stainable material (PABM)

PABM was not a significant function of cell number in any of the three investigated coccolithophores, suggesting that free TEP contributed significantly to PABM. The highest PABM normalized to cell number was observed in C. quadriperforatus RCC1168 (Fig. 2h). On average over all pCO2 treatments, cell number-normalized PABM was 275 ± 195, 168 ± 94, and 59 ± 10 pg GXeq cell−1 for C. quadriperforatus RCC1168, C. leptoporus RC1141, and C. leptoporus RC1130, respectively. Cell number-normalized PABM dynamics as a function of pCO2 differed for the three strains (Fig. 2h). Cell number-normalized PABM was not a function of pCO2 for C. leptoporus RCC1130 (Table 3), but was positively correlated with pCO2 for C. quadriperforatus RCC1168 (Table 3). And although the level of uncertainty is high, cell number-normalized PABM of C. leptoporus RC1141 may have exhibited an optimum-type relationship with pCO2 with the peak at 643 ppm (Fig. 2h).

Table 3 Correlation matrix for selected parameters of strain RCC1130, RCC1141, and RCC1168 individually and of all tested strains combined
Full size table

Cell number-normalized PABM was not a function of coccolith morphology in C. leptoporus RCC1141 or RCC1130, whereas PABM of C. quadriperforatus RCC1168 was significantly correlated with coccolith morphology (Table 3). While the fraction of slightly malformed and malformed coccoliths decreased with increasing cell number-normalized PABM, the fraction of extremely malformed coccoliths increased, resulting in significant negative correlations between PABM and coccolith malformation (Table 3). In C. leptoporus RCC1141 cell number-normalized PABM was a positive function of cellular POC and PIC, but a negative function of growth rate (Table 3). In C. quadriperforatus RCC1168, cell number-normalized PABM was a positive function of cellular POC, but a negative function of growth rate (Table 3). Cellular POC and PIC as well as growth rate were not correlated with cell number-normalized PABM in C. leptoporus RCC1130 (Table 3).

Coccolith morphology

Coccolith morphology was sensitive to increasing pCO2 in all strains (Fig. 4; Table 4, Online Resource 1). For each strain, the highest percentage of normally formed coccoliths occurred at the lowest pCO2 level tested, while the highest percentage of extremely malformed coccoliths occurred at the highest pCO2 level. Both C. leptoporus RCC1141 and C. quadriperforatus RCC1168 showed the highest percentage of normally formed coccoliths at 229 ppm, with a subsequent decline in coccolith morphology. Both strains had less than 10 % normally formed coccoliths at ambient pCO2. In C. leptoporus strain RCC1141, 89 % of coccoliths were normally formed at 196 ppm, and only 9 % were normally formed at the ambient pCO2. At the highest level of 1110 ppm, 0 % of coccoliths were normally formed, while 95 % were extremely malformed. This strain showed the most sensitivity with the lowest percentage of normal coccoliths observed through all strains and pCO2 levels. In C. quadriperforatus, only 42 % of coccoliths were normally formed at 229 ppm, and 25 % were extremely malformed. At the highest pCO2 level of 1338 ppm, 96 % of coccoliths were extremely malformed, while 0 % were normally formed. When comparing strains at ambient pCO2, C. leptoporus RCC1130 had the least amount of coccolith malformation; 85 % of coccoliths were normally formed, while only 1 % were extremely malformed. However, increased pCO2 still led to coccolith malformations; at 1255 ppm, 0 % of coccoliths were normally formed and 93 % were extremely malformed (Fig. 4a).

Table 4 Morphology data of coccoliths from three strains grown over a range of pCO2
Full size table
Fig. 4
figure 4

a Percentage of individual coccoliths that are normal (black), slightly malformed (gray), malformed (white), and extremely malformed (striped) for C. quadriperforatus strain RCC1168, C. leptoporus strain RCC1141, and C. leptoporus strain RCC1130, as a function of pCO2. b Representative SEM images of coccospheres from each strain at different levels of pCO2. Scale bars represent 2 m

Full size image

Strain- and species-specific physiological differences

Under ambient conditions, strains differed with respect to both mean cell diameter (including coccosphere) and the percentage of carbon allocated to calcium carbonate (Table 5). C. quadriperforatus had the largest cell diameter, with a mean ± standard deviation of 20 ± 2 µm, n = 30. The mean diameter of the smaller C. leptoporus strains were 13 ± 1 µm, n = 35 for RCC1141 and 12 ± 1 µm, n = 32 for RCC1130 (Table 5). C. quadriperforatus exhibited higher cellular PIC and POC concentrations than C. leptoporus strains, resulting in higher TPC content per cell (Fig. 2b).

Table 5 Cell diameter and percentage carbon allocated to calcite (%PIC) of Calcidiscus strains at ambient pCO2
Full size table

In C. quadriperforatus RCC1168, 74 ± 4 % of carbon was allocated to calcium carbonate. C. leptoporus strain RCC1141 had a similar  % PIC of 76 ± 3 %, but C. leptoporus strain RCC1130 had a lower value of 42 ± 10 %. C. leptoporus strain RCC1130 had the highest percentage of normally formed coccoliths at ambient pCO2 (89 %), while less than 10 % of coccoliths were normally formed in both C. leptoporus RCC1141 and C. quadriperforatus RCC1168 in this treatment (Fig. 4b).

Discussion/conclusions

Calcidiscus leptoporus and Calcidiscus quadriperforatus are heavily calcified coccolithophore species that contribute substantially to calcium carbonate production and export to the sediments (Baumann et al. 2004). However, few studies have examined the response of this species to future predicted changes in seawater carbonate chemistry (Langer and Bode 2011; Fiorini et al. 2011; Pedrotti et al. 2012). The well-studied coccolithophore species Emiliania huxleyi has been shown to exhibit strain-specific variable responses to OA (Langer et al. 2009). Based on the large amount of genetic and physiological variability present in the Calcidiscus genus (Quinn et al. 2003), we sought to investigate variability in its response to OA. We examined intraspecific differences in response to OA in two strains of C. leptoporus and the closely related, recently classified species C. quadriperforatus.

Strain- and species-specific responses to ocean acidification

Members of the coccolithophore genus Calcidiscus differ in their response to elevated pCO2, which affects multiple aspects of cell physiology. Variability in response to OA has been previously reported among strains of the coccolithophore E. huxleyi (Langer et al. 2009), and among coccolithophore species including E. huxleyi, C. leptoporus, and Syracosphaera pulchra (Langer et al. 2006; Fiorini et al. 2011), suggesting widespread inter- and intraspecific diversity of coccolithophores in response to OA (Langer et al. 2006). This variability may have important implications for adaptation of the Calcidiscus genera, and particular Calcidiscus strains and species, to future ocean conditions.

The well-studied coccolithophore E. huxleyi is physiologically diverse, consisting of several distinct morphotypes (Brand 1982; Young and Westbroek 1991; van Bleijswijk et al. 1994; Young et al. 2003, 2011), and also exhibits extensive genetic diversity both between and within populations (Medlin et al. 1996, Iglesias-Rodriguez et al. 2006; Read et al. 2013). It has been suggested that this genetic diversity allowed for the expansion of this prolific coccolithophore into widespread and variable environments and that it may underlie the ability of E. huxleyi to adapt to future ocean changes such as ocean acidification (e.g., Lohbeck et al. 2012; Benner et al. 2013). In the present study, we observed that physiologically and genetically variable Calcidiscus strains show variable responses to ocean acidification, suggesting that certain Calcidiscus strains may have greater potential than others to withstand future ocean changes. In combination with the E. huxleyi studies, our results lend support to the hypothesis that Calcidiscus coccolithophores may be able to adapt to changes in seawater chemistry. As our study examined only short-term physiological responses among strains, we can only hypothesize that genetic diversity among or within strains may allow for long-term adaptation to future OA.

In all strains examined here, coccolith morphology was impaired and PIC:POC ratio decreased with increasing pCO2. Other ecologically and biogeochemically relevant physiological parameters were also affected, depending on the strain, including PIC and POC production and content, growth rates, and production of extracellular particulate alcian blue-stainable material (PABM), either as transparent exopolymer particles (TEP) or as cell coatings. In C. leptoporus strain RCC1130, no significant difference in the carbon production and content, growth parameters, or cell number-normalized PABM production was observed between pCO2 treatments. Coccolith morphology in RCC1130 declined with increasing pCO2, but was less impacted at ambient pCO2. In contrast, C. quadriperforatus RCC1168 appeared more sensitive to changes in the carbonate system: POC and PIC content and production, growth rate, and PABM changed with increasing pCO2. C. leptoporus strain RCC1141 responded to increased pCO2 with increases in particulate organic and inorganic carbon, increases in POC production, and decreases in growth rate.

Possible explanations that could account for the strain- and species-specific responses observed in this study include the geographic origin of the isolated strains, experimental variability, and genetic and morphological variability. The three strains examined in this experiment were grown in the same experimental conditions, and experimental variability is not a plausible explanation for the variable responses observed. Geographic origin of isolation also does not appear to explain the observed variability. Three strains of C. leptoporus, including strain RCC1141 and RCC1130 from the present study and RCC1135 from the studies of Langer et al. (2006, 2012) and Langer and Bode (2011), were all isolated from the same region in the coastal waters of South Africa in the South Atlantic Ocean but showed variable responses to OA; RCC1141 and RCC1135 show a greater sensitivity to OA in terms of calcification and morphology than RCC1130. The C. quadriperforatus strain examined in the present study, RCC1168, was isolated from the Mediterranean Sea, and despite the distance in geographic isolation from the other strains examined, its sensitivity to OA was similar to the C. leptoporus strains RCC1141 and RCC1135. A similar conclusion was reached in Langer et al. (2009), where the responses of four strains of E. huxleyi to OA did not seem to be related to their origin of isolation. In addition, overall genetic differences between the strains did not correspond to differences in sensitivity to OA. C. leptoporus and C. quadriperforatus are genetically distinct species (Saez et al. 2003); however, C. leptoporus RCC1141 responded more similarly to the C. quadriperforatus strain than to the other C. leptoporus strain RCC1130. Potential differences in gene expression of OA responsive genes have not been examined for these strains, but might greatly improve our understanding of the factors responsible for the high variability in sensitivity to OA.

In our study, the more lightly calcified strain RCC 1130 was more resilient to the effects of OA in terms of growth rate (increased growth rate with increased pCO2) and impaired coccolith morphology than the other more heavily calcified strains (RCC 1168 and RCC1141). Based on these observations, we hypothesize that cellular calcite quota may be a predictor of sensitivity of Calcidiscus to OA. In ambient pCO2 conditions, RCC1130 has a PIC:POC ratio of less than one-third that of the other two strains (Fig. 2), meaning that less of the total cellular carbon was allocated to calcium carbonate. Strain RCC1130 was also examined in Fiorini et al. (2011), where a PIC:POC ratio of 1.5 was observed, a value higher than in the present study but still lower than any other strain examined to date (Table 5). Although the differences in PIC:POC between this study and Fiorini et al. (2011) may be attributable to culturing conditions (i.e., light intensity and light/dark exposure period were higher in the present study), these findings present an interesting hypothesis regarding calcification and responses to OA. Although it has been shown that increasing pCO2 shifts E. huxleyi community composition from more- to less-calcified species and morphotypes (Beaufort et al. 2011; Collins et al. 2014), important exceptions have also been identified. A heavily calcifying strain of the R morphotype was found associated with more acidic waters (Beaufort et al. 2011). A prior study by Iglesias-Rodriguez et al. (2008) also showed resilience of an R morphotype E. huxleyi strain to ocean acidification. To determine whether such a correlation existed in Calcidiscus, many more strains in both laboratory and environmental settings would need to be examined.

Effect of OA on the morphology of coccoliths

Coccolith morphology might be an important factor in the survival of Calcidiscus coccolithophores and may also influence carbon export to ocean depths and sequestration time. In this study, all strains showed extreme sensitivity in coccolith morphology in response to elevated pCO2, and at the highest pCO2 tested in all strains, more than 90 % of coccoliths were malformed. These results are consistent with prior studies conducted by Langer et al. (2006, 2011). In comparison, studies examining the effects of OA on E. huxleyi coccolith morphology have yielded variable results, with some studies showing negative effects of increased pCO2 on coccolith morphology (e.g., Langer et al. 2011 [quantitative], Riebesell et al. 2000 [qualitative]), and others showing no change (e.g., Iglesias-Rodriguez et al. 2008 [qualitative]). Our results suggest that the Calcidiscus genus is more sensitive to OA in terms of coccolith morphology than E. huxleyi, though field experiments and examination of more strains would be needed to further test this hypothesis. Furthermore, in the two sensitive Calcidiscus strains examined here (RCC1141 and RCC1168), coccolith morphology was compromised at the “ambient” treatment as compared to the low pCO2 levels, which may suggest that if these strains are unable to adapt to OA, their coccolith morphology may be impaired in future oceans.

It is unclear how impaired morphology, and calcite content in general, may affect survival and competition of natural coccolithophore populations. Since calcification, on the cellular level, is an energetically expensive and highly conserved process, retention of this trait likely provides an evolutionary advantage to coccolithophores. Laboratory studies has failed to conclusively identify a single rationale for coccolith production, though several potential mechanisms have been examined including protection from predators and UV irradiation (see review by Paasche 2002). The molecular mechanisms underlying coccolithophore calcification are similarly poorly understood, though much progress has been made (Paasche 2002; Brownlee and Taylor 2004; Mackinder et al. 2010). The recent publication of many transcriptomic studies have provided valuable insight into cellular pathways in E. huxleyi that may be involved in calcification as well as in response to OA (e.g., Richier et al. 2010; Mackinder et al. 2011; Rokitta et al. 2012; Benner et al. 2013; Lohbeck et al. 2014), although the lack of molecular tools available for coccolithophores hinders the progress of this field, highlighting the need for further research. Notably, all of these studies have been conducted using E. huxleyi, and similar studies in Calcidiscus could provide insightful comparisons. We hypothesize that, regardless of the molecular process underlying the calcification, impairments in coccolith morphology resulting from OA, which were observed in this study even with very modest increases in pCO2 levels, would be detrimental to the survival of Calcidiscus if the same effects occur in natural populations.

The roles of coccolith strength and composition in response to OA (i.e., resistance to dissolution) are not well understood. Although cells continue to grow and calcify despite coccolith malformations, it is possible that malformed coccoliths are weaker (i.e., can be crushed more easily or may dissolve more quickly) or in some way disadvantage the cell. In this case, Calcidiscus coccolithophores may experience negative impacts as a result of malformed coccoliths in future oceans. Normally formed Calcidiscus coccoliths are very resistant to dissolution (McIntyre and McIntyre 1970; Berger 1973), but it is unclear whether malformed coccoliths are similarly resistant. Langer et al. (2006) discussed the infrequent occurrence of malformed coccoliths in the fossil record, noting that since both malformed and normal coccoliths have been found in sediment traps (Kleijne 1990), adaptation in terms of coccolith morphology over time is a more likely explanation than selective dissolution. Even though Kleijne (1990) found malformed coccoliths in sediment traps, only normal coccoliths of C. leptoporus were present. The presence of malformed coccoliths in sediment traps does not exclude the possibility that malformed coccoliths might dissolve more readily than normally formed coccoliths. Furthermore, Kleijne (1990) did not look at the ratios of normal:malformed coccoliths from surface to deep traps to determine whether the ratio is stable through the water column or whether the percentage of malformed coccoliths decreases with depth compared to normal coccoliths. Our data, as well as data from Langer et al. (2006, 2011), suggest that it is important to test whether malformed coccoliths (especially of C. leptoporus and C. quadriperforatus) dissolve more easily than normal coccoliths since the observed responses of coccolith morphology to OA might affect future calcium carbonate sedimentation and ballasting. These changes might also influence air–sea CO2 exchange (see below).

PABM, TEP, and aggregation

Coccoliths of E. huxleyi and Pleurochrysis carterae are embedded in a polysaccharide layer (Marsh et al. 1992; Young et al. 1999; Hirokawa et al. 2005, Engel et al. 2004). The organic surface coating of coccolithophores embeds the coccoliths and facilitates their specific morphology (Westbroek et al. 1984; Young et al. 1999; Hirokawa et al. 2005). Pedrotti et al. (2012) determined TEP concentration of C. leptoporus strain AC370 (= RCC1130) to be 22.9 fg TEP-C µm−3 of cell under ambient conditions, whereas we determined 36.8 ± 5.8 fg PABM-C µm−3, suggesting that roughly one-third of the particulate extracellular polysaccharides was associated with the cell surface coating and the remainder with TEP. As TEP abundance and size were determined microscopically, the organic layer of coccoliths was not included in final measurements. C. quadriperforatus strain RCC1168 showed an increase in cell number-normalized PABM concentration and content with increasing pCO2, suggesting differences in regulation of the polysaccharides that embed coccoliths of C. quadriperforatus.

Increased PABM production due to ocean acidification could have different causes. Overconsumption of carbon and subsequent excretion of carbon-rich polysaccharides have been suggested to lead to increased TEP production under ocean acidification conditions, especially during diatom blooms (Arrigo 2007; Riebesell et al. 2007). However, an increase in PABM could also suggest an increase in the organic layer surrounding the coccoliths rather than an increase in free TEP. The organic layer embedding coccoliths is essential for their production (van Emburg et al. 1986); thus, an increase in the organic layer could potentially be associated with shifts in coccolith production. The organic layer also protects the coccolith from dissolution to some extent (Henriksen et al. 2004), and a thicker organic layer under increased pCO2 could strengthen the protection under more acidifying conditions. The positive correlation of the calcium carbonate-normalized PABM and pCO2 (Table 3) in strain RCC1168 strengthen the suggestion that the material of the organic layer surrounding each coccolith might have increased, but investigations that differentiate between TEP and the cell-associated polysaccharide coating are needed to determine whether increased pCO2 leads to a thicker or stronger organic layer protecting the coccoliths from dissolution.

At a high, yet environmentally realistic pCO2 (1110 ppm), we observed aggregation of cells in the C. leptoporus strain RCC1141. Langer and Bode (2011) observed aggregation of C. leptoporus strain RCC1135 at pCO2 higher than 1500 ppm. They concluded that pCO2 was the environmental parameter causing aggregation and suggested that an increase in TEP could be the reason for aggregation, but did not determine TEP concentration. TEP facilitate aggregation by increasing particle abundance and by increasing stickiness of particles (Logan et al. 1995). Assuming a constant physical environment, particle abundance and size determine the probability of particle collisions and high stickiness facilitates attachment after collision (see Burd and Jackson (2009) for review). PABM concentration was not appreciably higher in the treatment where aggregation occurred. Thus, assuming that the ratio of TEP to cell coating remained constant, either a change in stickiness or an increase in particle numbers or sizes (e.g., from cell debris or detached coccoliths which were not measured) could explain the observed aggregation. Our study thus does not support the idea suggested by Langer and Bode (2011) that increasing pCO2 leads to an increase in TEP and subsequent aggregation.

Future environmental impacts

Our results suggest a high functional diversity within Calcidiscus. Variability in the response to OA was high between different strains of the same species and between species. This high functional diversity within a population or genus may allow coccolithophores to adapt successfully to changing environmental conditions. Shifts among strains within a community could allow Calcidiscus coccolithophores to persist through environmental changes when time scales are too short to allow adaptation of favorable genotypes. The ability to acclimatize to changing conditions would in this scenario lie with the population (i.e., strain), rather than within a species.

Since the least sensitive strain, C. leptoporus RCC1130, calcifies less than the other strains and also has a growth rate unaffected by (or even enhanced by) OA, an increase in its abundance in nature as compared to other strains could result in an overall decrease in coccolithophore calcite production. For example, we can compare the two C. leptoporus strains RCC1130 and RCC1141, which are of similar size and were isolated from the same region. At ambient pCO2, RCC1141 has a faster growth rate and produces roughly two times the amount of PIC per cell (Figs. 2, 4). However, at 924 ppm pCO2, a level currently predicted for oceans in the year 2100, the lightly calcified RCC1130 has a faster mean growth rate and its PIC content is less than 20 % of that of RCC1141. This assumes that grazing pressure does not determine the relative abundance of both. In some regions, Calcidiscus coccolithophores are responsible for ≥50 % of calcite production and export to ocean sediments (Baumann et al. 2004; Ziveri et al. 2007). In that case, if a lightly calcified Calcidiscus strain such as RCC1130 were to become dominant, overall calcite production could be reduced by ≥40 %. The possibility of a strain such as RCC1130 becoming dominant is also supported by the fact that it is more resilient to OA than the other Calcidiscus strains examined in terms of PIC:POC ratio and morphology (see below).

Regardless of competition among strains in terms of growth rate and resilience to OA, a decrease in the PIC:POC ratio among heavily calcified strains may also lead to changes in ocean–atmosphere CO2 exchange. PIC:POC ratio is one of the most important indicators of coccolithophore impacts on marine carbon cycles and ocean–atmosphere CO2 exchange, because calcification has the effect of counteracting CO2 drawdown during POC production (Rost and Riebesell 2004; Findlay et al. 2011). Changes in the PIC:POC ratio could also affect the ratio at which PIC and POC are exported to the deep sea. A decrease in calcification relative to organic carbon production, which has been predicted for pelagic calcifying organisms as a group (Kroeker et al. 2010), could therefore potentially create a negative feedback effect on rising atmospheric pCO2 on short timescales (Zondervan et al. 2001; Rost and Riebesell 2004). PIC:POC ratio decreased with elevated pCO2 in the heavily calcified Calcidiscus strains RCC1141, RCC1168, and RCC1135 (Langer et al. 2006), while the lightly calcified RCC1130, which has a naturally lower PIC:POC, was not significantly affected (Fig. 2). Thus, even if Calcidiscus biodiversity were to remain constant in the face of OA, the overall PIC:POC ratio may decrease. If, as discussed above, strains with a low PIC:POC ratio become the dominant Calcidiscus members, the end result is also an overall decrease in PIC:POC ratio, and potentially a decline in the PIC to POC export ratio.

This study suggests that changes in seawater carbonate chemistry may have potentially large impacts on both the distribution and the calcite contribution of Calcidiscus coccolithophores to the global carbon cycle. If Calcidiscus community structure does not change (i.e., less heavily calcified strains continue to exist alongside more heavily calcified strains, despite differences in sensitivity), then the more heavily calcified strains may see decreases in PIC:POC, which could in turn affect the sinking rates of particulate carbon to the deep ocean, and ocean–atmosphere CO2 exchange. In all possible scenarios, future ocean carbonate chemistry will not be conducive to the formation of normally formed coccoliths in Calcidiscus coccolithophores, and if malformed coccoliths dissolve more readily than normal ones, the marine carbon cycle could be influenced by increased dissolution and less calcite sequestration in deep-sea sediments.