O. marina is a globally distributed heterotrophic dinoflagellate that inhabits coastal and shallow waters (Watts et al., 2011). It occurs in coastal habitats as well as intertidal pools and is a comparably well characterized model organism to study the trades of marine protozoa (Montagnes et al., 2011). O. marina is 20–30 μm in length (Lowe et al., 2011) and moves fast (90–179 μm s^–1) (Boakes et al., 2011). It feeds by phagocytosis and is able to engulf encountered prey items within <15 s (Öpik and Flynn, 1989; Höhfeld and Melkonian, 1998). We used this model organism because it is easy to keep in culture and can be raised to high concentrations. The ingestion rate of O. marina increases with increasing prey density and maximizes under sufficient food concentrations, when the ingestion rate is limited by the predators capability to process the food (Type II functional response) (Roberts et al., 2011).

A starting culture of O. marina (SAG 21.89) was kindly provided by Dr. Urban Tillmann (Alfred Wegener Institute, Bremerhaven). The dinoflagellates were fed with Rhodomonas baltica. The cryptophyte R. baltica as well as the coccolithophore species were cultured under the same basic conditions: Sterile filtered artificial seawater (Kester et al., 1967) was enriched with 256 μmol kg^–1 NaNO₃, 16 μmol kg^–1 NaH₂PO₄, 10 nmol kg^–1 SeO₂, vitamins and trace metals according to the f/2 medium (Guillard and Ryther, 1962) and 2 mL kg^–1 of natural North Sea water (autoclaved and 0.2 μm filtered) to prevent potential limitation by other micronutrients that are not included in the f/2 receipt (Bach et al., 2011). Algal cultures were raised in a 12:12 h light-dark cycle with a photon flux density of 230 μmol photons m^–2 s^–1 (measured with a LI-COR LI-250A light meter) at 18°C. O. marina was inoculated into a growing culture of R. baltica and raised for a couple of days until high cell concentrations could be observed with the naked eye (The dinoflagellate cells appear in veil-like formations). Thereupon, the culture was placed dark at 18°C until all cells of R. baltica were eaten up, which was determined by flow cytometry (as described further below). The culture was further kept dark for about one week to starve the dinoflagellates. The concentration of starved cells was determined by flow cytometry. In this way, O. marina was raised to concentrations of ∼10 × 10³ cells mL^–1.

The coccolithophores E. huxleyi (B92/11), G. oceanica (RCC 1303) and P. carterae (unknown strain number) were grown separately in a large volume (5 L) of medium and raised to high cell densities for the 3 predation experiments (∼275 × 10³ E. huxleyi cells mL^–1, ∼220 × 10³ P. carterae mL^–1, ∼130 × 10³ G. oceanica mL^–1), in order to have a certain margin for the dilution steps that followed later, when the concentrations between the treatments were adjusted and the grazer cultures was added (described below) (Figure 1). Thereupon, the coccolithophore cultures were incubated in the dark for 36 h to stop the growth of the cells. The first 12 h of the dark incubation corresponded to the regular 12-h night cycle during which cell division took place. The goal of the subsequent 24 h of darkness was to deplete the energy reserves of the coccolithophore cells through respiration, in order to reduce the ability of the cells to build a new layer of coccoliths in the dark after the decalcification treatment was completed. All further steps during the dark incubation were conducted under low light conditions (<0.01 μmol photons m^–2 s^–1) to keep energy inputs via photosynthesis at a minimum during handling.

After 36 h in the dark we conducted the “decalcification step,” which was done with half of the population to remove the coccosphere of the treated cells (Figure 1). The large culture was split into two smaller bottles (2.3 L), one of which was treated with acid and base to remove the coccoliths. We added 2.5 mL 1M hydrochloric acid (HCl) L^–1 to the cultures of E. huxleyi and P. carterae to dissolve the coccospheres. The cells of G. oceanica were decalcified by adding 3 mL of 1M HCl L^–1 because this species is more heavily calcified. After the addition of the acid, the culture bottle was mixed for 1 min. Previous testing showed that the pH decreased to 4.3 during the brief acidification treatment. After this minute, the pH was brought back to the value before the acid addition by adding 1M sodium hydroxide (NaOH) solution. The bottles were rotated for 5 min to ensure that all flocculation from the addition of NaOH was dissolved. The decalcification procedure was conducted with a large culture volume (2.3 L) to keep the headspace at a minimum relative to the volume of the culture. In this way a potential degassing of CO₂ during the low pH/high pCO₂ period into the headspace of the volume was minimized. We tested this procedure in previous experiments to work out an appropriate volume of acid to be added to the cultures, which ensured the complete removal of the coccoliths (evaluated by microscopy) while keeping the decline of cell abundances at a minimum. However, cell concentrations always decreased to some extent due to the addition of the HCl and NaOH (by 6% for E. huxleyi, 18% for P. carterae, 7% for G. oceanica, of the target concentrations reported above). This was likely due to the contact with the highly concentrated acid and base at the time these were added. After the initial loss, the cell concentrations remained stable, which was determined by repeated flow cytometry measurements (as described below) over a period of 1 h. As the last step, the second (untreated) bottle with the calcified cells was diluted to an approximately equal concentration as in the acid-base treated bottles with 0.2 μm filtered artificial seawater. This facilitated the subsequent subdivision of both cultures into the replicate bottles of the individual treatments.

The content of both bottles containing either calcified or decalcified coccolithophores was split into eight replicate bottles (250 mL), respectively (Figure 1). Additionally, we established a mixed treatment by adding both, decalcified and calcified cells in a ∼ 1:1 mixture into eight replicate bottles. Thus, overall, we had a “decalcified” a “calcified” and a “mixed” treatment. Thereafter, an equal amount of the O. marina culture was added to the replicate bottles, some of which, however, were diluted with the same amount of 0.2 μm filtered artificial seawater instead and served as controls, to monitor the coccolithophore concentrations in the absence of the grazer. We took great care to keep the O. marina culture in a homogenous suspension during the addition by repeated mixing. All 8 bottles were mounted on a plankton wheel to prevent sedimentation of the algae. The experiments took place in complete darkness to minimize re-calcification of the decalcified cells and to prevent growth of the prey algae. Prey and predator abundances were sampled regularly, whereby the sampling took place under low light conditions as described above (Figure 1).

The coccolithophore concentrations were measured with a flow cytometer (Accuri C6, Becton Dickinson) at a flow rate of 66 μL min^–1 from 1 mL subsamples that were taken from the replicate bottles and transferred into Eppendorf tubes. The coccolithophore population was identified based on the chlorophyll fluorescence signal (FL-3) versus the forward-scatter signal (FSC) using the BD Accuri C6 Software. Calcified cells scatter more light at right angles than decalcified cells so that both could be distinguished on the basis of their side-scatter signal (SSC) (Olson et al., 1989).

To enumerate the concentrations of O. marina, we took 20 mL subsamples, which were fixed with Lugol’s iodine (1% final concentration). In the experiment that was conducted with E. huxleyi the cell number of O. marina was counted with an inverted microscope (Zeiss Axiovert 100) using 10 mL Utermoehl sedimentation chambers. The numbers of O. marina in the experiments with G. oceanica and P. carterae were counted with Sedgewick counting chambers using a light microscope (Carl Zeiss).

The cell size of the calcified and decalcified coccolithophores was measured via microscopy before the start of the experiment. 3.5 mL subsamples were taken from the initial cultures with calcified or decalcified cells before they were further split into the replicates. Subsamples were transferred onto slides for an inverted microscope (Zeiss Axiovert 100). Images were taken under 400-fold magnification with an Axiocam 105 color and cell diameters were measured with help of the ZEN 3.0 lite software (blue edition) (Carl Zeiss Microscopy GmbH). In addition, 3.5 mL subsamples were taken from single replicates that contained O. marina and either calcified or decalcified coccolithophores after 24, and 72 h, in order to observe grazer and prey interactions.

The diameter of 20 calcified and decalcified cells of each species was determined and tested for significant differences in mean diameter with a Welch’s unequal variances t-test. The counts of O. marina between the start (5 h after the start in case of the E. huxleyi experiment) and end of each experiment were used to calculate growth rates of O. marina for each replicate: μ = (ln(x_t)-ln(x_t0))/(t-t₀), where x = concentration of O. marina, t = time (in days). Differences in mean growth rates between the treatments were compared using an ANOVA. Data were tested for normal distribution and homogeneity of variances using Shapiro-Wilk and Levene’s tests, respectively. In case the null hypothesis of the ANOVA (mean growth rates are equal) was rejected (p < 0.05), a Tukey’s HSD test was used to compare the mean growth rates between treatments. Ingestion rates were calculated by dividing the difference in the prey cell concentration by the average predator concentration for a certain time interval, with the average predator concentration x¯ = (x_t – x_t0)/(ln(x_t) – ln(x_t0)) (Heinbokel, 1978).

Predation preferences of O. marina on calcified vs. decalcified coccolithophores within the mixed treatments were analyzed using the Chesson’s index α for depleting food densities (Manly et al., 1972; Chesson, 1983):

where n = cell concentration, i = ingested cells, and t = time. The index ranges from 0 to 1 and was calculated for every replicate. With only two prey types present, a value of α = 0.5 implies no preference for one of the two prey types and α < > 0.5 denotes that more cells of the respective type were consumed. A significant deviation of the mean α from 0.5 was tested with a Student’s t-test. Statistics were done using R 3.6.1 (R Core Team, 2019), and the packages car (Fox and Weisberg, 2019), and multcomp (Hothorn et al., 2008). Graphs were done with ggplot2 (Wickham, 2016).

O. marina showed no growth when fed with calcified E. huxleyi and G. oceanica and only marginal growth when calcified P. carterae served as prey. It can be assumed that at the end of all three experiments, the prey concentrations in the calcified treatments were still sufficient to ensure feeding saturation of the grazers (Figure 2, compare Supplementary Figure 1 for E. huxleyi). Hence, the ingestion rates of O. marina on the different calcified coccolithophore species can be compared.

Overall, the mean ingestion rates of O. marina on the calcified coccolithophores were lower when calculated for the entire period of 48 h compared to the respective ingestion rates calculated for the initial 24 h of the experiments, suggesting that the digestion of prey particles that had already been ingested limited further food uptake (unless in the second half of the experiments the predator cells just ate less for other reasons). An average of 57 ± 5.3 (SD) calcified E. huxleyi predator^–1 day^–1, 26 ± 6.7 (SD) P. carterae, and 23 ± 7.0 (SD) G. oceanica pred. ^–1 d^–1 were consumed during the first 24 h of the experiments. Between 24 and 48 h, additional 33 ± 3.4 E. huxleyi, 17 ± 4.1 P. carterae, and 6 ± 6.7 (SD) calcified G. oceanica pred. ^–1 d^–1 were taken up. Ingestion rates calculated between 0 and 48 h were 21% lower in the E. huxleyi experiment, 18% lower for P. carterae and 36% lower for G. oceanica compared to the respective rates calculated for the period between 0 and 24 h. However, in the G. oceanica experiment, the decalcified cells were also still abundant after 48 h (Figure 2F), although the predator concentrations increased (Figure 2I). Here, O. marina ingested 35 ± 7.6 decalcified G. oceanica pred. ^–1 d^–1 during the first 24 h and 22 ± 1.4 cells pred. ^–1 d^–1 in the period between 24 to 48 h, when the ingestion rates were corrected with the calculated average predator concentrations during both intervals. Consequently, the ingestion rate of O. marina on decalcified G. oceanica were 37% lower in the second interval compared to the first 24 h.

The decalcified cells of all three coccolithophore species could initially be distinguished from the calcified cells based on the strength of the side scatter signal measured with the flow cytometer. Both cell types appeared in two distinct clusters (Figure 3). However, a clear distinction between calcified and decalcified E. huxleyi was only possible until 24 h after the start of the experiment. Thereafter, the cluster of the decalcified cells overlapped with the cluster of the calcified cells, thereby impeding their differentiation in the flow cytograms (Figure 3). As a consequence, an increasing fraction of the initially decalcified cells was enumerated as calcified cells, which is why the concentrations of the latter increased between 24 to 48 h after the start (Figure 4A). More calcified than decalcified E. huxleyi cells were consumed during the first 24 h [α = 0.58 ± 0.003 (SD), t (3) = 53.29, p < 0.0001] and the calcified cells were eaten up completely after 48 h, whereas some decalcified were still left (Figure 4B).

Calcified and decalcified P. carterae could be distinguished from each other throughout the experiment in the mixed treatment based on the SSC and remained stable in concentration when predators were absent (Figure 4C). More decalcified than calcified P. carterae were consumed during the first 24 h of the experiment [α = 0.71 ± 0.09 (SD), t(4) = 5.63, p = 0.0049] and the decalcified cells were completely gone after 48 h (Figure 4D).

The decalcified G. oceanica were able to produce new coccoliths in the dark whereby the SSC of the cells increased. These re-calcified cells formed an intermediate cluster in the flow cytogram that was located in between those formed by the decalcified and the calcified cells, respectively (Figures 3H,I). Nevertheless, the cluster of calcified G. oceanica could be distinguished from the cluster of decalcified cells within the first 24 h of the experiment, during which O. marina ingested more decalcified than calcified G. oceanica [α = 0.73 ± 0.05 (SD), t (4) = 10.89, p = 0.0004] (Figures 4E,F). The intermediate cluster of re-calcified cells in the cytograms was more scattered when predators were present compared to the controls (Figures 3H,I). The mean SSC of the cluster of decalcified cells increased less over the course of the experiment when O. marina was present. This was also the case in the treatment that contained decalcified cells only, indicating that the grazers preferred the re-calcified cells (Supplementary Figure 2).

In the presence of O. marina, additional particle clusters could be observed on the cytograms, which were lower in red fluorescence than the coccolithophore cells, but showed similar FSC and SSC signals indicating they were of similar size (Figure 3). The additional particle cluster did not appear in the controls, but only in the presence of predators. Particles were more abundant in the cultures with calcified coccolithophores and likely represent egested food items.

The decalcified coccolithophores were smaller than the respective calcified cells: The mean diameter of decalcified E. huxleyi [3.77 ± 0.23 (SD) μm] was significantly smaller than the mean diameter of the calcified cells [4.81 ± 0.55 (SD) μm] [t (26) = 7.88, p < 0.0001]. This was also the case for P. carterae [decalcified = 8.50 ± 0.73 (SD) μm; calcified = 10.52 ± 1.02 (SD) μm; t (34) = 7.22, p < 0.0001] and G. oceanica [decalcified = 5.41 ± 0.74 (SD) μm; calcified = 7.01 ± 0.45 (SD) μm; t (31) = 8.27, p < 0.0001]. Calculations of spherical volumes (V = 4/3πr³) from the measured diameters showed that the volume (organic cell + coccosphere) of all three calcified species was roughly twice as large as the volume of the decalcified cells (organic only) (∼58 vs. ∼28 μm³ for E. huxleyi, ∼610 vs. ∼322 μm³ for P. carterae, ∼180 vs. ∼83 μm³ for G. oceanica). Thus, more decalcified cells could be taken up per individual predator simply due to their smaller size. The flow cytometry data suggest that the calcified E. huxleyi suffered less from the prolonged dark period than the decalcified cells (Supplementary Figure 3). The measurements showed that the chlorophyll fluorescence of the decalcified cells declined stronger and further indicate that the decalcified cells decreased slightly in size over the course of the experiment. However, this effect was small compared to the absolute difference between both treatments (Supplementary Figure 3).

Observations that were done during (and after) the experiments from single replicates that contained either calcified or decalcified cells showed that more decalcified than calcified cells fit into an individual O. marina (Figure 5). Moreover, calcified cells inside the predator kept their shape and many of the O. marina cells that had ingested several calcified coccolithophores looked deformed (Figure 5A). It also needs to be mentioned that P. carterae is a motile species. At the high concentrations at which P. cartarae was grown prior to the start of the experiment a large fraction of cells was not moving in both treatments. However, the decalcified cells became sticky and we could often observe pairs of cells moving stuck together. The potential reduction of mobility of these decalcified cells could facilitate the uptake by O. marina.

Live O. marina were highly motile and thus difficult to observe and to photograph. However, they occasionally stopped their continuous spiral movement and remained motionless for a short period, during which some individuals sank to the bottom of the microscope slide and could be photographed. Pictures had to be taken quickly, which in part explains the lack of focus in some of them. The relatively few observations do not allow quantitative statements and, of course, can be challenged with respect to their representativeness regarding the bulk population of O. marina. Nevertheless, they provided important insights into prey digestion. After 72 h some individuals of O. marina that fed decalcified coccolithophores showed small inclusions (Figures 5D,H) that resembled coagulated prey items under progressive digestion as described by Öpik and Flynn (1989). We were not able to discover similar patterns in cells of O. marina that were fed with calcified coccolithophores, suggesting limited digestion of calcified cells. However, the lack of this observation may also have been due to the fact, however, that remnants of calcified cells were obscured by freshly engulfed cells, which were still abundant at the end of the experiments in contrast to the decalcified cells (Figure 2). We once observed an individual of O. marina egesting a relatively undigested calcified cell of E. huxleyi (Figure 6). The entire egestion process took about 15 min, during which the predator barely moved. In contrast, we observed another specimen that was packed with several calcified cells of E. huxleyi that rounded up its shape. It egested two virtually empty coccospheres within minutes and moved quickly out of sight (Figure 7).

Öpik and Flynn (1989) described O. marina as voracious predator. It feeds by engulfing several prey particles one after another until there is no further space to take up more. Ingested food particles are then assimilated successively (Öpik and Flynn, 1989; Roberts et al., 2011). The microscopic observations showed that the coccolithophores E. huxleyi, P. carterae and G. oceanica fit well into the prey size spectrum of the predator. Within the pure cultures that contained either calcified or decalcified coccolithophores, the decalcified cells disappeared faster due to predation. Furthermore, O. marina was able to grow when feeding on decalcified coccolithophores but not, or at reduced rates, when feeding on calcified cells. Growth inhibition through calcification was most pronounced when O. marina was fed with the most heavily calcified species (G. oceanica) and least pronounced when fed with the least calcified one (P. cartarae). This suggests that growth inhibition due to calcification is positively correlated with the relative proportion of calcite to organic carbon in the prey (compare Figure 8). We propose 3 possible mechanisms to explain how the coccosphere could have reduced grazer growth. These will be discussed consecutively in the following but they are not mutually exclusive.

It has been shown that O. marina is size-selective (Hansen et al., 1996; Strom et al., 2012). Our results show that O. marina consumed more of the calcified than decalcified E. huxleyi. In contrast, more decalcified P. carterae and G. oceanica were ingested. Furthermore, the results of the flow cytometry measurements indicate that re-calcified cells of G. oceanica were primarily eaten. The results suggest that the differences in ingestion rates between calcified and decalcified coccolithophores within the mixed treatments were more likely due to the size of the prey algae than due to calcification. With regard to the prey size, O. marina preferred the larger calcified E. huxleyi (∼4.8 μm), which was the smallest of the three coccolithophore species that was tested in our experiments, but the smaller decalcified cells of P. carterae (∼8.5 μm), which was the largest species (calcified ∼10.5 μm) (Figure 8). This is consistent with previous findings that O. marina feeds optimally on prey sizes > 4 μm (Roberts et al., 2011).

The results suggest that calcification does not prevent O. marina from ingesting coccolithophores and confirm previous findings that the coccosphere does not deter protozoan grazers (Harvey et al., 2015; Strom et al., 2018; Mayers et al., 2020). It remains a key question, however, whether other microzooplankton species that are better adapted to natural phytoplankton communities would be able to select against calcified species. Protozoa that are capable to avoid calcareous cells and to select for more easily digestible food items would potentially have a clear advantage as they likely grow faster. The ability of O. marina to chemoreception is well developed. It has been shown that the grazer is able to select prey items based on their food quality (Meunier et al., 2012). Moreover, O. marina has been shown to feed on artificial particles that are flavored with organic molecules (Hammer et al., 2001), and is able to distinguish between artificial particles that are coated with different carbohydrate-binding proteins (Wootton et al., 2007).

Prey concentrations and encounter rates are much lower under natural conditions than in our experiments. In nature, protozoa usually need to browse large volumes of seawater to be able to cover their nutritional demand (Kiørboe, 2011), suggesting that they rarely have a choice to select their food. It has been argued that selection likely favors the evolution of chemosensory receptors in protozoa in order to seek for prey along chemical gradients (Breckels et al., 2011). The question is whether selection favors the evolution of appropriate receptors to detect an inorganic shell surrounding an otherwise attractive prey item in such a dilute environment.

Our results indicate that coccolithophore calcification can reduce the feeding efficiency and growth of phagotrophic protozoa. The benefit of this effect for coccolithophores is likely depending on their dominance within the bulk phytoplankton community. In a scenario where the calcified coccolithophore is a minor component of the phytoplankton community accessible to a non-selective protozoan, the decrease in predation pressure due to ingested calcite will primarily benefit the non-calcareous phytoplankton in the community. This is because competing non-calcifying species do not contribute to the substantial energetic burden associated with calcification (Monteiro et al., 2016), whereas they still benefit from the reduced predation pressure. However, reduced growth of the grazer through calcification will benefit coccolithophores that dominate the phytoplankton community and attenuate the top-down control on their proliferation (Harvey et al., 2015).