Significant CO₂ fixation by small prymnesiophytes in the subtropical and tropical northeast Atlantic Ocean

Abstract

Global estimates indicate the oceans are responsible for approximately half of the carbon dioxide fixed on Earth. Organisms ⩽5 μm in size dominate open ocean phytoplankton communities in terms of abundance and CO₂ fixation, with the cyanobacterial genera Prochlorococcus and Synechococcus numerically the most abundant and more extensively studied compared with small eukaryotes. However, the contribution of specific taxonomic groups to marine CO₂ fixation is still poorly known. In this study, we show that among the phytoplankton, small eukaryotes contribute significantly to CO₂ fixation (44%) because of their larger cell volume and thereby higher cell-specific CO₂ fixation rates. Within the eukaryotes, two groups, herein called Euk-A and Euk-B, were distinguished based on their flow cytometric signature. Euk-A, the most abundant group, contained cells 1.8±0.1 μm in size while Euk-B was the least abundant but cells were larger (2.8±0.2 μm). The Euk-B group comprising prymnesiophytes (73±13%) belonging largely to lineages with no close cultured counterparts accounted for up to 38% of the total primary production in the subtropical and tropical northeast Atlantic Ocean, suggesting a key role of this group in oceanic CO₂ fixation.

Introduction

Looking inside the phytoplankton ‘black box’ to determine the distribution of carbon biomass and group-specific primary production is fundamental to understanding the roles of these groups in the global carbon cycle. The contribution of different phytoplankton communities to primary production has been extensively studied using size fractionation approaches (Teira et al., 2005; Pérez et al., 2006), and revealed a major role of the smallest phytoplankton (<2–5 μm) in the open ocean. Direct flow cytometric sorting of these smallest picoplankton cells coupled with radioactive tracer experiments have shown that although cyanobacteria dominated carbon fixation into biomass in the open ocean, nevertheless eukaryotic cells contributed significantly in spite of their lower abundance (Li, 1994). Subsequent studies conducted in both coastal (Worden et al., 2004) and open ocean waters (Goericke, 1998; Grob et al., 2007), which did not involve cell sorting, supported Li's observations. Moreover, studies based on size-fractionated pigment analysis also suggest a wide distribution of these small photosynthetic eukaryotes (Latasa and Bidigare, 1998; Wright et al., 2009). Recent progress using molecular approaches has begun to reveal their diversity (Moon-van der Staay et al., 2000; Díez et al., 2001; Vaulot et al., 2008; Worden and Not, 2008) showing a dominance of prymnesiophytes, pelagophytes and prasinophytes in marine environments (for example, see Moon–van der Staay et al., 2001; Romari and Vaulot, 2004; Not et al., 2004; Fuller et al., 2006a, 2006b; Viprey et al., 2008; Liu et al., 2009) and lately of chrysophytes (Fuller et al., 2006b; McDonald et al., 2007; Lepère et al., 2009). However, we still have a poor understanding of the phylogenetic affiliation of the smallest cells (for example, see Liu et al., 2010), and particularly in directly linking taxonomic identity with ¹⁴C primary production measurements. In this study, to determine the precise contributions of different phytoplankton groups to CO₂ fixation into biomass in the subtropical and tropical northeast Atlantic Ocean, we used a modified NaH¹⁴CO₃ radiotracer incubation flow cytometric sorting approach (Li, 1994) on board ship with natural communities sampled in surface waters (between 5 and 20 m) in which the majority of CO₂ fixation is thought to occur (Marañón et al., 2000; Poulton et al., 2006; Morán, 2007). In conjunction with these carbon fixation measurements, the differentiated phytoplankton groups were phylogenetically affiliated using molecular techniques, including fluorescence in situ hybridization (FISH). In doing so, we reveal that small prymnesiophytes, mostly of taxonomic lineages lacking close cultured counterparts, can contribute significantly to CO₂ fixation in the ocean.

Materials and methods

Sampling and experimental design

Samples were collected on board the Royal Research Ship Discovery during cruise D326 in the subtropical and tropical northeast Atlantic Ocean between 12 January and 1 February 2008 at 18 stations encompassing an area between 12–26°N and 24–36°W (Figure 1). Seawater samples were collected at dawn with 10 or 20 l Niskin bottles mounted on a conductivity–temperature–fluorescence–depth profiler, with either trace-iron free titanium frame (10 l) or metallic frame (20 l). All samples were pre-filtered through 100 μm pore-size mesh when being decanted from the Niskin bottles, directly into 20 l hydrochloric acid washed polycarbonate carboys. Gas tight high-performance liquid chromatography glass vials (8 ml) were then filled with 7.8 ml of sampled water and inoculated with either 74 kBq NaH¹⁴CO₃ for determining CO₂ fixation into total phototrophic biomass, or 7.4 MBq NaH¹⁴CO₃⁻ for assessing the contribution of the different groups of phototrophic microorganisms to CO₂ fixation. The vials were then incubated for up to 11 h—an average daylight period at the time of sampling. In situ light conditions on deck were monitored to be approximately 10% of direct sunlight, while the on deck incubation temperature varied <8.5%, on average, from in situ conditions. A set of samples was also incubated in the dark to ascertain dark CO₂ fixation. Up to 10 vial replicates incubated in the light and 5 vial replicates incubated in the dark were used for each experimental determination of total CO₂ fixation by phytoplankton. At each time point (∼3 h), three of the replicate vials were used for CO₂ fixation measurement, two as duplicates for light conditions and one for dark conditions. Samples were fixed with 1% (w/v) final concentration paraformaldehyde (PFA—Sigma-Aldrich, Poole, UK) to preserve cell integrity (Zubkov and Leakey, 2009) to enable quantitative comparisons with flow cytometrically sorted cells (see below). Fixed cells were then harvested onto 0.2-μm pore-size polycarbonate filters and washed twice with deionized water. A filter was placed in a scintillation vial and 1 ml of 1% (v/v) HCl (Sigma-Aldrich) was added to securely purge the residue of inorganic carbon. Subsequently, 5 ml of scintillation cocktail was added to each vial. The vial content was mixed and left to equilibrate before radioassaying. Radioactivity retained in cells, harvested on filters, was measured using a liquid scintillation counter Tri-Carb 3100 (Perkin Elmer, Cambridge, UK) on board the ship.

Figure 1

Schematic representation of the D326 cruise track. Circles indicate those stations at which CO₂ fixation experiments were performed. The black circle indicates the station (16400A) at which a sample used for clone library construction was collected.

Cell enumeration and flow cytometric sorting

Phytoplankton cells were enumerated by flow cytometry (FACSort—Becton Dickinson, NJ, USA) in seawater samples fixed with 1% (w/v) PFA final concentration to enable quantitative comparisons with flow cytometrically sorted cells. Previous work has shown no loss of either prokaryotic or eukaryotic cells occurs upon fixation (Zubkov and Burkill, 2006). Synechococcus cells were counted unstained (Zubkov et al., 2000) while eukaryotic phytoplankton and Prochlorococcus were counted after staining with SYBR Green I (Sigma-Aldrich, 1:5000 dilution of the commercial stock; see Zubkov et al., 2000, 2007). Multifluorescence 0.5-μm reference beads (fluoresbrite microparticles—Polysciences, Eppelheim, Germany) were used in all analyses as an internal standard for both fluorescence and flow rates. PFA-fixed, ¹⁴C-labelled Synechococcus and Prochlorococcus cells were flow-sorted from unstained samples. PFA-fixed, ¹⁴C-labelled eukaryotic cells were flow-sorted from SYBR Green I DNA-stained, unconcentrated samples. Flow sorting of both types of samples was completed within 10 h of fixation. The FACSort instrument was set at single-cell sort mode (the highest sorting purity of the instrument). Three to five proportional numbers (for example, 200, 400, 600, 800 and 1000 eukaryotes; 1000, 2000, 3000, 4000, 5000 and 6000 Synechococcus; 3000, 6000, 9000, 12 000, 15 000 and 18 000 Prochlorococcus) of cells for each target population were sorted. Sorted cells were collected onto 0.2-μm pore-size polycarbonate filters (Whatman, UK) and then processed as described above for the total samples. Mean cellular tracer incorporation by phytoplankton cells was determined by regression analysis (r²>0.97, P<0.05). To evaluate differences in CO₂ fixation rates among the different phytoplankton groups a two-tailed t-test was used (P<0.05). To assess sorting recovery (Zubkov and Tarran, 2008) total ¹⁴C fixation was determined in samples used for flow sorting by filtering six sub-samples (for example, 100, 150, 200, 300, 300 and 450 μl) onto 0.2-μm pore-size polycarbonate filters, washed twice with deionized water and radio-assayed as described above.

Fluorescent in situ hybridization

Seawater samples were amended with 0.05%. (v/v) pluronic solution (Sigma-Aldrich) final concentration to minimize clumping of concentrated cells. Samples (5 l) were then gently filtered through 0.2-μm pore-size CellTrap units (MEM-TEQ Ventures Ltd, Wigan, UK). Concentrated cells were washed off following the manufacturer's recommendations, fixed for 1 h at 4 °C with 1% (w/v) PFA, flash frozen in liquid nitrogen and then stored at −80 °C. Before sorting, samples were slowly defrosted on ice for 1 h and then stained with SYBR Green I (see above) for sorting eukaryotes or left unstained for sorting cyanobacteria. Samples were diluted with sterile seawater to obtain a flow cytometric counting rate of total particles <1000 particles per second. For each target group, 10 000–30 000-pigmented cells were sorted by flow cytometry based on endogenous phycoerythrin, chlorophyll a and stained DNA fluorescence signals (see above). Cells were then harvested onto 0.2-μm pore-size polycarbonate filters (Whatman), dehydrated in an ethanol series (50% (v/v), 80% (v/v) and 100% (v/v), 3 min each), dried at room temperature and stored at −80 °C until analysis. The 16S ribosomal RNA (rRNA) oligonucleotide probes used were PRO405 to target all known Prochlorococcus clades as well as the clade-specific probes 645LL, 645HLI and 645HLII (see West et al., 2001). For Synechococcus, the probes used targeted the following specific lineages: clade I (SYN1006), clade II (SYN1006RS), clade III (SYN262), clade IV (SYN635) or more generally clades I–VII and X (SYN1258). The nuclear 18S rRNA oligonucleotide probes used were EUK1209r, NCHLO01 and CHLO02 to target all eukaryotes and the class-specific probes PRAS01, PRAS02, PRAS05 (targeting prasinophytes), and CRYPTO13, PRYM02 and PELA01 targeting cryptophytes, prymnesiophytes and pelagophytes, respectively (Giovannoni et al., 1988; Simon et al., 1995, 2000; Biegala et al., 2003; Not et al., 2002, 2004). The chrysophyte-specific probe (CHRYSO1037) targeted the chloroplast 16S rRNA gene, as used previously for dot blot hybridization analysis (Fuller et al., 2006b) and validated here for tyramide signal amplification (TSA)-FISH analysis (see below). The protocol used for in situ hybridization with horseradish peroxidase-labelled probes has been previously described in Not et al. (2002) for the eukaryote-specific probes and in West et al. (2001) for the cyanobacterial-specific probes. In brief, before hybridization, cyanobacterial cells only were permeabilized by adding 30 μl lysozyme solution (5 mg ml⁻¹ dissolved in 100 mM Tris-HCl, pH 8, 50 mM EDTA—Sigma-Aldrich) and incubated at 37 °C for 30 min. Filters were then rinsed with MilliQ water, dehydrated in an ethanol series (50% (v/v), 80% (v/v) and 100% (v/v), 3 min each) and then dried at room temperature. The formamide (Sigma-Aldrich) concentrations used in hybridization buffers containing horseradish peroxidase-labelled probes were as follows: CHRYSO1037, 20%; EUK1209r, NCHLO01, CHLO02, PRAS01, PRAS02, PRAS05, CRYPTO13, PRYM02, PELA01, 40%; 645LL, 645HLII, 50%; 645HLI, SYN1006, SYN1006RS, SYN262, SYN635, SYN1258, 55%. Oligonucleotide probes (2 μl of 50 ng μl⁻¹ stock) were added to the hybridization buffer (variable % formamide, 0.9 M NaCl, 20 mM Tris-HCl, pH 7.5, 0.01% (w/v) sodium dodecyl sulphate [SDS]—Sigma-Aldrich) and incubated at 35 °C for 3–4 h. Filters were then washed twice for 20 min at 37 °C in washing buffer (56 mM NaCl, 5 mM EDTA, 0.01% (w/v) sodium dodecyl sulphate, 20 mM Tris-HCl, pH 7.5). Samples were equilibrated in TNT buffer (100 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.05% (v/v) Tween 20—Sigma-Aldrich) for 15 min at room temperature. TSA (individual fluorescein Tyramide reagent pack—Perkin Elmer) was performed for 30–40 min at room temperature in the dark in 10 μl of TSA mix following the manufacturer's recommendations. Filters were incubated twice at 55 °C for 20 min in TNT buffer and then dried at room temperature. Cells were counterstained with propidium iodide (final concentration 10 μg ml⁻¹) (Foulon et al., 2008) to visualize the nuclei. This allowed verification of labelling eukaryotic cells. Filters were then mounted in anti-fading reagent AF1 (Citifluor, London, UK). Cells were counted within 1 week (Not et al., 2002). The hybridized cells were counted under an Axioskop 40 (Zeiss, Jena, Germany) epifluorescence microscope equipped with a mercury light source and a Plan-Apochromat × 100 (Zeiss) objective. The green (fluorescein isothiocyanate) and red emission (propidium iodide) fluorescence produced by the different fluorochromes was collected between 510 and 550 nm and above 585 nm. Up to 350 positive cells were counted for the same sample and specific probe (data not shown). The average cell size of flow-sorted Euk-A and Euk-B cells was estimated using the fluorescein isothiocyanate signal by measuring at least 500 cells from each group in different samples.

Chrysophyte probe validation and optimization for whole-cell hybridization

The chrysophyte probe used in this study was originally designed by Fuller et al. (2006b) for use in dot blot hybridization analysis and its specificity to chrysophytes has therefore been validated. However, probe specificity was re-checked using our in-house ARB database containing recently published and unpublished plastid 16S rRNA gene sequences as well as by BLASTn analysis. In silico testing without mismatches revealed that 94.1% of the eukaryotic sequences (64 of 68 retrieved) were affiliated with chrysophytes. When the four ‘other’ sequences were inserted into our ARB database, it became apparent that these environmental sequences also grouped with chrysophytes. This analysis also showed that the CHRYSO1037 probe encompassed both marine and freshwater Chrysophyceae, including groups without a cultured representative.

Conditions for in situ hybridization were optimized using different formamide concentrations in the hybridization buffer together with varying the NaCl concentration in the washing buffer. A marine chrysophyte enrichment culture obtained from the Porcupine Abyssal Plain Observatory in the North Atlantic (L Jardillier and DJ Scanlan, unpublished data) was used as a ‘positive’ control. Marine representatives of various other photosynthetic eukaryote classes were used as negative controls (Prasinophyceae: RCC827, RCC1124, RCC745; Trebouxiophyceae: RCC289; Chlorophyceae: RCC6; Cryptophyceae: RCC20; Bacillariophyceae: RCC70; Dinophyceae: RCC88; Pelagophyceae: RCC100; Bolidophyceae: RCC205; Prymnesiophyceae: RCC192). Bright and clear FISH signals were obtained with the ‘positive’ enrichment control while no hybridization was obtained in any of the negative control cultures using the optimized probe conditions. Moreover, a no-probe control showed there was no signal coming from potential endogenous horseradish peroxidases (data not shown).

DNA extraction, clone library construction and phylogenetic analysis

DNA was extracted from 30 000 flow-sorted Euk-B cells (station 16400A) concentrated on board following the same procedure as for the FISH method and filtered onto 13 mm diameter 0.2-μm. pore-size polycarbonate filters (Whatman). DNA was subsequently extracted using a protocol adapted from Fuller et al. (2003) and Neufeld et al. (2007). Briefly, cells were resuspended in 50 μl lysis buffer (0.75 M sucrose, 400 mM NaCl, 20 mM EDTA, 50 mM Tris-HCl [pH 9.0]—Sigma-Aldrich), and then 6 μl 10% (w/v) sodium dodecyl sulphate, 10 μl 10 mg ml⁻¹ proteinase K (Roche, Welwyn Garden City, UK) added. The mix was incubated at 37 °C for 30 min, 55 °C for 10 min and then 2 μl 20 mg μl⁻¹ glycogen (Roche) added to aid DNA recovery. After 5 min at room temperature, 150 μl 100% (v/v) ethanol was added and the DNA precipitated by centrifugation. The pellet was washed in 80% (v/v) ethanol and finally resuspended in 10 μl 10 mM Tris-HCl [pH 8.0], 0.1 mM EDTA. Plastid 16S rRNA gene sequences were amplified using primers PLA491F (Fuller et al., 2006a) and OXY1313R (West et al., 2001). PCR amplification was performed in a total reaction volume of 50 μl using 200 μM deoxynucleotide triphosphates, 1.2 mM MgCl₂, 0.4 μM of primers and 2.4 U Taq polymerase in 1 × enzyme buffer (Qiagen, Crawley, UK) with 1 mg ml⁻¹ bovine serum albumin (Roche). Amplification conditions comprised 95 °C for 5 min followed by a step at 80 °C for 1 min when Taq polymerase was added. Then, 30 cycles were performed, of 95 °C for 30 s, 60 °C for 30 s and 72 °C for 40 s ended by a final extension at 72 °C for 6 min. PCR gene amplicons were cloned into pCR2.1-TOPO (Invitrogen, Paisley, UK). In all, 96 clones were chosen randomly and sequenced at the NERC Molecular Genetics Facility (Edinburgh, UK). Sequences were screened for the presence of chimeric artefacts using Chimera check from the Ribosomal Database Project II and Pintail software (http://www.bioinformatics-toolkit.org). Sequence alignment was performed using the ARB programme (Ludwig et al., 2004). The phylogenetic tree was constructed from sequences >1200 nucleotides in length using Jukes–Cantor correction and a maximum frequency filter for plastids that excluded alignment positions with sequence ambiguity or missing data. Shorter sequences were added by parsimony using the same filter. Bootstrap analysis was performed using the ARB parsimony bootstrap algorithm (Ludwig et al., 2004). The sequences reported in this paper have been deposited in Genbank under the following accession numbers: FJ797579–FJ797591 (see also Supplementary Table 1).

Results

Phytoplankton community composition

Photosynthetic eukaryotes (2.2±1.2 × 10³ cells ml⁻¹) were numerically the least abundant component of the phytoplankton, followed by Synechococcus (1.5±1.6 × 10⁴ cells ml⁻¹) and then Prochlorococcus (average cell number 2.1±0.8 × 10⁵ cells ml⁻¹), (Figure 2a, Table 1). Within the eukaryotes, Euk-A were approximately twice as abundant (1.4±1.1 × 10³ cells ml⁻¹) as Euk-B (0.6±0.2 × 10³ cells ml⁻¹). Euk-B represented 0.18–0.43% of the total oxygenic phototrophic community and 0.04–0.10% of the total microbial community (Figure 2b).

Figure 2

Composition of small phytoplankton in the studied area—(a) average cell abundance of the identified phytoplankton groups: Prochlorococcus, Synechococcus, and eukaryotes Euk-A and Euk-B; mean group % contribution to the total phytoplankton community in terms of (b) cell numbers and (c) biomass.

Table 1 Location, physical parameters, abundance and phytoplankton composition at the D326 stations sampled

On average Euk-A cells were 1.8±0.1 μm in size and Euk-B cells 2.8±0.2 μm (Table 1). The size of cyanobacteria (Prochlorococcus 0.5±0.2 μm; Synechococcus 1.0±0.3 μm) were determined by size fractionation on earlier cruises in the same area (Zubkov et al., 2000). A general biomass conversion factor of 0.22 pg C μm⁻³ was applied to biovolumes estimated from the cell sizes of the groups studied assuming a spherical shape (Table 1). Proportional to their numerical abundance, Prochlorococcus contributed the most to total phytoplankton biomass (51±19%) followed by Synechococcus (20±21%) (Figure 2c). The Euk-B group comprised 18±4% of total phytoplankton biomass owing to their relatively large cell sizes in spite of their low cell number. Although twice more abundant than the Euk-B group, the Euk-A group contributed less (11±8%) to phytoplankton biomass (Figure 2c).

Phytoplankton contribution to CO₂ fixation

Incorporation of ¹⁴C tracer into total phototrophic biomass over the time of the incubation, as evidenced by regression analysis (r²⩾98%, P<0.05), followed a simple sigmoidal curve, (Figure 3) indicating that phytoplankton cells were not physiologically impaired during the incubation process, for example, by the slight shifts in light and temperature conditions inherent to the experimental set up. There was no significant dark-driven CO₂ fixation, controls incubated in the dark being close to background (Figure 3). Nevertheless, the dark-fixation values were subtracted from light mediated CO₂ fixation rates. The average light CO₂ fixation rate across all 20 stations was 6.5±3.0 mg C m⁻³ d⁻¹, with a maximum of 12.3 mg C m⁻³ d⁻¹ observed close to the Cape Verde Islands (Table 2).

Figure 3

Sigmoid regression approximation (r²⩾98%, P<0.05) of total phytoplankton ¹⁴C uptake (Bq) (i) in the dark and (ii) in in situ simulated light conditions during the time course (h) of a typical tracer experiment.

Table 2 CO₂ fixation rates at stations along the D326 cruise track in the tropical North Atlantic

A second set of 100-times heavier ¹⁴C-labelled samples was used to determine group-specific contribution to CO₂ fixation. The combined CO₂ fixation of the total photosynthetic eukaryotes, Prochlorococcus and Synechococcus fitted well the total CO₂ fixation measured (101±7%) for the whole unsorted phytoplankton (Table 2). This indicated close to 100% recovery of flow-sorted cells (Zubkov and Tarran, 2008) as well as an absence of any other major CO₂-fixing groups in labelled samples, which remained unsorted. Hence, a CO₂ fixation budget for labelled samples could be ascertained. Owing to its high abundance, Prochlorococcus contributed 45±17% to CO₂ fixation in spite of the low cell-specific rates (1.2±0.6 fg C cell⁻¹ h⁻¹, Figures 4a and b, Table 2). Eukaryotes, on average, contributed approximately a third to total CO₂ fixation (34±8%) measured on the whole unsorted phytoplankton largely because of high cell-specific rates (98±50.0 fg C cell⁻¹ h⁻¹). Finally, Synechococcus contributed the least to primary production (21±13%) because of moderate abundance and cell-specific CO₂ fixation rates (9.5±4.3 fg C cell⁻¹ h⁻¹). Within the eukaryotes, the Euk-A group was responsible for between 7% and 19% of the total CO₂ fixed, because of the relatively low CO₂ fixation rates of this group (54.3±19.0 fg C cell⁻¹ h⁻¹). In contrast, Euk-B had high CO₂ fixation rates (196±72 fg C cell⁻¹ h⁻¹, Table 2) and in spite of their low abundance were responsible for 13–38% of the total CO₂ fixed. The biomass-specific CO₂ fixation rates (Figure 4c, Table 2) were not significantly different (t-test, P<0.05) between the different functional groups analysed here (0.09±0.04 mgC gC⁻¹ h⁻¹) although the Prochlorococcus rate was lower (0.06±0.03 mgC gC⁻¹ h⁻¹).

Figure 4

(a) Contribution of the phytoplankton groups Prochlorococcus, Synechococcus, and eukaryotes Euk-A and Euk-B, to total phytoplankton CO₂ fixation (b) cell-specific CO₂ fixation rates (fgC cell⁻¹ h⁻¹) averaged over all stations sampled and (c) biomass-specific CO₂ fixation rates (mgC gC⁻¹ h⁻¹) averaged at stations sampled wherein data were available for all phytoplankton functional groups.

Phytoplankton taxonomic composition

To ascertain the taxonomic composition of the different eukaryotic groups, that is, Euk-A and Euk-B, as well as Prochlorococcus and Synechococcus cyanobacteria, 10 000-pigmented cells from each group were sorted by flow cytometry from concentrated seawater samples at selected stations (see Materials and methods and Table 1) and identified by TSA-FISH (Schönhuber et al., 1997) using class-specific oligonucleotides targeting the nuclear 18S rRNA, except for the chrysophytes were a chloroplast-targeted oligonucleotide probe was used (see Materials and methods). The groups observed by flow cytometry in concentrated samples (Supplementary Figure 1) were identical to those in non-concentrated samples used for the tracer work (see also Zubkov and Tarran, 2008) showing that cell integrity was unaltered by concentrating and processing of samples. For the eukaryotes, the Euk-B group was dominated by prymnesiophytes (73±13%) (Figure 5a), with only minor contributions of pelagophytes (2±2%) and chrysophytes (8±8%) (Table 1). The Euk-A group comprised an assemblage of pelagophytes (36±15%), chrysophytes (26±15%) (Figure 5b) and prymnesiophytes (16±11%).

Figure 5

Fluorescence microscope image of a typical cell that was positively labelled with (a) the prymnesiophyte-specific probe or (b) the chrysophyte-specific probe. Green colour shows the positive signal of the horseradish peroxidase (HRP)-labelled probes (a) in the plastids and (b) in the whole cell while the red colour corresponds to the nucleus stained with propidium iodide.

Prasinophytes and cryptophytes, when detected, accounted for <4% of cells in either the Euk-A or Euk-B group. Phylogenetic analysis of 16S rRNA gene sequences obtained from a clone library constructed from flow-sorted Euk-B cells at one location (station 16400A) also showed a dominance of prymnesiophytes. Thus, of the 96 clones analysed, 64% were affiliated with prymnesiophytes including many sequences unrelated or poorly related (env SOL16400_NE12 and env SOL16400_NF03) to known cultured species (Figure 6). A more detailed phylogenetic characterization of the Euk-A and Euk-B groups will be presented elsewhere. As evidenced from the FISH data, there was remarkably little variation in taxonomic composition and size of the Euk-A and Euk-B groups (Table 1) indicating this community was largely homogeneous over the course of the study. Averaged over all stations Prochlorococcus comprised mainly (67±12%) cells belonging to the HLII lineage while Synechococcus was composed largely of members of clade II (41±26%) and clade III (31±28%) (Table 1) (see Fuller et al., 2003 for clade designations).

Figure 6

Neighbour joining tree of 16S rRNA gene sequences from marine algal plastids including new sequences derived from flow-sorted Euk-B cells at station 16400A. Bootstrap analysis was performed with ARB parsimony bootstrap (Ludwig et al., 2004). •: values > 90%; ○: 70–90%; values <70% are not shown. Environmental sequences are as follows: Solas: clones from the D326 cruise; AS: clones from the Arabian Sea (Fuller et al., 2006a); MC and MAMA: clones from the Gulf of Naples (McDonald et al., 2007); DYF: clones from the northwest Mediterranean Sea Dyfamed station (Marie et al., 2006); Biosope: clones from the southeast Pacific Ocean (Lepère et al., 2009).

Discussion

Assessment of phytoplankton group-specific CO₂ fixation by direct radiotracer incorporation has made little progress since the seminal study of Li (1994). In slight contrast to the Li study, however, here we used flow cytometric sorting of fixed cells to determine the precise contribution of different functional groups to CO₂ fixation into biomass, because we believe this provides more reproducible estimates compared with sorting of live cells (see Larsen et al., 2008). We should point out though, that while measurement of phytoplankton CO₂ fixation using a ¹⁴C radiotracer has been used for decades (Steemann-Nielsen, 1952) offering high precision (Irwin, 1991), the method does have some drawbacks including the low specific activity of labelling because of the high natural concentration of (bi)carbonate in seawater. Indeed, ¹⁴C added as dissolved inorganic carbon gradually enters pools of particulate and dissolved organic matter. Therefore, in long incubations the dynamics of the labelled carbon cannot accurately represent all transformations between organic and inorganic carbon pools (for example, respiration). Nonetheless, control sortings of heterotrophic bacterioplankton showed no measurable uptake of tracer into biomass (unpublished data) and hence, we feel justified that the ¹⁴C radiotracer method can accurately trace inorganic C assimilation into individual phytoplankton cells. However, as potential light respiration of fixed carbon by phytoplankton cells could occur, the reported specific rates of CO₂ fixation into cell biomass should be treated as lower or net estimates.

The total CO₂ fixation rates reported here (Figure 4, Table 2) are comparable to previous measurements (for example, see Landry et al., 2000; Pérez et al., 2006; Morán, 2007). Moreover, the large contribution of small phytoplankton (that is, Prochlorococcus, Synechococcus, Euk-A and Euk-B) to CO₂ fixation reported here (Figure 4) is also in agreement with previous size-fractionation studies (for example, see Marañón et al., 2001; Pérez et al., 2006). Indeed, Goericke (1998) observed that the contribution to chlorophyll a and primary production of phytoplankton <3 μm in size was almost constant over the year at a station southeast of Bermuda representing as much as 79% of the total.

Despite Euk-B being the Euk-B being the least abundant phytoplankton group, its contribution to oceanic CO₂ fixation reached 38% of total primary production (Table 2). Given the average cell size of the different functional groups, we might have expected that smaller cells would have competed more efficiently for resources (Raven, 1998) because of their high surface area to volume ratios. However, the opposite result confirms that CO₂ uptake is more dependent on biovolume than nutrient uptake rates. Interestingly, the lower biomass-specific CO₂ fixation rates of Prochlorococcus (Figure 4c) compared with the other functional groups analysed might be explained if Prochlorococcus cells can use harvested light not only for CO₂ fixation but also for directly producing adenosine triphosphate to fuel nutrient uptake including the acquisition of organic molecules (Zubkov, 2009). Certainly, it is clear that the Euk-B cells are important carbon fixers because of their high contribution to CO₂ fixation in spite of their low cell number. In addition, it is possible that, in spite of their small size, they can participate directly or indirectly to carbon export from the surface ocean (Richardson and Jackson, 2007). This facet may be more pronounced given that small phytoplankton <5 μm in size can also obtain significant carbon intake through bacterivory (Zubkov and Tarran, 2008).

While members of the Prasinophyceae, especially Micromonas spp., are now widely recognized as being the dominant component of the eukaryotic picophytoplankton in coastal environments (Not et al., 2004; Vaulot et al., 2008), much less is known of the composition of these communities in open ocean waters. The low proportion of prasinophytes observed in this study (Table 1) is consistent then with Micromonas spp. cell numbers decreasing rapidly in open ocean waters (Not et al., 2008). However, our use of the PRAS02 probe that targets all Mamiellales except Bathycoccus spp. could lead to the slight underestimation of prasinophytes here, and especially of this latter genus.

While pelagophytes appear to be widely distributed in these environments, though generally found at lower abundance than in this study (Bidigare and Ondrusek, 1996; Díez et al., 2001; Moon–van der Staay et al., 2001; Wilmotte et al., 2002; Romari and Vaulot, 2004; Fuller et al., 2006b; McDonald et al., 2007), there are few reports of members of the class Chrysophyceae in marine systems. While some ‘photosynthetic’ chrysophytes have been observed by microscopy, these are not abundant and usually refer to cells larger than 5 μm (Throndsen, 1996; Novarino et al., 1997). The exception is the order Parmales that encompass a group of small (<5 μm in diameter), siliceous cells, each with a chloroplast (Booth and Marchant, 1987; Bravo–Sierra and Hernández–Becerril, 2003). However, the chrysophytes observed by microscopy and TSA-FISH here had two chloroplasts, no apparent siliceous deposits in the outer membrane (that is, their 90° light scatter was similar to other eukaryotic cells) and were smaller than 2.5 μm (Figure 5b). Moreover, plastid 16S rRNA gene sequences of photosynthetic chrysophytes have recently been reported from various marine environments (Fuller et al., 2006a, 2006b; McDonald et al., 2007; Lepère et al., 2009) highlighting the potentially wide distribution and ecological importance of this class. Furthermore, 18S rRNA gene sequences of marine chrysophytes forming new groups without cultured representatives and comprising putative photosynthetic organisms have also been retrieved in several recent studies (Worden and Not, 2008; Shi et al., 2009).

Prymnesiophytes were clearly the main component of the Euk-B fraction as determined by FISH analysis. Plastid 16S rRNA gene sequences obtained through clone library construction of sorted Euk-B cells allowed us to ascertain more accurately the composition of the Euk-B group. Subsequent phylogenetic analysis (Figure 6) showed that these prymnesiophytes were closely related to environmental sequences retrieved from the southeast Pacific Ocean encompassing some of the most oligotrophic waters on earth (Lepère et al., 2009) suggesting these lineages have a wide geographic distribution. Moreover, the majority of these sequences were taxonomically affiliated with lineages with no close cultured counterpart. While prymnesiophytes appear widespread in most mesotrophic and oligotophic waters (Bidigare and Ondrusek, 1996; Moon-van der Staay et al., 2000; Díez et al., 2001; Wilmotte et al., 2002; Fuller et al., 2006b; Lepère et al., 2009) and can contribute from 30% to 50% of total photosynthetic standing stock across the world ocean (Liu et al., 2009), cultured representatives of the size fractions reported here are still rare. The prymnesiophytes observed in this study are likely not calcified as shown both by microscopy (Figure 5a) and their flow cytometry signature (that is, their 90° light scatter was similar to other eukaryotic cells) (Supplementary Figure 1) reinforcing the assertion that these prymnesiophytes include novel groups. Indeed, Young et al. (2005) have suggested coccolithophorids represent less than half the existing diversity of Haptophyta. Other studies are beginning to describe the extensive diversity within Haptophyta <3 μm in size (Worden and Not, 2008; Liu et al., 2009). Moreover, recent work (Urein et al., 2007; Zubkov and Tarran, 2008) suggests that oceanic mixotrophy by plastid-containing protists is dominated by small Haptophyta-like cells, consistent with both photoautotrophic and mixotrophic trophic modes being observed in cultured prymnesiophytes (for example, Chrysochromunlina sp. and Prymnesium parvum see Kawachi et al., 1991; Jones et al., 1995).

The key role of prymnesiophytes in marine CO₂ fixation we show here, is entirely consistent with new estimates of depth-integrated relative abundance of 19′-hexanoyloxyfucoxanthin, an accessory pigment considered characteristic of Haptophyta (though also present in a few other Heterokont algae, see Andersen, 2004), which indicate haptophytes dominate chlorophyll-a-normalized phytoplankton standing stock in the modern ocean (Liu et al., 2009). Further group-specific CO₂ fixation data collected at basin scales will be critical to assess the significance of these prymnesiophytes in global primary production. These data will also help to validate models that estimate ocean productivity using chlorophyll a-based fluorescence measurements (Behrenfeld et al., 2001). The next crucial step, beyond identification and assessment of their metabolic activities, will be to address the autoecology of the dominant photosynthetic eukaryotes by dissecting the biotic and abiotic factors that control their proliferation in the Ocean (Amann and Fuchs, 2008). Thus, the revealed major role for the prymnesiophyte-dominated phytoplankton group in photosynthetic CO₂ fixation in the subtropical and tropical northeast Atlantic Ocean should be considered an important stepping stone towards understanding the Oceans’ productivity.