Diazotrophs: a non-negligible source of nitrogen for the tropical coral Stylophora pistillata

Tropical scleractinian corals obtain nitrogen via a variety of mechanisms including (i) symbiosis with unicellular dinoflagellates – zooxanthellae – that can take up and retain dissolved inorganic nitrogen (ammonium and nitrate) from the surrounding waters (Falkowski et al., 1993; Marubini and Davies, 1996; Muscatine, 1990), (ii) recycling of animal wastes by the zooxanthellae and subsequent transfer to the coral host as amino acids (Ferrier, 1991), ammonium or urea (Furla et al., 2005), (iii) ingestion of nitrogen-rich sediment particles (Mills and Sebens, 2004; Mills et al., 2004) and plankton (Ferrier-Pagès et al., 2003; Houlbrèque et al., 2004), and (iv) dinitrogen (N₂) fixation by symbiotic diazotrophic communities (Lema et al., 2012; Lesser et al., 2007, 2004; Shashar et al., 1994). Among these potential nitrogen sources, the contribution of free-living planktonic diazotrophs to the nitrogen metabolic requirements of the coral holobiont has not been quantified yet.

Pelagic N₂ fixation is regarded as the most important external source of fixed nitrogen to the ocean (Gruber and Galloway, 2008), fuelling primary production in oligotrophic waters and balancing the global ocean nitrogen reservoir (Mahaffey et al., 2005). In the New Caledonian lagoon, planktonic diazotrophs are very abundant and diverse (Biegala and Raimbault, 2008; Turk-Kubo et al., 2015), accounting for among the highest N₂ fixation rates in the world (Bonnet et al., 2015). Diazotrophic cyanobacteria retain >75% of the N₂ they fix intracellularly (Benavides et al., 2013a; Berthelot et al., 2015), and thus their direct ingestion could provide an important source of nitrogen for corals. Here, we investigated N₂ fixation within the coral holobiont and, for the first time, the ingestion of ¹⁵N-labelled cultured and natural diazotrophic assemblages from the New Caledonian lagoon, quantifying their contribution to the nitrogen balance of one of the major scleractinian corals: Stylophora pistillata.

Stylophora pistillata samples were collected in the New Caledonian lagoon (22.328°S, 166.410°E; sampling licence issued by the ‘Province Sud’, Government of New Caledonia). Thirty terminal branch portions (5 cm) were cut from five parent colonies, hung on nylon wires and suspended for 3 weeks in a 20 l aquarium supplied with 100 µm-filtered seawater, renewed at a rate of 16.5 l h⁻¹ and mixed using a submersible pump (Aquarium system, micro-jet MC 320, Mentor, OH, USA). Temperature (26±0.1°C) and salinity (35.69±0.02) were kept constant using heaters connected to electronic controllers (±0.2°C) and routinely verified using a YSI MPS 556 probe (YSI, Yellow Springs, OH, USA). Corals received a constant irradiance of 120±10 µmol photons m⁻² s⁻¹ (12 h:12 h) using four Aquablue plus neon bulbs (blue–white, 15,000 K, Giesemann, Germany).

After 3 weeks of acclimation, S. pistillata branches were incubated in quintuplicate 1 litre sterilized glass containers, filled with filtered (0.2 µm) lagoon seawater and continuously stirred with magnets. The corals were suspended from the top of the containers with the nylon wires. Nitrogen assimilation by the coral holobiont was tested by comparing two mechanisms: (i) the direct uptake of N₂ fixed by endosymbiont diazotrophs (experiment 1) and (ii) the ingestion of planktonic diazotrophs. In the latter case, we tested both a cultured diazotroph strain (experiment 2) and a natural diazotroph assemblage (experiment 3).

For experiment 1 we used the dissolved ¹⁵N₂ method (Mohr et al., 2010) as outlined in Großkopf et al. (2012). The quintuplicate glass containers were filled to 20% volume with ¹⁵N-enriched seawater, 80% unenriched filtered seawater, and incubated for 4 h.

In experiment 2, we incubated corals with a culture of unicellular diazotrophs isolated from the lagoon and cultured in autoclaved filtered seawater supplemented with nutrients in the same proportion as for the YBCII medium (Chen et al., 1996). The strain used was a 5–6 µm unicellular diazotroph classified as Cyanothece, which is regularly found in the New Caledonian lagoon (Turk-Kubo et al., 2015). The day before experiment 2 started, Cyanothece cells were pre-labelled with ¹⁵N₂ by incubating triplicate 20 ml culture aliquots (∼10⁶ cells ml⁻¹) with 20% volume ¹⁵N₂-enriched seawater (as explained above). These aliquots were incubated for 24 h at 27°C, 100 µmol photons m⁻² s⁻¹ on a 12 h:12 h light:dark cycle. After 24 h, triplicate 3 ml subsamples were filtered through pre-combusted (450°C, 5 h) GF/F filters (Whatman, St Louis, MO, USA) to obtain the initial ¹⁵N-enrichment of the culture (source pool). The rest of the culture aliquots were concentrated by gentle filtration onto 2 µm polycarbonate filters. The filters were added to 50 ml filtered seawater and vortexed to obtain a concentrated suspension of ¹⁵N-labelled cells (∼9×10⁵ cells ml⁻¹); 20 ml of this concentrate was added to the quintuplicate glass containers before filling them with filtered seawater and hanging the corals from the top. In parallel, triplicate containers with the same cell concentration but without corals were used as a control.

In experiment 3, we tested the ingestion of diazotrophs using a natural planktonic assemblage. We note that as Cyanothece is a usual member of the planktonic diazotrophic community in the New Caledonian lagoon (Turk-Kubo et al., 2015), the natural sample used may be partly composed of Cyanothece cells. Unfortunately, we did not perform strain-specific quantitative analyses that allow us to report the abundance of Cyanothece in our sample. Plankton was sampled from the lagoon the day before the experiment by prefiltering seawater through a 100 µm mesh to exclude larger cells. The sample was split into eight 4.3 l polycarbonate bottles: three were immediately filtered to obtain the initial ¹⁵N-enrichment of the source pool and five were incubated with 20% volume ¹⁵N₂-enriched seawater for 24 h as described above. At the end of the incubation, the content of the bottles was filtered onto 0.2 µm polycarbonate filters which were placed in 50 ml filtered seawater and vortexed to resuspend the cells. Aliquots of 5 ml were kept for cell counts (flow cytometry, see below), while the remaining 45 ml was added to the incubation containers with corals.

In all experiments, the temperature was checked for optimal conditions before attaching the coral branch to the top of the containers. Subsequently, the containers were closed hermetically using a glass slide. At the end of each experiment, corals were rinsed six times with filtered seawater to remove cells potentially adhered to the coral surface (Houlbrèque et al., 2003) and subsequently transferred to Ziploc bags and stored at −20°C until analysis.

where ¹⁵N_target,Tf is the ¹⁵N enrichment of the target pool (tissue or zooxanthellae) at the end of the incubation, ¹⁵N_target,0f is the corresponding ‘time zero’ ¹⁵N enrichment (natural ¹⁵N enrichment of the target pool at the start of the incubation). The source pool varies between the three experiments and both the ¹⁵N₂-labelled (¹⁵N_{enriched source}) and non-labelled (¹⁵N_{non-enriched source}) atom % values are considered in Eqn 1 as follows: (i) in experiment 1 the source pool is seawater and the isotopic enrichment of the filtered seawater added to the containers, (ii) in experiment 2, it is the Cyanothece culture and (iii) in experiment 3 it is the natural planktonic assemblage. t is the incubation time in hours and PN_target/A is the mass (µg) of particulate nitrogen of the target pool per area (cm⁻²) of coral skeleton surface (µg N cm⁻² h⁻¹).

The tissue was removed from the skeleton using an air-pick (Rodolfo-Metalpa et al., 2010) and homogenized with a Potter tissue grinder. Zooxanthellae and tissue were separated by centrifugation (Houlbrèque et al., 2003) and both phases were frozen, lyophilized, ground and encapsulated in tin cups for ¹⁵N analysis. The GF/F filters concentrating planktonic samples and coral samples were analysed using an elemental analyser coupled to an isotopic ratio mass spectrometer (EA-IRMS, Integra CN, SerCon Ltd, Cheshire, UK).

In experiments 2 and 3, the concentration of prey cells was quantified at the start and end of the incubation period to check for their ingestion by the corals. The abundance of Cyanothece cells was determined by epifluorescence microscopy, while the abundance of picoplankton was determined with a FACSCalibur flow cytometer (BD Biosciences, Franklin Lakes, NJ, USA) as previously described (Gasol and del Giorgio, 2000).

Significant differences in the number of Cyanothece cells between the start and the end of the incubation period were checked by the non-parametric Mann–Whitney test. We also checked for differences in the abundance of Cyanothece between containers containing coral branches (N=5) and control containers without Cyanothece cells (N=3). To improve the robustness of the statistical analysis, the number of cells in each control and coral sample was counted 20 times both at the start and at the end of the incubation. Statistical significance was accepted at P<0.05.

While scleractinian corals live in close association with diverse diazotrophic communities (Grover et al., 2014; Lema et al., 2012), to the best of our knowledge, this is the first study using the new dissolved ¹⁵N₂ method (Großkopf et al., 2012; Mohr et al., 2010) to measure N₂ fixation rates in corals (experiment 1), and to quantify the amount of nitrogen incorporated through the ingestion of planktonic diazotrophs (experiments 2 and 3). Over the three experiments, significant ¹⁵N enrichments were found in the zooxanthellae (Fig. 1) but never in the tissue, for which results were below the detection limit (i.e. less than three times the standard deviation of the coral tissue natural ¹⁵N abundance).

For coral incubations in ¹⁵N₂-enriched seawater (experiment 1), we measured a mean (±s.d.) nitrogen assimilation of 0.33±0.14 µg N cm⁻² h⁻¹ in zooxanthellae (Fig. 1). These results confirm the presence of diazotroph communities within corals (Lema et al., 2012, 2014; Olson and Lesser, 2013; Olson et al., 2009), and reinforce previous reports that use the acetylene reduction method (Lesser et al., 2007; Shashar et al., 1994; Williams et al., 1987). Zooxanthellae loss due to thermal stress may decrease nitrogen acquisition capacity in corals (Ferrier-Pagès et al., 2010). In such bleaching events, coral resilience can be enhanced by alternative fixed nitrogen sources such as that provided by endosymbiont diazotrophs (Cardini et al., 2014, 2016).

In a similar experiment, Grover et al. (2014) did not find significant ¹⁵N enrichment either in the zooxanthellae or in the tissues of Red Sea corals after 24 h of incubation. They reported ¹⁵N enrichment ranging between 0.0002% and 0.0015% (considerably lower than those measured here in experiment 1: 0.012–0.219%; data not shown). Their lack of significant enrichment might be due to (i) the use of the ¹⁵N₂ bubble method (Montoya et al., 1996), which may significantly underestimate N₂ fixation rates (Mohr et al., 2010; Großkopf et al., 2012), (ii) longer incubation periods than ours (24 versus 4 h), leading to increased oxygen concentrations in the incubation containers and inhibiting N₂ fixation (oxygen deactivates the nitrogenase enzyme; Postgate, 1982) or (iii) Red Sea corals harbour fewer or less active diazotrophs than those of New Caledonia. Our results probably indicate a significant heterogeneity in the ability of the coral versus the zooxanthellae to assimilate nitrogen. While both bear the enzymatic machinery to assimilate ammonium (the primary by-product of N₂ fixation), zooxanthellae account for most of the uptake of dissolved inorganic nitrogen from the environment and can fix 14–23 times more nitrogen than the host (Pernice et al., 2012).

When corals were incubated with Cyanothece cells (experiment 2), the average nitrogen assimilation rate in zooxanthellae was 0.04±0.01 µg N cm⁻² h⁻¹ (Fig. 1), revealing for the first time that one scleractinian coral feeds on planktonic diazotrophs. The consumption of Cyanothece was confirmed by a significant decrease in their concentration while corals were present (1.27×10⁴ to 1.04×10⁴ cells ml⁻¹, N=5, P=0.02; Fig. 2). No significant changes in Cyanothece concentration were measured over the same incubation period in control containers without corals (N=3, P=0.70). Considering that corals feed half the time, our nitrogen assimilation rates (0.48 µg N cm⁻² day⁻¹) represent a third of the daily nitrogen input associated with the ingestion of non-diazotrophic picoplankton and nanoplankton estimated for the same coral species (1.4 µg N cm⁻² day⁻¹; Houlbrèque et al., 2004). The lack of significant ¹⁵N enrichment within the tissue in experiment 2 is consistent with the preferential transfer of prey nitrogen to the zooxanthellae (Piniak and Lipschultz, 2004). The high ¹⁵N enrichment in zooxanthellae measured in our short incubation period (4 h) might be due to direct uptake of prey nitrogen digested within the coelenteron, probably as ammonium, rather than host assimilation. However, other studies have observed complete prey digestion after even shorter incubations (e.g. 3 h in Galaxea fascicularis; Hii et al., 2009), suggesting that at least part of the Cyanothece cells could have already been digested after 4 h when nitrogen recycling (i.e. synthesis into host macromolecules, catabolism, excretion and uptake via zooxanthellae) should have started. It is likely that digestion had only begun when our incubations were stopped, impeding detectable ¹⁵N enrichment in the tissue.

Nitrogen assimilation rates in zooxanthellae measured when corals were fed with natural plankton (0.76±0.15 µg N cm⁻² h⁻¹, experiment 3; Fig. 1) were ∼20-fold higher than those measured when corals were fed with Cyanothece (experiment 2). Despite the fact that Cyanothece is a common diazotroph in the lagoon (Turk-Kubo et al., 2015), and hence is likely to be present in the planktonic assemblage used in experiment 3, we did not perform quantitative analyses to infer their abundance from the number of nifH gene copies. During the experiment, the concentration of all plankton groups decreased (from 0.11×10⁴ to 0.04×10⁴ cells ml⁻¹, 1.88×10⁴ to 1.22×10⁴ cells ml⁻¹ and 0.20×10⁴ to 0.08×10⁴ cells ml⁻¹, respectively, for Synechococcus, Prochlorococcus and picoeukaryotes; Fig. 3), except for bacteria, being statistically significant for Prochlorococcus and picoeukaryotes (P<0.05). In terms of the number of prey ingested (normalized to surface and polyp considering 50 polyps cm⁻² for S. pistillata), Cyanothece (experiment 2) were quantitatively the major group [19.7(±6.31)×10⁴ cells cm⁻² h⁻¹ or 3.93(±1.26)×10³ cells polyp⁻¹ h⁻¹], followed by picoeukaryotes [7.37(±2.86)×10⁴ cells cm⁻² h⁻¹ or 1.47(±0.57)×10³ cells polyp⁻¹ h⁻¹] and Prochlorococcus [4.29(±1.54)×10⁴ cells cm⁻² h⁻¹ or 0.86(±0.31)×10³ cells polyp⁻¹ h⁻¹].

The rates of ingestion for Cyanothece are in the same range as those obtained for other cyanobacteria on the same coral species (0.3×10⁴ cells polyp⁻¹ h⁻¹; Houlbrèque et al., 2004), but natural plankton nitrogen assimilation rates were ∼20-fold higher. Under natural conditions, corals extend their tentacles and swell their coenosarc to prepare for feeding as a result of their ability to sense the diffuse cloud of metabolites that surrounds plankton swarms (Hellebust, 1965). The release of dissolved organic compounds by Cyanothece is known to be negligible (Benavides et al., 2013a; Berthelot et al., 2015), and thus may not provide the metabolite cloud needed to chemically trigger feeding in S. pistillata. Under this scenario, in experiment 2, Cyanothece might have been only ‘mistakenly’ trapped by adhesion on polyps and mucus. Conversely, interactions between the different plankton groups included in the natural planktonic assemblage of experiment 3 may provide a greater release of dissolved organic compounds (estimated to represent ∼20% of primary production; Marañón et al., 2004), providing a chemical signal similar to the one experienced by corals in situ and favouring the ingestion of diazotrophs. While the significant decrease of Cyanothece cells during experiment 2 renders their ingestion by the coral unequivocal, the ¹⁵N enrichment measured in experiment 3 could be due to different processes: (i) direct uptake of diazotrophs digested within the coelenteron (as in experiment 2), (ii) uptake of ¹⁵N-enriched dissolved nitrogen compounds fixed by the diazotrophs and released extracellularly (Benavides et al., 2013b; also as in experiment 2) and/or (iii) ingestion of non-diazotrophic plankton enriched in ¹⁵N as a result of diazotroph-derived nitrogen transfer (Bonnet et al., 2016).

In experiments 2 and 3, planktonic cells were concentrated to monitor as quickly as possible a significant change in cell concentrations due to coral ingestion without enclosing corals for too long. While the increased concentration of cells could enhance prey ingestion rates (Anthony, 1999; Ribes et al., 2003), the concentrations used here are in the range of those usually found in the lagoon (Conan et al., 2008; Jacquet et al., 2006; Turk-Kubo et al., 2015). Moreover, the use of short incubations also minimized possible changes in the planktonic community (death due to enclosure or grazing of picoplankton by nanoplankton).

We show evidence of ¹⁵N enrichment in zooxanthellae by direct N₂ fixation and verify the use of widespread diazotrophic cells like Cyanothece in coral lagoons by one major coral species: S. pistillata. We estimate that the uptake of natural plankton (<100 µm) including diazotrophic cells would bring 0.76±0.15 µg N cm⁻² h⁻¹, which is six times the daily nitrogen input calculated for the picoplankton and nanoplankton in Houlbrèque and Ferrier-Pagès (2009). The high abundance and activity of diazotrophs in the New Caledonian lagoon suggests that these cells represent a non-negligible source of nitrogen for corals. Additional experiments are needed to decipher through which pathways and mechanisms planktonic diazotrophs transfer nitrogen to corals.

We wish to thank R. Farman and the technical staff of the Aquarium des Lagons (Noumea) for their welcome and assistance in tank maintenance. We would also like to thank R. Rodolfo-Metalpa for providing the coral branches used in this study. Finally, we are grateful to M. Caffin and the Regional Flow Cytometry Platform for Microbiology (PRECYM) of the Mediterranean Institute of Oceanography (MIO) for support with flow cytometry analysis.