Diazotroph-Derived Nitrogen Assimilation Strategies Differ by Scleractinian Coral Species

Reef-building corals generally thrive in nutrient-poor tropical waters, where among other elements, nitrogen (N) availability often limits primary productivity. In addition to their close association with endosymbiotic dinoflagellates of the family Symbiodiniaceae, enabling an effective use and retention of dissolved inorganic nitrogen (DIN), scleractinian corals have developed strategies to acquire new N: (1) They can ingest N-rich sediment particles and preys (from picoplankton to macro-zooplankton) via heterotrophy, including diazotrophs [plankton fixing dinitrogen (N2) and releasing part of this nitrogen—Diazotroph-Derived N (DDN)—in seawater], a pathway called “heterotrophic nutrition on diazotrophs”; (2) Symbiotic diazotrophs located in the coral holobiont have the molecular machinery to fix N2, a pathway called “symbiotic N2 fixation”. Here we used the 15N2 isotopic labeling in a series of incubations to investigate the relative contribution of each of these DDN transfer pathways in three worldwide distributed coral species: Acropora muricata, Galaxea fascicularis, and Pocillopora damicornis. We show that N provision via “symbiotic N2 fixation” is negligible compared to that obtained via “heterotrophic nutrition on diazotrophs,” with DDN assimilation rates about a thousand times lower for P. damicornis and G. fascicularis, or assimilation rates via “symbiotic N2 fixation” almost nil for A. muricata. Through heterotrophic feeding on planktonic diazotrophs, only G. fascicularis and P. damicornis can successfully obtain N and fulfill a large part of their N requirements (DDN asimilation rates: 0.111 ± 0.056 and 0.517 ± 0.070 μg N cm–2 h–1 in their Symbiodiniaceae, respectively). Whereas this contribution is again negligible for A. muricata. They also largely consume the picoplankton that likely benefit from this DDN (Prochlorococcus and Synechococcus cells; respectively, 2.56 ± 1.57 104 and 2.70 ± 1.66 104 cell h–1 cm–2 for G. fascicularis; 3.02 ± 0.19 105 and 1.14 ± 0.79 104 cell h–1 cm–2 for P. damicornis). The present study confirms the different dependencies of the three tested species regarding heterotrophy, with P. damicornis and G. fascicularis appearing highly efficient at capturing plankton, while A. muricata, considered as mainly autotroph, does not rely on these food resources to meet its N and energy needs.


INTRODUCTION
Reef-building corals generally thrive in nutrient-poor tropical waters, where among others, nitrogen (N) availability often limits primary production (Howarth, 1988). The highly efficient uptake and recycling of nutrients by corals can explain this paradox (Wild et al., 2004;De Goeij et al., 2013). The close association between the coral animal host and its endosymbiotic dinoflagellate of the family Symbiodiniaceae (LaJeunesse et al., 2018) enables an effective use and retention of nutrients and photosynthates (Muscatine and Porter, 1977). Together with their algal symbionts, corals are associated with a variety of other microorganisms, including viruses, protists, Bacteria, Archaea and fungi, an assemblage termed the coral holobiont (Rohwer et al., 2002;Bourne et al., 2016;van Oppen and Blackall, 2019), which also participate in nutrient supply and recycling (e.g., Rosenberg et al., 2007;Rädecker et al., 2015). Among nutrients, N is an essential element of many fundamental molecules of living organisms such as nucleic acids (DNA, RNA) and amino acids (proteins, enzymes) and represents an essential basis for all organic activity. Scleractinian corals developed strategies to acquire new N via different pathways (Goreau et al., 1971;Rädecker et al., 2015). First, algal symbionts efficiently take up dissolved inorganic N (DIN) from surrounding waters and translocate it to coral tissues (Falkowski et al., 1993). Corals can also assimilate dissolved organic N compounds (DON), such as urea and amino acids from the environment (Muscatine, 1990;. Then, the animal host can ingest N-rich sediment particles and prey via heterotrophy (Houlbreque, 2004;. It is well known that corals feed on mesozooplankton (200 µm −2 cm; which they catch by using their tentacles (Sebens et al., 1996;Ferrier-Pagès et al., 2003;Houlbreque, 2004;Goldberg, 2018), but are also able to trap pico-and nanophytoplankton (0.2 µm-20 µm; Houlbreque, 2004;Tremblay et al., 2012;Goldberg, 2018;Sangmanee et al., 2020). Pico-and nanophytoplankton supplies a substantial proportion of the metabolic requirements of most scleractinian corals (Goldberg, 2018;Sangmanee et al., 2020) and represents a continuous available source of food in the reefs (Tremblay et al., 2012). Among plankton, diazotrophs (dinitrogen (N 2 )fixing prokaryotes), are particularly abundant in coral lagoon waters (Turk-Kubo et al., 2015;Messer et al., 2017;Saulia et al., 2020), and heterotrophic nutrition on diazotrophs represent a potential source of N for corals (Benavides et al., 2016;Meunier et al., 2019). This plankton fixes N 2 (Tilstra et al., 2018) transform it into a bioavailable N form (NH 4 + ), and releases part of the recently fixed N (Diazotroph-Derived N, DDN) in seawater, providing available N for the development of the planktonic food web Bonnet et al., 2016). Hence, when corals feed on plankton, it can ingest both planktonic diazotrophs, but also plankton that developed on DDN from planktonic diazotrophs (for simplicity, this pathway will be called hereafter "heterotrophic nutrition on diazotrophs"). Finally, symbiotic diazotrophs located in the coral tissue, coral mucus layer and skeleton, have the molecular machinery to fix N 2 (Lema et al., 2012;Bednarz et al., 2019) and this pathway has been evidenced as a N source in several corals species (Santos et al., 2014;Cardini et al., 2015;Tilstra et al., 2019). While recent works have shown that N acquisition through symbiotic N 2 fixation greatly varies in relation to environmental factors, species, but also corals' metabolic status (Rädecker et al., 2015;Bednarz et al., 2017;Benavides et al., 2017;Lesser et al., 2018Lesser et al., , 2019, heterotrophic nutrition on diazotrophs has only been demonstrated for one coral species, Stylophora pistillata in Benavides et al. (2016) and Meunier et al. (2019).
The general aim of the present study was to determine how DDN acquisition through symbiotic N 2 fixation and heterotrophic nutrition on planktonic diazotrophs vary according to coral species. We used the 15 N 2 isotopic labeling in a series of incubation experiments to investigate the relative contribution of each DDN transfer pathway in three coral species: Acropora muricata, Galaxea fascicularis and Pocillopora damicornis. Widely distributed around the world, these three species have a different heterotrophic capacity. While the genus Acropora has always been considered largely autotrophic (Conti-Jerpe et al., 2020), Galaxea and Pocillopora are conversely considered with higher heterotrophic capacities (Wijgerde et al., 2011;Toh et al., 2013;Conti-Jerpe et al., 2020;Lyndby et al., 2020).

Study Site and Coral Collection
The study was performed in March 2019 (austral summer). The three coral species tested (Acropora muricata, Galaxea fascicularis, and Pocillopora damicornis) differ in the size of their polyps. A total of 20 apexes (5 cm long) of each the coral species were collected between 3 and 5 m depth at the Crouy reef, New Caledonia (22 • 15,883 S; 166 • 19,658 E, sampling license issued by the "Province Sud, " Government of New Caledonia). Coral fragments were transported in individual zip-lock bags to the Aquarium Des Lagons (Nouméa) in a cooler containing freshly collected seawater. Fragments were allowed to recover prior to the start of the incubations, for 24 h in a 20 L aquarium supplied with filtered (porosity 40 µm) seawater, renewed at a rate of 16.5 l h −1 and mixed using a submersible pump (Aquarium system, microjet MC 320, Mentor, OH, United States). Seawater was maintained under a constant temperature of 26 • C and coral fragments received a constant irradiance of 120 ± 10 µmol photons m −2 .s −1 (photoperiod 12 h: 12 h light: dark) using two Aquablue plus neon bulbs (blue-white, 15.000 K, Giesemann, Germany).

Measurement of Symbiotic N 2 Fixation Rates: 15 N 2 Incubation Experiments in Filtered Seawater
We used the 15 N 2 -labeling technique (dissolved 15 N 2 method, Mohr et al., 2010;Grokopf et al., 2012) as described in Meunier et al. (2019), to quantify symbiotic N 2 fixation and the potential transfer of DDN to the Symbiodiniaceae fraction. Ten coral fragments from each species were individually incubated for 8 h (from noon to 8 pm, local time) in the dark, into 1 L glass bottles containing 15 N 2 -enriched 0.2 µm-filtered seawater continuously stirred with magnets. The 15 N 2 -enriched seawater was prepared as described in Meunier et al. (2019) and its 15 N isotopic enrichment was measured by MIMS (Membrane Inlet Mass Spectrometry; Kana et al., 1994) after the 8 h of incubation. All beakers were immersed in a water bath to maintain the temperature constant (26.1 ± 0.3 • C) throughout the experiment.

Measurement of Heterotrophic Nutrition on Diazotrophs Diazotrophy-Feeding Experiments
In parallel experiments, we quantified the assimilation of DDN via heterotrophic nutrition on diazotrophy in the same three coral species. As described in Meunier et al. (2019), we prelabeled with 15 N 2 , 5 sets of 4.3 L of seawater containing the natural plankton assemblage (collected at the same site as corals). After 24 h of incubation with 15 N 2 , this plankton was collected onto 0.4 µm polycarbonate filters, and vortexed in a 50 mL seawater aliquot to resuspend cells. Each concentrated 15 N 2enriched natural plankton aliquot was provided to 5 coral fragments per species, each of them being incubated for 8 h, in the dark in separated glass beakers (15 beakers in total + controls, see below), continuously stirred with magnets. The natural plankton assemblage was collected at the same site as corals, thus planktonic concentration in the beakers is comparable to natural plankton densities of the sampling site. Three additional control beakers (without corals) were also incubated to quantify pico-and nanoplankton abundance fluctuations in the absence of coral predation (internal grazing, natural cell growth and/or cell death) at the beginning and at the end of the incubations. Incubations were carried out in the dark in order to maximize heterotrophy as, on the reef, coral feeding is exacerbated at night (with polyps more deployed and prey more available) (reviewed by Houlbrèque and Ferrier-Pagès, 2009).
Triplicate aliquots of the 15 N-labeled plankton given to corals were also kept for further IRMS analyses to measure the 15 N 2 enrichment of plankton. Triplicate seawater samples were also collected from each beaker at the beginning and the end of the incubation period for further flow cytometry analyses. Samples were fixed with paraformaldehyde (final concentration of 0.1%) for 30 min at room temperature in the dark, then stored at -80 • C. Prochlorococcus sp. (0.6 µm), Synechococcus sp. (1-2 µm), picoeukaryotes and heterotrophic bacteria abundance were analysed with a FACSVerse Flow Cytometer (Becton Dickinson, CA, United States) as described in Meunier et al. (2019), to quantify the respective ingestion rates by each coral species. Heterotrophic bacteria were detected by diluting the medium 10fold with filtered seawater (0.2 µm) and staining with SYBR R Green I (DNA). For Prochlorococcus sp., Synechococcus sp., and picoeukaryotes abundances, quantification was done according to the level of red fluorescence corresponding to Chlorophyll a content, and orange fluorescence representative of pigment contents such as phycoerythrin (especially to discriminate between Synechococcus sp. and other pico-and nanoplankton) and also by size-sorting using 1 µm green calibration beads (Marie et al., 1997(Marie et al., , 1999.

Sample Preparation
At the end of the incubation, corals were rinsed six times with filtered seawater to remove cells potentially adhered to the coral surface (Houlbrèque et al., 2003), transferred to ziplock bags and stored at -20 • C until analysis. The coral tissue was removed from the skeleton using an air-pick and 20 mL of 0.45 µm filtered seawater and homogenized with a Potter tissue grinder (Rodolfo-Metalpa et al., 2010). Algal symbionts and tissue fractions were separated by centrifugation (Houlbrèque et al., 2003). The pellets containing algal symbionts were dried and ground. Coral tissue solutions were filtered (15 mL) on pre-combusted GF/F filters under low pressure. The filters and the algal symbionts crushed were encapsulated in tin cups for 15 N analyses using an elemental analyzer coupled to an isotopic ratio mass spectrometer (EA-IRMS, Integra CN, SerConLtd, Cheshire, United Kingdom). The coral tissue fraction was filtered onto GF/F filters that were also analyzed by EA-IRMS. 10 mL subsample was taken from each homogenate and processed for algal symbionts density and total chlorophyll concentrations (see below). N assimilation rates into coral tissue and algal symbionts were calculated as described in Benavides et al. (2016).

Ingestion Rates of Pico-and Nanoplankton
Ingestion rates were assessed by means of the clearance rate, according to previous studies on corals (Houlbrèque et al., 2004;Tremblay et al., 2012), and calculated using the equations of Ribes et al. (1998), which take into account the natural growth and death of the prey during incubations. Ingestion rates were expressed as the number of prey organisms ingested and normalized to the skeletal surface area (cm 2 ). Carbon (C) and N content of ingested preys were estimated using conversion factors from the available literature (summarized in Table 1).

Statistical Analyses
All tests were performed using R version 3.6.1 within RStudio (Version 1.1.456, 2018). Given the small sample size, nonparametric Kruskal-Wallis tests were first used to test for significant differences in algal symbionts densities, total chlorophyll content and picoplankton ingestion rates between species. When significant differences were found, Wilcoxon-Mann-Whitney tests were performed to test for pairwise differences between each species (Table 2). To look for potential differences in DDN assimilation rates by coral species, DDN transfer pathways and compartments, three-way ANOVAs including interactions were conducted. When the ANOVA determined a significant difference, a Tukey's honest significant difference test (HSD) was used to test for pairwise differences while taking into account the interactions between the different variables ( Table 3). Given the very large discrepancy in measured values of assimilation rates between the two DDN transfer pathways, a separated two-way ANOVA was used to test the effect of coral species, compartments and DDN assimilation rates on the "symbiotic N 2 fixation" pathway values alone ( Table 4).
The ggplot2 package (Wickham, 2008) was used to create the box plot figures. Throughout the manuscript, values given are expressed as mean ± SE. Statistical significance was accepted at P < 0.05.

Nitrogen Assimilation Through the Heterotrophic Nutrition on Diazotrophs
Very low DDN assimilation rates through heterotrophic nutrition on diazotrophs were measured within A. muricata tissue (0.028 ± 0.008 µg N cm −2 h −1 ) and no assimilation was detected in their algal symbionts. There was no significant difference between both compartments in A. muricata (twoway ANOVA, diff = -0.003, adj. p-value = 0.99, Table 4). G. fascicularis and P. damicornis DDN assimilation rates were significantly higher in the symbionts compared to the tissue (two-way ANOVA, diff = + 0.1, adj. p-value = 0.014 and diff = + 0.5, adj. p-value < 0.001, respectively, Table 4).

Ingestion Rates of Picoplankton
In the control beakers without coral colonies, cell concentrations across all plankton groups remained stable or decreased except for bacteria that increased slightly. Synechococcus and Prochlorococcus were the most abundant organisms in the incubation media at the beginning of the experiment (5.10 ± 0.39 10 4 and 8.37 ± 0.90 10 4 cell mL −1 , respectively) compared to other pico-and nanoplankton taxa (3.10 ± 0.48 10 3 cell mL −1 ). In the beakers containing coral colonies of P. damicornis and G. fascicularis, the populations of Prochlorococcus and Synechococcus decreased but not those of picoplankton and bacteria (Supplementary Figure 1). Prey increase in beakers containing A. muricata colonies, thus the calculated ingestion rates were negligible (Supplementary Figure 1). Hence, Figure 1C represents only the ingestion rates of Prochlorococcus and Synechococcus by G. fascicularis and P. damicornis. Prochlorococcus decreased by 72% in beakers containing P. damicornis and was the most significantly ingested prey (3.02 ± 0.19 10 5 cell h −1 cm −2 ; Mann-Whitney-Wilcoxon test; p = 0.01, Table 2). G. fascicularis actively and equally fed on Prochlorococcus and Synechococcus, with no significant differences between the ingestion rates (2.56 ± 1.57 10 4 and 2.70 ± 1.66 10 4 cell h −1 cm −2 ; Mann-Whitney-Wilcoxon test; p = 0.9, Table 2). Synechococcus decreased by 24 and 43% in beakers containing the coral colonies of P. damicornis and G. fascicularis, respectively. The corresponding calculated ingestion rates were 1.14 ± 0.79 10 4 and 2.70 ± 1.66 10 4 cell h −1 cm −2 , respectively.

DISCUSSION
By simultaneously quantifying the DDN assimilation rates in three coral species through heterotrophic nutrition on planktonic diazotrophs and symbiotic N 2 fixation, this study allows a quantification of the relative contribution of each of these pathways, as well as the distribution of the DDN between two coral compartments (tissue and symbionts). In both P. damicornis and G. fascicularis, N assimilation rates through heterotrophy on planktonic diazotrophs were found to be a thousand times higher than those obtained via symbiotic diazotrophs. The high assimilation rates obtained here for two species, as well as those previously obtained for S. pistillata using the same methodology (Benavides et al.,  Meunier et al., 2019), demonstrate for the first time that N assimilation originating from natural plankton fixing N 2 or plankton that developed on diazotrophy, seems to be widespread among scleractinian corals. The present study shows that corals with higher heterotrophic capacities compared to Acropora, are capable of directly consuming plankton that developed on N 2 fixation. Although we did not measure the grazing of diazotrophs per se, we can hypothesize that these planktonic organisms are both diazotrophs, that are abundant in the New Caledonian lagoon (Turk-Kubo et al., 2015;Bonnet et al., 2016), but also picoplankton that is well known to benefit from this DDN in the New Caledonian lagoon (Prochlorococcus and Synechococcus cells; Berthelot et al., 2016;Bonnet et al., 2016).

DDN Assimilation From Planktonic Diazotrophs
Colonies of A. muricata, considered to be mainly autotrophic, assimilated very little N through planktonic diazotrophs, whereas the two species G. fascicularis and P. damicornis greatly benefited from this N source by assimilating, respectively, two or four times more DDN through this pathway, compared to A. muricata. After 8 h of incubation in the dark, in G. fascicularis and P. damicornis, the totality of the assimilated DDN through planktonic diazotroph predation occurred within the algal symbionts compartment. DDN assimilation rates from planktonic diazotrophy measured in G. fascicularis (0.11 ± 0.05 µg N cm −2 h −1 ) and P. damicornis (0.517 ± 0.07 µg N cm −2 h −1 ) symbionts are in the same range as those previously measured using the same methodology within S. pistillata after 4 h of incubation also in the dark (0.76 ± 0.15 µg N cm −2 h −1 ) (Benavides et al., 2016). DDN assimilation rates in P. damicornis (belonging to the same family, i.e., Pocilloporidae, as S. pistillata) after 8 h of incubation are, however, slightly lower than those of S. pistillata. This suggests that either: (i) S. pistillata colonies exploit more planktonic diazotrophs as a source of N; (ii) a longer incubation (8 h instead of 4 h for S. pistillata) means that more N was transferred from the symbionts to the other coral compartments; or (iii) the planktonic community used at the time of the experiment (more or less abundant, with varying degree of N 2 fixation activity according to the incubation conditions) influence the DDN transfer to the corals. Regardless, the capacity of DDN assimilation through planktonic diazotrophs in corals from Pocilloporidae family (P. damicornis and S. pistillata) are twice the one of G. fascicularis.

DDN Contribution to Coral Nitrogen Daily Needs
N budgets within the coral holobiont remained so far poorly resolved (sensu Fiore et al., 2010;Pernice et al., 2012;Rädecker et al., 2015). Our results indicate for the first time that G. fascicularis and P. damicornis fulfilled a large part of their N requirements through heterotrophic feeding on planktonic diazotrophs or plankton which has benefited from diazotrophy. For example, in colonies of S. pistillata, the dissolved inorganic N uptake rates (NH 4 + and NO 3 − at in situ concentrations) are at least 2 ng N cm −2 h −1 (Grover et al., 2002(Grover et al., , 2003. For the plankton density tested, we estimate that the amount of N coming from planktonic diazotrophy for G. fascicularis and P. damicornis (2.51 and 3.44 ng N cm −2 h −1 , respectively) would therefore be almost equivalent to the N contribution through inorganic N uptake.

Picoplankton Ingestion Rates
Consistent with 15 N data, the higher heterotrophy degree of P. damicornis and G. fascicularis is confirmed by the plankton consumption rates. While no ingestion was observed for A. muricata, both G. fascicularis and P. damicornis showed uptake rates ranging from 0.06 to 3.0 10 5 cell h −1 cm −2 . Both species did not ingest bacteria or picoeukaryotes but consumed significant numbers of picocyanobacteria: Synechococcus sp. and Prochlorococcus sp. Our study, focusing only on picoplankton grazing is difficult to compare with previous work interested in pico-nanoplankton as a whole, but ingestion rates of Prochlorococcus and Synechococcus measured here in P. damicornis and G. fascicularis are well within the range of cyanobacteria concentrations ingested by S. pistillata, G. fascicularis, Montipora digitata, and Porites lutea (Houlbrèque et al., 2004;Tremblay et al., 2012;Sangmanee et al., 2020) in dark conditions also. According to Table 1, Prochlorococcus and Synechococcus ingestion supplied P. damicornis with 3.12 ng N cm −2 h −1 , which is 4 times higher than for G. fascicularis (0.78 ng N cm −2 h −1 ). While for P. damicornis, Prochlorococcus was quantitatively the major prey ingested, as already demonstrated in previous studies for S. pistillata (Benavides et al., 2016;Meunier et al., 2019), G. fascicularis consumed both Prochlorococcus and Synechococcus. McNally et al. (2017) had been the first to highlight a selective grazing on specific picoplankton taxa (Synechococcus) by Porites astreoides. These observed differences in coral species diet highlight that corals may have selective prey acquisition strategies. Sizes of Synechococcus and Prochlorococcus being comparable, the mechanisms at the origin of this selection remains to be elucidated.
tissue and algal symbionts, respectively. Even if we used the same technique as Bednarz et al. (2017), different 15 N 2 enrichments of enriched seawater (10 atom% vs. 30 atom% in our study) and different incubation times may have affected both the diazotrophic activity and the transfer of DDN into the different coral compartments, justifying the small differences observed between both studies. Although the assimilation rates through symbiotic N 2 fixation measured in our study are low, they are higher in the predominantly heterotrophic corals, G. fascicularis (Wijgerde et al., 2011;Conti-Jerpe et al., 2020) and P. damicornis (Lewis and Price, 1975;Sebens et al., 1996;Lyndby et al., 2020), compared to the predominantly autotrophic A. muricata (Conti-Jerpe et al., 2020).
These conclusions are the opposite of those outlined by Pogoreutz et al. (2017b), showing higher symbiotic N 2 fixation rates in autotrophic corals than in the heterotrophic ones. However Pogoreutz et al. (2017b) considered P. verrucosa and S. pistillata as autotrophic species while corals from Pocilloporidae family are considered as heterotroph in all studies on coral nutrition, showing in particular a group I feeding strategy (feeding by tentacle capture, Lewis and Price, 1975;Ferrier-Pagès et al., 2003;Houlbrèque et al., 2003Houlbrèque et al., , 2004Houlbrèque et al., , 2015Houlbreque, 2004;Lyndby et al., 2020). Results of Pogoreutz et al. (2017b) showed that physiological differences in symbiotic N 2 fixation rates among coral species tested align with relative nifH gene copy. So we can hypothesize that more than the functional group (autotrophic vs. heterotrophic), it is the abundance and the diazotrophic community associated with each species that seems to influence N uptake by symbiotic N 2 fixation. A. muricata and P. damicornis host contrasted communities in the Great Barrier Reef (GBR) (Lema et al., 2012), which might harbor different diazotrophic activity. Thus in our study, differences in the diazotroph community associated with those two species, may explain the observed differences in DDN assimilation via symbiotic N 2 fixation.
It should be noted that in the present study, N 2 fixation rates by symbiotic diazotrophs were obtained at night, and are therefore maybe slightly lower than rates measured during the day. Bednarz et al. (2017Bednarz et al. ( , 2018, as well as Tilstra et al. (2019), found that the ecological dependence of corals on diazotrophy might depend on depth, and so on light availability, with a decrease in relative nifH gene copy numbers with increasing depth. This may be linked to reduce photosynthate translocation by Symbiodiniaceae to the coral host, thus decreasing the fueling of the energy-expensive activity of coral-associated diazotrophs (Pogoreutz et al., 2017a). However, no study until now has measured day/night differences in N 2 fixation in corals. Like the rates measured during the day by Cardini et al. (2015), or during day/night cycles by Bednarz et al. (2017), N 2 fixation rates by symbiotic diazotrophs in our three tested species remain low and very negligible compared to what is contributed by diazotrophic plankton. P. damicornis colonies preferentially accumulate N in their symbionts than in their tissue, confirming that symbiotic diazotrophs have a close relationship with algal symbionts which rapidly benefit from the fixed DDN (Lesser et al., 2007;Olson et al., 2009;Pogoreutz et al., 2017b). Unlike P. damicornis, in G. fascicularis, a significant proportion of DDN coming from symbiotic diazotrophs is also found in the tissue compartment.

CONCLUSION
The present study investigated the patterns of DDN acquisition in three different coral species and allowed us, among other things, to identify whether these patterns changed according to their dependence on heterotrophy. P. damicornis and G. fascicularis appear to be highly efficient at capturing plankton (either diazotrophs and/or picoplankton that developed on diazotrophy), while A. muricata does not rely on these food resources to meet its N and energy needs. As climate change destabilizes symbiosis, thus blocking the main source of energy and nutrients, the heterotrophic ability of corals as an alternative nutrient resource plays an essential role in their survival (e.g., Sebens et al., 1996;Grottoli et al., 2006;Anthony et al., 2009;Tremblay et al., 2016;Conti-Jerpe et al., 2020;Fernandes de Barros Marangoni et al., 2020;Rädecker et al., 2021). From our results, we can expect that coral species such as P. damicornis and G. fascicularis, which are able to meet their N equally by planktonic diazotrophy and dissolved inorganic sources, will be able to maintain their energy reserves to cope with climate change. On the opposite, corals that are mainly autotrophic such as A. muricata may not adapt and survive, as heat stress events will become more frequent and intense. A previous study (Meunier et al., 2019) has demonstrated that bleached colonies of Stylophora pistillata were able to increase their feeding rate precisely on these planktonic diazotrophs but also on Synechococcus cyanobacteria, whose cells are rich in N (Bertilsson et al., 2003;Jacquet et al., 2006). So our study reveals that coral species that highly depend on heterotrophy could, in the event of bleaching, rely not only on the supply of nutrients and energy from meso-macroplankton but also on a major supply of N by planktonic diazotrophs and picoplankton. Taking advantage of the consumption of planktonic diazotrophs or picoplankton that have developed on diazotrophy could be one of the main strategies for coral recovery facing bleaching, as both the activity and geographical distribution of diazotrophs will likely increase with future rising sea surface temperature (Breitbarth et al., 2007;Levitan et al., 2007;Hutchins et al., 2017).

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/ Supplementary Material, further inquiries can be directed to the corresponding author/s.

AUTHOR CONTRIBUTIONS
VM, FH, and SB conceived the ideas and designed the methodology. VM and FH performed the experiments and led the writing of the manuscript. OG and CL carried out some of the laboratory analyses. VM and AR performed the statistical analyses. All authors contributed critically to the drafts and gave the final approval for publication.

FUNDING
VM was the beneficiary of a Ph.D. grant from LabEx-Corail (MACADAM project). This work was also funded by the EC2CO/BIOEFECT program (TOUCAN project).

ACKNOWLEDGMENTS
We thank the Race for Water Foundation for allowing us to carry out these experiments on board their vessel and all the crew members for their help. We are especially grateful to A. Tilstra and L. Montilla for critical reading and valuable comments on this manuscript.