Nutrient depletion and heat stress impair the assimilation of nitrogen compounds in a scleractinian coral

Nitrogen is an essential nutrient for marine life because it is a key component of proteins, amino acids and nucleic acids, which are essential for the growth and functioning of all organisms (Nelson and Cox, 2017). It therefore plays a vital role in supporting the productivity of marine ecosystems (Bristow et al., 2017). However, nitrogen is often a limiting nutrient in many oceans, especially in coral reef ecosystems, which can be regions of extremely low nutrient availability (Sheppard et al., 2017) and are often referred to as ‘marine deserts’.

Despite this nutrient limitation, coral reefs form the base of one of the most productive ecosystems in the world (Roberts et al., 2002). This success is due to the symbiosis between several animal taxa (such as corals, giant clams and jellyfish) and endosymbiotic dinoflagellate algae of the family Symbiodiniaceae, which, with other microorganisms such as bacteria, form a meta-organism called a holobiont (Rohwer et al., 2002; Knowlton and Rohwer, 2003). The endosymbiotic dinoflagellates photosynthesise and provide the host with the majority of its carbon requirements (photosynthates; Muscatine et al., 1997) and receive respiratory carbon dioxide and nitrogenous waste, in the form of ammonium, in exchange (Lipschultz and Cook, 2002; Wang and Douglas, 1998). The holobiont can also actively scavenge the small amounts of inorganic (ammonium and nitrate) or organic (urea and amino acids) nitrogen dissolved in reef water (Grover et al., 2002, 2003, 2006, 2008).

The availability and form of nitrogen in reef water can directly influence the growth and productivity of all symbiotic organisms on reefs, including scleractinian corals, which are the main reef builders. Nitrogen limitation can reduce coral growth and increase vulnerability to stressors such as heat stress (Ezzat et al., 2019), which triggers coral bleaching, resulting in the loss of symbionts from coral tissues (Oakley and Davy, 2018). On the contrary, nitrogen enrichment has different effects depending on the form of nitrogen and whether phosphorus is available in seawater: whereas ammonium and urea have beneficial effects on holobiont physiology, bleaching and thermal tolerance (Béraud et al., 2013; Biscéré et al., 2018; Chase et al., 2018; Fernandes de Barros Marangoni et al., 2020; Roberty et al., 2020), excessive nitrate levels can have detrimental effects on corals and also induce bleaching (Nordemar et al., 2003; Wiedenmann et al., 2013; Fernandes de Barros Marangoni et al., 2020). The effects of nitrate are further exacerbated under low levels of dissolved inorganic phosphorus (DIP), as symbionts are replenished in nitrogen but starved for phosphorus. Under these conditions, phosphorus starvation decreases photosynthetic efficiency (Wiedenmann et al., 2013) and leads to a higher susceptibility to bleaching (Rosset et al., 2017). More broadly, maintaining a balance of carbon, phosphorus and nitrogen acquisition within the coral holobiont is essential for its health, as cycles of nutrients are all linked together. Indeed, under balanced conditions, photosynthate translocation from the symbionts to the host increases the host's assimilation of waste ammonium into amino acids (Rees, 1987; Rees and Ellard, 1989; Wang and Douglas, 1998). This process restricts symbiont growth by reducing the availability of ammonium to the symbionts and serves as a mechanism for the host to control symbiont density (Wang and Douglas, 1998; Xiang et al., 2020). However, excessive nitrogen availability or heat stress events can disrupt this nutrient recycling: the coral loses its primary source of organic carbon and the ability to recycle nitrogen, which ultimately leads to coral death (Oakley and Davy, 2018; Morris et al., 2019).

Although nitrogen acquisition and regulation are vital to coral physiology, and several studies (as mentioned above) have assessed the effects of nitrogen enrichment on corals, limited research has focused on the capacity of the coral holobiont, both under normal growth conditions and during heat stress, to use the different forms of nitrogen available in seawater. Most studies have primarily investigated the uptake of dissolved inorganic nitrogen such as ammonium and nitrate (Godinot et al., 2011a,b; den Haan et al., 2016; Rädecker et al., 2021), whereas only a few studies have investigated the uptake rate of other forms of dissolved organic nitrogen, such as urea (Grover et al., 2006) and dissolved free amino acids (DFAAs; Grover et al., 2008). However, concentrations of dissolved organic nitrogen are usually much higher in reef waters than inorganic forms (Miyajima et al., 2007; Tanaka et al., 2011), and can represent an important source of nitrogen for corals. The availability of this organic source can even increase during thermal stress, owing to the higher release of dissolved organic matter by photoautotrophs (Niggl et al., 2009; Thornton, 2014). Another poorly investigated aspect of the nitrogen cycle in corals is the allocation of nitrogen in its different available forms between the symbionts and host during a heat stress event (Godinot et al., 2011a; Rädecker et al., 2021). As stated above, the internal nitrogen balance within the host and symbionts can determine symbiont density and the translocation of carbon to the host (Dubinsky et al., 1990; Dubinsky and Jokiel, 1994), with implications for coral bleaching. Finally, the effect of nitrogen/nutrient depletion on the capacity of coral holobionts to rapidly take up newly available nitrogen from seawater remains unexplored. However, this process is of great significance in phytoplanktonic species, which have demonstrated the ability to promptly increase their nitrogen uptake rate following a period of nitrogen starvation (Krasikov et al., 2012; Berthold and Schumann, 2020). As increased water temperatures can lead to increased stratification and nutrient depletion in the open ocean (Sarmiento et al., 2004; Moore et al., 2013), coral reefs located close to atolls may experience both heat stress and nutrient depletion concurrently (Wiedenmann et al., 2023).

The present study aimed to examine several under-studied aspects of the nitrogen cycle in corals. We first assessed, using stable isotopes, the assimilation and allocation of dissolved inorganic carbon and different forms of nitrogen within the coral holobiont, under different temperature regimes (normal and heat stress). We hypothesised that heat stress would cause decreases in carbon assimilation by the symbiont and subsequent decreased photosynthate translocation to the host owing to impaired photosynthesis and symbiont expulsion. As symbiont photosynthesis drives the assimilation of inorganic nitrogen (Grover et al., 2002, 2003), the assimilation rates would therefore decrease during heat stress owing to bleaching. Conversely, the host would increase its assimilation of organic nitrogen from the surrounding seawater to compensate for the reduced amount of nitrogen assimilated by the symbionts. We then focused on the combined effects of temperature and short-term nutrient depletion on the capacity of corals to take up nitrogen again when available. We hypothesised that nutrient-depleted conditions would enhance nitrogen assimilation, as seen in phytoplankton, except during thermal stress. This study provides new perspectives on the importance of nitrogen for the future of coral reefs.

A total of 180 nubbins of the same approximate size (3–4 cm long) were generated by cutting terminal portions of the branches of five colonies (36 nubbins per colony) of the coral Stylophora pistillata Esper 1797 harbouring Symbiodinium A1 (Ezzat et al., 2017; Martinez et al., 2022a), originating from the Gulf of Aqaba (Red Sea, Jordan). The nubbins were attached to labelled nylon threads and distributed randomly in nine independent flow-through tanks (four nubbins per colony in each aquarium). The tanks were supplied with oligotrophic Mediterranean seawater [0.5 µmol l⁻¹ dissolved inorganic nitrogen (DIN), 0.2 µmol l⁻¹ DIP and 0.26 µmol l⁻¹ N-urea] pumped from a depth of 50 m in front of the Centre Scientifique de Monaco (flow rate of 10 l h⁻¹) and filtered through sand filters. Metal halide lamps (Philips, HPIT 400W, Distrilamp, Bossee, France) mounted above the aquaria provided an irradiance of 200±10 µmol photons m⁻² s⁻¹ on a photoperiod of 12 h:12 h light:dark. The light intensity was measured using a ULM-500 data logger (Heinz Walz GmbH, Germany) with a quantum PAR sensor. Submersible resistance heaters (Visi-ThermH Deluxe, Aquarium Systems, Sarrebourg, France) connected to Elliwell PC 902/T controllers kept the water temperature constant at 26±0.5°C, while submersible pumps mixed the water. Salinity was measured using a conductivity metre (Meter LF196) and irradiance with a 4π quantum sensor (Li-Cor, LI-193SA). The tanks were cleaned weekly to prevent algal proliferation and the nubbins were fed twice a week with Artemia salina nauplii during the healing period. Feeding ceased 3 weeks before the experiments to avoid interference with nitrogen uptake (Grover et al., 2002).

After the healing period, the first step of the experiment consisted of dividing the tanks into three different temperature treatments of 26°C (control temperature), 30°C and 34°C, with three tanks per treatment, and with the temperatures adjusted at a rate of 0.5°C day⁻¹ (Fig. 1). The temperature was then maintained for 2 weeks, except for the corals in the 34°C treatment, which displayed visible bleaching after just 4 days and were used for the experiments on day 5.

The second step of the experiment started at the end of the temperature treatments (5 days or 2 weeks). A first set of 26 nubbins from each thermal treatment was sampled for the measurements described below. They were termed ‘nutrient replete’ because they were maintained in the same seawater they were raised in (containing 0.5 µmol l⁻¹ DIN and 0.2 µmol l⁻¹ DIP).

Five nubbins were immediately used to determine their rates of photosynthesis and respiration. Then, 20 nubbins were used to assess the assimilation rates of dissolved nitrogen and carbon by incubating them with ¹⁵N-labelled compounds and ¹³C-labelled sodium bicarbonate, as described below. One nubbin per treatment (six in total) was used to determine the natural ¹³C and ¹⁵N abundance (isotope ratio control).

A second set of 26 nubbins sampled from each thermal treatment was first placed for 24 h in aquaria containing nutrient-depleted seawater. To deplete seawater of DIN and DIP, a separate tank of 100 l was prepared that contained two large healthy colonies of S. pistillata, and the water was recycled, without any input of new water, for 48 h before the large colonies were removed and the experimental nubbins were added. As healthy colonies of S. pistillata rapidly assimilate nitrogen and phosphorus (Grover et al., 2002, 2003; Godinot et al., 2009), the DIN in the nutrient-depleted tank was below 0.2 µmol l⁻¹ and DIP was not measurable using a nutrient analyser according to the protocols of Tréguer and Le Corre (1975). Using the same technique and coral species, Blanckaert et al. (2023) found that 24 h under nutrient-depleted conditions induced phosphorus deficiency in the coral, whereas 72 h impaired coral metabolism. Therefore, a timeframe of 24 h was chosen to examine the effects of nutrient depletion without the confounding effect of impaired metabolism. After the 24 h in nutrient-depleted seawater, the nubbins were processed as described above (five nubbins for photosynthesis/respiration, 20 for ¹⁵N and ¹³C uptake rates, and one for natural isotopic abundance). These will be termed ‘nutrient-depleted’ nubbins. The remaining nubbins were frozen and used for additional measurements not presented in this paper.

Nubbins for each temperature and treatment were placed individually in 250 ml beakers of 0.22 μm-filtered seawater (FSW) enriched with 0.3 mmol l^{−1 13}C-sodium bicarbonate and 1 µmol l⁻¹ of a ¹⁵N-labelled nitrogen source: five beakers (with nubbins from different colonies) with ¹⁵N-ammonium chloride, five with ¹⁵N-urea and five with ¹⁵N-DFAA (all 98% atom, Sigma-Aldrich). A further five nubbins per treatment were incubated with FSW enriched with just 0.3 mmol l^{−1 13}C-sodium bicarbonate and one nubbin per treatment was placed in a beaker with only FSW. The beakers were immersed in a water bath to maintain constant temperatures of 26°C, 30°C or 34°C, depending on the treatment, and each contained a magnetic stirrer to homogenise the medium. The nubbins were incubated for 5 h at a constant irradiance of 200 µmol photons m⁻² s⁻¹. At the end of the incubations, the nubbins were removed and rinsed with FSW for 15 min, frozen at −80°C for 1 h and then stored at −20°C until analysis.

Coral tissue was removed from the skeleton using a Water Pick with MilliQ water, and the resulting slurry was homogenised with a Potter tissue grinder. A 500 µl aliquot was then taken and frozen at −20°C for subsequent analysis of protein content. Next, the extracted slurry was centrifuged at 2000 g for 10 min at 4°C to separate the host and symbionts, and the symbionts were rinsed twice in MilliQ water to avoid tissue contamination. The tubes containing the different fractions were then frozen at −80°C and freeze-dried using a Heto lyophilizer (CT 60), also at −80°C.

Between 600 and 700 µg of the freeze-dried powders were added to tin-weighing capsules and analysed using an Integra II isotope ratio mass spectrometer (Sercon, UK) to determine the ¹⁵N/¹⁴N and ¹³C/¹²C isotopic ratios. The ratios measured for the ¹⁵N and ¹³C incubations were compared against the natural ratio measured in the unenriched coral and algal fractions to determine the percentage of ¹⁵N and ¹³C enrichment. This was then converted into uptake rates using the equation of Dugdale and Wilkerson (1986) and presented in Grover et al. (2002). However, as visible tissue sloughing was observed in the 34°C treatment, the uptake rates were normalised per mg of protein instead of surface area. The C:N ratio of the coral and algae was also calculated, by dividing the total mass of carbon by the total mass of nitrogen.

At the same time point as the incubations, five nubbins from the replete and the depleted tanks (one from each colony) were removed per temperature. The nubbins were placed individually in 60-ml plexiglass chambers in 0.45 µm FSW at the corresponding temperatures and stirred using a stir bar. The concentration of oxygen in the chambers was measured using an oxygen sensor (polymer optical fibre, PreSens) that was connected to an Oxy-4 channel fibre-optic oxygen metre (PreSens). The oxygen concentration was recorded until there was a change of 10% using the Oxy4v2-30fb software at both 200 µmol photons m⁻² s⁻¹ and in the dark to determine rates of net photosynthesis (P_n) and respiration (R), respectively. Gross photosynthesis (P_g) was calculated as the sum of the absolute values of P_n and R. The nubbins were then frozen at −20°C until subsequent analysis as described below.

The coral tissue was removed from the skeleton using a Water Pick and homogenised using a Dounce tissue grinder, and aliquots were taken for different measurements: 100 µl for symbiont density, 5 ml for chlorophyll content and 500 µl for protein content, which were frozen at −20°C. The symbiont density was measured using a Z1 Coulter Particle Counter (Beckman Coulter) in triplicate and averaged. The chlorophyll aliquot was first centrifuged at 8000 g for 10 min at 4°C to separate the coral and symbiont fractions. The symbiont pellet was then resuspended in 5 ml of pure acetone to extract chlorophyll and stored at 4°C in the dark for 24 h. The extract was further centrifuged at 11,000 g for 15 min at 4°C and the absorbance of the supernatant was measured at 630, 663 and 750 nm using a Safas UVmc² spectrophotometer. The equations of Jeffrey and Humphrey (1975) were then used to calculate the concentrations of chlorophylls a and c₂, which were summed to determine total chlorophyll. The protein aliquots from both the corals used for the physiological measurements and those used for the incubations were thawed, and host protein content was determined using the bicinchoninic acid assay (Smith et al., 1985). Finally, coral skeletons were used to determine the surface area by dipping the skeleton in paraffin wax at 65°C and determining the mass of wax on each fragment. These were then used to create a calibration curve of measured surface area against wax mass, and masses were converted to surface area for each coral, following Veal et al. (2010).

All results presented were normalised per mg total protein content, apart from the total protein content, which was normalised per cm² of surface area. The proportion of the total carbon and nitrogen assimilated by the symbiont was also calculated. The interquartile range (IQR) method was used to identify and exclude outliers that fell outside the range of quartile 1–2×IQR to quartile 3+2×IQR. The data were then analysed using mixed-effects models with predictor variables of temperature and nutrient condition, and random effects of tank and colony. The model quality (including normality, homogeneity of variance and collinearity) was assessed using the performance package in R (Lüdecke et al., 2021), and the data were log or reciprocal transformed when required. In cases where the variance of the tank or colony was 0, the factor was excluded and the model was simplified. When significant effects were detected (P<0.05), pairwise comparisons were conducted using estimated marginal means, and adjusted using the Tukey method to control for multiple comparisons. A regression was also conducted to determine the relationship between the total carbon and nitrogen assimilated by the holobiont in both the ammonium and DFAA treatments. Results are presented as means±s.e.m unless otherwise indicated.

Symbiont density decreased with temperature (n=27, F=71.8, P<0.0001; Fig. 2A) but the effect varied depending on nutrient condition (F=11.5, P<0.001). At 26°C, symbiont density in the nutrient-replete treatment was 14% higher than in the depleted treatment (P<0.05). However, at 30°C, symbiont density in the replete treatment decreased by 34% (P<0.0001) to become 23% lower than that in the nutrient-depleted condition at the same temperature (P<0.05); it then did not change between 30°C and 34°C. The contrary was observed in the nutrient-depleted treatment, with no significant change between 26°C and 30°C, but a decrease of 36% between 30°C and 34°C (P<0.0001), to reach a density similar to the nutrient-replete condition at 34°C.

Protein content also decreased with temperature (n=27, F=36.3, P<0.01; Fig. 2B), and again, the effect varied with nutrient treatment (F=4.71, P<0.05). Significant decreases were seen in the replete condition, when the protein content declined by 42% between 30°C and 34°C (P<0.05), and in the depleted condition, when it declined by 45% between 26°C and 30°C (P<0.05).

Total chlorophyll per mg protein was significantly affected by temperature (n=27, F=24.9, P<0.0001; Fig. 2C), with decreases between 30°C and 34°C of 39% (P<0.05) and 53% (P<0.0001) in the replete and depleted conditions, respectively. However, chlorophyll content increased at first in the depleted condition, by 44% between 26°C and 30°C (P<0.05).

Gross photosynthesis decreased with temperature (n=27, F=59.02, P<0.0001; Fig. 2D) and was affected by nutrient treatment (F=5.84, P<0.05). However, there were no significant differences between the replete and depleted conditions at any temperature. In comparison, the impacts of thermal stress were much more marked, with gross photosynthesis collapsing to approximately zero in both the replete (P<0.001) and depleted treatments at 34°C (P<0.0001). Respiration rate increased with temperature (n=27, F=17.5, P<0.0001; Fig. 2E), but only in the nutrient-replete condition (F=3.83, P<0.05), where it increased by 64% between 26°C and 30°C (P<0.01). As a result of the increase in respiration and the decrease in gross photosynthesis, the P_g:R ratio decreased (n=28, F=200.0, P<0.0001; Fig. 2F) by 52% and 48% between 26°C and 30°C in the nutrient-replete and depleted treatments, respectively (P<0.0001), and decreased to zero or near-zero in both conditions at 34°C (P<0.0001).

Across all temperatures and treatments, and for both the host and symbiont fractions, the natural isotopic abundance of ¹³C was stable, with an average of 1.076±0.0003 atom % C. For the control and all of the incubations, the symbiosis collapsed at 34°C, with carbon assimilation decreasing in the symbionts and host by over 70% (P<0.001).

In the control, the assimilation of ¹³C decreased with temperature in the nutrient-replete treatment in both the symbionts (n=26, F=247.1, P<0.0001; Fig. 3A) and host (n=28, F=117.9, P<0.0001; Fig. 3E). It decreased in the symbionts by 23% (P<0.05) between 26°C and 30°C, whereas in the host there were no significant changes between 26°C and 30°C. Proportionally, the symbionts assimilated between 60.9% and 70.8% of the total carbon assimilated by the holobiont at all temperatures, and there was no difference due to temperature or nutrient condition (Fig. 4A).

For the corals incubated with ammonium, the assimilation of ¹³C decreased with temperature in both the symbionts (n=28, F=40.7, P<0.01; Fig. 3B) and host (n=27, F=70.1, P<0.0001; Fig. 3F). However, unlike in the control treatment, the effect of temperature on ¹³C assimilation varied with nutrient regime in both the symbiont (F=6.65, P<0.05) and host (F=7.92, P<0.01). In the symbionts, this was due to the replete treatment having a higher carbon assimilation at 26°C and a lower assimilation at 34°C than the depleted condition, although these differences were not significant. In the host, the assimilation in the nutrient-replete treatment decreased by 56% (P<0.001) between 26°C and 30°C, while assimilation in the depleted treatment remained the same. Out of the total carbon assimilated by the holobiont, the symbionts assimilated between 59.2% and 72.6% at all temperatures, and there was no effect of temperature or nutrient regime (Fig. 4B).

The assimilation by the corals incubated with urea decreased with temperature in both the symbionts (n=29, F=42.9, P<0.0001; Fig. 3C) and host (n=29, F=97.7, P<0.0001; Fig. 3G). In the symbionts, there was no significant change until the decrease at 34°C, whereas the assimilation by the host decreased by 34% between 26°C and 30°C in the nutrient-replete condition and did not change significantly in the depleted condition. In comparison, the symbionts assimilated between 61.8% and 73.7% of the total carbon assimilated by the holobiont, and this did not vary with the temperature or nutrient regime (Fig. 4C).

In the DFAA incubation, ¹³C assimilation also decreased with temperature in both the symbionts (n=29, F=136.6, P<0.0001; Fig. 3D) and host (n=30, F=84.0, P<0.0001; Fig. 3H), and the effect of temperature varied with the nutrient regime in the host (F=8.14, P<0.01) but not the symbionts. In the symbionts, there was again no significant change until the collapse at 34°C. However, in the host, the pattern differed between the nutrient-replete and depleted treatments, as the assimilation in the replete treatment decreased by 55% (P<0.001) between 26°C and 30°C, while assimilation in the depleted treatment remained the same. The proportion of the holobiont carbon that was assimilated by the symbionts was affected by both temperature (n=25, F=15.3, P<0.001) and nutrient regime (F=6.31, P<0.05), and there was an interaction between temperature and nutrient regime (F=9.39, P<0.01). This was because, in the nutrient-replete condition, the proportion assimilated by the symbionts increased from 58.6±3.14% to 72.6±2.71% between 26°C and 30°C (P<0.05), before decreasing to 50.2±2.72% at 34°C (P<0.001), whereas the proportional assimilation did not change with temperature in the nutrient-depleted condition and remained between 61.7% and 69.2% (Fig. 4D).

The natural abundance of ¹⁵N across all temperatures and treatments, and for both the host and symbiont, was also stable, at 0.37±0.0002 atom % N. Similar to the ¹³C results, the assimilation of ¹⁵N decreased significantly between 30°C and 34°C to almost zero in all conditions (P<0.0001), apart from in the host when incubated with DFAA.

The assimilation of ¹⁵N-ammonium by both the symbionts and host decreased with increasing temperature (symbionts: n=28, F=166.2, P<0.0001; host: n=27, F=68.5, P<0.0001; Fig. 5A,B), and was significantly higher in the symbionts in the nutrient-replete treatment than in the depleted treatment (n=28, F=6.51, P<0.05). This was mainly because the assimilation of ¹⁵N-ammonium by the symbiont in the nutrient-replete treatment at 26°C was 2.7-fold higher than that in the nutrient-depleted treatment (P<0.05), while the nutrient regimes were not different at the higher temperatures. The assimilation in the replete condition in both the symbiont and host decreased at 30°C, by 74% and 72%, respectively (P<0.01 for both), to reach the same level as seen in the nutrient-depleted condition (P<0.01). Out of the total ¹⁵N-ammonium assimilated by the holobiont, the proportion assimilated by the symbionts decreased significantly with temperature (N=28, F=187.3, P<0.001; Fig. 6A), decreasing between 30°C and 34°C from 60.8±2.19% to 23.5±2.19% in the nutrient-replete condition and from 56.8±2.43% to 25.9±2.43 (P<0.0001 for both) in the depleted condition.

In the urea incubation, the assimilation of ¹⁵N-urea by the symbionts was very low compared with that by the host or with the symbiont's assimilation of ¹⁵N-ammonium/¹⁵N-DFAA. Moreover, it decreased with temperature (n=30, F=76.0, P<0.0001; Fig. 5C) and was higher on average across temperatures in the nutrient-replete condition than in the depleted condition (F=9.66, P<0.01). The assimilation was 33% higher in the nutrient-replete than in the depleted treatment at 30°C (P<0.05), and then decreased at 34°C by 87% and 95% in the nutrient-replete (P<0.0001) and depleted conditions (P<0.001), respectively. In the host, there was also an effect of temperature on ¹⁵N-urea assimilation (n=29, F=221.2, P<0.0001; Fig. 5D), though unlike in the symbionts, assimilation first increased at 30°C before decreasing at 34°C. In the nutrient-replete treatment, assimilation increased by 4.1-fold at 30°C before decreasing by 95% at 34°C, whereas assimilation in the depleted treatment increased by 3.3-fold at 30°C before decreasing by 99% at 34°C (P<0.0001 for all comparisons). Assimilation was also higher in the nutrient-replete treatment than in the depleted treatment (F=61.5, P<0.0001) at every temperature (P<0.05), by an average of 2.4-fold. Relative to the total assimilation of ¹⁵N-urea by the holobiont, the proportion assimilated by the symbionts decreased with temperature (n=24, F=25.0, P<0.0001, Fig. 6B), with a decrease between 26°C and 30°C in the nutrient-replete treatment from 33.07±4.18% to 11.14±4.86% (P<0.05), and from 42.2±4.18 to 16.81±3.71% in the depleted treatment (P<0.01).

Assimilation of ¹⁵N-DFAA by the symbionts decreased with temperature (n=27, F=199.0, P<0.0001; Fig. 5E) and was also higher in the nutrient-replete treatment than in the depleted treatment (F=83.5, P<0.0001). The assimilation in the nutrient-replete treatment at 26°C was 3.5-fold higher than in that in the depleted treatment (P<0.0001), and then decreased at 30°C by 61% (P<0.001), which was still 2.2-fold higher than that in the depleted treatment at the same temperature (P<0.01). The assimilation of ¹⁵N-DFAA by the host was also affected by temperature (n=29, F=250.8, P<0.0001; Fig. 5F) and was higher in the nutrient-replete treatment than the depleted one (F=128.8, P<0.0001). The effect of temperature also varied with nutrient regime (F=10.5, P<0.001). This was because, in the nutrient-replete treatment, assimilation first decreased by 95% at 30°C, before increasing 4-fold at 34°C, whereas in the nutrient-depleted condition, the assimilation decreased by 88% at 30°C (P<0.0001 for all comparisons), but then did not change at 34°C. The assimilation in the nutrient-replete treatment was also 4-fold higher at 26°C and 3.9-fold higher at 34°C (P<0.0001 for both) than that in the depleted treatment. Out of the total ¹⁵N-DFAA assimilated by the holobiont, the proportion assimilated by the symbionts was affected by temperature (n=25, F=805.5, P<0.0001; Fig. 6C) and the effect varied with nutrient treatment (F=22.9, P<0.0001). The proportional assimilation of the symbiont followed a similar pattern in the nutrient-replete and depleted conditions, first increasing between 26°C and 30°C from 48.2±2.2% to 89.0±2.0% in the replete condition, and from 46.7±1.82% to 81.1%±1.82 in the depleted condition, before decreasing at 34°C to 18.7±1.99% and 31.0±1.97% (P<0.0001 for all comparisons) in the replete and depleted conditions, respectively. However, the increase at 30°C was larger in the replete treatment, causing it to be significantly higher than in the depleted treatment at 30°C (P<0.05), and the subsequent decrease at 34°C was greater, causing it to be significantly lower at 34°C (P<0.001).

As both the host and symbiont showed a similar pattern of nitrogen and carbon assimilation in the ammonium and DFAA incubations, these data were combined, and a regression analysis was conducted to determine the relationship between nitrogen and carbon assimilation. A significant positive relationship was found in both the nutrient-replete (n=29, F=6.75, R²_adj=0.51, P<0.0001; Fig. 7A) and depleted treatments (n=27, F=4.29, R²_adj=0.39, P<0.001; Fig. 7B). In the nutrient-replete treatment, 12.4 ng N was assimilated per 1000 ng C, whereas in the depleted treatment, only 5.2 ng N was assimilated per 1000 ng C. The effect of temperature and nutrient treatment on the C:N ratio of the host and symbiont was also examined. The C:N ratio of the symbiont was slightly higher than that of the host at 26°C in both the nutrient-replete and depleted conditions, with 5.00±0.06 and 4.88±0.06, respectively, compared with the host, with 4.65±0.07 and 4.52±0.06, respectively. However, the C:N ratio in the symbiont varied with temperature (n=119, F=3504.6, P<0.0001; Fig. 8A), first decreasing at 30°C in the nutrient-replete and depleted conditions to 4.31±0.05 and 4.24±0.05, respectively, before increasing at 34°C by 3.3-fold and 3.5-fold to reach 14.58±0.55 and 14.77±0.56, respectively (P<0.0001 for all comparisons). In contrast, the C:N ratio in the host did not change with temperature, remaining between 4.51±0.06 and 4.74±0.07 (Fig. 8B).

This study investigated the effect of heat stress and nutrient depletion on the assimilation of carbon and nitrogen by S. pistillata. We found that corals exposed to nutrient (nitrogen and phosphorus)-depleted waters for 24 h reduced their capacity to assimilate dissolved inorganic and organic nitrogen, likely because of a lack of energy and/or lack of other nutrients such as phosphorus. Heat stress also caused the assimilation of ammonium and DFAA to decrease, while slightly increasing the assimilation of urea, and ultimately causing the symbiont to become extremely nitrogen limited, as confirmed by the high C:N ratios. Because, in nature, heatwaves induce a stratification of surface waters in the open ocean and a marked decrease in the availability of dissolved nutrients (Moore et al., 2013), the present study suggests that some coral reefs, such as those located close to atolls, will suffer strong nutrient starvation in the future from the combined effects of thermal stress and nutrient depletion.

When kept under nutrient-replete (control) conditions, corals were able to assimilate all forms of dissolved nitrogen tested. The holobiont (host and symbionts together) assimilated ammonium and DFAA to a similar extent when supplied in seawater at the same concentration (1 µmol l⁻¹). Although ammonium and DFAA were taken up rapidly, consistent with previous observations (Grover et al., 2002, 2008), urea was assimilated at a lower rate. This is because urea cannot be directly metabolized but needs to be degraded first to ammonia (NH₃) and bicarbonate through the action of urease, and subsequently protonated to become ammonium (NH₄⁺). Ammonium resulting from this pathway remains mainly in the animal tissues compared with ammonium that has been directly acquired (Grover et al., 2006), perhaps because the ammonia produced through urease is used to neutralise protons created during calcification to form ammonium, before being reincorporated into urea via the ornithine cycle (Crossland and Barnes, 1974). In addition, urea uptake can be inhibited by relatively low ammonium concentrations in seawater (Molloy and Syrett, 1988). These observations have important ecological implications because, on the reef, corals will be unable to fully exploit the urea excreted by fish schools (Walsh et al., 2001; Crandall and Teece, 2012) unless it is first recycled into ammonium by the microorganisms present in the seawater or sediment.

Pre-incubation in nutrient-depleted water for 24 h impaired the assimilation of nitrogen by the holobiont. Indeed, when the nutrient-depleted corals were re-incubated with 1 µmol l⁻¹ nitrogen (in the various forms) added to seawater, we observed a significant decrease in ammonium and DFAA uptake rates by the holobiont and in urea uptake rates by the host. This decrease in uptake rates is surprising, as we would have expected the opposite, as seen in phytoplankton cells, which increase their nitrogen uptake during nutrient pulses (Berthold and Schumann, 2020). Because nitrogen uptake is closely linked to the presence of other nutrients such as carbon and phosphorus or vice versa (see discussion below; Fitzsimons et al., 2020; Frost et al., 2023), a lack of other nutrients in the depleted seawater may have caused the observed decrease in nitrogen assimilation. For example, corals have a high demand for phosphorus (Ezzat et al., 2016; Rosset et al., 2015) as it is a key component of adenosine triphosphate (ATP) and is used in the synthesis of phospholipid membranes (Nelson and Cox, 2017). In the nutrient-depleted condition, the lack of phosphorus likely decreased the availability of ATP, in turn decreasing the energy available for nitrogen assimilation. Coastal coral reefs can be exposed to elevated levels of DIN, which can result in phytoplankton blooms that deplete the reef waters in DIP (D'Angelo and Wiedenmann, 2014), whereas coral reefs in the open ocean, such as atolls, can experience nutrient-depleted conditions during heatwaves, as high seawater temperatures increase stratification, inhibit vertical mixing and decrease nutrient supply to surface waters (Moore et al., 2013). Therefore, corals can rapidly become limited in nitrogen and phosphorus, which is known to enhance bleaching (Béraud et al., 2013; Rosset et al., 2017; Han et al., 2022). In turn, nitrogen deficiency can cause reduced photosynthetic capacity, and perhaps increased consumption of carbon (e.g. glucose) by the symbionts, so reducing the potential for translocation of photosynthetic products to the host (Li et al., 2021) and further compromising the symbiosis (Courtial et al., 2018). Regions that experience stratification and heat stress simultaneously are thus likely to be threatened by more frequent or intense bleaching events, unless corals can compensate, potentially through increased heterotrophy to acquire additional nitrogen (Grottoli et al., 2006; Seemann et al., 2013), though heterotrophic nutrient assimilation can also be impaired by nutrient depletion (Ezzat et al., 2019). Moreover, recent evidence suggests that corals may have the potential to compensate through digestion of their symbionts, though this strategy would not be sustainable in the longer-term (Wiedenmann et al., 2023).

As observed in phytoplankton and mentioned above, carbon and nitrogen assimilation were correlated in S. pistillata (when urea was excluded from the analysis) with different physiological implications depending on whether corals were maintained in nutrient-replete or depleted conditions. First, in the nutrient-replete condition, ca. 12.4 ng of nitrogen was assimilated per 1000 ng of carbon. Nitrogen assimilation from ammonium and DFAAs thus amounted to approximately 1% of carbon assimilation, which is in the range measured for plants (Shaw and Cheung, 2018). The fact that nitrogen assimilation is linked to carbon assimilation is not surprising. Ammonium is assimilated by the symbionts using the glutamate synthetase/glutamine oxoglutarate aminotransferase (GS/GOGAT) cycle (D'Elia et al., 1983; Roberts et al., 1999, 2001) and by the host using GS and glutamate dehydrogenase, with GS employing ATP to catalyse the amination of glutamate to produce glutamine (Miller and Yellowlees, 1989; Wang and Douglas, 1998; Pernice et al., 2012; Cui et al., 2019). It has been proposed that organic carbon enhances ammonium assimilation by providing energy and carbon backbones for amino acid synthesis (Rees, 1987; McAuley, 1995; Cui et al., 2019), and increasing the activity of GS (Wang and Douglas, 1998). This explains why light, and the resulting increased availability of photosynthetically fixed carbon, increases the assimilation of ammonium (Ezzat et al., 2017). Similarly, light has been shown to enhance the assimilation of amino acids, with photosynthetic products promoting the incorporation of nitrogen into proteins (Al-Moghrabi et al., 1993). This nutrient-replete condition thus corresponds with the normal carbon and nitrogen cycles observed in corals. On the contrary, in the nutrient-depleted condition, carbon was assimilated at the same rate as in the nutrient-replete condition, whereas half the amount of nitrogen was assimilated. This situation could potentially lead to an uncoupling of photosynthesis and growth (Berman-Frank and Dubinsky, 1999). Further research is required to understand the potential implications for coral reefs.

Heat stress significantly reduced ammonium and DFAA assimilation by both the symbionts and the coral host. This reduction can be explained by the parallel decrease in symbiont density, photosynthetic rate and carbon assimilation. Because photosynthates provide a source of energy and enhance metabolic rates and the incorporation of nitrogen into amino acids (Muscatine and D'Elia, 1978), the decreased carbon fixation would have reduced the ability of the symbionts to assimilate ammonium and DFAA, causing the decreases seen with temperature. The same mechanism can potentially explain the low assimilation of ammonium and DFAA in the host at high temperatures, as reduced carbon fixation also corresponds to decreased carbon translocation to the host (Hughes et al., 2010). A similar heat-induced decrease in ammonium and nitrate uptake rates was observed previously for several coral species (Godinot et al., 2011a; Ezzat et al., 2016), together with a decreased gene expression of GS in S. pistillata (Rädecker et al., 2021), suggesting that increasing temperature significantly impairs the ability of corals to take up inorganic nitrogen from the surrounding environment. In fact, the optimal uptake of nitrogen in S. pistillata is known to be at 29°C, which is also the optimal temperature for other physiological processes in this coral species (Godinot et al., 2011a). Although the proportion of ammonium assimilated by the symbionts remained the same between 26°C and 30°C, the proportion of DFAA assimilated almost doubled. Similarly, when the coral Oculina patagonica was incubated in the dark for 2 weeks, the majority of heterotrophically acquired amino acids were transferred to the symbionts (Martinez et al., 2022b). This may have been caused by the symbionts becoming more heterotrophic in response to stress, or by the host being unable to assimilate amino acids due to a lack of energy caused by the decreased carbon translocation.

Although the assimilation of all other nitrogen sources decreased with temperature, the assimilation of urea by the host increased at 30°C. Urea is assimilated primarily by the host (Grover et al., 2006), whereas for ammonium and DFAA, assimilation is driven by the symbionts' production of photosynthetic carbon. This explains why the assimilation of urea did not decrease with symbiont density. This could also indicate a possible compensatory response by the host, as it loses the ability to assimilate other nitrogen sources. Therefore, urea can become an important source of nitrogen during thermal stress. The carbon from urea has previously been found to be incorporated into the coral skeleton, indicating a role in calcification (Crossland and Barnes, 1974). In healthy corals, carbon from photosynthesis is used for calcification, whereas bleached corals incorporate much less of this carbon into their skeletons, likely owing to the decreased translocation from the symbionts to the host (Hughes et al., 2010). Therefore, heat-stressed corals may have an increased demand for carbon for calcification, which they could satisfy by increasing their assimilation of urea.

Although the total nitrogen assimilated by the symbionts from the surrounding seawater decreased by 66% at 30°C, the symbiont density and carbon assimilated by the symbionts only decreased by 34% and 18%, respectively. This indicates that it is not just the decrease in carbon assimilation or the decrease in symbiont density that impaired the assimilation of nitrogen. Several hypotheses can be put forward to explain this lack of a relationship between carbon and nitrogen assimilation during thermal stress. The first hypothesis follows the theory of Rädecker et al. (2021), who proposed that during the early stages of heat stress, the respiration rate of the coral host increases, resulting in increased energetic demand and the subsequent catabolism of amino acids for energy. This would result in increased ammonium production, which would be available to the symbionts, removing the nitrogen limitation that was impeding their growth. If the symbionts were nitrogen replete through internal nitrogen recycling, as per this theory, they would have a reduced demand for external nitrogen, which would explain why nitrogen assimilation decreased more than carbon assimilation in our study. This is supported by the decrease in the C:N ratio in the symbiont at 30°C, which suggests that the symbiont becomes more replete in nitrogen. Another hypothesis is that the nitrogen limitation of the symbionts increases during heat stress, as indicated by the over 3-fold increase in the C:N ratio at 34°C and by the lower assimilation of carbon observed at 30°C, because nitrogen deficiency has been shown to increase energy consumption and decrease energy production, growth and investment into photosynthate transport (Li et al., 2021), as observed in the present study. Nitrogen limitation can occur when the enzymes needed to transport and assimilate nitrogen have exceeded their thermal limit, as might be the case for S. pistillata, because the optimal temperature for physiological processes in this species is 29°C (Godinot et al., 2011a). The energy limitation of the host and symbionts, as suggested by the decreased carbon assimilation, may also reduce the energy investment into nitrogen assimilation, resulting in nitrogen limitation in both the host and symbionts. This would also explain why incubation with ammonium prior to heat stress can prevent coral bleaching (Fernandes de Barros Marangoni et al., 2020), as the symbionts would have greater stores of nitrogen that prevent them from becoming extremely nitrogen limited. When considered together, the above observations suggest that the symbionts become nitrogen limited during heat stress.

Corals are able to thrive in regions of low nutrient availability owing to their efficient processes of nitrogen assimilation and recycling. However, our results indicate that this key ability is severely impaired by heat stress events. Nutrient depletion, as can be seen in stratified waters during marine heatwaves, decreased the assimilation of all forms of nitrogen studied. Similarly, heat stress decreased the assimilation of ammonium and DFAA, and led to increased C:N ratios in the symbionts. Therefore, nitrogen limitation is of major concern for corals during heat stress events. Corals can potentially acquire nitrogen from increased heterotrophic feeding on plankton during heat stress (Anthony and Fabricius, 2000; Grottoli et al., 2006), though not all species have this ability (Ferrier-Pagès et al., 2010); there is also evidence that they could digest their own symbionts under nutrient-limited conditions (Wiedenmann et al., 2023). Our results highlight that a further compensatory mechanism warrants consideration, relating to the increase in the host's assimilation of urea observed at 30°C. As fish release urea (Walsh et al., 2001; Crandall and Teece, 2012), this suggests that a beneficial management technique may be to prohibit or restrict fishing on coral reefs during heat stress events, and so at least partially counter the reduced ability of the corals to assimilate other sources of nitrogen. However, much more work is needed to test this recommendation, as well as further elucidate the complex inter-partner fluxes that underlie our physiological observations.

The authors thank M. I. Marcus, A. C. A. Blanckaert and C. Rottier for their technical help in the laboratory, L. Woods for help with the statistical analysis, and D. Desgré and E. Tambutté for maintenance of the aquaria.