Symbiont regulation in Stylophora pistillata during cold stress: an acclimation mechanism against oxidative stress and severe bleaching

Nutritional mutualistic symbioses are widespread in terrestrial and aquatic ecosystems and involve a diversity of symbiont taxa and animal hosts (Selosse et al., 2017). By having access to a mix of food sources, mutualistic associations can adapt to a wide range of environmental conditions. This is the case for reef-building corals, which grow in nutrient-poor tropical environments thanks to their symbiosis with Symbiodiniaceae dinoflagellates. This association is the basis of coral reef ecosystems, which are one of the most biodiverse and productive ecosystems on Earth (LaJeunesse et al., 2018). Despite their ecological importance, coral reefs are highly vulnerable marine ecosystems in a changing climate (Ainsworth et al., 2016), and coral bleaching – the breakdown of the symbiotic relationship between corals and their dinoflagellate algae – is a response of corals to stressful conditions (Weis, 2008). Widespread coral bleaching and mortality, leading to coral reef decline, have been associated with climate-change-driven increases in sea surface temperature (SST) (Hughes et al., 2018), which is particularly severe during episodes of the El Niño–Southern Oscillation (ENSO) (Claar et al., 2018). The prevailing mechanism for bleaching under these conditions invokes an overproduction of reactive oxygen species (ROS) by the endosymbionts and/or the coral host. Excessive production of ROS causes major cellular damage, such as membrane oxidation, protein denaturation and damage to the genetic material (Suggett and Smith, 2020). Host and symbionts exhibited adaptations for neutralizing ROS and preventing cellular damage, which include an arsenal of enzymatic and non-enzymatic antioxidant defenses. Bleaching occurs when the antioxidant capacity of the holobiont becomes overwhelmed by the overproduction of ROS under stressful conditions (reviewed in Suggett and Smith, 2020).

Coral bleaching and mortality events have also been related to decreases in SST. For example, corals of Florida reefs experienced, under the influence of the 2010 North Atlantic Oscillation, one of the coldest winters recorded, which caused widespread bleaching and mortality (Lirman et al., 2011). The same was observed for pocilloporid corals of the Galapagos reefs during a La Niña event in 2007 (Glynn et al., 2018). Such cold stress events are expected to increase in frequency owing to a changing climate (Kim et al., 2014; Glynn et al., 2018). Other local-scale physical processes such as upwellings, cold currents, wind-driven mixing and even tropical cyclones can reduce SST by several degrees in a few hours (Carrigan and Puotinen, 2014; Glynn and Stewart, 1973; Riegl et al., 2019; Green et al., 2019). Coral reefs are frequently exposed to these local episodes, especially to upwellings, as seen in the Great Barrier Reef (Andrews and Gentien, 1982), Central Pacific islands (Kelly et al., 2014), Pacific coast of Central America (D'Croz and O'Dea, 2007), Florida Keys (Leichter et al., 1996), Caribbean (Diaz-Pulido and Garzon-Ferreira, 2002) and Seychelles (Novozhilov et al., 1992). Although cooling events can be beneficial to corals during periods of increased SST by creating thermal refuges and reducing bleaching intensity (Glynn, 1996; Riegl and Piller, 2003; West and Salm, 2003; Riegl et al., 2019; Johnson et al., 2020; Sawall et al., 2020; Storlazzi et al., 2020), this is not always the case. For example, around the northern island of the Galapagos reefs, internal waves, forcing cool water upwards above the thermocline, led to a 12°C drop in temperature over a 6-day period, and induced coral bleaching and mortality (Banks et al., 2009; Glynn et al., 2017a,b). In addition to reduced SST, upwellings and wind-driven mixing are also associated with nutrient enrichment of surface waters by up to 15 µmol l⁻¹ dissolved inorganic nitrogen and 1.2 µmol l⁻¹ phosphorus (D'Croz and O'Dea, 2007; Stuhldreier et al., 2015; Radice et al., 2019; Riegl et al., 2019; Johnson et al., 2020). Importantly, despite the fact that nutrient enrichment and ratios play a major role in the oxidative metabolism and extent of bleaching that corals undergo during heat stress (Wiedenmann et al., 2013; Marangoni et al., 2020), the combination of nutrient enrichment and cold stress on the physiology of reef-building corals has been poorly studied (Johnson et al., 2020).

Cold stress can reduce coral growth rates (Glynn, 1977; Roth et al., 2012) and alter their feeding capacity (Palardy et al., 2005; Radice et al., 2019). Also, cold-stress-induced bleaching has several similarities to heat stress-induced bleaching with respect to Symbiodiniaceae photophysiology. In particular, it has been shown to be light dependent, and to induce symbiont loss after reducing photosystem II quenching and inducing photodamage (Saxby et al., 2003; Thornhill et al., 2008; Kemp et al., 2011; Roth et al., 2012). In addition, it has been reported that the heat-tolerant symbiont of the genus Durusdinium (formerly clade D) associated with Montastrea cavernosa is also cold tolerant (Silverstein et al., 2017). Although symbionts experienced photodamage, corals did not bleach, suggesting that symbiont loss can be decoupled from photochemical impairment. This later study put forward several hypotheses for the lack of bleaching occurrence, such as reduced production of ROS (McGinty et al., 2012), more effective antioxidant capacity of Durusdinium symbionts to detoxify ROS (Krueger et al., 2014), or a capacity, like some more parasitic symbionts, to repress host immune responses (Detournay et al., 2012).

Overall, because cold stress creates physiological symptoms in symbionts similar to those observed when they are heat stressed (see above), it has been suggested that biochemical mechanisms underpinning cold stress in corals may be related to an overproduction of ROS and consequent cell damage (Saxby et al., 2003; Hoegh-Guldberg et al., 2005). Bleaching has been shown to be even more intense when cold stress is combined with high irradiance, reinforcing the idea that oxidative stress might be involved (Saxby et al., 2003). However, up to now, this hypothesis had yet to be tested.

The effects of cold temperature stress (alone and combined with nutrient enrichment) on coral oxidative status need further investigation. Knowledge of how the variability of the physical environment can affect coral physiology, and ultimately coral reefs, is fundamental to understanding the future effects of global climate change on corals (Hoegh-Guldberg et al., 2005). Because upwelling or transient cold-water inputs on reefs have been suggested to create thermal refuges during the warm season, a better understanding of how short and long exposures to cold stress affect the physiology and biochemistry of reef-building corals is needed. Therefore, this study measured the physiological (photosynthesis and respiration rates, symbiont density, chlorophyll concentration, protein biomass and growth) and biochemical traits (ROS production, antioxidant capacity and lipid damage) of the coral Stylophora pistillata experimentally exposed to short- and long-term cold stress. We also combined short-term cold stress and nutrient enrichment (nitrate and phosphate) to mimic upwelling conditions. Results presented here further our knowledge on the biochemical mechanisms underpinning cold-stress-induced bleaching in reef-building corals. They show that, in the conditions tested here, the rapid expulsion of symbionts by the host during cooling conditions may be an acclimation mechanism to avoid oxidative stress and, ultimately, severe bleaching. Also, our results show that upwelling conditions (short-term cold stress+nutrient enrichment) can provoke more severe oxidative stress in corals than cold stress alone.

Nine Stylophora pistillata Esper 1797 colonies, originally sampled in the Gulf of Aqaba under CITES permit number DCI/89/32 and then cultured at the Centre Scientifique de Monaco (CSM), were used to generate a total of 108 nubbins (12 per colony). Nubbins were attached to nylon threads, identified, and kept in several independent 20-liter experimental tanks supplied with seawater (flow rate of 10 l h⁻¹). Metal halide lamps (Philips, HPIT 400W, Distrilamp, Bossee, France) provided irradiance of 200 µmol photons m⁻² s⁻¹ (12 h:12 h light:dark). Seawater temperature was kept at 26±0.5°C using submersible resistance heaters (Visi-ThermH Deluxe, Aquarium Systems, Sarrebourg, France) and salinity values were constant at 38 PSU. Coral nubbins were left to recover for 3 weeks in the experimental tanks and fed twice a week with Artemia salina nauplii. Submersible pumps ensured proper water mixing. Aquaria were cleaned weekly to avoid algal proliferation.

After recovery, two cold stress experiments were conducted. In the first experiment (experiment 1: long-term cold stress), we submitted corals to lower temperature conditions based on the mean seasonal temperature experienced by corals in the Gulf of Aqaba/Eilat (spring: 22.4±0.8°C; summer: 25.8±1.15°C; autumn: 25.5±0.9°C; winter: 22.3±0.7°C). The temperature treatments applied were based on long-term temperature time series (from January 2004 to December 2018) collected by the Israel National Monitoring Program at the Gulf of Eilat (available from meteo-tech.co.il). Nubbins were therefore divided in two groups (two tanks per condition, with nine nubbins each): (i) control condition, with natural seawater maintained at the same conditions as the acclimation period, and (ii) cold stress condition, in which temperature was gradually lowered from 26±0.5°C to 22±0.5°C over a 2-week period (0.3°C per day) and maintained constant for a 2-week period.

In the second experiment (experiment 2: short-term cold stress) we applied the same temperatures as in experiment 1 for comparison purposes. We aimed to simulate an acute cold stress condition as might happen when corals are subjected to a cold current or a wing-mixing event as well as a condition that may be experienced by corals during upwelling (fast drop in temperature and nutrient enrichment). Thus, nubbins were divided into four groups (two tanks per condition, with nine nubbins each): (i) control, with natural seawater maintained at the same conditions as the acclimation period; (ii) cold stress; (iii) nutrient enrichment (nitrate and phosphate, see below); and (iv) cold stress+nutrient enrichment. For this experiment, the cold stress was applied by lowering the temperature from 26±0.5°C to 22±0.5°C over a 5-h period (0.8°C per hour, as described by Roth et al., 2012) and maintained constant for 43 h. The nutrient enrichment was started at the same time as the cold stress and maintained until the end of the experiment. To generate enrichment conditions in the tanks [3 µmol l⁻¹ nitrate (N), 0.3 µmol l⁻¹ phosphate (P)], stock solutions of nitrate (as NaNO₃) and phosphate (as NaH₂PO₄) were pumped at a constant flow rate from a batch tank using peristaltic pumps (REGLO Digital, ISM 833, ISMATEC^®). Nitrate and phosphate enrichment were defined according to levels reported during upwellings affecting different reef environments (Eidens et al., 2014; Johnson et al., 2020; Radice et al., 2019; Stuhldreier et al., 2015). Nutrient concentrations in the experimental tanks were monitored using an Autoanalyzer (Alliance Instrument, AMS, France) according to Aminot and Kérouel (2007), and remained constant in the tanks throughout the whole experiment.

Net photosynthesis (P_net) and respiration (R) rates were measured (n=6 per treatment from different colonies) at the end of both experiments. Measurements were taken at the same temperature and light level as in the experimental conditions. Nubbins were incubated in 50 ml glass chambers, filled with 0.22 µm filtered seawater (FSW), continuously stirred with stirring bars. P_net and R rates were assessed at 200 and 0 µmol photons m⁻² s⁻¹, respectively, using Unisense oxygen-sensors connected to a fiber-optic oxygen meter (Presens, Regensburg, Germany). Calibration of the optodes (100% and 0% oxygen) was done using air- and nitrogen-saturated FSW. Data were expressed as µmol O₂ h⁻¹ cm⁻² and µmol O₂ h⁻¹ per symbiont for P_net and R.

Chlorophyll (chl a and chl c₂), symbiont density and coral host protein content were determined on the same samples used for photosynthesis determination, as described by Hoogenboom et al. (2010). Briefly, coral tissue was removed from the skeleton using an airbrush and collected in 10 ml of 0.45 µmol l⁻¹ FSW. The tissue slurry was then homogenized using a potter grinder. Sub-samples were used to quantify the symbiont density on a Z1 Coulter Particle Counter (Beckman Coulter), and to determine the coral host protein content using the Bradford Protein Assay Kit (23200, Thermo Fisher Scientific, USA). Chlorophyll concentration was measured in accordance with Jeffrey and Humphrey (1975). A 5 ml volume of the tissue slurry was centrifuged (8000 g, 10 min), the supernatant was discarded, and the symbionts (pellet) were resuspended in 5 ml acetone for chl a and c₂ extractions. Data were normalized to surface area (cm²) using the wax-dipping method (Stimson and Kinzie, 1991).

Calcification rates were determined for corals (n=6 per treatment from different colonies) exposed to long-term cold stress (experiment 1) using the buoyant weight technique originally described by Jokiel et al. (1978). Growth was calculated using the equation [(M_F–M_I)/(M_I×number of weeks)]×100, where M_F is the mass of each nubbin at the end of the experiment and M_I is the mass at the beginning of the experiment. Growth rates are presented as percent mass increase per week.

Coral nubbins (n=6 per treatment and measurement, from different colonies) were cut into small pieces (0.5 cm²) and homogenized by ultrasound (20 kHz, Vibra-Cell™ 75185, Bioblock Scientific, France) using 300 µl of the specific homogenization buffer for each analysis (as described below). We employed 20 pulses (2 s each) in order to disrupt the coral host tissue and release the endosymbiotic cells without breaking them. The remaining skeleton was discarded; the homogenized holobiont solution was centrifuged (5 min, 2000 g, 4°C) to separate the coral host tissue (supernatant) and symbiotic algae (pellet). The coral host tissue phase was collected and put in a new Eppendorf tube. Samples were again submitted to ultrasound (70 kHz, 20 s) and centrifuged according to each analysis (see below). The remaining pellet containing endosymbiotic cells was washed by thoroughly vortexing three times with homogenizing buffer followed by centrifugation (5 min, 2000 g, 4°C) in order to eliminate the remaining coral tissue in the pellet. Then, 60 µl of homogenizing buffer was added to the clean pellet for a final ultrasound homogenization (70 kHz, 20 s). Finally, the symbiont cell homogenate was also centrifuged according to each analysis. To ensure there was no contamination by symbiont cells in the coral host homogenate, we performed chorophyll analysis using an aliquot of the homogenate, as previously described by Marangoni et al. (2019). The total protein content in each sample was determined according to Bradford (1976) using the Comassie (Bradford) Protein Assay Kit (23200, Thermo Fisher Scientific) according to the manufacturer's instructions.

The quantification of intracellular ROS was performed at the end of both experiments for all treatments. ROS were also measured over time in experiment 2 for the control and cold stress treatment (after 5, 24 and 48 h). Measurements were carried out using the fluorescence technique described by Aguiar et al. (2008), with some modifications (Marangoni et al., 2020). Each sample was homogenized in a buffer containing Tris-HCl 100 mmol l⁻¹ (pH 7.7), ethylenediaminetetraacetic acid 2 mmol l⁻¹ and MgCl₂ 5 mmol l⁻¹, and then centrifuged (20,000 g, 20 min, 4°C). The protein content was adjusted to a final concentration of 0.8 mg ml⁻¹ for the coral host and endosymbiont homogenates to obtain the best fluorescence curves over time. In a flat-bottom black microplate, 10 µl of sample was added to a medium containing HEPES 30 mmol l⁻¹, KCl 200 mmol l⁻¹ and MgCl₂ 1 mmol l⁻¹ (pH 7.2). Finally, 10 µl of the fluorescent probe H₂DCFDA 16 µmol l⁻¹ (Invitrogen) was added. Fluorescence (excitation: 488 nm; emission: 525 nm) was measured (every 5 min up to 50 min) using a spectrofluorometer (Xenius, Monaco). Results were expressed as fluorescence units per minute (FU min⁻¹).

Determination of TAC of soluble low molecular weight antioxidants was measured at the end of both experiments for all treatments using the OxiSelect^TM Total Antioxidant Capacity (TAC) Assay Kit, according to the manufacturer's instructions. The antioxidant net absorbance values of the samples were compared with a known uric acid standard curve, with absorbance values being proportional to the sample's total antioxidant capacity. Absorbance readings were performed in a 96-well flat-bottom transparent microplate using spectrofluorometry (Xenius, Monaco). Data were normalized to the total protein content in the sample in each well and expressed as μmol l⁻¹ copper reducing equivalents (CRE) mg⁻¹ protein.

The oxidative damage to lipids (lipid peroxidation, LPO) was determined at the end of both experiments for all treatments according to Oakes and van der Kraak (2003). Each sample was homogenized in KCl (1.15%) solution containing 35 µmol l⁻¹ butylated hydroxytoluene (BHT), and centrifuged for 10 min (10,000 g, 4°C). Samples (20 µl) were added to a reaction mixture containing BHT solution (35 µmol l⁻¹), 20% acetic acid (pH 3.5), TBA (0.8%), sodium dodecyl sulfate (8.1%) and ultrapure water, and then incubated in a water bath at 95°C for 30 min. After cooling, n-butanol was added with thorough vortexing. Mixtures were centrifuged (300 rpm, 15°C, 10 min) and the immiscible organic layer was collected and added to a 96-well flat-bottom black microplate. Fluorescence (excitation: 515 nm; emission: 553 nm) was measured using a spectro-fluorometer (Xenius, Monaco). Data were normalized to the total protein content in the samples in each well and expressed as nmol MDA mg⁻¹ protein.

The effects of long-term cold stress (experiment 1) and short-term cold stress (experiment 2) on the physiological and oxidative stress parameters were evaluated using Student's t-test and two-way ANOVA (cold stress and nutrients as factors), respectively. For experiment 2, ROS production over time in the control and cold stress conditions was also evaluated using two-way ANOVA (temperature and time as factors). If indicated, ANOVA was followed by a post hoc Student–Newman–Keuls (SNK) test. Homogeneity of variance and data normality were checked prior to the analysis using Levene’s and Shapiro–Wilk tests, respectively. Data were log transformed to meet assumptions when necessary. In all cases, the significance level adopted was 95% (α=0.05). Results were expressed as means±s.e.m.

Physiology measurements showed that corals exposed to 30 days of cold stress exhibited a significant decrease in host protein content (P<0.03), and lower symbiont density (P<0.001) and chlorophyll concentration (P<0.001) compared with control corals. Calcification rates were not different between conditions (P=0.70) (Fig. 1). Net photosynthesis per unit skeletal surface area was significantly lower under cold stress (P<0.04). Since respiration rates did not differ between conditions (P>0.06) (Fig. 2), there was an overall decrease in gross photosynthesis per unit skeletal surface area. Such a decrease was mainly due to a loss of symbionts. Indeed, while net photosynthesis per symbiont cell decreased by 69% under cold stress (P<0.04), respiration rates increased by 60% (P=0.06). As a consequence, gross photosynthesis per symbiont cell remained practically unchanged between cold and control conditions. With respect to the oxidative stress parameters, a significant decrease was observed in ROS production in the symbiont fraction under cold stress (P=0.04). In turn, only a marginal difference in ROS level was observed in the coral host tissue (P=0.058). Symbionts and coral host TAC and LPO levels were not different between conditions (P>0.1) (Fig. 3).

After 48 h of exposure, cold stress, nutrient enrichment and the combined treatment (cold stress+nutrients) had different effects on coral physiology. Corals under cold stress showed a significant decrease in symbiont density (P<0.013) compared with the control (Fig. 4). On the contrary, nutrient enrichment did not change the symbiont density, but induced a significant increase in chlorophyll concentration (P<0.028) and in host protein content (P<0.006) with respect to all other conditions. Corals exposed to the combined treatment (cold stress+nutrients) suffered a significant decrease in symbiont density compared with the control (P<0.01), together with a lower chlorophyll concentration (P<0.028) and host protein content (P<0.02) compared with all other conditions (Fig. 4). Also, net photosynthesis per unit skeletal surface area (Fig. 5) was significantly reduced under cold stress and in the combined treatment (P<0.04). Because respiration rate did not differ between conditions (Fig. 5), there was an overall decrease in gross photosynthesis per unit skeletal surface area, mainly owing to a loss of symbionts. Net photosynthesis per symbiont cell showed a slight increase in corals under nutrient enrichment, but it was only significantly different compared with corals in the combined treatment (P<0.01). Also, respiration rate per symbiont cell was significantly higher under nutrient enrichment (P<0.001).

Regarding the biochemical responses, ROS were measured after 5, 24 and 48 h in the control and cold stress treatments, while for the other treatments, ROS were measured only after 48 h. ROS production in the control condition did not change over time for either symbionts or host. In the cold stress condition, both symbionts and host fractions generated higher ROS levels after 24 h compared with their respective controls (P<0.02), with the coral host fraction also presenting a significant increase in ROS levels over time (P<0.01). However, these levels returned back to control levels after 48 h, with symbionts also showing a significant decrease over time (P<0.02) (Fig. 6). With respect to the 48 h measurements in the other conditions, a significant increase in ROS production was only observed in the symbiont fraction in the combined treatment (cold stress+nutrients) (P<0.04). In accordance, TAC also increased in the symbionts in the same treatment (P≤0.008). A small increase in TAC was observed in the coral host tissue; however, it was only statistically different compared with the nutrient treatment (P=0.05). Decreased levels of LPO in the symbionts and in the coral host were observed in the nutrient treatment compared with all other conditions (P<0.02). Lower levels of LPO were also observed in the symbionts in the combined treatment compared with the control condition (P<0.03). In turn, increased LPO levels were detected in the combined treatment in the coral host tissue compared with all other conditions tested (P≤0.02) (Fig. 7).

This study assessed how short- and long-term excursions of cold waters affect the physiology and biochemical processes related to oxidative stress in a reef-building coral. Here, we show direct evidence that the mechanisms underpinning cold stress-induced bleaching are related to ROS overproduction. Furthermore, this study is also one of the first to provide evidence that upwelling conditions (acute cold stress+nutrient enrichment) can provoke more severe oxidative stress and bleaching in corals than cold stress alone.

Our results clearly show that both short- and long-term cold stress induce bleaching in S. pistillata, as previously observed for most coral species tested (Muscatine et al., 1991; Roth et al., 2012; Higuchi et al., 2015; Schoepf et al., 2019), except for the cold-resistant Acropora millepora (Nilsen et al., 2020), which showed increased symbiont density and chl a content after a 10-week exposure to 23°C (4°C decrease in the mean ambient temperature). The intensity of the bleaching response was related to the length of the stress, as the observed decrease in symbiont density and photosynthetic rate in S. pistillata was twice as high in the long-term compared with the short-term stress. In the long-term stress, corals lost up to 70% of their photosynthetic capacity, following a 60% decrease in chlorophyll pigments and a 40% decrease in symbiont density. Similar results were observed in a previous study (Roth et al., 2012), in which Acropora yongei showed a decline in dinoflagellate density of ∼20% after acute cold stress (5 days, 21°C) and of ∼40% after longer exposure (20 days, 21°C). Our results tend to confirm a general response (bleaching) of corals to cold stress, although this response may vary depending on the intensity and the duration of the stress (Pörtner, 2002), and on the symbiont type hosted by coral species considered (Silverstein et al., 2017).

The transient increase (within the first 24 h) in ROS levels observed here in corals and their symbionts subjected to short-term stress suggests that the observed bleaching is due to photoinhibition of photosystem II, likely following a temperature-dependent reduction in enzyme activities operating in the Calvin–Benson cycle (Somero, 1995; Jones et al., 1998). Indeed, decrease in enzyme activities induces a reduction in photosynthetic electron transport rate, which, combined with continued light energy absorption, can cause a buildup of excess light energy and ROS production and damage to photosystem II (Saxby et al., 2003; Roth et al., 2012; Schoepf et al., 2019).

However, corals showed a very rapid biochemical and physiological adjustment to decreasing temperature by expelling a portion of their symbionts (probably the most photosynthetically compromised symbionts) to avoid excess ROS in the holobiont, and also energy expenditure in antioxidant defenses that can be very costly (Halliwell, 2007; Lesser, 2006). Indeed, after 48 h, ROS levels were significantly lower in the symbionts (∼25%) and had returned to control levels in the coral host. Under long-term cold stress, the reduced symbiont density also resulted in lower ROS production at the end of the experiment (∼36%). Importantly, no signs of oxidative stress (oxidative damage or TAC fluctuations) were observed in either symbiotic partner after 48 h or after longer exposure to cold stress. It is possible that the partial expulsion of the most photosynthetically compromised symbionts and decreased concentrations of photosynthetic pigments is related to acclimation to a new redox balance to maintain homeostasis under cooler conditions. Interestingly, Higuchi et al. (2015) observed a reduction in symbiont density followed by decreased activity of antioxidant enzymes [superoxide dismutase (SOD) and catalase (CAT)] after 10 days of exposure to cold stress in the temperate coral host Acropora pruinosa. Symbiotic cnidarians naturally experience elevated P_O2 within their tissues (a pro-oxidant condition) as a result of photosynthetically produced oxygen (Dykens and Shick, 1982; Lesser, 2006). SOD and CAT act in concert with other enzymes to inactivate ROS, and their induction in the presence of increased ROS is well known (Lesser, 2006). Findings from the present study along with those reported by Higuchi et al. (2015) reinforce our statement that the expulsion of compromised symbionts during cold stress decreases the amount of ROS in the holobiont, and therefore results in lower demands on the antioxidant systems of the host tissue. Such a trade-off may prevent corals from experiencing severe oxidative stress, which can ultimately be lethal, at a cost of receiving less energy from photosynthesis for the maintenance of the holobiont's physiological needs.

In addition to expelling excess symbionts in order to adjust their redox balance, corals may also increase their levels of photoprotective pigments such as carotene and xanthophylls (Roth et al., 2012) in order to resist cold stress for as long as 9 months, although with reduced metabolism compared with corals under control conditions (Schoepf et al., 2019). Here, the reduced symbiont density and rates of photosynthesis impaired tissue growth and maintenance, as S. pistillata showed a significant decrease in protein content at the end of the long-term experiment. This observation can be linked to that of Roth and Deheyn (2013), who demonstrated that green fluorescent proteins, which are an important component of the total soluble proteins in corals (Leutenegger et al., 2007), can be rapidly degraded and or depleted in corals under cooling conditions. However, in our experiment with S. pistillata, there was an apparent reallocation of resources to calcification (Hoogenboom et al., 2015; Wall et al., 2017), the rates of which were maintained constant throughout the long-term thermal stress.

Nutrient enrichment had an opposite effect on the physiology of the host and symbionts of S. pistillata, regardless of whether it was applied alone or in combination with cold stress. Nutrient enrichment alone enhanced both host (increased protein) and symbiont (increased photosynthetic efficiency and respiration) metabolism, because nutrients are essential, and often limiting, for coral growth (Ezzat et al., 2015; Krueger et al., 2020). In this experiment, where we used the levels of nitrate and phosphate enrichment often measured during upwellings (Eidens et al., 2014; Johnson et al., 2020; Radice et al., 2019; Stuhldreier et al., 2015), and in which phosphorus is not limiting compared with nitrate, the nutrient enrichment led to beneficial effects on the oxidative status of corals, as evidenced by the decreased LPO levels in both symbionts and coral host. Previous studies have indeed shown that balanced nutrient availability (in terms of the N:P ratio) is important for the stability of the coral–dinoflagellate symbiosis during heat stress (Wiedenmann et al., 2013; Rosset et al., 2017), whereas increased concentrations of nitrate lead to enhanced oxidative stress, energy deficit and bleaching in corals (Marangoni et al., 2020).

The results of the ‘upwelling condition’ point to a disruption of the symbiosis by the combined effect of cold stress and nutrient enrichment. In such conditions, even a balanced N:P ratio (or a non-limiting phosphorus concentration) did not prevent corals from undergoing an oxidative imbalance with an overproduction of ROS in the symbionts, followed by increased TAC, in both the coral host and symbionts. The higher levels of TAC indicate that excessive ROS production in this condition stimulated the antioxidant machinery of corals. The overall efficiency of cellular antioxidants to counteract ROS is a sensitive measurement for revealing an organism's susceptibility to oxidative stress in many stressful conditions (Gorbi and Regoli, 2003), including symbiotic corals (e.g. Marangoni et al., 2017, 2019; Ayalon et al., 2019; Levy et al., 2020). Here, this compensatory response in the symbionts was sufficient to decrease the occurrence of LPO. However, an increase of LPO in the coral host was still observed, and may have affected enzyme activity and ATP production, and have initiated apoptosis (Ayala et al., 2014). Therefore, corals maintained under upwelling conditions experienced the highest loss in symbionts and chlorophyll content, as well as the highest decrease in photosynthetic capacity compared with the other conditions. Such decreased productivity under upwelling conditions was also measured by Eidens et al. (2014) in Tayrona National Park (Colombian Caribbean) on the genera Orbicecella and Pseudodiploria.

Overall, this study has shown that in all cases, cold stress impairs the autotrophic capacity of corals by inducing bleaching and decreasing photosynthetic capacity. Nevertheless, during heat waves (e.g. El Niño), corals under upwelling influence may benefit from cooler conditions compared with their neighbors as well as from increased heterotrophy (Riegl and Piller, 2003; Riegl et al., 2019). Nutrient levels and ratios may also play a major role in the extent of bleaching that corals undergo during upwelling. For instance, and contrary to other studies (e.g. Eidens et al., 2014), Stuhldreier et al. (2015) reported increased net primary production in corals along the Pacific coast of Costa Rica under upwelling conditions. Although these authors reported similar nitrate and phosphate concentrations to the ones used here, they also measured high concentrations of ammonium in the upwelled seawater. As ammonium is preferentially assimilated by corals over nitrate (Grover et al., 2008), it may have prevented the deleterious effects of nitrate. In addition, ammonium has been shown to moderate the deleterious effects of thermal stress by favoring the oxidative status and energy metabolism of the coral holobiont (Marangoni et al., 2020). In this context, it is possible to infer that the nutrient conditions reported by Stuhldreier et al. (2015) were beneficial for corals during cooling conditions. However, if corals experience an oxidative imbalance during upwelling seasons – even if to a slight degree – owing to less favorable N:P ratios or nitrogen sources (as tested here), they could become more vulnerable to anthropogenic stressors commonly present in coastal zones, and which are also known to favor oxidative stress and cause bleaching (e.g. chemical pollutants) (van Dam et al., 2011; Marangoni et al., 2017; Fonseca et al., 2017). In this context, further in situ studies of upwelling-influenced coral reefs are needed to assess both the nutrient composition of the seawater and the physiological and biochemical responses of corals to cold stress and nutrient addition.

This study provides the first evidence that partial bleaching observed in some tropical corals facing lower temperatures may consist of an acclimation mechanism to reach a new redox balance to maintain homeostasis. In this context, we suggest that the maintenance of a lower number of healthier symbionts may prevent corals from enduring severe oxidative stress and bleaching. Also, our findings indicate that corals under upwelling conditions (lower temperature+increased nutrient levels) can be more susceptible to oxidative stress and, ultimately, bleaching. Considering that upwelling regions are considered to be possible refugia for reefs in a time of changing climate, future studies under field conditions should be undertaken. A more profound understanding of the physiological and biochemical responses of corals inhabiting upwelling-influenced coral reefs could be of great importance for the future of coral reef conservation and restoration purposes.

Special thanks to Prof. Oren Levy for supporting comments.