On November 2–3, 2018, 14 specimens of reef-building corals ( Table 1 ) belonging to nine genera (seven families) were collected by SCUBA in the Luhuitou fringing reef (3 m depth) at the Tropical Marine Biological Research Station in Hainan, Chinese Academy of Sciences (18°14′N, 109°28′E). Moreover, three bleached Acropora spp. (A. florida, A. nana, and A. intermedia) were also sampled after a tropical cyclone on December 28, 2018. Images were collected using a camera (TG-6, Tokyo, Olympus, Japan) before sampling. All coral specimens were identified by taxonomists at the South China Sea Institute of Oceanology, Chinese Academy of Sciences. From October 2018 to January 2019, the temperature (°C) and light intensity (lux) of the sampling area were monitored continuously at 2-h intervals using a HOBO logger (UA-002-64, Onset Computer, Bourne, MA, USA) ( Figure 1 ). The HOBO logger was cleaned weekly to reduce the interference of algae, sediment, and other fouling organisms as much as possible.

Healthy (n = 8) and bleached (n = 3) samples from different colonies (3–5 cm fragments) of each coral species were collected and immediately transported to the laboratory, where they were divided into two subsamples. One subsample (healthy, n = 5; bleached, n = 3) was freeze-dried in a vacuum freezing dryer (Biosafer-21A, Duhong Biological Technology Co., Ltd., Guangzhou, China), crushed into 0.1–0.3-mm pieces, and then frozen at −80°C until analyses of the total lipid content and FA composition. The other subsample (healthy, n = 3) was stored at −20°C until analyses of the Symbiodiniaceae density and corallite area.

The Symbiodiniaceae density in the coral tissues was evaluated as described by Martínez-Castillo et al. (2020) with some modifications. Coral tissues were removed from the coral fragment using jets of pressurized air in filtered seawater (~1 ml) until only the white coral skeleton was visible. The tissue slurry was homogenized using a hand-held blender (MY-10, Jingxin Technology, Jiangsu, China) for 45 s and centrifuged (3,000×g, 5 min; Microfuge 20, Beckman Coulter, Brea, CA, USA); the supernatant was then removed. The process was repeated three times. The number of Symbiodiniaceae per slurry volume was counted using hemocytometer plates with four replicate counts. The total number of Symbiodiniaceae per area was calculated based on the volume of the homogenate and coral surface area (×10⁶ cells cm⁻²). The coral surface area was measured using a structured light 3D scanner (JTscan-MS, Guangzhou Jeatech Electronics Technology Co., Ltd., Guangdong, China).

Given that the shape of coral corallite is irregular, the area of the corallite (mm²) was measured as described by Conti-Jerpe et al. (2020). Images of the skeleton of each sample were captured under a microscope (CX31, Olympus, Japan) after determining the Symbiodiniaceae density, and ImageJ software (version 1.8.0; NIH, Bethesda, MD, USA) was used to calculate the area of the coral corallite (mm²) for 10 replicates.

To ensure the accuracy of the measurement of the total lipid content and prevent the oxidation of FAs, total lipid content and FA composition were assessed separately. The crushed samples (approximately 0.5 g) were packed in a filter paper, and lipids were extracted using a Soxhlet extractor (AG-SXT-06, Shanghai Ouge Electronics Co., Ltd., Shanghai, China). Petroleum ether (boiling point, 60°C–90°C) was used as the solvent, and a thermostat-heating mantle was used to maintain 90°C for 5 h. The total lipid content was considered as the difference in sample weight between before and after the extraction.

Total lipid content was measured in crushed whole coral samples (skeleton, animal tissue, and Symbiodiniaceae), normalized to the dry weight, and expressed as a percentage (g lipid/g dry weight).

Approximately 1-g sample (dry weight) from each sample was analyzed. The samples were homogenized, and each lipid fraction was extracted using chloroform/methanol (2:1, v/v) containing 0.01% butylated hydroxytoluene as an antioxidant, as previously described by Liu et al. (2019). The chloroform layer was separated from the methanol layer and dried to a constant weight under a stream of nitrogen to obtain the total lipid content.

Fatty acid methyl esters (FAMEs) were obtained by esterification with 2 ml methyl esterification reagent (hydrochloric acid/methanol, 1:5, v/v) at 90°C for 3 h, as described by Liu et al. (2018). The upper phase was dried under nitrogen and resuspended in hexane.

FAMEs were quantified by injecting 1 μl of sample into a gas chromatograph instrument (GC-2010 Plus; Shimadzu, Kyoto, Japan) equipped with a flame-ionization detector (GC-2010; Shimadzu, Kyoto, Japan) and an RTX-WAX fused-silica capillary column (length, 30 m; internal diameter, 0.25 mm; thickness, 0.25 μm; Phenomenex, Torrance, CA, USA). The gradient temperature program was set as follows: (i) initial temperature of 60°C for 1.0 min; (ii) increase at a rate of 10°C min⁻¹ to 190°C; (iii) increase at 2.0°C min⁻¹ to 260°C; and (iv) hold at 260°C for 0.6 min. The FAMEs were identified and quantified by comparing the retention times (identification) and peak areas (quantification) with the 37-FAME Mix calibration solution (Supelco, Bellefonte, PA, USA).

R software version 3.5.2 (The R Project for Statistical Computing, Vienna, Austria) was used for statistical analyses. The distribution of the data was evaluated for normality using the Kolmogorov–Smirnov test (p > 0.05). Levene’s test (p > 0.05) was used to assess the homogeneity of variance. Data of massive and branching samples were then evaluated by t-test analysis to identify significant differences (p < 0.05). The correlation of total lipid content between the Symbiodiniaceae density and corallite area was determined using Pearson’s product-moment correlation with the package “corrplot,” and the FA composition of corals between massive and branching corals was characterized using multivariate principal component analyses with the package “vegan” (Oksanen et al., 2010). Graphs were prepared using the ggplot2 package (Wickham, 2016).

To facilitate a comparison of FA compositions, we calculated the unsaturation index (UI) and the unsaturated-to-saturated fatty acid ratio (U/S) as reported by Snyder et al. (2012) and Wallaert and Babin (1994). The UI and U/S algorithms were as follows:

where monoenes, dienes, and trienes … are FAs containing 1, 2, 3…double bonds, respectively; % is the weight percentage; UFA indicates unsaturated FA; and SFA indicates saturated FA.

The Symbiodiniaceae density and corallite area are shown in Tables 2 , 3 . The Symbiodiniaceae density ranged from 1.96 × 10⁶ (A. nana) to 1.02 × 10⁶ cells cm⁻² (Platygyra acuta). The average Symbiodiniaceae density of the massive and branching corals were 1.50 ± 0.26×10⁶ and 1.55 ± 0.36×10⁶ cells cm⁻², respectively. There was no significant difference (p > 0.05) in the Symbiodiniaceae density between the massive and branching corals.

The corallite area of massive corals (7.19 ± 6.24 mm²) was significantly larger (p < 0.05) than that of the branching corals (0.69 ± 0.25 mm²). Platygyra acuta showed the largest corallite area (17.35 ± 5.11 mm²), followed by Coelastrea aspera (10.33 ± 2.02 mm²), and Pocillopora damicornis showed the smallest area (0.44 ± 0.12 mm²).

The total lipid content of the 14 species of scleractinians are shown in Table 2 . Among all species, the highest total lipid content was observed in P. acuta (10.81 ± 0.96%), followed by Galaxea fascicularis (9.61 ± 1.16%), whereas P. damicornis showed the lowest total lipid content (2.90 ± 0.20%). The total lipid content varied with the colony morphology. Overall, the branching corals (3.60 ± 0.36%) showed lower lipid content than massive corals (7.45 ± 2.87%) ( Table 2 ). Moreover, the total lipid content of branching corals was positively correlated with the Symbiodiniaceae density but unrelated to the corallite area. In contrast, the total lipid content of massive corals was positively correlated with the corallite area but not with the symbiodiniaceae density ( Table 3 ).

Compared to healthy corals, a significant decrease in total lipid content was found in the bleached coral samples ( Figure 2 ), which accounted for only approximately 25% of the total lipid content of the healthy group. The average total lipid content was 3.84 ± 0.49% for healthy corals and 1.01 ± 0.16% for bleached corals.

Most corals maintain an endosymbiotic association with algae of the family Symbiodiniaceae (Radice et al., 2019). However, the symbiotic relationship is fragile and may collapse because of adverse environmental changes, provoking coral bleaching (whitening of corals caused by the loss of symbiotic dinoflagellates) and even mortality (Hughes et al., 2017; Tagliafico et al., 2017).

Lipids are involved in most biochemical and physiological processes in corals; thus, these may reflect the health status of corals. We found that bleached corals lost more than 50% and 75% of their original SFAs and total lipid content, respectively, which is common following bleaching events (Yamashiro et al., 1999; Imbs and Yakovleva, 2012). Symbiodiniaceae inherently contain considerable amounts of lipids and SFAs (Chen et al., 2015; Conlan et al., 2017). Therefore, the loss of symbiodiniaceae directly leads to decreased levels of lipids and SFAs in corals. Under bleached conditions, corals may rely more heavily on stored lipid to sustain their normal physiological functions (Tolosa et al., 2011). Similar reductions in lipid levels of bleached corals in the Caribbean (Porter et al., 1989), Okinawa (Yamashiro et al., 2005), and Hawaii (Rodrigues et al., 2008) were attributed to decreases in translocated carbon from their symbionts. Moreover, marine organisms tend to preferentially metabolize SFAs to protect long-chain PUFAs (such as 20:4n6 and 22:6n3) from oxidation during starvation (Tocher, 2003; Liu et al., 2020). Thus, a large quantity of SFAs, as the preferred source of metabolic energy, was oxidized to provide energy for bleaching corals.

Interestingly, C_20–22 PUFAs were significantly reduced in bleached corals, whereas C₁₈ PUFAs were increased significantly. The reduction in PUFAs in corals was reported following bleaching events (Tolosa et al., 2011; Tagliafico et al., 2017). Bachok et al. (2006) showed that completely bleached coral lost more than 70% of their PUFAs. Symbiodiniaceae transfer sufficient PUFAs for the hosts’ requirements under ideal conditions (Chen et al., 2015), whereas in bleaching situations, the host just acquires PUFAs from heterotrophic feeding (Grottoli et al., 2006). Moreover, recent works showed that Cnidaria, including scleractinian corals, contain genes for de novo biosynthesis of PUFAs (Kabeya et al., 2018; Monroig and Naoki, 2018). C₁₈ PUFAs, such as 18:2n6 and 18:3n3, are key precursors in the biosynthesis of C_20–22 PUFAs in marine invertebrates. Therefore, C₁₈ PUFAs were abundant in bleached corals, indicating that corals could obtain PUFAs from biosynthesize or heterotrophic assimilation, but with low efficiency. Furthermore, 18:3n-6 and 18:2n6 and their metabolites can affect the expression of genes associated with immunity and apoptosis (Rocker et al., 2019), such as the aryl hydrocarbon receptor (AhR). Overall, considering the large consumption of lipids and SFAs and the inability to rapidly replenish essential FAs (long-chain PUFAs), long-term bleaching of corals inevitably leads to coral death.

Indeed, the nutritional status of corals can affect their resistance and resilience to bleaching. Branched corals have low total lipid content but high surface areas to volume ratio; hence, branched corals have lower levels of stored lipid than do massive corals. Such levels significantly influence the timing of onset of bleaching and subsequent mortality (Anthony and Willis, 2002; Grottoli et al., 2004). Conlan et al. (2017) suggested that corals with “full” initial lipid stores survive twice as long as those with low lipid stores under high bleaching rates. Therefore, branching corals with low initial lipid stores are disadvantaged in the face of environmental stress. Moreover, the ability to obtain extra lipids from the environment is also extremely important for corals under adverse conditions (Houlbrèque and Ferrier-Pagès, 2009; Grottoli et al., 2014). Heterotrophy accounts for 0%–66% of the fixed carbon incorporated into coral skeletons and can meet 15%–35% of the daily metabolic requirements in healthy corals and up to 100% in bleached corals (Houlbrèque and Ferrier-Pagès, 2009; Conti-Jerpe et al., 2020). Yamashiro et al. (2005) found that massive corals were relatively rich in lipids even after a bleaching event, which was attributed to increased feeding rates. Massive corals, predominantly heterotrophic or mixotrophic, are less susceptible to bleaching and show higher resilience to stress than branching corals do.

FAs are the primary constituents of the cell and subcellular organelle membranes in organisms. Differences in FA compositions can affect membrane-associated physical features and biological functions such as fluidity, permeability, and related enzyme activity (Tchernov et al., 2004; Ernst et al., 2016; Liu et al., 2020). We found that massive corals have a higher composition of UFAs than that of branching corals. Compared to SFAs, homologous UFAs have a lower melting point and occupy a larger space within the membrane lipid bilayer to enhance their fluidity and stability (Hazel, 1979; Yeagle, 1989; Liu et al., 2018). Increased biomembrane fluidity may facilitate membrane-bound enzyme activity and material exchange (Conlan et al., 2020; Liu et al., 2020) and help minimize the influence of environmental stress. Tchernov et al. (2004) demonstrated that high levels of unsaturated in the membrane are a key determinant of thermal-stress sensitivity in symbiotic algae of cnidarians. Furthermore, PUFAs (such as 20:4n6, and 22:6n3) exert immune and reproductive functions and are used as precursors of some hormones (Berge and Barnathan, 2005). Therefore, massive corals with more PUFAs may have a survival advantage during environmental changes.