Introduction

Lipids are essential compounds in marine organisms and source of energy (Bergé and Barnathan 2005; Lee et al. 2006; Parrish 2013). In comparison to other compounds, e.g. macromolecules, such as proteins and carbohydrates, lipids have the highest energy content (approximately 17 kJ g−1, 18 kJ g−1, and 39 kJ g−1, respectively). Additionally, lipids are important to maintain cell membranes, are components of fat-soluble antioxidants and hormone (e.g. ecdyson), and regulate anti-inflammatory and immune responses (Ackman 1999; Tocher 2003; Lee et al. 2006; Parrish 2013). Thus, lipids play a crucial role in the physiology of marine organisms (e.g. respiration, cell renewal, and reproduction) and also display biochemical and ecological conditions of the marine environment (Arai et al. 1993; Ward 1995; Tarrant 2005; Wang et al. 2013). Lipid biomarkers have the potential to serve as an advanced diagnostic tool for coral health. Subtle changes in physiological condition can manifest as large variations in lipid biomarker concentrations as well as relative ratios of certain classes of lipid. Thus, coral health can be assessed well before changes in mortality, growth rates, occurrence of diseases, and community changes are observed (Rocker et al. 2019). One major component of lipid molecules are fatty acids (FAs), which occur either as free FAs, or chemically bound to different lipid classes. In particular, free fatty acids (hereafter “FA/FAs”) are essential components of lipids in many marine organisms (Bergé and Barnathan 2005; Parrish 2013). They are transferred throughout the food web without experiencing changes in different trophic levels, which encourages their use as effective biomarkers in ecological studies (Parrish et al. 2000; Alfaro et al. 2006). Indeed, many studies consider FAs as signature lipids in trophic biomarkers (Volkman 1999) of various marine organisms including bacteria, diatoms, dinoflagellates, plankton, macroalgae, invertebrates, fishes, and marine mammals (Rajendran et al. 1993; Parrish et al. 2000; Falk-Petersen et al. 2002; Howell et al. 2003; Bergé and Barnathan 2005; Lee et al. 2006; Kelly et al. 2008; Shin et al. 2008; Ju et al. 2011; Sardenne et al. 2017). Furthermore, FA profiles was shown to display a strong taxon specificity (Volkman 1999).

In corals, lipids are one of the main biochemical compounds and a crucial dietary source for energy storage. As such, lipids take up 10–50% of the dried coral tissue (Joseph 1979; Harland et al. 1992, 1993; Al-Lihaibi et al. 1998; Yamashiro et al. 1999, 2005; Oku et al. 2003b; Seemann et al. 2013). Indeed, FA are vital to coral metabolism and stress tolerance (Imbs et al. 2015). Moreover, FA profiles of different coral species have been used as chemotaxonomic biomarkers as they are suggested to be species-specific (Latyshev et al. 1991; Imbs et al. 2007, 2010b, 2016; Lopes et al. 2016). For example, octocorals can be identified by their synthesis of 24:5n-3 and 24:6n-6, as these FA are absent in hexacorals (Svetashev and Vysotskii 1998; Imbs et al. 2010a). Symbiotic corals can be identified by high level of 18:3n-6 (GLA) and 18:4n-3 (SDA), as these FAs are present in symbiotic algae, while they are very low in non-symbiotic corals (Papina et al. 2003; Imbs et al. 2007, 2010b; Imbs 2013; Lopes et al. 2016). Imbs et al. (2007) classified five different families (Acroporidae, Faviidae, Fungidae, Pocilloporidae, and Poritidae) of the scleractinian corals based on their specific polyunsaturated fatty acids (PUFAs), as they found that several FAs can only be detected in certain taxonomic groups of corals. Most scleractinian corals have symbiotic relationships with dinoflagellates, namely the family Symbiodiniaceae (hereafter ‘symbionts’) (LaJeunesse et al. 2018), which are mostly photosynthetic that guides carbon dioxide fixation (Wooldridge 2014). Their photosynthesis contribute significantly to the supply of biomolecules used to feed themselves while the nutrients are also translocated to the coral host for growth, reproduction, and metabolism (Muscatine et al. 1981). The main photosynthates transferred from symbionts to the coral host tissue include lipids, glycerol, glucose, and amino acids (Gates et al. 1995; Papina et al. 2003; Whitehead and Douglas 2003; Reynaud et al. 2009), where excess organic carbon is stored as lipids in the host tissue (Patton et al. 1977; Patton and Burris 1983). Of the lipids, FAs in particular, is one of the most important nutrients transferred across the plant-animal interface in the aquatic food web (Dalsgaard et al. 2003; Allan et al. 2010). It has been reported that photosynthetic organisms (e.g. plants, macro–microalgae, dinoflagellates, etc.) synthesize essential PUFAs, such as 18:2n-6 or 18:3n-3, which are precursors that, in turn, metabolize several n-6 (e.g. 18:3n-6, 20:4n-6) and n-3 PUFAs (e.g. 18:4n-3, 20:5n-3, 22:6n-3) through their respective pathways of FA biosynthesis (Chen et al. 2015; Revel et al. 2016). Animals cannot synthesize these precursor FAs due to the lack of certain metabolic enzymes (Δ12, Δ15 desaturases) (Papina et al. 2003; Bachok et al. 2006; Revel et al. 2016). Thus, the PUFAs synthesized by symbionts can act as biomarkers to understand the coral–symbiont relationship among coral species.

On the other hand, certain essential FAs cannot be synthesized in higher trophic positions, although they are incorporated into heterotrophic species in the marine ecosystems (Arai et al. 2015). Moreover, FAs mirror nutritional input, making it possible to use them as biomarkers to trace diet and quantify feeding relationships (Dalsgaard et al. 2003; Bay et al. 2013; Mies et al. 2018). For example, 20:5n-3 is abundant in diatoms and 22:6n-3 is abundant in dinoflagellates. Both are highly conserved in the marine food web, which allows food source tracing (Viso and Marty 1993; Scott et al. 2002; Dalsgaard et al. 2003). The ratio of 20:5n-3 to 22:6n-3 (EPA:DHA) has been used to define trophic level (autotrophy or heterotrophy) and feeding type (herbivore, omnivore, carnivore) in zooplankton (Graeve et al. 1994; Dalsgaard et al. 2003). Additionally, carnivorous zooplankton tend to have higher PUFAs than herbivorous species, hence the PUFA:SFA ratio offers an index of carnivory (Cripps and Atkinson 2000; Dalsgaard et al. 2003; Stevens et al. 2004). A few coral studies successfully applied these FA indices to define trophic level in coral reef ecosystems (Tolosa et al. 2011; Seemann et al. 2013; Salvo et al. 2017; Radice et al. 2019; Rocker et al. 2019).

Although FA profiles have shown to present excellent opportunities for chemotaxonomic and trophic biomarkers in symbiotic corals, applying FA profiles to corals and symbionts separately, as opposed to a two-organism unity, remains understudied. Specifically, symbiotic corals are able to obtain essential FAs either via autotrophy which are being translocated from symbionts, or via heterotrophy through external food sources, e.g. phytoplankton, zooplankton, particulate organic matter (Sebens et al. 1996; Ferrier-Pagès and Gattuso 1998; Anthony and Fabricius 2000; Yahel et al. 2004; Rocker et al. 2019). Furthermore, corals have a high trophic plasticity, switching between autotrophic and heterotrophic feeding strategies to cope with environmental changes (Goreau et al. 1971; Porter 1976; Muscatine et al. 1989; Anthony and Fabricius 2000; Ferrier-Pagès et al. 2003; Fabricius 2005). Therefore, their symbiotic relationship and trophic plasticity provide challenges to the application of FA biomarkers.

In this study, we characterized FA profiles from five symbiotic coral species belonging to five different genera (Acropora, Pavona, Turbinaria, Favites, and Platygyra) and four different families (Acroporidae, Agariciidae, Dendrophyllidae, Faviidae). To do so, we first separated symbionts from the coral host tissue, and then investigated the interaction between symbionts and the host tissue using FA biomarkers. We then selected these FA biomarkers from the coral host fraction, in particular specific FAs and their ratios (EPA:DHA, PUFA:SFA) to examine chemotaxonomy and trophic level (autotrophy vs. heterotrophy) of each coral species. We tested the following hypotheses: (1) the five coral target species can be classified by distinct FA profiles; (2) symbiotic corals have a relative trophic level (more autotrophic or more heterotrophic) which can be defined by certain FA biomarkers.

Materials and methods

Coral sampling

In May 2016, five different symbiotic coral species (Acropora samoensis of the family Acroporidae, Pavona decussata of the family Agariciidae, Turbinaria peltata of the family Dendrophyllidae, Favites abdita and Platygyra carnosa of the family Faviidae) were collected from four mesocosm coral tanks at the Swire Institute of Marine Science (SWIMS) located at the marine reserve area of Cape d’Aguilar in Hong Kong SAR. Open circulation flow-through tanks (≈ 5 l min−1) at SWIMS were used and seawater was supplied directly from the bay after a primary step of physical filtering (sand filter). The seawater included natural food sources such as plankton and particulate organic matter, hence additional feeding was not necessary. The samples from species grown in mesocosm tanks as opposed to species from a natural site were taken to ensure that the corals have not been disturbed by sedimentation or unexpected weather conditions (e.g. heavy rain, typhoon) and all colonies were exposed to the same light conditions and food sources, which enabled to achieve our study purposes. Samples were collected in May 2016. Hong Kong has alternating wet/dry seasons; the wet season is from March to October, characterized by higher temperatures, and the dry season is from November to March. We collected our samples during the transition time between dry and wet season, with an average water temperature of 26.0 °C. Branching and foliaceous corals (A. samoensis, P. decussata) were collected by scissors, and plate corals (T. peltata) or boulder corals (F. abdita, P. carnosa) were collected by hammer and chisel. Each nubbin was taken from different colonies (n = 4) in each tank. The collected samples were immediately rinsed with deionized water and kept frozen at − 70 °C until further analyses.

Coral–symbiont separation

For lipid analysis, the coral fragments were rinsed with buffer (5 mM EDTA) with 20 μl ml−1 butylated hydroxytoluene in distilled water to prevent FA oxidation and development of artefacts. Coral tissue was extracted from the skeleton using an airbrush with the same buffer solution and collected in 50 ml tubes through a funnel. Slurry tissue, which included coral host tissue and symbionts was homogenized on ice with a tissue grinder for 30 s. The homogenate was centrifuged for 5 min at 4 °C at initially 300 rcf, increasing up to 800 rcf (depending on tissue thickness, amount of mucus, and species) (Table S1) to separate symbionts (pellets) and coral tissue (supernatant). One centrifugation process was difficult to differentiate the two fractions thus further separation was required. The supernatant was transferred to another 50 ml tube and centrifuged in deionized water three more times for 5 min at 4 °C with different velocity of centrifugation depending on the species, to eliminate any remaining symbionts. A drop of supernatant was observed under a microscope to verify no symbiotic cells remained in the coral host tissue sample (Fig. S1); nevertheless minor contamination from symbionts and coral host fraction may exist in the supernatant. The pellet was re-suspended in 20–30 ml of DI water and centrifuged at 10 rcf for 30 s to remove any carbonate residue. The supernatant was immediately transferred to another tube and then centrifuged three times for 5 min at 4 °C at varying velocity of centrifugation depending on the species (Table S1). After the third centrifugation, the pellet was recollected. Both coral host tissue and symbiont samples were kept in the freezer at − 80 °C. Prior to chemical analyses, the samples were freeze-dried and kept as powder.

Extraction lipid and fatty acids

Total lipid (TL) was extracted from dried samples of the coral host tissue and symbionts separately by Folch solution (2:1; chloroform:methanol). Folch et al. (1957) method can be applied for a wide range of sample types (Couturier et al. 2020), where the extraction efficiency is high and sample matrix effect is minimal. Extracted TL was dried under nitrogen gas on a 37 °C heating block. The dried TL was weighed to determine TL content as relative proportion (%) of dry weight of symbiont and coral host, respectively. To the dried TL, aqueous 1 M potassium hydroxide (KOH) and internal standard (19:0; nonadecanoic acid, 1 μg l−1 in dichloromethane) were added, heated for 60 min at 70 °C oven to hydrolyze the esterified FA. The total FAs were then collected by liquid–liquid extraction using hexane:diethyl ether (9:1 v/v), and the solvent was evaporated immediately under nitrogen gas and derivatized with boron trifluoride (BF3) at 70 °C for 30 min to form fatty acid methyl esters (FAMEs). FAMEs were extracted three times with hexane:diethyl ether (9:1 v/v) and then pooled. The extracted FAMEs were dried under nitrogen gas and re-suspended in dichloromethane for gas chromatography mass spectrometry (GC–MS) analysis set at electron ionization (EI) mode. A set of external standards (37 Component FAME mix, Supelco, USA) was used to identify and quantify each FAME. GC–MS (Agilent 5977A mass selective detector interfaced with an Agilent 7890B gas chromatograph) was performed using an SP-2560 capillary column (100 m × 0.25 mm, 0.2 μm film thickness, Sigma–Aldrich, USA) to detect the FAs. Helium was used as the carrier gas with a flow rate of 1 ml min−1. A volume of 1 μl of the derivatized sample was injected into the GC with a split ratio of 1:100. The oven temperature was programmed to ramp from 100 °C to 140 °C at 10 °C min−1 and from 100 °C to 240 °C at 4 °C min−1. Each FAME was identified using the target and qualifier ion listed in Table S3. FAMEs were quantified by relating the peak area of the individual FAME with the peak of the internal standard. For quantification, the dry weight of TL was used to normalize FAME concentration (μg mg−1).

Statistics

ANOVA test was performed using the software SPSS (ver. 19.0, IBM, NY, USA). Statistical differences among the five species were determined for TL and FA ratios by one-way analysis of variance (ANOVA) and Tukey’s post hoc test for pairwise comparison. Values of at least p < 0.05 were considered significant in the analysis. Further, we performed multivariate analyses on square root transformed value of the 19 FA profiles using the software PRIMER version 5. Hierarchical cluster analysis was conducted to assess the degree of similarity between symbionts and tissue of their coral hosts. FA profiles were compared among the five coral species by non-metric multidimensional scaling (nMDS) and analysis of similarity (ANOSIM) on the Bray–Curtis similarity index. The contribution of each fatty acid to these clusters was determined by similarity percent (SIMPER) analysis. A principal component analysis (PCA) was performed among coral species to highlight certain FA profiles.

Results

Total lipid and fatty acids of five different coral species

Overall, TL in symbionts showed to be much higher than in host tissue of the five coral species (average 30.7 ± 4.2% in symbionts, 13.5 ± 3.0% in coral host tissue) (Fig. 1). TL of symbionts in P. carnosa showed the highest value (36.1 ± 5.5%) compared to four other species, but it was not significantly different among the species (p > 0.05). There was no significant difference in TL of coral host tissue between A. samoensis (8.9 ± 1.4%) and P. decussata (12.4 ± 3.9%). TL in host tissue of A. samoensis was significantly lower compared to T. peltata (p < 0.05), F. abdita (p < 0.005), and P. carnosa (p < 0.005). There were no significant differences between P. decussata, T. peltata, F. abdita, and P. carnosa.

Fig. 1
figure 1

Total lipid content (%) in symbionts and their host tissue from five different coral species. Statistically significant differences (p < 0.05) in each symbiont and coral host tissue are indicated by asterisk (*): * indicates p < 0.05, ** indicates p < 0.005, ns indicates no significance (one-way ANOVA, Tukey’s test) between species

Full size image

Out of the 37 FAMEs, we evaluated 26 types of long chain FAs (> 16C) where a total of 19 were detected (Table 1). The most predominant FA in both symbionts and coral host tissue for all species was 16:0. The second most dominant SFA in both symbionts and coral host tissue for all species was 18:0. The 16:0 in symbionts showed to be higher than the host tissue except for those of P. carnosa, whereas 18:0 constantly showed to be higher in the host tissue than in symbionts. The major MUFA for all species was 18:1n-9, followed by 16:1. The highest concentration of 18:1n-9 was found in both symbionts and the host tissue of F. abdita. A. samoensis was characterized by the highest concentration of total PUFA in both symbionts and the host tissue, 142.0 ± 38.0 μg mg−1 and 58.1 ± 11.0 μg mg−1, respectively (Table 1). However, P. decussata was characterized by the lowest value in both symbionts and the host tissue, 49.4 ± 13.7 μg mg−1 and 17.4 ± 1.4 μg mg−1, respectively. In general, the major PUFAs were 18:4n-3, 20:5n-3, 22:6n-3, 18:3n-6, and 20:4n-6. PUFAs namely 18:4n-3, 20:5n-3, 22:6n-3, and 18:3n-6 in symbionts for all species showed higher concentrations than the coral host tissue. The highest concentrations of PUFAs (18:4n-3, 20:5n-3, 22:6n-3, 18:3n-6) except for 20:4n-6 in the host tissue were observed in A. samoensis. However, the lowest value of 18:4n-3 was found in F. abdita (0.3 ± 0.1 μg mg−1), for 20:5n-3 in P. carnosa (0.6 ± 0.2 μg mg−1), and for 22:6n-3 and 18:3n-6 in T. peltata (1.5 ± 0.2 μg mg−1, 0.8 ± 0.1 μg mg−1, respectively). The concentration of 18:4n-3, 20:5n-3, and 22:6n-3 in symbionts of A. samoensis showed a much higher value (23.4 ± 7.3 μg mg−1, 30.9 ± 22.0 μg mg−1, and 44.1 ± 16.7 μg mg−1, respectively) compared to the symbionts of other species (average 7.1 ± 2.2 μg mg−1, 9.9 ± 2.1 μg mg−1, and 18.4 ± 3.2 μg mg−1, respectively). In addition, 18:3n-3 in symbionts of A. samoensis showed to be the highest (16.2 ± 3.7 μg mg−1) whereas it was not detectable in P. decussata and T. peltata. The concentration of 20:4n-6 in the host tissue showed higher levels than 20:5n-3 and 22:6n-3 among all species. In particular, 20:4n-6 in the host tissue of T. peltata, F. abdita, and P. carnosa had the highest concentration among all PUFAs, where P. carnosa had the highest level among all species.

Table 1 Fatty acid profiles in symbionts and their host tissue from five different coral species
Full size table

In this study, we applied FA ratios namely 20:5n-3 to 22:6n-3 (EPA:DHA) and PUFA:SFA as biomarkers for coral feeding or determining trophic level. The ratio is an indicator for trophic level (higher value indicates relatively autotrophic, lower value indicates relatively heterotrophic feeding strategies) (Rocker et al. 2019). The ratio of EPA:DHA was significantly higher (one-way ANOVA, Tukey’s test, p < 0.05) in T. peltata and A. samoensis (0.9 ± 0.1, 0.8 ± 0.4, respectively) compared to the other three species (Table 1). PUFA:SFA has been suggested as indicator for carnivory (higher value indicates relatively carnivorous, lower value indicates relatively herbivorous). PUFA:SFA showed the highest value (one-way ANOVA, Tukey’s test, p < 0.05) in A. samoensis (0.9 ± 0.2), followed by P. carnosa (0.7 ± 0.1), F. abdita (0.5 ± 0.1), T. peltata (0.4 ± 0.1) while P. decussata had the lowest value (0.3 ± 0.03).

Multivariate analysis of coral host and symbiont fatty acid profiles

Cluster analysis was conducted on 19 different FAs (as noted in Table 1) to evaluate whether FA profiles from the symbionts and their coral host tissue are statistically separate. As shown, two groups were formed at 75% similarity level (Fig. 2) where Group I is composed of all coral host tissues except for host tissue from A. samoensis, whereas Group II is composed of the symbionts from the five species and including the host tissue of A. samoensis.

Fig. 2
figure 2

Hierarchical cluster analysis based on a Bray–Curtis similarity matrix of fatty acid profiles (μg mg−1) from five different symbiotic coral species (n = 4): ACR Acropora samoensis, PAV Pavona decussata, TUR Turbinaria peltata, FAV Favites abdita, PLA Platygyra carnosa. (S) indicates symbionts, (H) indicates coral host tissue. Group I and Group II derived at 75% similarity level

Full size image

The nMDS plot based on 19 FA profiles clearly revealed a distinct grouping among the five different species. ANOSIM analysis also confirmed a clear separation among the groups (R = 0.999; p < 0.05) (Fig. 3). SIMPER analysis showed that each group had high similarity (> 90%) within the group (Table S4). In all species groups, two main FAs namely 16:0 and 18:0 explained 25.1–39.8% of the group similarity. In the T. peltata, F. abdita, and P. carnosa group, 20:4n-6 explained 13.4%, 10.8%, and 13.5%, respectively, of the group similarity. While in the A. samoensis group, 18:3n-3 explained 9.5% of the group similarity (Table S5). Dissimilarity between A. samoensis and T. peltata showed to be the highest (30.3%), which was explained by the contribution of the following FA, in decreasing order of importance: 18:3n-3, 20:1n-9, 16:0, 18:3n-6, 22:6n-3 (Table S5). Moreover, the dissimilarity between F. abdita and P. carnosa (same family: Faviidae) was the lowest (12.6%), which is explained by the contribution of the following FAs, in decreasing order of importance: 22:5n-3, 20:4n-6, 18:3n-3, 18:1n-9.

Fig. 3
figure 3

Non-metric multidimensional scaling (nMDS) plot of fatty acid profiles of host tissue from five different symbiotic coral species: ACR Acropora samoensis (Black up-point triangle), PAV Pavona decussata (Black down-point triangle), TUR Turbinaria peltata (black circle), FAV Favites abdita (black diamond), and PLA Platygyra carnosa (black square). (H) indicates coral host tissue. Each symbol indicates one coral species. Dotted lines (--) indicate homogenous groups as determined by ANOSIM (R = 0.998, P < 0.05)

Full size image

In PCA, PC1, PC2, and PC3 accounted for 48.5%, 17.7% and 13.2%, respectively (only PC1 and PC2 are shown in Fig. 4). PC1 separated A. samoensis and P. decussata (positive scores) from the other species (negative scores), explaining 48.5% of the variability between the FA profiles of all species. The main FAs driving this distinction for A. samoensis were 18:3n-3, 20:5n-3, 22:6n-3, 20:3n-6, and 20:1n-9 (Fig. 4). A. samoensis had the highest PC1 scores, whereas T. peltata had the lowest. In contrast, PC2 separated P. decussata (negative scores) from the other species (positive scores), explaining 17.7% of the variation in FA profiles. In particular, P. decussata was characterized by 18:2n-6, F. abdita and P. carnosa were characterized by 20:4n-6.

Fig. 4
figure 4

Principal component analysis (PCA) of fatty acid profiles from five different coral species

Full size image

Discussion

Our results showed TL in holobionts (unity of symbionts and coral host tissue) ranged from 19 to 26% in average. Indeed, a similar trend was found in Okinawan corals (14–37%), Caribbean corals (12–32%) and Red Sea corals (12–32%) (Harland et al. 1993; Yamashiro et al. 1999). TL in symbionts separated from coral holobionts were much higher (27–36%) than in coral host tissue (9–16%). Culture experiments on five different marine dinoflagellate taxa have shown that they store 6–16% lipid, in particular Symbiodinium microdiriaticum, which is one of the symbiotic dinoflagellates associated with corals, contained 15% TL (Mansour et al. 1999). Therefore, it is likely that TL contents in the symbionts and in their coral host separately show distinct different values. However, insofar there are no such studies comparing TL between symbionts and their coral host tissue. TL in corals indicates contribution of energy storage and nutritional status (Harland et al. 1992; Imbs et al. 2010b; Imbs 2013). In addition, symbionts can translocate lipids to their coral host to maintain physical fitness (Papina et al. 2003; Treignier et al. 2009; Imbs et al. 2014). However, there was no significant variation found in TL of symbionts separated from the five target species, which indicates that symbionts are limited in their ability to store a certain quantity of lipid; this depends on the given environmental condition regardless of the coral host species. According to previous study, it is known that the same Symbiodiniaceae genus (C1) is hosted in all of our target coral species (Wong et al. 2016), which further supports that the environment is the driving variation in TL of symbionts across coral host species. Thus, we hypothesized that lipid production might not be different in symbionts from either coral species. However, TL in the host tissue of F. abdita and P. carnosa were significantly higher than in the host tissue of other species, which can be explained by the feeding strategy of these two species, i.e. they may feed more than other species. In fact, fed corals showed to have higher lipid contents than unfed corals (Al-Moghrabi et al. 1995; Treignier et al. 2008). On the other hand, TL in A. samoensis has shown to be significantly lower than TL in T. peltata, F. abdita, P. carnosa, which reversely may indicate A. samoensis relies on relatively less feeding.

Hierarchical cluster analysis confirmed that FA profiles and their abundance in coral host tissue and symbionts are significantly different, with the exception of A. samoensis host tissue, which is categorized in the group with symbionts. Indeed, FA profiles in A. samoensis host showed subtle differences compared to its symbionts. Our results have shown that both symbionts as well as the host tissue of A. samoensis have considerably higher concentrations of PUFA when compared to other species. Other studies have shown that among symbiotic corals, PUFAs were the major FAs transferred from symbionts to the coral host tissue (Al-Moghrabi et al. 1995; Papina et al. 2003). Although SFAs can be translocated from symbionts to coral host tissue, the variation in concentration of both tissues from the five species was not obvious because corals can synthesize SFAs by themselves, masking a clear distinction (Treignier et al. 2009; Revel et al. 2016). On the other hand, higher PUFA concentrations in the host of A. samoensis compared to other species could also mean higher feeding rates on external food sources. However, the Acropora species has been found to be relatively autotrophic in studies assessing FA profiles and stable isotopic values (Seemann 2013; Seemann et al. 2013; Conti-Jerpe et al. 2020). In addition, higher concentrations of 18:4n-3 and 18:3n-6 in this species further indicates a stronger reliance on symbionts as these FAs are known to be biomarkers for the presence of symbionts within symbiotic coral species (Imbs et al. 2010b; Imbs 2013). Thus, we suggest that A. samoensis relies relatively more on FA from symbionts rather than on external feeding.

Although our target species are associated with the same genus of symbionts (Wong et al. 2016), higher concentration of major PUFAs in symbionts of A. samoensis when compared to the other species could potentially be explained by different physiology that depends on the species of coral host due to specific coral morphology (i.e. growth rate, tissue thickness, polyp size) and the absence of disturbances of light penetration (i.e. particulate matter or sedimentation) in our coral tanks, which is beneficial to autotrophic species. Theoretically, symbionts in A. samoensis perform the best biochemistry via photosynthesis, where surplus FA can be transferred to the host tissue. In fact, the concentration of each FA in symbionts and the host tissue alike was very similar to each other in this species. Therefore, we confirm that A. samoensis is the most autotrophic species among the five target species. However, the PUFA:SFA ratio, which is an indicator of carnivory, was the highest in A. samoensis, which contrasts our above interpretation. Although the highest PUFA was observed in A. samoensis, SFA did not vary within the five species. This indicates that PUFA is the determinant for the high value of PUFA:SFA ratio in A. samoensis which in turn, is responsible for the contradicting indications on trophic level. Therefore, applying FA ratios to determine trophic level in corals has its limitations due to the mixture of possible food sources.

The diversity of FA profiles exhibited by our five target coral species is in accordance with previous coral FA studies (Imbs et al. 2007, 2010b, 2014). These studies confirmed that hard corals can be distinguished on family or subclass level on the basis of PUFA profiles. In this study, we took this even further and managed to clearly separate all our five corals on genus level by multivariate analyses using 19 FA profiles including not only PUFAs but also SFAs and MUFAs. However, we should be careful when interpreting the data since FA profiles in corals can be modulated by external nutritional input (Parrish et al. 2000; Dalsgaard et al. 2003; Alfaro et al. 2006) as well as by environmental factors, such as season, site, depth, and temperature (Imbs 2013; Seemann et al. 2013; Rocker et al. 2019). Furthermore, our coral samples were collected in flow mesocosm tanks, where sea water comes from the marine protected area of Hong Kong after being primarily filtered through a sand filtration system to remove suspended matter and sedimentation. This implies, coral colonies are exposed to more stable environmental conditions when compared to wild corals in the bay. Regarding season, we sampled in May that falls within the transition period between dry and wet season when it is climatically less extreme. In addition, the average water temperature was 25.7 °C, which is close to the all-year-average water temperature (26.0 °C in 2016). In contrast, Imbs et al. (2007) collected their coral samples in the shallow waters of Vietnam in the South China Sea. Despite the same sampling region, season, depth etc., FAs in corals from natural habitats can be influenced by a variety of food sources (Imbs et al. 2010b). Therefore, we believe that the consistent and stable environmental condition in our coral tanks was the key to distinguish distinct groups among the five coral species.

The five distinct species groups are associated to 16:0 and 18:0 which are known to be transferred from glucose, symbionts and dietary sources (Imbs et al. 2011; Revel et al. 2016). Numerous studies on FA profiles of hard corals have also documented that these FAs are the most abundant FAs since their first report by Meyers (1977) (Latyshev et al. 1991; Harland et al. 1993; Yamashiro et al. 1999; Imbs et al. 2007, 2010b; Imbs 2013). The 16:0 in host cnidarian tissue can be de novo synthesized from symbiont photosynthesis-driven glucose (Oku et al. 2003a; Revel et al. 2016), while a high 16:0 in symbionts is originated from their photosynthesis products. In P. decussata, 16:0 in both symbionts and their host tissue showed the highest concentration, suggesting 16:0 in this species relies on symbionts. However, 16:0 in host tissue can be originated from both symbionts and dietary sources, such as zooplankton which is known to have abundant 16:0 (Lee et al. 2006). An elevated concentration of 16:0 in the P. carnosa host tissue than in its symbionts, as opposed to the other four species, suggests that P. carnosa obtains additional 16:0 from external food sources. Upon obtaining 16:0, the biosynthesis pathway 16:0 → 18:0 → 18:1n-9 can be consistently metabolized in both symbionts and their coral host tissue (Treignier et al. 2009; Revel et al. 2016). However, the coral host cannot synthesize further from 18:1n-9 to 18:2n-6, which is the precursor of n-6 PUFA pathway, due to the lack of ∆12-desaturase enzyme (Dunn et al. 2012; Matthews et al. 2018). Therefore, we assume that the coral host stores energy in the form of 16:0 or 18:0 in lipids, but it is species-specific and depends on the nutrition.

Although, F. abdita and P. carnosa belong to the same family (Faviidae), their FA profiles differed. The dissimilarity between FA profiles of F. abdita and P. carnosa was the lowest (12.6%), which indicate that these two species have comparable FA profiles relative to any other combination of species. Both species were also characterized by the highest concentration of 20:4n-6, and further exhibited the highest concentration among all the PUFAs. Unfortunately, there are very few studies on 20:4n-6 in corals. Rocker et al. (2019) documented that symbiont density had a negative correlation with 20:4n-6 concentration, which implies that this FA in the coral host tissue might not be derived from symbionts, hence 20:4n-6 suggests to be a potential indicator of coral feeding. Accordingly, a few studies surmised that 20:4n-6 in corals might originate from external food sources e.g. phytoplankton and/or herbivorous zooplankton (Seemann et al. 2013; Imbs et al. 2016; Kim et al. 2021). Indeed, high 20:4n-6 content in phytoplankton has been documented (Jónasdóttir 2019) Subsequently, higher 20:4n-6 in P. carnosa, F. abdita and T. peltata than in other species show that these species are more heterotrophic. In addition, P. decussata is characterized by a negative relationship with 20:4n-6, which confirms the relative autotrophy of this species.

On the other hand, the most dissimilar species coupling with P. carnosa and/or F. abdita was A. samoensis. This species was determined by the predominance of 16:0 and 18:0, followed by 18:3n-3. Especially, 18:3n-3 in A. samoensis was the most abundant of all the PUFA, and it should be noted 18:3n-3 plays a key role in further synthesizing essential n-3 PUFA. However, this FA cannot be biochemically synthesized by coral hosts due to their lack of ∆15-desaturase enzyme, indicating that 18:3n-3 originates from symbionts via photosynthesis and subsequently transferred to the coral host. Moreover, the high concentration of 18:3n-3 in the host tissue of A. samoensis indicates that this species is heavily dependent on autotrophy through its symbionts. The highest dissimilarity (30.3%) of FA profiles among the five coral species was in between A. samoensis and T. peltata, and 18:3n-3 contributed to this separation the most. In contrast to the high concentration of 18:3n-3 in A. samoensis, no 18:3n-3 could effectively be detected in T. peltata. In previous studies, on a different species of the same genus, T. reniformis, no 18:3n-3 was observed in the representatives under starving conditions, but the concentration was high in corals fed with zooplankton (Artemia salina nauplii) (Tolosa et al. 2011). Although we did not quantify the extent of food availability in our coral tanks, a lack of zooplankton is possible. However, T. peltata showed a high concentration of 20:4n-6, which is an indicator of heterotrophic feeding (mainly phytoplankton) as previously discussed. This incompatible result is due to the abundance of different food sources (here, zooplankton vs. phytoplankton), which causes a variation of 18:3n-3 and 20:4n-6 in this species. On the other hand, symbionts of T. peltata showed the highest concentration of 18:3n-3 among all five species indicating a low translocation rate from symbionts to the host tissue in this species and might be responsible for our finding, giving the unique characteristic of T. peltata. Moreover, exceptional values of 18:3n-3 due to species-specific characteristics have been reported by Papina et al. (2003). They observed low 18:3n-3 in Montipora digitata and postulated species-specific differences to be responsible.

Besides a high concentration of 18:3n-3, A. samoensis was also characterized by high concentrations of 20:5n-3 and 22:6n-3. These FAs are known to be essential PUFAs in marine organisms, including corals, as they act as functional FAs for growth, reproduction, and physical fitness (Wacker and von Elert 2001; Pernet et al. 2002; Figueiredo et al. 2012). It is known 22:6n-3 is highly conserved through the food web (Scott et al. 2002; Dalsgaard et al. 2003), whereas 20:5n-3 is the dominant FA in symbionts and transferred to the coral host tissue (Revel et al. 2016). Thus, the ratio of 20:5n-3 to 22:6n-3 (EPA:DHA) has been applied to define trophic level (low value means heterotrophy, high value means autotrophy) (Rocker et al. 2019). According to the EPA:DHA ratio, A. samoensis and T. peltata are defined to be relatively autotrophic, whereas P. decussata, F. abdita, and P. carnosa are relatively heterotrophic species. This ratio confirms our previous conclusions drawn from FA profiles that A. samoensis is autotrophic, F. abdita and P. carnosa are heterotrophic. However, this ratio contrasts with P. ducussata and T. peltata when compared to our findings, as well as to investigation from a previous study (Treignier et al. 2008). We may explain this controversy with the fact that corals are polytrophic, thus they can switch their trophic strategies depending on specific environmental conditions such as food sources and amount or symbiont density influenced by water quality (Dalsgaard et al. 2003; Seemann et al. 2013; Rocker et al. 2019). The most plausible reason is that not only 20:5n-3, but also 22:6n-3 in the coral host tissue can be either transferred from symbionts (Revel et al. 2016) or taken up externally through planktonic food sources (20:5n-3 enriched in diatom, 22:6n-3 enriched in dinoflagellate) (Dalsgaard et al. 2003; Figueiredo et al. 2012; Revel et al. 2016). Although we did not measure FA compositions in our mesocosm tanks, a recent study by Kim et al. (2021) showed that both FAs are enriched in zooplankton from HK waters.

In summary, TL and FA profiles proved to be a powerful tool as chemotaxonomic biomarkers at genus level of symbiotic corals, especially when the host tissue and symbionts are investigated separately. This confirms our first hypothesis, FA profiles being species-specific, enabling a classification of symbiotic corals when applying coral host tissue from symbionts separately. In addition, our study provides new insights into the interaction of coral-symbionts endosymbiosis and species-specific trophic strategies. We hypothesized that the trophic level of symbiotic corals can be deduced from specific biomarkers. However, we realized that interpretation based on the accepted and widely used FA ratio indicators: EPA:DHA or PUFA:SFA to define specific trophic levels needs to be handled with more caution in research application and data interpretation. Moreover, 20:5n-3/22:6n-3n ratio and PUFA/SFA ratio themselves should not be mistaken for universal indices. Since FA in coral host tissue can be driven by different sources such as symbionts and external food sources, interpretations may not be as straightforward as currently applied. Hence, our second hypothesis is true only when rephrased as certain FA biomarkers giving indications on the relative trophic level of symbiotic corals. Given these limitations of FA ratios, we suggest that it to be more appropriate if these FA ratios were only applied to examine response to environmental change within species. Going forward, our study provides important FA baseline data that builds the foundation to future investigation on impacts of environmental changes on nutrition and metabolism in symbiotic corals.