Introduction

Coral reefs support high biodiversity (Knowlton et al. 2010) and provide a range of ecosystem services (Moberg and Folke 1999; Woodhead et al. 2019), so it is important to understand how coral biomineralisation will respond to expected future environmental change. Coral skeletons are composite materials of aragonite and biomolecules (e.g. lipids, proteins, polysaccharides) which are embedded within the aragonite crystallites (Cuif et al. 1997, 2005; Dauphin et al. 2006; Drake et al. 2020). The skeleton can form from an amorphous CaCO3 (ACC) precursor produced intracellularly in the layer of epidermal coral cells adjacent to the skeleton (Mass et al. 2017; Sun et al. 2020). In addition, skeletal growth may occur by direct attachment of Ca2+ and CO32− ions epitaxially to the crystallites in the existing skeleton at an extracellular calcification site, sandwiched between the skeleton and the base of the polyp, which is at least, partly isolated from seawater (Sun et al. 2020).

The skeletal biomolecules are produced by the coral holobiont (Falini et al. 2015) and synthesised by the coral calicoblastic cells, the bottom tissue layer at the base of the organism (Puverel et al. 2007). The organic matrix contains proteins which can be rich in acidic amino acids (aspartic acid and glutamic acid) and which have Ca2+ binding properties (Puverel et al. 2005). This organic matrix is inferred to exert control over the biomineralisation process as biomolecules affect CaCO3 nucleation (Picker et al. 2012), CaCO3 polymorph (Falini et al. 2013), precipitation rate (Kellock et al. 2020, 2022), crystal morphology (Tong et al. 2004; Reggi et al. 2016) and material properties (Kim et al. 2016).

Increased seawater pCO2 can influence coral skeleton morphology by increasing or reducing calyx size (Tambutté et al. 2015; Allison et al. 2022, respectively), by increasing or reducing the coenosteum area which connects adjacent corallites (Allison et al. 2022; Scucchia et al. 2021, respectively) or by altering the corallite morphology (Scucchia et al. 2021; Allison et al. 2022). It is unclear how these morphological changes are mediated, but some observations are suggestive. The organic content (Coronado et al. 2019), protein content (Tambutté et al. 2015) and amino acid content (Kellock et al. 2020) of skeletons are increased in tropical corals cultured at high seawater pCO2 (an ocean acidification scenario).

Massive Porites spp. corals are an important reef building coral species in the Indo-Pacific Ocean (Veron 1993). The most prevalent amino acid in the Porites spp. coral skeletal organic matrix is aspartic acid (Cuif et al. 1999; Kellock et al. 2020). This biomolecule inhibits aragonite precipitation at the aspartic acid concentrations estimated at the coral calcification site (Kellock et al. 2020), and the extent of inhibition is affected by the seawater saturation state (Ωaragonite). Observations of fluorescent dyes at the external coral calcification site indicate that the pH of the calcification media is elevated above that of seawater (Venn et al. 2011, 2013, 2019) and decreases at high seawater pCO2 (Venn et al. 2013, 2019, 2022). This likely results in a reduction in calcification media Ωaragonite which could affect the influence of the biomolecules at the calcification site.

The present study explores the effects of seawater pCO2 and temperature on the amino acid composition and aragonite lattice disorder of massive Porites spp. skeletons. We analyse the skeletal amino acids of a suite of corals cultured at 25 °C and 28 °C and at seawater pCO2 of 180, 400 and 750 µatm (Cole et al. 2018), respectively, representing approximate CO2 concentrations under glacial, early twenty-first century and projected future conditions (Gattuso and Lavigne 2009). Prior to culturing, the corals were collected from a reef in Fiji where seasonal seawater temperatures typically range from ~ 26 to 29 °C (Wu et al. 2013) and the upper temperature during culturing is likely to be below the thermal stress threshold for these specimens. We use Raman spectroscopy to analyse the coral skeletons and additional specimens cultured at the same temperatures but at seawater pCO2 260 µatm (representing pre-industrial CO2 concentrations) and an additional set of massive Porites spp. corals cultured at varying seawater pCO2 and analysed previously for skeletal amino acids (Kellock et al. 2020). Raman spectroscopy has been used to identify the CaCO3 polymorph (Nehrke and Nouet 2011), to explore CO3 disorder in the CaCO3 lattice (Nehrke and Nouet 2011; Kamenos et al. 2013; DeCarlo et al. 2017), map the distribution of skeletal organic material (Nehrke and Nouet 2011; DeCarlo et al. 2018) and infer Mg content in biogenic carbonate (Kamenos et al. 2013; DeCarlo et al. 2017). We compare the organic contents and CO3 disorder in the different coral skeletons to identify how seawater pCO2 and temperature influence these aspects of the coral biomineralisation process. We also analyse the Raman spectra of a suite of synthetic aragonite precipitated over variable temperature and Ω to identify how these parameters influence disorder in a similar but non-biological system. We compare these to the spectra from the coral samples to yield insights into the environmental factors controlling coral skeleton Raman signature.

Methods

Coral samples

Coral culturing

Full details of the coral culturing are provided in Cole et al. (2018) and the supplementary material. In this research, we study the skeletons of massive Porites spp. corals cultured over seawater pCO2 and temperature by Cole et al. (2018) (reported as experiment 2). In brief, 4 individual coral heads (defined as separate genotypes on the basis that they were collected from spatially separate colonies) were imported into the UK. Each head was divided into sub-colonies (~ 12 cm in diameter) and at least one piece of each was cultured at 28 °C and ~ 180, 260, 400 and 750 seawater pCO2. Seawater pCO2 was gradually adjusted over a 1-month period and then corals were maintained at these conditions for 4 months and then stained with alizarin red S to create a marker in the skeleton. Calcification (along with photosynthesis and respiration) of each sub-colony was measured 3–4 times over the next 5 weeks (the experimental period) while the corals were maintained at constant conditions. Thereafter the corals were stained again and the seawater temperature was reduced to 25 °C over 4 weeks. The corals were acclimated at this temperature for 4 weeks, stained again and then physiological rates were measured over another 5-week experimental period. Physiological rates are reported in Cole et al. (2018).

At the end of the study the corals were sacrificed, and the skeletons cleaned by submergence in 3–4% sodium hypochlorite for ≥ 24 h with intermittent agitation. The skeletons were rinsed and dried and sawn into slices. The skeleton deposited within each of the 25 and 28 °C experimental periods was sampled from the maximum growth axis using a microdrill with a bit of diameter 100 µm. The skeleton was crushed to a powder with an agate mortar and pestle. Aliquots of the drilled coral powders were analysed for amino acid content and Raman analysis. Seawater [PO43−] in the first week of the 5-week experimental period in the 400 µatm pCO2 treatment at 28 °C was below the limit of detection (0.00 µM, Cole et al. 2021) and lower than observed in typical reef sites (Szmant 2002). The culture seawater for this treatment was discarded and replaced with seawater sourced from the remaining 3 reservoirs bubbled to bring it to seawater pCO2 of 400 µatm before use. The geochemistry of these specimens was therefore only analysed in the skeleton deposited at 25 °C.

Amino acid analysis

Amino acid analysis was conducted on the skeletons of 3 of the 4 coral genotypes (termed G4, G6 and G7 in Cole et al. (2018)) and in individuals cultured at 180, 400 and 750 µatm at 25 °C and at 180 and 750 µatm at 28 °C following the method of Tomiak et al. 2013. Skeletal samples were subject to an extensive cleaning programme to remove any remnants of the coral tissue and endolithic organisms, e.g. algae (Le Campion-Alsumard et al. 1995), and to leave only intra-crystalline material.

All reagents are analytical grade unless otherwise reported. In brief, < 20 mg of powdered skeletal sample (< 100 μm) was accurately weighed into a plastic microcentrifuge tube and bleached using 50 μL 12% NaOCl mg−1 for 48 h, followed by the removal of the bleach and sequential rinsing with 18.2 MOhm H2O and methanol. Samples were dried overnight. To hydrolyse the peptide bonds in the samples, < 10 mg was accurately weighed into a 2-mL sterile glass vial (Wheaton) and 20 μL mg−1 7 M HCl was added. After a flush with nitrogen, the vials were heated at 110 °C for 24 h. Upon removal, samples were dried in a centrifugal evaporator overnight. Samples were rehydrated and analysed using reverse-phase HPLC with fluorescence detection, following a modified method of Kaufman and Manley (1998). This enables quantification of L and D isomers of 12 amino acids. As asparagine and glutamine undergo deamination during the preparation process, they will contribute to observed concentrations of aspartic acid and glutamic acid. However, proteomics suggests that the contributions of asparagine and glutamine to coral skeletal proteins are small. For example, 56 aspartic acid and 5 asparagine residues occur in an acidic amino acid-rich protein (CARP3) isolated from a coral skeleton (Mass et al. 2013). Similarly, 24 glutamic acid and 1 glutamine residues occur in the same protein (Mass et al. 2013).

We report aspartic acid/asparagine as Asx and glutamic acid/glutamine as Glx. Other amino acids are abbreviated as follows: serine (Ser), L-threonine (L-Thr), glycine (Gly), L-arginine (L-Arg), alanine (Ala), valine (Val), phenylalanine (Phe), leucine (Leu) and isoleucine (Ile). All samples were run in duplicate alongside standards and blanks. Analyses of replicate drilled coral powders (n = 5) indicate that the standard deviation (1σ) of repeat analyses is ± 125, 34, 51 and 13 pmol mg−1 for total amino acid, Asx, Gly and Glx, respectively, equivalent to coefficients of variation of 8%, 4%, 20% and 7%, respectively (Kellock et al 2020). For the remaining amino acids, standard deviation (1σ) of [amino acid] of replicate analyses was always < 12 pmol mg−1. The contribution of each amino acid to the total amino acid was calculated as [amino acid]/[total amino acid], with both quantities in pmol mg−1. Inclusion of the stain did not affect the mol% of different amino acid groups (Kellock et al. 2020).

Raman spectroscopy

Raman spectroscopy was used to analyse all the coral skeletons cultured at 25 °C (4 coral genotypes: G4, G5, G6 and G7 × 4 pCO2 treatments), those cultured at 180, 260 and 750 µatm at 28 °C and all the synthetic aragonite precipitates. Raman spectra were also collected on 11 additional Porites spp. skeletons, cultured in our laboratory at 25 °C and 180, 400 and 750 µatm seawater pCO2 (Cole et al. 2018 reported as experiment 1 with coral genotypes defined as G1, G2 and G3). Two sub-colonies of coral genotype G3 were cultured at 400 and 750 µatm. These are defined as replicates 1 and 2 in each case.

Raman spectra were collected with a Renishaw In-Via Qontor Raman Microscope using a NIR 300 mW 785 nm solid state laser. Spectra were collected in static mode from 100 to 1311 cm−1 with the laser set at 5% full power with a 1200 cm−1 grating and a spectral resolution of ~ 0.3 cm−1. The laser spot was focused onto a particle of sample powder and data were collected for 2 s for each acquisition. 10 acquisitions were repeated on each spot and summed to yield the final spectrum. Twelve to fifteen spots were analysed on each sample powder. All the Raman spectra had lattice mode peaks at ~ 153 and 206 cm−1 (DeCarlo et al. 2018) and a doublet v4 peak (Urmos et al. 1991), indicative of aragonite. The spectra were processed using OriginLabs peak processing software and the peak centre and the full width half maxima (FWHM) of the v1 peak were estimated following a Voigt fit (Meier 2005). This ν1 band results from symmetric C–O stretching in the planar carbonate group (Bischoff et al. 1985) and broadening of this peak is indicative of increased local disorder around the CO32− group. Measured FWHM were corrected to true FWHM using the instrument spectral resolution (Nasdala et al. 2001).

We analysed a milled synthetic aragonite powder as an in-house reference material at frequent intervals during each day of Raman analyses. This aragonite was precipitated at pHNBS 8.445 and Ω = 11 and ground in an agate ball mill with water. The peak centre of this reference material drifted by > 0.7 cm−1 over the first 2 h of the session (Fig. S1a) but the FWHM of the peak was stable (Fig. S1b). After this 2-h drift period, the peak centre population was normally distributed (Shapiro–Wilk test for normality, p = 0.68, n = 50). The FWHM population of this reference material was normally distributed over the entire session (p = 0.98, n = 100).

Particle size can influence Raman spectroscopy (Gómez et al. 2019). To test the effect of particle size on the FWHM of the aragonite v1 peak, we compared the Raman spectra of the milled aragonite particles with that of the same aragonite sample before milling. The aragonite particles produced during the laboratory precipitation had approximate diameters of ~ 70–120 µm (Fig. S2a) while the milled aragonite had dimensions of ~ 3–5 µm (Fig. S2b). We collected Raman spectra of both the ground and unground aragonite particles and found that v1 FWHM in the ground aragonite (4.52 ± 0.07, 1 s, n = 14) was not significantly different from that of the unground aragonite (4.54 ± 0.06, 1 s, n = 13) as tested by t test for equal means (p = 0.37). The coral skeletal samples measured by Raman had dimensions within the range represented by the milled and unmilled reference material so we do not consider particle size influences FWHM in this study.

Corals were stained with alizarin red S while living to create temporal markers in the skeleton (see Cole et al. 2021 for images of a stained skeleton). To test if the stain influenced the Raman spectrum, we drilled a coral skeleton powder along a stain line and compared it to the skeleton deposited immediately underneath the stain. The mean FWHM of the v1 peak in the stained skeleton (FWHM = 4.08 ± 0.10 cm−1, 1 s, n = 12) was not significantly different from that of the unstained skeleton (FWHM = 4.06 ± 0.09 cm−1, 1 s, n = 15) as tested by t test for equal means (p = 0.55).

Preliminary spectra of the same powders were collected using a 532-nm laser with an 1800 grating in extended mode from 102–3202 cm−1 in an attempt to use the Raman instrument to identify variations in the fluorescence associated with skeletal organic matrix between skeletons (DeCarlo et al. 2018). These preliminary analyses demonstrate that fluorescence (mean signal intensity at 2400–2700 cm−1 normalised to the v1 peak) is about 3 × higher in the stained skeleton compared to the unstained skeleton (Fig. S3a). Furthermore, C–H bonding peaks are not visible in the Raman spectra in either the stained or unstained skeleton (Fig. S3b). Due to the high signal from the alizarin red S stain, we did not undertake further analyses of these samples using the 532-nm laser.

Synthetic aragonite

To investigate the effects of Ω and temperature on the peak centre and FWHM of the aragonite Raman spectrum v1 peak, we analysed synthetic aragonite precipitated over varying Ω and temperature using a Metrohm Titrando 902 titrator (Castillo Alvarez et al. 2024). We precipitated the aragonite from seawater in the laboratory using a pH stat titrator which dosed Na2CO3 and CaCl2 solutions to replace the ions removed during aragonite formation. To explore the effect of Ω, aragonite was precipitated at pHNBS 8.545, T = 25 °C and Ω = 4, 7, 10, 13 and 18. To explore the effect of temperature, aragonite was precipitated at Ω = 14 and T = 10, 15, 20, 25 and 30 °C. Measurements of the [CO32−] and pH of the extracellular calcification media in corals suggest the media has a Ωaragonite of about 12 (Sevilgen et al. 2019).

Results

The supplementary information contains all data for skeletal amino acid concentrations (Table S1), Raman data of coral skeletons and synthetic aragonites (Tables S2 and S3, respectively).

Amino acids in cultured coral skeletons

Asx is the most common amino acid in the coral skeletons, typically making up about half of the total amino acid content (Fig. 1). Glx, Gly and Ser typically make up ~ 11, 15 and 6%, respectively, of the total skeletal amino acid (Fig. 1). Histidine is rarely detected in the coral skeletons and we removed this amino acid from the analysis. Skeletal [amino acid] differs considerably between different coral genotypes cultured at the same conditions (Fig. 1).

Fig. 1
figure 1

Concentrations of total and key amino acids (p mol mg−1) in the skeletons of the 3 coral genotypes (G4, G6 and G7) cultured over varying seawater pCO2 and temperature. Typical standard deviations of repeat analysis of replicates of skeletal powders are shown by error bars (Kellock et al. 2020)

Full size image

We used a paired t test to test for variations in the amino acid concentrations between temperature treatments. For this test we matched the [amino acid] in the 25 °C and 28 °C regions of each skeleton (same genotype and same seawater pCO2) and pooled all the seawater pCO2 treatments. For example, we pair the total amino acid concentration for coral genotype 4 at 25 °C and 180 µatm with the total amino acid concentration for coral genotype 4 at 28 °C and 180 µatm. We create similar pairs for all the corals skeletons and then test for variations in skeletal [amino acid] at 25 °C and 28 °C. Similarly, to test for the influence of seawater pCO2 on skeletal composition, we matched the [amino acid] in the skeleton of each genotype cultured at 25 °C or 28 °C with that of the same genotype cultured at the same temperature but different pCO2 and used paired t tests to test for difference in [amino acid] between pCO2 treatments; for example, we pair the total amino acid concentration for coral genotype 4 at 25 °C and 180 µatm with the total amino acid concentration for coral genotype 4 at 25 °C and 750 µatm and the total amino acid concentration for coral genotype 4 at 28 °C and 180 µatm with the total amino acid concentration for coral genotype 4 at 28 °C and 750 µatm, etc.

[Asx], [Glx], [Gly], [Ser] and [total amino acids] are significantly lower at 25 °C compared to 28 °C, over all seawater pCO2 treatments (Table 1; Fig. 1). [Asx], [Glx], [Gly], [Ser], [Ala], [L-Thr] and [total amino acids] are significantly lower at 180 µatm compared to 400 µatm. [Ser] is also significantly reduced at 180 µatm compared to 750 µatm, but otherwise skeletal amino acid concentrations do not vary significantly between pCO2 treatments (Table 1).

Table 1 p values of paired t tests to test for differences in [amino acid] between seawater pCO2 and temperature treatments
Full size table

To test if the contribution of each amino acid to the total amino acid content of the skeleton is influenced by temperature or seawater pCO2, we calculate the molar proportions of each amino acid in the skeleton as [amino acid]/[total amino acid] with both quantities in pmol mg−1. We use paired t tests to test for differences between temperature and seawater pCO2 treatments as before. Variations in the relative contributions of the amino acid to total amino acid are almost always insignificant (Table 2), with the exception that the contribution of Glx to total amino acid is significantly higher at 28 °C compared to 25 °C, while the contribution of Ala to total amino acid is significantly lower at the higher temperature (Fig. 2).

Table 2 p values of paired t tests to test for differences in the proportional contribution of each amino acid to total amino acids between seawater pCO2 and temperature treatments
Full size table
Fig. 2
figure 2

Changes in the molar contribution of a Glx and b Ala to the total skeletal amino acids as a function of pCO2 and temperature. Contributions are calculated as [amino acid]/[total], with both amounts in pmol mg−1. Typical standard deviations of repeat analysis of replicates of skeletal powders are shown by error bars (Kellock et al. 2020)

Full size image

Skeletal amino acids, coral calcification rate and calcification media pH

To further explore controls on skeletal amino acids, we conduct multiple linear regression analyses modelling the skeletal amino acid concentrations as a function of coral calcification rate and coral calcification media pH. For this exercise the data from the present study are combined with skeletal amino acid data from a previous experiment, in which Porites spp. were cultured at 25 °C in our laboratory (Kellock et al. 2020). Calcification rates from the current study and the previous experiment are already published as experiment 2 and experiment 1, respectively, in Cole et al. (2018). Coral calcification media pH are derived from skeletal δ11B measurements of the same coral skeletons (Allison et al. 2018, 2021 for experiments 1 and 2, respectively). Calcification is significantly related to all detected skeletal amino acids and total amino acids but calcification media pH is not (Table 3).

Table 3 p values of multiple regression analysis models to predict concentrations of each coral skeletal amino acid from coral calcification rate and calcification media pH (pHCM) and ANCOVA p values to compare linear relationships between calcification rate and [amino acid] at 25 ºC and 28 ºC. p values ≤ 0.05 are highlighted in bold
Full size table

We use one-way ANCOVA tests to compare linear relationships between calcification rate and skeletal concentration of each amino acid at 25 °C and 28 °C (Table 3). We observe no significant differences in the relationships between calcification rate and concentrations of skeletal amino acids at the 2 temperatures, with the exception that [Glx] is offset to a higher concentration at comparable calcification rates at 28 °C compared to 25 °C (Fig. 3). These relationships are also illustrated for Asx and Ala in Fig. 3.

Fig. 3
figure 3

Relationships between skeletal a [Asx], b [Glx] and c [Ala] and calcification rate in corals cultured at 25 and 28 °C. At 25 °C, skeletal amino acid from this study is combined with data from a previous experiment in which Porites spp. were cultured at 25 °C in our laboratory (Kellock et al. 2020). Corals from all pCO2 treatments are pooled for this analysis. All calcification data are reported in Cole et al. (2018)

Full size image

Raman spectroscopy of coral skeletons

FWHM values for the carbonate ʋ1 Raman mode of the coral skeletons are shown in Fig. 4. To test the influence of seawater temperature on FWHM, we use a paired t test matching the FWHM of each individual coral at 25 °C and 28 °C for each pCO2 treatment. At 28 °C the FWHM are significantly narrower than at 25 °C (p = 0.015). To test for variations in FWHM between pCO2 treatments, we pool all coral genotypes and test the effect of pCO2 at each temperature separately by one-way ANOVA. We observe no significant effect of seawater pCO2 on FWHM at 25 °C or 28 °C (p = 0.469 and 0.324, respectively).

Fig. 4
figure 4

FWHM of the ʋ1 peak in the aragonite Raman spectrum of the corals cultured at various pCO2 and temperature. The typical standard deviation of multiple analyses (n = 12 to 15) of a single sample is shown

Full size image

Further we determine the FWHM of the ʋ1 peak for each coral as a function of coral calcification rate, pHCM and skeletal Asx (the most common amino acid in the skeleton, Fig. 5). None of these parameters is significant related to FWHM at 25 °C of 28 °C (Table 4).

Fig. 5
figure 5

The FWHM of the Raman spectrum ʋ1 peak for each coral as a function of a coral calcification rate (µmol CaCO3 cm−2 d−1), b pHCM (total scale) and c skeletal [Asx]

Full size image
Table 4 Coefficients of determination (r2) and p values for relationships between Raman aragonite spectra ʋ1 peak FWHM and coral parameters. All pCO2 treatments are pooled for this analysis
Full size table

Raman spectroscopy of synthetic aragonite

We observe positive relationships between Raman aragonite spectra FWHM and both Ωaragonite and water temperature in the synthetic aragonite precipitates (Fig. 6a, b). Temperature and Ωaragonite are positively related to aragonite precipitation rate in these experiments (Castillo Alvarez et al. 2024) and we plot aragonite FWHM as a function of precipitation rate for experiments conducted at variable Ωaragonite and at 25 °C or at variable temperature and Ωaragonite = 14 (Fig. 6c). One-way ANCOVA indicates that relationships between aragonite precipitation rate and FWHM do not vary significantly between the 2 datasets (p equal means = 0.38, p equal slopes = 0.96).

Fig. 6
figure 6

FWHM of the ʋ1 peak in the aragonite Raman spectrum of the synthetic aragonite precipitates cultured over variable a Ω and b temperature. In c data from experiments at variable Ω and temperature are combined onto one graph as a function of aragonite precipitation rate. The typical standard deviation of multiple analyses (n = 12 to 15) of a single sample is shown. Lines indicate best fit linear relationships and p values and coefficients of variations are shown. In c the line fits the entire dataset (experiments over a range of Ω and temperature)

Full size image

Discussion

Environmental change and coral skeletal amino acids

Asx, Glx and Gly are the most common skeletal amino acids as reported previously in Porites spp. and many other corals (Cuif et al. 1999; Puverel et al. 2005). In the present study, skeletal [Asx], [Glx], [Gly], [Ser] and [total amino acids] are significantly lower at 25 °C compared to 28 °C but the contribution of Asx, Gly and Ser to total skeletal amino acid is not altered significantly by temperature (Fig. 1; Table 1). Calcification increases by 66% on average at 28 °C compared to 25 °C in the 3 coral genotypes analysed in the present study (Cole et al. 2018) although we note that the effect of temperature on calcification rate is seawater pCO2 dependent (Cole et al. 2018). However, calcification rate is significantly inversely related to all detected skeletal amino acids and total amino acids (Table 3, illustrated for Asx, Glx and Ala in Fig. 3). Although this seems counterintuitive (given that skeletal [amino acids] increase at high temperature when calcification is usually higher), the inverse relationship between skeletal [amino acid] and calcification reflects the large variation in calcification rate between the coral genotypes. For example, genotype G7 contains low concentrations of skeletal [amino acid] (Fig. 1) but has high calcification rates compared to genotype G6, which has a lower calcification rate but higher skeletal [amino acid] (Supplementary data, Table 1). These data show that skeletal [amino acid] is significantly higher in skeletons from corals with relatively low calcification rates.

Our observation that skeletal [amino acid] usually increases at higher temperature when calcification rate also increases (on average), indicates that the skeletal [amino acid] does not reflect dilution of the biomolecules by the CaCO3 produced each day, i.e. resulting in high skeletal [amino acid] at low calcification rates (as hypothesised by Kellock et al. 2020). Rather, higher amounts of skeletal protein must be produced at higher temperatures. Although more skeleton is also precipitated at these higher temperatures, the mass of protein incorporated into the skeleton, as a function of skeletal mass, still increases.

Temperature does not affect the relative contribution of each amino acid to [total amino acid], with the exception that the proportion of Glx is lower at 25 °C compared to 28 °C, while the proportional contribution of Ala to total amino acid is higher at the low temperature (Table 2). These changes reflect alterations in the protein compositions of the skeletal organic matrix deposited at the 2 temperatures. Many studies on the proteomic responses of corals to temperature focus on the effects of heat stress (e.g. Mayfield et al. 2018; Stuhr et al. 2018), rather than on temperature increases below the stress threshold such as tested here (Cole et al. 2018). However, several studies indicate that the concentration and composition of coral tissue biomolecules is temperature dependent. For example, seasonal temperature variations altered the genomic response of Acropora millepora (Wuitchik et al. 2019), and temperature significantly influenced the physiological profile of Porites spp. corals living along an environmental gradient (McLachlan et al. 2021), including affecting tissue biomass and soluble protein and lipid content. Seasonal variations in coral tissue lipid content were positively correlated with seawater temperature in Goniastera aspera (Oku et al. 2003) and temperature significantly affected the tissue lipid compositions of the coral analysed in the present study (von Xylander et al. 2023).

The amino acids for coral protein synthesis can be sourced from seawater, from heterotrophy or are synthesised by the coral or by the algal symbionts (Grover et al. 2008; Ferrier-Pagès et al. 2021). Glutamic acid, glutamine, aspartic acid, and Ala are probably synthesised both by Porites australiensis and its symbionts (Shinzato et al. 2014). However, glutamic acid and Ala (along with Ser and Thr) compositions varied significantly between the skeletons of symbiont-bearing and non-symbiotic corals, such that glutamic acid is relatively enriched and Ala depleted in the skeletons of symbiotic corals (Cuif et al. 1999). This disparity may indicate that glutamic acid is predominantly sourced from the symbionts and Ala from the host. In support of this, glutamate in the sea anemone Aiptasia pulchella (a close coral relative) is predominantly synthesised in the symbionts (Swanson and Hoegh-Guldberg 1998). In the present study, the shift to higher skeletal [Glx] at higher temperatures (Fig. 2) may indicate an increased contribution of symbiont amino acids to the host amino acid pool, although we note that both net and gross photosynthesis are lower at 28 °C compared to 25 °C in these corals (Cole et al. 2018).

There are significant inverse relationships between coral calcification rate and skeletal [amino acid] in all the amino acids tested with the exception of His (Table 3). It is unclear if the skeletal organic matrix acts to promote or inhibit the formation of the aragonite mineral. The morphology and polymorph of CaCO3 precipitated in vitro is influenced by the presence of the organic matrix extracted from coral skeletons (Falini et al. 2013). Aspartic acid, glutamic acid and Gly suppress aragonite precipitation at ≥ 0.2 mM at the approximate pH and dissolved inorganic carbon conditions of the coral calcification site (Kellock et al. 2020, 2022; Castillo Alvarez et al. 2024), although lower concentrations of aspartic acid (1–10 µM) may promote aragonite formation to a minor degree (Kellock et al. 2022). The influence of amino acids on calcite precipitation is affected by peptide chain length (Elhadj et al. 2006) and further work is required to clarify how coral skeletal proteins (rather than their amino acid constituents) affects aragonite formation.

With respect to seawater pCO2, skeletal concentrations of Asx, Glx, Gly, Ser, Ala, L-Thr and total amino acid are lower at 180 µatm compared to 400 µatm, but otherwise the only difference between seawater pCO2 treatments is that [Ser] is reduced at 180 µatm compared to 750 µatm. Concentrations of skeletal, Asx, Glx, Ala and total amino acids were positively correlated with seawater pCO2 in a previous study (Kellock et al. 2020). Skeletal [amino acid] varies by more than × 2 between different coral genotypes cultured under the same environmental conditions in the present study (Fig. 1) and in another experiment in our aquarium (Kellock et all. 2020). This large variation between coral genotypes, combined with our small sample size, makes it difficult to resolve an influence of seawater pCO2 on skeletal amino acids.

Raman spectroscopy of coral aragonite CO3 disorder

The decrease in coral skeleton aragonite Raman spectra ʋ1 peak FWHM between corals cultured at 28 °C compared to 25 °C is small (0.05 cm−1, on average), but statistically significant. The smaller FWHM reflects an increase in local order around the CO3 group with higher temperature. We examine the Raman spectra of synthetic aragonites precipitated over variable Ω and temperature (Castillo Alvarez et al. 2024) to explore the influence of these parameters on aragonite CO3 group disorder. However, we note that these abiogenic precipitations do not fully encompass the conditions of the coral calcification media which contains complex biomolecules and enzymes (Tambutté et al. 2011).

The response of aragonite CO3 disorder to Ω and temperature in synthetic aragonite

The Raman analyses of the synthetic aragonite precipitates demonstrates that the FWHM of the aragonite ʋ1 peak is significantly positively related to Ωaragonite of the precipitating seawater (Fig. 6a). The FWHM is also positively related to temperature (r2 = 0.41) but this relationship is not significant due to the small number of precipitates (n = 6). Both temperature and Ω influence aragonite precipitation rate (Castillo Alvarez et al. 2024) and we observe no significant difference in the precipitation rate versus FWHM relationship between experiments conducted over variable temperature or Ω (Fig. 6c). This suggests that aragonite precipitation rate influences the FWHM rather than Ωaragonite or temperature. The origin of this effect is unclear, but rotational disorder of the CO3 group may reflect disturbance of the aragonite lattice to accommodate trace elements (DeCarlo et al. 2017; Farfan et al. 2022) in place of the mineral host ions (Ca2+ and CO32−). If mineral precipitation rates are rapid, then these impurity ions are more likely to be entrapped in the mineral (the growth entrapment model, Watson 2004), thus increasing disorder.

Our observation that Ω influences FWHM agrees with DeCarlo et al. (2017), although the magnitude of the effect is far smaller in the present study. DeCarlo et al. (2017) reported a logarithmic relationship between solution Ω and FWHM equivalent to an increase in FWHM of ~ 0.05 cm−1/Ω from Ω = 10 to 15 and 0.04 cm−1/Ω from Ω = 15 to 20. This is considerably higher than observed in the present study (0.011 cm−1/Ω). The Ca2+ consumed from the seawater during aragonite formation was not replaced during the titrations in the DeCarlo et al. (2017) study, leading to large variations in solution [Ca2+] and Ω within each precipitation (> 70%, Holcomb et al 2016). This [Ca2+] drop almost certainly affects aragonite precipitation rate. Further work is required to determine how this variability influences FWHM.

Causes of coral aragonite CO3 disorder

Although aragonite precipitation rate influences FWHM in synthetic aragonite (Fig. 6), we do not observe significant relationships between coral calcification rate and FWHM in the skeletal samples (Fig. 5, Table 4). Calcification rates are highly variable between different coral genotypes and pCO2 treatments (Cole et al. 2018) but, for the 4 coral genotypes cultured here at both temperatures, calcification increases by 39%, on average, at the higher temperature. This temperature increase is associated with a narrowing of the Raman ʋ1 FWHM, which is contrary to our observation that higher temperatures increase the FWHM in synthetic aragonite. We hypothesise that the increase of Raman ʋ1 FWHM in synthetic aragonite precipitates at high precipitation rates reflects an increased uptake of trace elements at a rapidly advancing crystal surface (Section “The response of aragonite CO3 disorder to Ω and temperature in synthetic aragonite”). However, high calcification rates in tropical corals do not necessarily reflect fast rates of crystal growth. Corals extend their skeletons along a growth axis, but the crystal fibres which make up the bulk of the skeleton are not orientated parallel to this axis (Wells 1956) so rapid skeletal extension does not necessarily equate with rapid fibre growth. Rapid coral calcification rates may reflect extension of many fibres, rather than rapid extension of individual fibres.

We do not observe significant relationships between skeletal Raman ʋ1 FWHM and coral calcification media pH, [Asx] or [total amino acid (Fig. 5, Table 4). To identify the potential origin of the change in coral skeletal Raman ʋ1 FWHM, we summarise the observed influences of temperature, Ωaragonite and [aspartic acid] on the FWHM of synthetic aragonite in Fig. 7. Increasing seawater temperature by 3 °C increases the FWHM in synthetic aragonite by 0.014 cm−1 (Fig. 6b) and, as already noted, this cannot explain the observed decrease in coral skeletal FWHM at higher temperature. Skeletal [Asx] and other amino acids are higher in corals cultured at 28 °C compared to 25 °C (Table 1). [Asx] can be highly variable between coral genotypes, but comparing the skeletal [Asx] for each coral specimen cultured at both 25 and 28 °C indicates that skeletal [Asx] increases by 23% on average at the higher temperature. Increasing seawater [aspartic acid] increases the Raman ʋ1 FWHM of synthetic aragonite (Kellock et al. 2022), but the effect is small, such that a 23% increase in seawater aspartic acid increases aragonite FWHM by ~ 0.006 cm−1 (shown in Fig. 7). It is unclear how biomolecules are incorporated into aragonite and coral skeletons. Coral skeletons are composed of nanograins (typically tens of nm in dimension) which are enveloped within materials inferred to be organic (Cuif et al. 2004; Drake et al. 2020; Tan et al. 2023). This suggests the skeleton is a nanocomposite composed of 2 materials intimately related at the nanoscale. However, inclusion of biomolecules within the aragonite structure cannot be discounted. In coral skeletons, Asx predominantly occurs in proteins rather than as the free amino acids used in the synthetic aragonite experiments (Kellock et al. 2022), however.

Fig. 7
figure 7

A comparison of the effect of a 3 °C increase in coral culturing temperature on the FWHM of the ʋ1 peak in the coral skeleton Raman spectrum with the influences of precipitation media temperature and Ω and [aspartic acid] in synthetic aragonite. Temperature and Ω influences are estimated from Fig. 6 (this study). The effect of aspartic acid is calculated from Kellock et al. 2022 which observed a linear relationship between aragonite ʋ1 peak FWHM and seawater log[aspartic acid] above [aspartic acid] = 10 µM

Full size image

These latter experiments suggest that the increase in skeletal amino acid at higher temperature is not responsible for the observed decrease in skeletal FWHM.

The decrease in skeletal ʋ1 peak FWHM at higher temperature could reflect a decrease in calcification media Ω (Fig. 6a), as hypothesised by DeCarlo et al. (2017). Comparisons of the effect of temperature increases on coral calcification and synthetic aragonite precipitation rate suggests that coral calcification media Ω decreases from 25 to 28 °C at low seawater pCO2 (180 and 260 µatm), but increases at 750 µatm seawater pCO2 (Cole et al. 2018; Allison et al. 2021). Coral skeletal boron geochemistry (B/Ca and δ11B) suggests that the effects of temperature on coral calcification media pH and Ωaragonite are complicated. Calcification media pH was significantly lower in massive Porites spp. cultured at 28 °C compared to 25 °C at low (180 µatm) but not at high (750 µatm) seawater pCO2 (Allison et al. 2021) and was unaffected by temperature in a range of coral species cultured at ~ 300–3300 µatm seawater pCO2 (Eagle et al. 2022). However, reduced temperatures were associated with lower calcification media pH at high pCO2 in cultured Porites spp. (Comeau et al. 2019) but with higher calcification media pH in corals spanning a range of localities in Western Australia (Ross et al. 2019) and in a long term (1939–2013) coral record from the Great Barrier Reef (D’Olivo et al. 2019). Similarly, boron geochemistry estimates of Ωaragonite may decrease at higher temperature in cultured corals (Eagle et al. 2022) but are either unrelated or positively related to temperature in tropical corals growing along a temperature gradient on reefs (Ross et al. 2019). A recent study found an offset between estimates of coral calcification media pH measured directly by the pH sensitive dye SNARF and inferred from skeletal δ11B (Allison et al. 2023), suggesting that other processes besides pH may influence skeletal δ11B. This will affect estimates of coral calcification media Ω from boron geochemistry. Our synthetic aragonite observations (Fig. 6) suggest a decrease in Ωaragonite of > 5 is required to generate the small decrease in FWHM observed in the corals cultured at 28 °C in this study. This is large in comparison with the observed decrease in calcification media Ω reported by Eagle et al. 2022 where Ω decreased by 1.1 on average in response to a 3 °C temperature increase. Further work is required to resolve the response of coral calcification media Ω to temperature change and to identity if this influences coral aragonite structural disorder.

Conclusions

This study shows that some skeletal [amino acids] of Porites spp. corals, grown in laboratory culture, are increased by rises in seawater temperature below the thermal stress threshold. Biomolecules control the polymorph, morphology and precipitation rate of calcareous minerals (Wolf et al. 2007; Kellock et al. 2020; Kim et al. 2016). In addition, the inclusion of amino acids within CaCO3 (calcite) improves the material hardness (Kim et al. 2016) and biomolecules are inferred to contribute to the superior material properties of CaCO3 biominerals compared to their abiogenic analogues (Deng et al. 2022). However, Raman spectroscopy indicates that disorder around the CO3 in the aragonite lattice of the coral skeletons is reduced at higher temperatures. This is contrary to observations in synthetic aragonite where increases in the temperature and [amino acid] of the precipitating media both increase CO3 disorder. Understanding the effect of biomolecules on aragonite structure and material properties is in its infancy, and future work will resolve if changes in skeletal protein content are likely to alter the physical resilience of coral skeletons.