Location: 
patch reef 13 in Kāne'ohe Bay, O'ahu, Hawai'i (21.4509, -157.7954) 

Physiological analyses:

A total of 276 fragments were collected from colonies of bleaching-resistant and bleaching-susceptible M. capitata and P. compressa (n= 7–10 pairs per phenotype per species) across eight timepoints from October 2019 to September 2022, incorporating the peak of the 2019 heatwave and three years (35 months) of recovery during non-heatwave years (Fig. S1, Brown et al. (2023)). Fragments (4–5 cm in length) were collected by hand using bone cutters, transported to the Hawai’i Institute of Marine Biology (HIMB) in ambient seawater and held in flow-through seawater aquaria for 24–72 hours until measurements of coral performance. For four timepoints (October 2019, October 2021, March 2022, and September 2022), metabolic rates were assessed via changes in oxygen evolution using oxygen optodes (PSt7, PreSens) connected to an optical analyser (OXY-10, PreSens) (Innis et al., 2021). Oxygen optodes were calibrated on each day with a 0% oxygen solution (0.01 g ml−1 NaSO3) and 100% air saturated seawater. Coral fragments were analyzed between 08:00 and 17:00 within 250 ml clear acrylic chambers on top of a magnetic stirrer to allow for continuous mixing. Seawater temperatures were replicated to those experienced on the reef by using ambient seawater and a water jacket to maintain temperatures within the incubation chambers. Temperature and dissolved oxygen concentrations were recorded every 3 seconds at increasing increments of light over a total of 45–50 minutes to determine maximum net photosynthesis (Pmax). Light levels ranged from 112–726 µmol m-2 sec-1 in October 2019  (Innis et al., 2021) (step size ~100 µmol m-2 sec-1; ~10 minutes step-1), and from 387–1800 µmol m-2 sec-1 in 2020–2021 (step size ~300 µmol m-2 sec-1; ~10 minutes step-1). After measurements were completed at the maximum light levels, the lights were turned off (0 µmol m-2 sec-1) to measure light-enhanced dark respiration (LEDR) as defined by (Edmunds and Davies, 1988). Photosynthesis-irradiance curves were fitted using the Platt model to extract Pmax and LEDR (Innis et al., 2021; Platt et al., 1981 ), 3). Upon completion of these in vivo analyses, coral fragments were flash frozen in liquid nitrogen and stored at −80°C until further processing.

Coral tissues containing symbionts were removed from skeletons with a waterpik using 50 mL of 0.1 M phosphate buffered saline (PBS) solution. The resulting holobiont homogenate was centrifuged at 4°C for 5 minutes at 2500 x g to separate the coral host fraction (supernatant) from the intracellular endosymbiont cells (family Symbiodiniaceae; pellet). Symbiodiniaceae were resuspended in PBS and cell abundances were quantified using a Millipore Guava flow-cytometer (Guava easyCyte 5HT) following the methodology of (Innis et al., 2021). Symbiodiniaceae cells were excited with a blue laser (488 nm) and identified by analyzing forward scatter and red autofluorescence in GuavaSoft 3.4 with the same gating for all samples. Symbiont densities were standardized to skeletal surface area (cm2), which was determined using the wax-dipping technique (Holmes, 2008; Stimson and Kinzie, 1991). Endosymbiont photopigments were extracted in 100% acetone for 24 hours and the concentration of chlorophyll-a was determined via absorbance at 630, 663, and 750 nm using the equation in (Jeffrey and Humphrey, 1975) and subtracting the absorbance at 750 nm from both A663 and A630: 

Chlorophyll a =(11.43 (A663-A750)) -(0.64 (A630-A750))

Photopigment concentrations were standardized to both skeletal surface area and symbiont densities. Endosymbiont community composition (proportion Durusdinium vs. Cladocopium) of M. capitata colonies was determined by Dilworth et al. 2021 from samples collected in July 2019 using quantitative PCR. 

Host tissue biomass was determined as ash-free dry weight (AFDW) from the coral fraction (i.e. symbionts removed as described above). First, 1 mL of the coral fraction was dried at 60°C for 24 hours until a constant weight was achieved. After the dry weight was recorded, the samples were then burned in a muffle furnace at 450°C for 6 hours. The samples were allowed to cool in the furnace before being weighed and the weight of the resulting ash was recorded. The difference between the ash weight and the dry weight was calculated to determine the AFDW of each sample. Host soluble protein content of each sample was determined via the Bradford method. Specifically, 10 μL of the coral fraction was pipetted into each of three wells of a 96-well plate (technical replicates), followed by the addition of 300 μL of Coomassie Plus Bradford reagent (Thermo Fisher Scientific) to each well. The plate was incubated for 10 minutes at room temperature followed by the collection of an absorbance reading at 595 nm on a spectrophotometer (BioTek PowerWave XS2). Each plate contained a set of internal standards with known concentrations of bovine serum albumin (0–2000 μg mL-1), which were used to generate a second-degree polynomial standard curve relating absorbance (x) to protein concentration (y). The standard curve equation was used to calculate the protein concentrations of the samples. The concentration of lipids in each sample was determined via a modified method of (Dunn et al., 2012). First, ~45 mL of the coral fraction was lyophilized to produce dry tissue. A subsample of the lyophilized tissue was weighed (~90 mg), and 2 mL of chloroform-methanol (2:1) was added. The sample was then mixed using a homogenizer at 15000 rpm for 5 seconds, and an additional 2 mL of chloroform-methanol was used to rinse the homogenizer into the sample, and the tubes were then vigorously vortexed before being placed at 4°C for 2 hours in the dark. After incubation, tubes were again vortexed and the chilled homogenates were then passed through 0.22 μm syringe filters into new pre-weighed tubes. An additional 1 mL of chloroform-methanol was passed through the filters into each tube to ensure the full passage of sample material. Next, 1 mL of 0.1 M KCl was added to the samples, which were then vortexed and placed at 4°C for at least 1 hour in the dark until two phases formed. The aqueous (top) phase was discarded, and 5 mL of 50% methanol was added to the organic (bottom) phase. The samples were then placed at 4°C for 1 hour in the dark, followed by removal of the aqueous (top) phase. Washes with 50% methanol were repeated twice more, and the remaining organic phases were dried under a fume hood until the solvent had completely evaporated, at which point the lipid pellet was weighed. Coral host biomass, protein and lipid content were standardized to skeletal surface area. 

The total antioxidant capacity of the coral homogenate was assessed using a kit from Cell Biolabs (STA-360) according to the manufacturer’s instructions. First, 20 µL of the coral fraction was transferred to each of two wells of a 96-well plate (technical replicates), followed by the addition of 1x reaction buffer (20 µL) from the kit. An initial absorbance reading was collected with a spectrophotometer at 490 nm. Next, 50 µL of 1x copper ion reagent were added to each well, followed by incubation for 5 minutes with gentle shaking. Following shaking, 50 µL of 1x stop solution was added to each well, and a second absorbance reading was collected at 490 nm. Each plate contained a set of standards with known concentrations (0–0.1 mM) of uric acid. For analysis, the initial absorbance readings were subtracted from the final absorbance readings for each standard well, and the resulting values were used to generate a linear standard curve relating the absorbance (x) to the concentration of uric acid (y). The standard curve equation was used with the average change in absorbance across the two technical replicates for each sample to first determine the concentration of uric acid equivalents in the samples, which was then converted to a concentration of copper reducing equivalents (CRE) using the equivalence of 1 mM of uric acid to 2189 μM CREs. Finally, the CRE values were normalized to the amount of protein in each well, for a final value of TAC expressed as µM CRE mg protein-1.

To assess the melanin content of the coral homogenate, a protocol was adapted from (Wall et al., 2021). First, 1.5 mL tubes were filled with approximately 20 mg of the dry coral fraction and the weight recorded. Next, 300 µL of 10 M NaOH was added to each tube, followed by vortexing for 20 seconds. Samples were left overnight at room temperature in the dark, then vortexed again for 10 seconds before being centrifuged at 7000 x g for 5 minutes. Following centrifugation, 100 µL of supernatant was transferred to each of two wells of a 96-well plate (technical replicates), which was then read for absorbance on a spectrophotometer at 490 nm. Each plate contained a set of internal standards with known concentrations (0–0.01 mg mL-1) of synthetic melanin (M8631, Sigma Aldrich) in 10 M NaOH, which were used to generate a linear standard curve relating absorbance (x) to melanin concentration (y). The standard curve equation was then used to calculate the concentration of melanin in each sample, which was standardized to the original weight of the dry coral host tissue used in the assay.

A protocol was adapted from (Wall et al., 2021; Mydlarz and Palmer, 2011; Fuess et al., 2018) to assess the content of prophenoloxidase (PPO) in each coral sample. First, 150 µL of the coral fraction was transferred to each of two wells of a 96-well plate (technical replicates). Next, 23 µL of 0.2 mg mL-1 trypsin was added to each well, and the plate was incubated for 5 minutes at room temperature with shaking. Following incubation, 60 µL of 10 mM L-1,3- dihydroxyphenylalanine (L-DOPA) was added to all wells, and the plate was again incubated at room temperature for 5 minutes with shaking. Next, the plate was read for absorbance on a spectrophotometer at 490 nm every minute for 15 minutes. The absorbance of each well at the start of the 15 minutes was subtracted from the absorbance of the same well at the end, and values were divided by 15 to determine the change in absorbance per minute, which were then averaged over the two technical replicates. Finally, these values were standardized to the amount of protein in each well for a final quantification of PPO activity expressed as the change in absorbance minute-1 mg protein-1

Wax-dipping was used to determine calcium carbonate (CaCO3) bulk density, where the skeleton was cleaned, dried to a constant mass and weighed, sealed with a coat of wax, dry weighed with the wax and then buoyant weighed in DI water at 20°C (Tambutté, et al. 2015; Brown and Mello-Athayde, 2022). The difference between dry weight (with wax) and buoyant weight (divided by the density of the DI water medium of 1 mg cm-3) was calculated to determine the total volume enclosed. The dry skeletal mass (wax free) was then divided by the total volume enclosed to yield bulk density (g cm-3).

Taxonomic Identifiers:
Montipora capitata, LSID (urn:lsid:marinespecies.org:taxname:287697)
Porites compressa, LSID (urn:lsid:marinespecies.org:taxname:207236)