This methodology describes this dataset and other datasets from this experiment. See "Related Datasets" section for data access and more details of each related dataset.

Study location and environmental conditions

The experiment was performed during the austral summer from mid-January to late March 2021 at Heron Island Research Station (HIRS), southern Great Barrier Reef (23 27°S, 151 55°E). Heron Reef is composed of five distinct geomorphological habitats characterized by diverse benthic communities and biogeochemical conditions (Phinn et al., 2012; Brown et al., 2018). This study focused on two distinct habitats, the reef flat (North Beach) and reef slope (North Bommie) (Brown et al. 2022 Figure 1). Semidiurnal tidal fluctuations on the reef flat result in higher variability in temperature and CO2 compared to reef slope habitats, and exposes reef flat corals to extreme temperature and CO2 conditions projected for future reefs (Rathbone et al., 2021; Camp et al., 2018; Brown et al., 2018)(Brown et al., 2022 Figure 1 and Figure S1). In-field measurements (temperature, photosynthetically active radiation, and nutrients) were recorded concurrently with the manipulative experiment at the same locations where corals were collected (8 January – 18 March 2021), whereas pCO2 was recorded over the same season, but in 2016 (8 January – 18 March 2016; Supp Methods). Long-term studies show remarkable consistency in pCO2 measurements recorded at the same location between years (Fabricius et al., 2020), suggesting pCO2 variability measured within these identical reef habitats over the same time period may be similar across years.

Seawater temperature (n=2 sensors) and photosynthetically active radiation (PAR) (n=2 sensors) were continuously recorded at 30 min intervals (Brown et al., 2022 Figure S1). PAR sensors were fitted with copper rings to prevent biofouling and were cleaned weekly. Additionally, weekly trips were made to North Beach and North Bommie at both low and high tide to collect seawater samples to determine nutrient concentrations (i.e., ammonium, nitrate, nitrite and phosphate) (Brown et al., 2022 Figure 1 and Figure S2). Seawater was collected at depth directly adjacent to sensors in a 500 mL glass bottle pre-rinsed with 0.1M HCl. Samples were brought to HIRS in the dark on ice after collection and immediately filtered (0.22 µM) into triplicate 20mL glass vials pre-rinsed with 0.1M HCl. Samples were stored at -20°C prior to analysis. Nutrient determination was done by Flow Injection Analysis (Lachat QuikChem 8500 Series 2 Flow Injection Analyser) at the Advanced Water Management Facility, University of Queensland. Seawater pCO2 conditions were determined by use of by use of Conductivity Temperature Depth units (SBE 16plusVS SEA-CAT) fitted with an auxiliary CO2 sensor (Optical CO2 sensor, AMT Analysemesstechnik GmbH) within the reef flat (n=1) and reef slope (n=1) over the same season in 2016 (Brown et al., 2022 Figure 1).

Sample collection, species identification and experimental design

Fragments of the coral Pocillopora damicornis were collected from the reef flat and slope locations within the same depth range (1–3 m) on 14 and 15 January 2021 (Brown et al., 2022 Figure 1). Four fragments were collected from each individual colony (genetic clones), totaling 96 fragments from 24 colonies (n = 12 per habitat) (Brown et al., 2022 Figure S2). One additional chip per colony was preserved in 100% ethanol and kept at -80°C for genetic analyses to confirm the collected coral specimens were all P. damicornis based on the mitochondrial ORF [cf. Schmidt-Roach et al., 2013 ] and identify the species of resident intracellular Symbiodiniaceae using the ITS2 rDNA and chloroplast minicircle psbA non coding region [cf. (Sampayo et al, 2009; LaJeunesse et al., 2011); full details in Brown et al., 2022 Supp Methods). All 96 collected coral fragments were standardized to a length of ~5 cm using bone cutters and randomly suspended with nylon fishing line from a bamboo stick (Brown et al., 2022 Figure S2). Six fragments were suspended from each stick and two sticks placed in each experimental treatment tank (33L; n = 12 fragments per tank). To minimize ‘tank effects’, the eight tanks were randomized across one outdoor table (n = 4 per treatment), with each set of coral fragments rotated into an adjacent tank of the same treatment every third day. Tanks and lids were covered with filters (Old Steel Blue #725, Lee Filters) to mimic the light environment at the collection sites (Brown et al., 2022 Figure 1, Figure S1, Figure S3). Noticeable paling was observed during the first week of the experiment, so light intensities were reduced with an additional shade cloth (Brown et al., 2022 Figure S1). All surfaces including exposed cut coral bases were cleaned every three days to remove any epilithic algae.

After 7 days of recovery from collection and handling, corals were exposed to two distinct treatments for 8 weeks: (1) stable pCO2 and (2) variable pCO2 (Brown et al., 2022 Figure 2), which were maintained following previously described methods (Eyre et al., 2018; Dove et al., 2013)(Brown et al., 2022 Supp Methods, Figure S3). Upstream CO2 was continuously recorded in treatment sumps (Brown et al., 2022 Figure 2) and within experimental tanks, seawater temperature (HOBO pendant logger) and photosynthetically active radiation (Odyssey PAR sensor) were continuously measured at 30 min intervals in each treatment by randomly rotating two probes per treatment between tanks (Brown et al., 2022 Figure S1). Weekly samples (n = 3 per treatment) were collected for total alkalinity (AT) and pHTotal at midday and midnight. AT was determined via the Gran titration method using 0.1 M HCl and pHTotal was determined via a high-precision glass pH electrode (DGi101-SC, Mettler Toledo) across replicated 20 g seawater samples (Kline et al., 2012). Acid concentration was calibrated at the beginning of each titration session using the certified reference materials from the Dickson Laboratory at Scripps Institute of Oceanography, USA. Salinity was measured via refractometer, and remained constant at 35.0 throughout the experiment. Parameters of the seawater carbonate chemistry, including carbonate, bicarbonate, aragonite saturation state, were calculated from our temperature, salinity, AT and pHTotal measurements using the seacarb package in R (Gattuso, 2021)(Brown et al., 2022 Table 1).

CO2 conditions were manipulated using a flow-through acidification simulation system. Seawater was pumped directly from the Heron Island reef flat into two very large treatment sumps (8,000 L). Sumps were maintained in the dark, and in combination with high flow rates, were used to effectively eliminate any interaction between the sump walls and the bulk of the water that passed through them (i.e., differential confinement effects) (Rivest et al., 2017). Conditions in each sump were manipulated using a computer-controlled feedback system that responded to conditions measured in experimental aquaria (SCIWARE Software Solutions). CO2 (Invensys Process Systems and CO2 -Pro CO2 regulators, ProOceanus) was manipulated by injection of air enriched to 30% CO2 or CO2 -free air. Sump, as opposed to tank, CO2 was controlled and monitored to allow organisms and their biology to influence in-tank CO2. Manipulated treatment seawater was then pumped from the sumps into flow-through tanks at an average rate of 4.17 L min-1, with circulation within each tank enhanced by a wave-maker (Nano 900, Hydor).

For more detailed information, please see: Brown et al. (2022).