Sampling locations: Six reefs within Kāne'ohe Bay, O'ahu, Hawai'i:

1.) USA: Hawaii HIMB: 21.436056, -157.786861
2.) USA: Hawaii Reef.11.13: 21.450806, -157.794944
3.) USA: Hawaii Reef.35.36: 21.473889, -157.833667
4.) USA: Hawaii Reef.18: 21.450806, -157.811139
5.) USA: Hawaii Lilipuna.Fringe: 21.429417, -157.791111
6.) USA: Hawaii Reef.42.43: 21.477194, -157.826889

Experiment conducted at the Hawai'i Institute of Marine Biology

Experimental Tank Setup: Treatment conditions (n=3 tanks treatment-1) were randomly assigned to twelve outdoor mesocosm tanks (122 cm x 122 cm x 30 cm; 510 L). Flow rates, measured daily with a graduated cylinder and timer, averaged 84.36 ± 1.20 mL second-1 (n=826), providing full mesocosm tank turnover every ~2 hours. Mesocosm tanks were 60% shaded from full irradiance, and photosynthetically active radiation was measured continuously with the Apex cosine corrected PAR Sensor (accuracy = ± 5%) that was cross calibrated to the Li-Cor cosine corrected PAR sensor (LI-193). Additionally, light values (PAR) from six different positions in each tank were compared to determine spatial differences within each tank. There was no significant difference in light between positions in each tank (n=4 position-1 tank-1; Table S2). Based on these results, light was measured in the center of the tank ~daily for the duration of the experiment using the Li-Cor 193 spherical underwater quantum sensor. To further reduce any potential position effects with respect to incoming water, heater position, or bubble stream, the positions of the coral fragments in the tank were changed weekly.

Experimental pCO2 conditions: Experimental pCO2 treatment conditions were based on measurements of conditions in Kāneʻohe Bay, Hawaiʻi (Drupp et al. 2011) and projected future lower pH conditions in embayments, which can have lower mean pH and higher diel fluctuations due to calcification, photosynthesis and respiration of the benthic community and increased residence time in the embayment (Jury et al. 2013); (Shaw, Hamylton, and Phinn 2016)). The daily fluctuating pH levels in the high pCO2 treatment tanks were maintained between 7.6-7.7 with an independent pH-stat feedback system in each tank and ambient conditions fluctuated between 7.9-8.0 (Fig. 1F). In order to generate the high pCO2 treatment, two 99.99% food-grade CO2 cylinders were connected to an automatic gas cylinder changeover system (Assurance Valve Systems, Automatic Gas Changeover Eliminator Valves #6091) to prevent an abrupt shortage of CO2 supply. CO2 was brought into the system on-demand through gas flow solenoids (Milwaukee MA955), based on the pH reading of a probe (Apex pH lab grade probes, Neptune Systems) in each tank via airlines plumbed into a venturi injector (Forfuture-go G1/2 Garden Irrigation Device Venturi Fertilizer Injector), which was connected to a water circulating pump (Pondmaster Pond-mag Magnetic Drive Water Pump Model 5). Gas injected into the system was either CO2 or ambient air and bubbling was constant due to the pressure driven pump moving water and gas through the Venturi injector. An Apex AquaController (Neptune Systems) environmental control system with a wifi base unit (Apex Controller Base Unit, Neptune Systems) was linked to 12 individual monitoring units (Apex PM1 pH/ORP Probe Module, Neptune Systems) with Apex pH Probes (accuracy = ±0.01 pH, Neptune Systems), which were used for microprocessor-control of a power strip (Apex Energy Bar 832, Neptune Systems) containing 12 individual solenoids. This pH-stat feedback system constantly monitored seawater temperature and pH conditions.

Experimental temperature conditions: Temperature treatment conditions were programmed to mimic the natural daily fluctuations (0.75 °C ± 0.06) of the environment at the collection sites in Kāneʻohe Bay, Hawaiʻi (NOAA Moku o Loʻe Buoy data from September 2018; Fig. 1G, Fig. S1). Based on these data, high temperature treatment fluctuated between ~29-30°C to reflect previous marine heatwaves in Kāneʻohe Bay, Hawaiʻi (+2°C above ambient temperature). Temperature was monitored with Apex Extended Life Temperature Probes (accuracy = ± 0.05 °C, Neptune Systems) and temperature loggers (HOBO Water Temp Pro v2, accuracy = ±0.21°C, resolution = 0.02°C, Onset Computer Corp) that were placed in each tank at the same height as the coral fragments for the duration of the experiment and logged temperature at 10 minute intervals. Temperature was separately controlled by submersible heaters (ProHeat D-1500 Heater Controllers, precision ± 1°C) due to the electrical demand of the heaters. Ambient temperature treatments were not controlled, and thus reflected the natural conditions of Kāneʻohe Bay.

Total Alkalinity and Carbonate Chemistry: Tank parameters (temperature °C, total scale pH, and salinity in psu) were measured ~twice daily using a handheld digital thermometer (Fisherbrand Traceable Platinum Ultra-Accurate Digital Thermometer, accuracy = ±0.05 °C, resolution = 0.001°C) and a portable multiparameter meter (Thermo Scientific Orion Star A series A325). A pH probe (Mettler Toledo InLab Expert Pro pH probe #51343101; accuracy = ±0.2 mV, resolution = 0.1 mV) and conductivity probe (Orion DuraProbe 4-Electrode Conductivity Cell Model 013010MD; accuracy = 0.5% of psu reading, resolution = 0.01 psu) were used with an Orion A Star meter to measured pH and salinity (psu), respectively. pH (total scale) was calculated from standard curves of pH (mV) across a range of temperature (°C) in a tris standard (Dickson Laboratory Tris Batch T27 Bottle 269, 236 and Batch T26 Bottle 198). 125 mL water samples were taken from each tank twice a week to measure carbonate chemistry using the total alkalinity method. An automated titrator (Mettler Toledo T50) was used to titrate water samples with salinity adjusted 0.1M hydrochloric acid (Dickson Laboratory Titrant A3, A14). A non-linear, least-squares procedure of the Gran approach (SOP 3b; Dickson et al. 2007) was used to calculate total alkalinity (TA; µmol kg−1 seawater). Accuracy was determined using certified reference material (Dickson Laboratory CO2 CRM Batch 132, 137, 176). Carbonate values, including aragonite saturation, carbon dioxide, carbonate, dissolved inorganic carbon, bicarbonate, pCO2, and pH, were calculated using SEACARB with total pH and total alkalinity given (flag=8; v3.2.16, Gattuso et al. 2015) in R Studio.