Experimental conditions and maintenance

Treatments were created in 8 tanks (Aqua Logic, San Diego), each holding 150 L of seawater and regulated independently for temperature, light, and pCO2.

Temperatures were maintained at 28.0°C, which corresponded to the ambient seawater temperature in the back reef when the study was conducted, and 30.1°C which is close to the maximum temperature in this habitat (Putnam and Edmunds 2011). pCO2 treatments contrasted ambient conditions (~ 408 micro-atm) and 913 micro-atm pCO2, with the elevated value expected to occur within 100 y under the "stabilization without overshoot" representative concentration pathway (RCP 6.0) (van Vuuren et al. 2011). pCO2 treatments were created by bubbling ambient air or a mixture of ambient air and pure CO2 that was blended continually and monitored using an infrared gas analyzer (IRGA model S151, Qubit Systems). A solenoid-controlled, gas regulation system (Model A352, Qubit Systems, Ontario, Canada) regulated the flow of CO2 and air, with pCO2 logged on a PC running LabPro software (Vemier Software and Technology). Ambient air and the elevated pCO2 mixture were supplied at ~ 10-15 L min-1 to treatment tanks using pumps (Gast pump DOA-P704-AA, see Edmunds 2011).

The temperatures and pCO2 levels created four treatments with two tanks treatment-1: ambient temperature-ambient pCO2 (AT-ACO2), ambient temperature-high pCO2 (AT-HCO2), high temperature-ambient pCO2 (HT-ACO2) and high temperature-high pCO2 (HT-HCO2). Treatment conditions were monitored daily, with temperature measured at 08:00, 12:00 and 18:00 hrs using a digital thermometer (Fisher Scientific model #150778, ± 0.05 °C), and light intensities at 12:00 hrs using a Li-Cor LI-193 sensor attached t 170 o a LI-1400 meter. Seawater within each tank was replaced at 200 ml/min with filtered seawater (50 micro-m) pumped from Cook’s Bay.

Carbonate chemistry and pH analysis

To evaluate dissolved inorganic carbon (DIC) conditions in the 8 tanks, total alkalinity (TA) and pH of the seawater were recorded every third day of the experiment. Seawater was collected between 07:00-09:00 hrs using stoppered glass bottles, equilibrated to room temperature (25.0°C), and processed within 2-3 hrs of collection. TA was determined using an open cell potentiometric titrator (Model T50, Mettler-Toledo, Columbus, OH) fitted with a DG115-SC pH probe (Mettler-Toledo, Columbus, OH) calibrated daily using NBS buffers (pH 4.00, 7.00 and 10.00, Fisher Scientific, 15-0787-8, ± 0.05 °C), and used to perform gran titrations using standard operating procedure 3 (SOP) of Dickson et al. (2007). Seawater pH was determined spectrophotometrically using the dye m-cresol purple (SOP 6b of Dickson et al. 2007), where pH was expressed on the total scale. The results of the gran titrations together with seawater salinity (YSI 3100 conductivity meter) and seawater temperature were used to calculate TA, pCO2, HCO3-, CO3 2- and aragonite saturation state (Omega) using CO2SYS (Lewis and Wallace 1998), with the constants of Mehrbach et al. (1973) and pH on the total scale.

To evaluate the accuracy and precision of TA analyses, certified reference materials (CRM, batch 105 from A. Dickson, Scripps Institution of Oceanography) were processed before each set of seawater samples. CRMs were evaluated with a mean error of 0.37% (~ 8 micro-mol kg-1, n = 11) relative to the certified values. The precision and accuracy of pH measurements were evaluated using standardized Tris buffers (Batch 5 from A. D 193 Dickson Laboratory, Scripps Institution of Oceanography) that were processed spectrophotometrically with m-cresol as described above. Percent average error from the known pH of the Tris buffer was 0.16% (0.01 pH units, n = 13).

Incubation schedule and dependent variables

On April 24th 2011, nubbins and cores were buoyant weighed (± 1 mg, Spencer-Davies 1989) and randomly placed in the mesocosm, with four taxa and two replicates per taxon in each tank. Over the following 24 h, seawater temperature and pCO2 were adjusted to target values. Corals remained in the treatments for 19 d, and were moved randomly within the tanks daily to eliminate position effects. Individual corals, along with the racks holding them, were cleaned every 5 d by wiping algal growth from walls of the tanks, racks, and PVC coral holders. On May 12th, the experiment ended and the corals were again buoyant weighed and the area of living tissue determined using aluminum foil (Marsh 1970).