Larval collection and settlement
Mature colonies of the brooding corals F. fragum and P. astreoides were collected from inshore patch reefs in Bermuda just prior to their predicted time of larval release in July 2007 (F. fragum), August 2007 (P. astreoides), and July 2008 (both species). Colonies were maintained at the Bermuda Institute of Ocean Sciences (BIOS) in outdoor flow-through seawater aquaria under near-ambient temperature and light conditions, and were held in either jars or mesh bags of aerated seawater during the nights of release to isolate the larvae. Zooxanthellate larvae were collected daily as they were released by the adults, and were settled on preconditioned tiles in small (0.5 L) plastic containers of seawater at the saturation state of each experimental aquarium (see below). Preconditioning of tiles was achieved by leaving racks of tiles on nearby reefs for 4-6 weeks, allowing them to obtain the biofilms and algae needed to induce larval settlement. After a settlement period of 24-48 h, the tiles containing metamorphosed primary polyps were transferred to the experimental aquaria. The polyps were grown for two weeks, after which the polyp tissue was removed by bleaching to reveal the underlying corallite. The skeleton of each polyp was removed from the tile and individually weighed using a micro balance. Since all skeletal carbonate retrieved from the experiments was formed under the experimental conditions, total corallite weight provides a direct measure of the amount of calcification (CaCO3 production) achieved by each polyp under the different experimental conditions. For statistical analysis, corallite weight data were square root transformed to meet assumptions and were analyzed using One-Way ANOVA followed by Multiple Comparison of Means TK, GT2, T’ tests (BIOMstat33).

Experimental conditions
Static, 30 L, glass-lidded aquaria containing reef seawater were pre-adjusted to a range of seawater saturation states (Table 1). In 2007, the aquarium seawater alkalinity was decreased by small additions of HCl and bubbled with lab air. The seawater pCO2, calculated from alkalinity and DIC, was approximately 450 ppmv. In 2008, the aquarium seawater pCO2 and DIC levels were set by bubbling with air from a compressor room separate from the lab, and with air+CO2 mixtures produced with pairs of mass flow controllers. The composition of the bubbling gas mixtures in 2008 was monitored daily using a Qubit infra-red CO2 analyzer and mean ppmv ± SD were: 394 ± 9 (ambient air; control), 753 ± 12 (mid CO2), and 2327 ± 23 (high CO2). The seawater temperature in all aquaria in each experiment was monitored using Hobo temperature loggers (Onset Corp.) Average seawater temperatures for the two week period were: 25 °C ± 0.5 (mean ± SD) for 2007 F. fragum; 28.5 ± 0.2 for 2007 P. astreoides; and 29.4 ± 1.3 for both species in 2008. The polyps were not fed during the two week experiments (apart from particulate matter initially present in the aquaria), and were kept on a 12/12 hr light-dark cycle with the maximum light levels achievable with the aquarium lights: mean (± SD) of 61 ± 6 µmol m-2 s-1.

The chemical conditions for all treatments in each experiment are summarized in the manuscript. Salinity was determined with an Autosal salinometer. Discrete water samples for analysis of salinity, alkalinity (Alk), and dissolved inorganic carbon (DIC) were collected weekly; the Alk/DIC samples were poisoned immediately after collection. Alkalinity and DIC were measured using a closed cell titration (inorganic carbon and alkalinity analyser) with non-linear curve fitting on ~100 mL samples, standardized using certified reference materials obtained from Dr. A Dickson (SIO). The pH (NBS) of the aquaria during all experiments was checked twice weekly using an Orion 3-star pH meter and calibrated electrode; the precision of replicate pH measurements was +/- 0.015 units. The measured seawater temperature, salinity, alkalinity, and DIC concentrations were used to calculate other carbonate system parameters ([HCO3-], [CO32-] and Omega), using a spreadsheet version of the CO2SYS program of Lewis and Wallace (1998), with the dissociation constants of Roy et al. (1993) and the aragonite solubility of Mucci (1983). The precision of the titrations was ± 0.2 % for both alkalinity and DIC in ambient seawater, but only ± 0.6 % and ± 1.7 %, respectively, in the most strongly acidified treatment. This resulted in an analytical uncertainty in calculated saturation state of roughly ± 0.5 % at ambient conditions and ± 16 % in the lowest Omega treatment.