To evaluate whether rates of calcification within C. nereostratum have changed or will change with ocean warming and acidification, we cultured C. nereostratum under experimental conditions mimicking past, present, and predicted future levels of ocean temperature and pCO2 in the region, then followed this four-month incubation period with measurements of skeletal density using micro-computed tomography (microCT). Small C. nereostratum colonies (~4-5 cm diameter) were live collected from Adak in 2015 and immediately transported to the Northeastern University Marine Science Center in Nahant, Massachusetts. There, all specimens were acclimated to laboratory conditions at 8.5 degrees C for two weeks, after which individual C. nereostratum colonies were attached to the underside of plastic Petri dishes using cyanoacrylate glue and then allowed to acclimate for an additional two weeks before being moved to experimental aquaria. Conditions were then incrementally modified to achieve target temperature and pCO2 levels (see below) over a one-week period. After reaching target conditions, each 42-L aquarium was dosed with 213 mL of calcein fluorescent dye (Western Chemicals Inc.), which was recirculated in the aquaria for three days and then flushed from the system. Coralline algae incorporate the dye into their skeleton, thus creating a distinct line that can be viewed via fluorescent microscopy to demark the region of new growth within each individual.

We employed four pCO2 conditions and three temperatures that, while factorially crossed, spanned pre-industrial, present-day, and projected year 2100 conditions (assuming an IPCC "business as usual" carbon emissions scenario; Pachauri and Meyer 2014). More extreme temperature (12.5 degrees C) and pCO2 (2800 micro-atm) conditions were also employed in the broader experiment but were not included in our study because they are not predicted to occur until year 2500, or later. For each treatment, we set values to average summertime conditions, the time when ~75% of C. nereostratum growth occurs (Adey et al. 2013).

All treatments (4 pCO2 concentrations x 3 temperatures, fully factorial) were housed on individual shelves and consisted of three 42-liter acrylic aquaria and one 65-liter sump (n = 3 tanks/treatment). The aquaria were connected to a sump via a common overflow and return line but were each independently and continuously replenished with new seawater—thereby establishing them as true experimental replicates. The sump contained a filter box with a nylon mesh particle filter and activated carbon, a protein skimmer (Eshopps PSK-75), and a return pump, all of which was connected to a water chiller (Coralife 1/4HP). Filtered natural seawater was added via Darhor manual flow controllers at a rate of 50 mL/min/tank, resulting in full replacement of treatment water every ~21 hours - sufficiently fast to prevent material depletion of the dissolved constituents of the seawater yet slow enough to allow the mixed gases being sparged into the experimental treatments to approach equilibrium with the seawater. Mixed gases were sparged into each tank with 91 cm long flexible bubblers at the rate of ~1 L/min via Darhor needle-valve gas flow controllers. Two 12,000K LED light arrays (Ecoxotic Panorama, Pro 24V) were mounted above each tank and set to an irradiance that mirrored average summer daylight irradiance at 10 m depth in the Aleutian Islands (~258 micro-E m-2 s-1; 12 hr light:12 hr dark cycle).

Over the course of the four-month experiment, we measured pH (Accumet AB15 pH meter with Accufet solid state probe), salinity (YSI3200 meter with K=10 conductivity electrode and temperature probe), and temperature (NIST traceable red spirit glass thermometer) in each tank every Monday, Wednesday, and Friday. The pCO2 of the gas mixtures was measured with a Qubit S151 infrared CO2 analyzer and calibrated with certified mixed CO2 from Airgas Incorporated. Every 10 days, we characterized the full carbonate system chemistry of the experimental treatments from measured total alkalinity, dissolved inorganic carbon, temperature, and salinity. For this, seawater samples were obtained in 250 mL borosilicate ground-glass-stoppered bottles and immediately poisoned with 100 micro-L of saturated HgCl2 solution to halt biological activity (Dickson et al. 2007). Total alkalinity was measured via closed-cell potentiometric Gran titration and dissolved inorganic carbon was measured with a UIC 5400 Coulometer on a VINDTA 3C (Marianda Incorporated) using Dickson certified seawater reference material. Seawater pCO2, pH, carbonate ion concentration ([CO32-]) bicarbonate ion concentration ([HCO3-]), aqueous CO2, and calcite saturation state were calculated with the program CO2SYS (Lewis and Wallace 1998), using Roy and colleague's (1993) values for the K1 and K2 carbonic acid constants, the Mucci (1983) value for the stoichiometric calcite solubility product, the seawater pH scale, and an atmospheric pressure of 1.015 atm.

At the beginning of the experiment we measured the buoyant weight of each specimen. We then scrubbed each specimen with a toothbrush and reweighed it every month and at the end of the experiment. With each weighing, we also photographed the specimen with a ruler and Reef Watch coral bleaching card in the field of view. We then measured the 2-d surface area of the photographed specimens (Image J, NIH). At the end of the experiment, all coralline algae were sectioned with a diamond lapidary saw (Inland Craft SwapTop 6.5” Diamond Trim Saw) and either frozen for genetic analysis or sectioned into 6 mm slices, rinsed in a series of two 90% Ethanol baths, and allowed to air dry for further examination of growth and skeletal density.

We measured the density of the calcified skeleton deposited by C. nereostratum during the four-month laboratory experiment via micro-computed tomography (microCT); see Chan et al. (2017) for methods development and analytical setup. In brief, samples were scanned in a GE Locus RS-9 (General Electric Health Care, London, Ontario) x-ray microCT at an energy of 90kVp and tube current of 450 micro-A. Two frames, each 4500 ms in duration, were averaged at 900 projection angles over a 360-degree rotation of the gantry to produce data that was processed into a 3D image with 20 micron isotropic voxel spacing. Only specimens raised in the experimental temperature and pCO2 treatments employed in the feeding assay (6 treatments, n = 3 specimens/treatment) were studied. For each specimen, three cuboid regions of interest (ROI) were then selected, focusing on the region of new growth as indicated by the calcein mark. ROI size was similar for all measurements (1445-1575 voxels); however, dimensions were adjusted depending on the amount of accretion incurred and to avoid overlap with the epithallus or tissues deposited prior to the experiment. Grayscale thresholding to eliminate non-calcified tissue was unnecessary, given that conceptacles were not present in the newly deposited tissue and intracellular pore spaces (6 microns) are smaller than the microCT voxel size (20 microns) and were therefore not resolved. However, an analysis employing thresholding (Chan et al. 2017) produced virtually identical results. We quantified the skeletal density within each ROI by calculating the fractional mineral content of the ROI (i.e., fractional composition of each voxel that is CaCO3), converting each value to units of pure crystal calcite (physical density: 2.71 g/cm^3), then averaging over all voxels in the ROI.