Methodology:

Five specimens of Acanella arbuscula and four specimens of Keratoisis sp. were collected from two stations (Davis Strait; 63.34533 degrees N; 58.19571 degrees W, 1311 m, 29th July 2021, Disko Fan; 67.96631 degrees N, 59.49381 degrees W, 889 m, 2nd August 2021) using a remotely operated submersible (Sub-Atlantic® Comanche, Forum Energy TechnologiesTM, USA) during the 2021 Amundsen expedition (AMD21 Leg 2; 15th July 2021 - 12th August 2021, CCGS Amundsen). Where possible, corals were sampled at or close to the basal internode (near the base of the specimen at the sediment surface). Any external debris and residing fauna were carefully removed from the collected colonies using tweezers before each specimen was sealed in a plastic Ziplock bag and frozen at -20 degrees Celsius. After 72 hours, the specimens were removed from the freezer and carefully cleaned with jets of re-circulated 0.45 micrometer membrane-filtered seawater (FSW) at 4 degrees Celsius using a WaterPikTM (Johannes & Wiebe 1970) before being placed back in -20 degrees Celsius. The cleaned skeleton portions were then sealed in new Ziplock plastic sample bags enclosed in Tupperware (Acanella arbuscula) or PVC vinyl tubing (Keratoisis sp.) before being transported to the University of Southampton, UK.

Reconstruction of biogenic structures was achieved using a Nikon XT custom designed micro-focus computed tomography system (based on the XT H 225 ST, Nikon Tring, UK) housed within the 3D X-ray Histology (XRH) facility (www.muvis.org). Coral specimens were securely positioned vertically in Perspex holding tubes with polystyrene bungs to ensure stability during rotation and scanning. Scans were conducted at 80 KVp using a Molybdenum target. Lower resolution scans (50 micrometer) used a 12 W power, while higher resolution scans (15 micrometer) reduced it to 6.9 W for a sharper X-ray focal spot. A water tube was also scanned under the same conditions as the samples to serve as a density reference. CT Pro 3D 6.6 or 6.7 software (Nikon Xtek, Tring UK) was used for all reconstructions. Due to the limited scanner field of view, multiple overlapping scan positions were used and merged after reconstruction to remove cone beam artifacts. This merging process was facilitated by a custom macro in Fiji, allowing alignment of the scans based on the same slice. The script could also adjust image contrast to compensate for any heel effect in the absence of density calibration. The final 32-bit volume was converted to an 8-bit volume using Fiji/ImageJ (v 1.53c; Schindelin, Arganda-Carreras, Frise et al., 2012) and then converted to stacked tagged image file format (TIFF) images to reduce data size and facilitate further processing.

More detailed methodology surrounding sampling and scanning can be found in Williams, Standish, Archambault et al. [in prep] and Williams, Katsamenis, Basford, et al. [in prep], respectively.

Data collection:

Nikon XT custom designed micro-focus computed tomography system (based on the XT H 225 ST, Nikon Tring, UK

Remotely operated submersible (Sub-Atlantic Comanche, Forum Energy TechnologiesTM, USA)

WaterPikTM (Johannes & Wiebe 1970)

CT Pro software (6.6/6.7; Nikon Xtek, Tring UK)

Fiji/ImageJ software (v 1.53c; Schindelin et al., 2012)

Data quality:

Coral specimens were securely positioned vertically in Perspex holding tubes with polystyrene bungs to ensure stability during rotation and scanning.

A water tube was also scanned under the same conditions as the samples to serve as a density reference. CT Pro 3D 6.6 or 6.7 software (Nikon Xtek, Tring UK) was used for all reconstructions.

During concatenation of separate scans, image contrast was adjusted to compensate for any heel effect in the absence of density calibration.

Conversion to stacked tagged image file format (TIFF) images was lossless.