Sample Collection and Maintenance
Three coral species (one octocoral and two scleractinians) were utilized in this experiment. The octocoral, Eunicea flexuosa, was collected from Wonderland Reef (24.558694, -81.503528) within the Florida Keys National Marine Sanctuary (FKNMS) under FL saltwater fishing permit (Permit #: I-H1R76333834 held by A.M. Reigel). Three axial branch tips were clipped from each of 10 healthy E. flexuosa colonies located at depths of ~5-8 meters. Branch tips were kept in seawater and immediately transported to the outdoor land-based nursery (CAOS) at Mote Marine Laboratory at Summerland Key, FL where they were placed into a shaded, temperature- and pH-maintained flow-through tank and allowed to acclimate for ~24 hours. The two hard coral species, Acropora cervicornis and Orbicella faveolata were provided by Mote Marine Laboratory's field- (A. cervicornis; Coordinates: 24.562747, -81.400455) and land-based (O. faveolata) nurseries as permitted under the FKNMS-2015-163-A3. All hard coral fragments were placed in the same flow-through CAOS tanks as the E. Flexuosa samples and allowed to acclimate for ~24 hours.

To develop a representative sponge community for the Florida Keys reefs, we collected 5 individuals of each of 6 sponge species (Niphates digitalis, Verongula rigida, Aplysina fulva, Aplysina cauliformis, Xestospongia muta, Callyspongia aculeata) from Wonderland Reef under FL saltwater fishing permit (Permit #: I-H1R76333834 held by A.M. Reigel). Sponges were kept in seawater and immediately transported to the lab where they were placed in a shaded CAOS flow-through tank to acclimate for ~24 hours. Corals and sponges were not in the same CAOS tanks during the acclimation period.

Coral and Zooxanthellae Separations
To prepare for downstream analyses the coral fractions, host and zooxanthellae, were manually separated. Scleractinian fragments were thawed and airbrushed with an aerosolized jet of 0.22 um filtered seawater to physically separate the coral tissue and skeleton and suspend the coral tissue material into a homogenate. To separate host tissue from zooxanthellae cells, the homogenate was centrifuged at 2000g for 3-5 minutes. Centrifugation formed a pellet comprised of zooxanthellae cells and a homogenate of host material. The host homogenate was pipetted into a separate sterile 50 ml Falcon tube. The homogenate, zooxanthellae pellet, and skeletal fragments were frozen and transported to Appalachian State University where they were stored at -20F until further processing. At Appalachian State, host homogenates and zooxanthellae pellets were thawed and checked for purity. Impure fractions were combined, homogenized with a tissue homogenizer (maximum speed for ~15 sec) to physically separate zooxanthellae cells from host tissue, and centrifuged (3000 x g, 6 min.) to pellet the zooxanthellae cells. Separated fractions were combined with original fractions each time and checked for purity under the microscope. The process was repeated until at least 80% purity was reached.

E. flexuosa fractions were separated using a different process. First, the frozen coral branches were lyophilized (Labconco™ FreeZone™ Bulk Tray Dryer) for 22-24 hours, until they were completely dry. Following lyophilization, the axial skeleton was removed and the tissue was ground up using a mortar and pestle (note: separate mortar and pestle sets were used for control and enriched samples). The ground tissue was weighed and then rehydrated in 10ml of MilliQ water in a sterile 15ml Falcon tube. Very quickly following rehydration, the sclerites (skeletal fragments) sank to the bottom of the tube and the remaining host homogenate was pipetted into a new tube taking care not to transfer the sclerites. The homogenate was homogenized using a tissue homogenizer for ~15 seconds at maximum speed and centrifuged at 4000g for 5 minutes to separate the fractions. The centrifugation step was repeated as necessary until the host homogenate and zooxanthellae pellets were pure. Following both octocoral and scleractinian fraction separations, 50 ul of pure zooxanthellae from each sample was transferred to a cryovial with 50 ul of 10% paraformaldehyde (PFA) to fix the cells and stored in the refrigerator for future zooxanthellae counts. The pure host homogenates and remaining zooxanthellae pellets were stored in the -20F freezer.

Coral Surface Area Measurements
Coral fragment surface area was calculated following two well-documented methods: Image J measurements and the aluminum foil method (Marsh 1970). ImageJ coral surface area measurements were completed utilizing planar photography and ImageJ software (Schneider et al. 2012). For A. cervicornis and O. faveolata, fragment surface area was measured using the frozen, airbrushed skeleton for each sample, while for E. flexuosa, the entire branch, prior to lyophilization as detailed above, was used. To obtain images of A. cervicornis and E. flexuosa, the fragments were held at an upright position, similar to their natural growth direction, and photographed from four sides (rotated 90°). O. faveolata are dome-shaped mounding corals, so photographs were only taken from above. A ruler was held in alignment with the fragments for scaling purposes. Photographs were individually uploaded to ImageJ and pixel dimensions were set using the straight-line tool and ‘set scale’ option. Using the polygon tool to drag an outline around the perimeter of the fragment, the enclosed area was calculated with the ‘measure’ function (in cm²). The area of all four sides was summed to estimate the surface area of the skeletal fragment.

The aluminum foil method was completed as documented in Marsh (1970), but briefly, small pieces of aluminum foil were cut and carefully measured (cm²) and weighed (g) to obtain a standard weight per unit of area (g/cm²) for the foil. The foil used in this study had a standard weight per unit of area of 0.00618 g/cm². Coral skeletal fragments or octocoral branches were carefully covered with aluminum foil and all excess foil was trimmed until there was no overlap. Each foil wrapping was carefully removed from the fragment and weighed. The fragment surface area was calculated using the standard weight per unit area of the aluminum foil (Surface area of coral fragment = mass of coral fragment foil*0.006180525794 g/cm²). The surface area calculations from both methods were compared for all fragments to ensure a relative consensus between the methods. Results were largely similar between methods and we chose to utilize the surface area estimates from the ImageJ method for downstream analyses as it resulted in slightly higher surface area values, which we deemed more conservative.

Zooxanthellae Counts
To obtain counts of zooxanthellae cells per coral fragment, 20 ul of the fixed zooxanthellae cells from each sample were stained with either 2 ul (E. flexuosa) or 5 ul (A. cervicornis and O. faveolata) of Trypan Blue to increase their visibility under the microscope. If the zooxanthellae cells were too numerous to be counted accurately, they were further diluted with 175 ul of MilliQ water. 10 ul of the dilution was then placed into the cell counting chamber of a hemocytometer (Marienfeld Superior Neubauer Improved Chamber) and cells were counted in 4 quadrants following the standard Neubauer protocol as suggested by Electron Microscopy Sciences (https://www.emsdiasum.com/microscopy/technical/datasheet/68052-14.aspx). We used the following calculations to obtain whole fragment zooxanthellae counts and densities for each coral sample:

Zooxanthellae cells/ml of dilution = (Zooxanthellae cell count*Dilution Factor*10,000 cells/ml)/4 quadrants

Total zooxanthellae cells/fragment = Zooxanthellae cells/ml of dilution*Total Volume of host homogenate

Zooxanthellae cells/cm² of coral fragment = Total zooxanthellae cells/fragment*fragment surface area (cm²)

Known Issues/Problems
Three coral samples (Ef-iso-T3-7, Ef-iso-T6-2, Ac-iso-I-2) were not utilized for zooxanthellae counts because either no zooxanthellae cells were seen in the sample or, during the host and symbiont separations, the ethanol cleaning step was not fully rinsed, which causes the zooxanthellae to burst and therefore they are unable to be counted. These details are noted in the "Notes" column of the dataset.