This study was conducted at an artificial reef near Looe Key reef in the Florida Keys (USA), depth 10 meters. The artificial reef was removed after the experiment was completed.

To examine differences in nutrient profiles across sponge species, an artificial reef was constructed with 5 rows, with 10 cinder blocks per row, the rows were arranged in a semi-circle, and rebar was run through each individual row and secured at the middle and ends to ensure that the rows would not move over the duration of the experiment. Species (n = 10 replicates of 10 species) were placed on the rows in roughly a Latin-square design to spread out any environmental variable that may exist on the artificial reef across replicates of species. The depth of the reef was approximately 7 meters and immediately surrounding the reef was sand on all sides. The rows of sponges were placed so that they were as equal as possible in exposure to light and water flow. Two pieces of 5-millimeter (mm) polystyrene mesh were placed on top of each of the cinder blocks and secured with cable ties. Ten individuals of each of the 10 focal species (Aiolochroia crassa, Aplysina cauliformis, Aplysina fulva, Amphimedon compressa, Iotrochota birotulata, Ircinia felix, Niphates digitalis, Callyspongia aculeata, Verongula rigida, and Xestospongia muta) were selected and individuals were attached to the polystyrene mesh using cable ties. The sponges were allowed to acclimate for 4 months. The sponge species were selected based on prevalence in the Caribbean and included four low microbial abundance (LMA) species (Amphimedon compressa, Iotrochota birotulata, Niphates digitalis, Callyspongia aculeata) while the remaining six species are considered high microbial abundance (HMA) sponges.

In Situ Water collection:
Samples for all nutrient analyses were collected using a modular vacuum setup (VacuSIP) which was implemented and modified from Morganti et al. 2016. The VacuSIP included poly ether ether ketone (PEEK) tubing (used commonly in HPLC instruments) that was placed over the sponge osculum (sponge exhalent seawater) or near the sponge (inhalant seawater) and positioned using tripods. The tubing was connected via syringe needle to pressurized 250 milliliter (mL amber glass bottles with Teflon septate caps. The 250 mL glass vials were pre-combusted within days of sampling (6 hours at 450 °Celsius) and pressurized manually to -15 psi and this pressure increased slightly at the depth of the artificial reef. VacuSIP lines were acid rinsed in 10% HCl. Pumping was confirmed using fluorescent dye before each collection and the dye was allowed to clear before sampling. Tubing was then positioned directly above the pumping oscula for small sponges or inside the oscula close to the sponge for larger sponges. VacuSIP lines were then attached to the appropriate 250 mL bottles in the crates by sticking the needles at the end of the line into the septa of the bottle. Tubing for inhalant water collection was inserted into the appropriate 250 mL bottles. The apparatus contained fewer lines than 250 mL collection bottles, so the apparatus was set up to fill half of the bottles in the collection crate for 120 minutes, then the lines were moved to new bottles to fill the remaining bottles for 90 minutes. Once at the surface, the 250 mL bottles for each individual sample were combined into 2 liter (L) bottles, one for inhalant and one for exhalent water samples, that were labeled and then stored on ice in coolers until they were filtered.

Water Filtration:
A Cole-Palmer Masterflex L/S Intertek fitted with a Masterflex L/S easyload II head and Multichannel Pump Head Cartridges was used to filter the samples at a rate of 40 mL per minute through High-Performance Precision Pump Tubing, PharMed® BPT, L/S 15 with a 0.2 micrometer (µm) Supor filter into 1L acid-rinsed polycarbonate bottles that were covered with aluminum foil. The filters were archived for future microbiome analysis and the filtrate was stored in 40 mL amber vials with Teflon septa. Filtrate for dissolved organic carbon (DOC) and total dissolved nitrogen (TDN) was acidified to ~pH 2 using 6N HCl and stored at 4°Celsius. Samples for DOC and TDN were sent to the UGA Stable Isotope Ecology Laboratory for analysis using a Shimadzu TOC-5000A Total Organic Carbon Analyzer. While DOC and TDN samples were collected, there was contamination in these samples from an unknown source and therefore, DOC and TDN were not used in any data analysis. Filtrate for fluorescent dissolved organic matter (fDOM) was not acidified and stored at 4°Celsius. The fDOM samples were shipped to a collaborator at the University of Hawaii (Dr. Craig Nelson) where they were stored until analysis. The investigator worked with the Nelson lab at UH to process and analyze the fDOM as samples described below.

fDOM Sample Processing:
Following the methods from Nelson et al. (2015), samples were analyzed with a Horiba Aqualog scanning fluorometer with 150 watts Xe excitation lamp, Peltier-cooled CCD emission detector, and simultaneous absorbance spectrometer. Quartz cuvettes of 1-centimeter (cm) diameter, which were DIW-leached and rinsed, were used to measure fluorescence. Samples were brought to room temperature while the Xe bulb warmed up.

Known problems or issues:
Not all replicates of each sponge were sampled. Dissolved Organic Carbon (DOC) and Total Dissolved Nitrogen (TDN) values are absent from the dataset due to contamination of samples.

Percent change in nutrients was calculated only for fDOM components, not for inorganic nutrients or for particulates captured on the GF/F filters.

Several samples were not analyzed for fDOM or for inorganic nutrients (broken or lost), and only a subset of samples were analyzed with the GF/F filters.