Sample Collection:
Ambient seawater and sponge tissue were collected from two geographically distant locations (~1203 kilometers apart): Conch Reef, Florida (24° 56.9' N, 80° 27.2′ W), and Carrie Bow Cay, Belize (16° 48.14' N, 88° 4.79′ W). Ten of the most common Caribbean coral reef sponges were sampled in Key Largo, Florida (Conch Reef), and eight of these species were sampled in Belize (Carrie Bow Cay) from depths of 13 to 23 meters in June and July 2016, respectively (see Supplemental File named "Microbial communities in sponge species" (Table_1_Finelli_Dataset.pdf)). Only apparently healthy sponge individuals (i.e., no evidence of disease, tissue damage, algal colonization, or epibionts) with a single osculum were sampled (except in the case of Agelas tubulata which had multiple oscula) (McMurray et al. 2018). All sponge tissue samples were collected in separate bags, preserved in 100% ethanol, and stored at -20°C until processing. Seawater samples (1 liter) were collected at each sampling site and day of sponge tissue sample collection, concentrated onto 0.2 micrometer (μm) filters, preserved in 100% ethanol, and stored at -20°C until processing.

Sponge barcoding
All ten sponge species represent common Caribbean coral reef species and were identified morphologically following Zea et al. (2014). To confirm identifications made in the field, PCR amplification of the partial mitochondrial cytochrome c oxidase subunit I (COI) gene was amplified using the forward primer LCO1490 and reverse primer HCO2198 for species barcoding (Folmer et al. 1994). PCR amplification reactions contained 0.5 μl of each primer 10 μM, 12.5 μl (0.5 units) of MyTaq™ Red Mix DNA polymerase (Bioline), 1 μl of DNA template, and PCR water for a total reaction volume of 25 μl. The thermocycler conditions included an initial denaturation step (95 °C, 1 minute) followed by 35 cycles of denaturation (95 °C, 15 s), annealing (45 °C, 15 s), and extension (72 °C, 10 s), with a final extension step (72 °C, 1 min) and 6 °C hold. The COI amplicons were used in a sequencing PCR with BigDye version 3.1 (Applied Biosystems) and a thermocycler program consisting of an initial denaturation step (96 °C, 1min), 25 cycles of annealing (50 °C, 5 s), extension (60 °C, 4 min), and denaturation (96 °C, 10 s), followed by a final annealing (50 °C, 5 s), extension (60 °C, 4 min), and 10 °C hold. Amplicons were cleaned using BigDye® XTerminator™ Purification Kit (Thermo Fischer), following the manufacturer’s protocol, and sequenced on an AB 3500 Gene Analyzer (Applied Biosystems) at the UNCW Center for Marine Science. Forward and reverse sequences were aligned in Geneious version 8.1.9 (Kearse et al. 2012) to create consensus sequences and compared to the GenBank database using the nucleotide Basic Local Alignment Search Tool (BLASTn). Sequence data were deposited in GenBank under the accession numbers MH297440 to MH297461.

Sample metadata
For each sponge tissue sample that was collected, metadata on sponge pumping rates, sponge volumes, and carbon fluxes were collected on in situ colonies (prior to tissue sampling) and processed as reported previously (McMurray et al. 2018). Briefly, sponge pumping rates were measured using an acoustic Doppler velocimeter (SonTek) (McMurray et al. 2014) and sponge tissue volume was estimated using measurements of the dimensions of each sponge (McMurray et al. 2018). Paired incurrent (ambient) and excurrent seawater samples (1.5 L) were collected via syringe and subsequently filtered (Whatman GF/F). POC on filters was quantified using a CE Elantech NC2100 elemental analyzer, and DOC in the filtrate of each sample was quantified using a Shimadzu TOC 5050 analyzer (McMurray et al. 2016). In Belize, an additional 40 mL of the filtrate from each incurrent and excurrent seawater sample was collected and stored frozen until quantification of NOx, NH4, and PO4 using a Bran + Luebbe AutoAnalyzer III following standard protocols (EPA 1997). The specific filtration rate (SFR, μmol C, or nutrient/s/L sponge) for each carbon and nutrient species (i.e., POC, DOC, NOx, NH4, and PO4) was calculated as: SFR ((Cin−Cex) x Q)/Vsponge where Cin and Cex are the incurrent and excurrent concentrations of each carbon pool or nutrient type (C/mL), Vsponge is the sponge tissue volume (L), and Q is the pumping rate for each sponge (mL/s); thus, positive values indicate net consumption and negative values indicate net production of a particular carbon pool or nutrient type.

DNA extraction and sequence processing
Ethanol-preserved tissue samples were dissected into 2 mm³ cubes that included interior and exterior sponge tissue and were extracted using the DNeasy® Blood & Tissue Kit (Qiagen) following the manufacturer protocols. Partial (V4) 16S rRNA gene sequences were amplified using the 515f forward primer and 806r reverse primer (Caporaso et al. 2011) and sequenced on an Illumina MiSeq platform at Molecular Research LP (Shallowater, TX). Illumina sequence reads were processed in mothur v1.38.0 (Schloss et al. 2009) using a modified version of the bioinformatics pipeline described in Weigel and Erwin (2016). Briefly, raw sequences (n = 13.9 million) were demultiplexed, quality-filtered, aligned, classified, and clustered into operational taxonomic units (OTUs) at 97% sequence identity (nOTU = 25,712). Sequence libraries for each sample were subsampled to the lowest read count (n = 15,825), and all data analyses were based on the subsampled dataset. Sequence data were deposited in the Sequence Read Archive (SRA) for microbial data (16S ribosomal RNA gene sequences, Illumina MiSeq platform) of the National Center for Biotechnology under Accession number SRP142647 (https://www.ncbi.nlm.nih.gov/sra/SRP142647).

GenBank Database for sponge data (Mitochondrial cytochrome oxidase subunit I genes, Sanger platform). Accession numbers MH297440 to MH297461. https://www.ncbi.nlm.nih.gov/nuccore/?term=Microbial+Symbionts%2C+Carbon+and+Nutrient+Cycling+in+Caribbean+Coral+Reef+Sponges