Grazing Rates:

We use the term ‘grazing’ to refer to the combination of direct ingestion by P. astreoides and/or grazing by coral-associated protista. Specific rates (units of d-1) were calculated for total picoplankton cells (from DAPI counts), Synechococcus (from flow cytometry), and the SAR11 clade and Rhodobacteracea (from FISH counts). Grazing rates were calculated using the standard equations of Frost (1972). The growth constant (k), or net apparent growth rate for each microbial group was calculated from each mesocosm as:

C2 = C1ek(t2-t1)

where C1 and C2 are the cell concentrations (cells ml-1) at t1 and t2. Grazing coefficients (g) were then calculated as the difference in k between the control and the coral addition mesocosms. Uncertainty in g was calculated using uncertainty propagation (Kline and McClintock 1953). The ANOVA and Tukey ad hoc statistics were done using the aov and tukeyHSD functioHere, wns in the stats package within the R programming language. The Honesty Significant Difference is tabled as the mean HSD value with its associated p value for significance at 95% confidence for all time points (n=3) and for the four groups – picoplankton, Synechococcus, SAR11 and Rhodobacteracea.

Microbial Processing:

The microbial composition of the DNA was assessed by targeting the V4 region of the SSU rRNA gene using modified primers, 515F and 806RB, as outlined by Apprill and colleagues (2015). Triplicate 25 ul PCR reactions were conducted per sample, and contained 1.25 U of GoTaq Flexi DNA Polymerase (Promega Cooperation, Madison, WI), 5X Colorless GoTaq Flexi Buffer, 2.5 mM MgCl2, 200 M dNTP mix, 200 nM of each barcoded primer, and 1 - 4 ng of genomic template. The reaction conditions consisted of an initial denaturation step at 95 degrees celsius for 2 min, followed by 27-34 cycles of 95 degrees celsius for 20 s, 55 degrees celsius for 15 s, and 72 degrees celsius for 5 min, concluding with an extension at 72 degrees celsius for 10 min. Reactions were carried out on a Bio-Rad thermocycler (Bio-Rad Laboratories, Inc., Hercules, CA). Reaction products (5 ul) were screened on a 1% agarose/TBE gel using a HyperLadder 50 bp standard (generally 5 ng ul-1) (Bioline, London, UK). Samples were optimized for the lowest number of cycles that resulted in an amplified PCR product detected on a gel. The three replicate reactions were purified using the QIAquick Purification Kit (Qiagen, Valencia, CA), and quantified using the Qubit fluorescent broad range dsDNA assay (Life Technologies, Grand Island, NY).

Amplicons were pooled in equimolar ratios and shipped to the University of Illinois W.M. Keck Center for Comparative and Functional Genomics and sequenced using 2x250 bp paired-end MiSeq (Illumina, San Diego, CA), as detailed previously (Kozich et al. 2013). Sequence analyses were conducted using mothur v.3.3.3 (Schloss et al. 2009) and included assembly of the paired ends, amplicon size selection (253 bp median size) and alignment to the SSU rRNA gene. Chimera detection was also conducted via UCHIME (Edgar et al. 2011) using mothur, and chimeric sequences (7–18% of sequences per sample) were removed. Taxonomic classification of sequences was conducted with the SILVA SSU Ref database (release 119) using the k-nearest neighbor algorithm. Sequences were grouped into operational taxonomic units (OTUs) using minimum entropy decomposition (MED) (Eren et al. 2015). Statistical analysis of these data was conducted using Primer (v.7, Primer-E Ltd, Ivybridge, UK). Bray Curtis similarity data was used to create Non-metric Multidimensional Scaling (NMDS) and Principle Coordinate (PCO) analyses, ANOSIMS and PERMANOVA tests with the OTU data. Significant differences between sequence percentages were compared using two-way ANOVA with Tukey’s post-hoc tests per time point using Prism 7 (GraphPad Software, Inc., La Jolla, CA). Additionally, the data were tested to determine if PCR cycle number impacted the representation of any OTUs using the BEST analysis with Primer, and did not identify any OTUs correlated with PCR cycles. Sequence data are available at NCBI’s SRA under accession PRJNA 312300.

Quantitative PCR (qPCR) was used to identify betaproteobacterial and archaeal ammonia monooxygenase subunit a genes (amoA) in seawater samples from the inflow, control and coral mesocosms (n = 39), and combined coral tissue plus mucus extracts from a subset of coral colonies in the coral mesocosms (n = 6). qPCR protocols were as described previously (Santoro et al. 2010), with minor modifications. Briefly, 20 ul reactions contained 10 ul mastermix (Bio-Rad ssoAdvanced SYBR), 200 nM of each primer, 1 ul of genomic DNA template, and 8.4 ul of nuclease- free water. Samples and no template controls were analyzed in triplicate along with duplicate sets of standards containing 10 to 105 templates, on a CFX96 real-time PCR machine (Bio-Rad Laboratories). The betaproteobacterial amoA qPCR assay used the 1F/2R primer set (Rotthauwe et al. 1997). The archaeal amoA assay used the Arch-amoAF/Arch-amoAR set (Francis et al. 2005) with an additional 2 mM MgCl2. The detection limit for each assay was approximately 10 genes per reaction, corresponding to a detection limit of 2 genes ml-1 seawater. amoA detection from coral tissue samples was analyzed in undiluted and 1:10 diluted samples; detection is reported as presence/absence.

DMO Notes:

-changed column names to comply with BCO-DMO standards
-filled in blank cells with "nd"
-added NCBI links for all accession numbers
-italicized species names
-separated acquisition and analysis/processing procedures as appropriate