Dataset: Coral SSU rRNA gene database

The coral-microbial database was built over time using the ARB software (Ludwig et al., 2004). Initially, the All-Species Living Tree Project (LTP) s95 database containing SSU rRNA gene sequences for type strains of bacteria, archaea and eukarya was used as the database backbone (Yarza et al., 2008). In 2009, we used the search criteria ‘coral + bacteria’ and ‘coral + archaea’ within BLAST (Altschul et al., 1990) to obtain coral-associated SSU rRNA gene sequences from studies that applied cultivation-dependent or independent approaches from the GenBank database (Benson et al., 2008). These sequences were aligned using SINA (Pruesse et al., 2007) and imported into ARB. This database was then revised by first searching for all sequences coded as ‘coral*’ in any field. From these, sequences that matched ‘eukaryot*’ in the field tax_slv were removed, and all others were marked. An initial search of these marked sequences for those that had *coral* in the field isolation_source was done and these were manually checked. For those that were isolated from a coral (soft or hard, tropical or deep sea, etc.) were assigned ‘coral’ in the remark field. For those that were not isolated from a coral all were assigned ‘checked’ in the remark field. At this stage, there were 1333 sequences remaining with the term *coral* in any field. These were manually checked for their description and isolation details and marked either ‘coral’ or ‘checked’ in the remark field. From these, a database was created that contained just the sequences marked ‘coral’ from our in-house (2009) database. This small curated database was then merged with the SSU_Ref111_SILVA_NR (Pruesse et al., 2007) database (released 19 July 2012) which created new names for the living tree sequences. All sequences that were brought in from the small living tree database were marked with an identifier and the SSU_Ref database was searched as above to locate sequences in the SSU_Ref database that were bacteria or archaea sequences derived from corals. Information was added to these sequences in the location, author, journal and host species fields and all were given the term ‘coral’ in the remark field.

Next, the ISI web of science electronic database was searched for any publications that contained the search term 'coral*' and 'bacteria*' or ‘coral*’ and ‘archaea* in ‘topic’ from 2010 to present. From these, the publications were manually checked and any manuscript that mentioned sequence data or appeared likely to contain Sanger sequence data was obtained. If sequence data was associated with a manuscript, the corresponding sequences were downloaded from NCBI (http://www.ncbi.nlm.nih.gov/) in fasta format, aligned using the online SINA aligner (Pruesse et al., 2007) and imported into arb. If replicates were located they were removed from arb. Newly imported sequences were manually curated to include as much metadata as possible. These included the location (in country field), host (in host field), author names (in author field) and any other accessible data (e.g. Journal).