Abstract
Symbiotic microalgae (Symbiodinium spp.) strongly influence the performance and stress-tolerance of their coral hosts, making the analysis of Symbiodinium communities in corals (and metacommunities on reefs) advantageous for many aspects of coral reef research. High-throughput sequencing of ITS2 nrDNA offers unprecedented scale in describing these communities, yet high intragenomic variability at this locus complicates the resolution of biologically meaningful diversity. Here, we demonstrate that generating operational taxonomic units by clustering ITS2 sequences at 97% similarity within, but not across, samples collapses sequence diversity that is more likely to be intragenomic, while preserving diversity that is more likely interspecific. We utilize this 'within-sample clustering' to analyze Symbiodinium from ten host taxa on shallow reefs on the north and south shores of St. John, US Virgin Islands. While Symbiodinium communities did not differ between shores, metacommunity network analysis of host-symbiont associations revealed Symbiodinium lineages occupying 'dominant' and 'background' niches, and coral hosts that are more 'flexible' or 'specific' in their associations with Symbiodinium. These methods shed new light on important questions in coral symbiosis ecology, and demonstrate how application-specific bioinformatic pipelines can improve the analysis of metabarcoding data in microbial metacommunity studies.