Improving eDNA-based protist diversity assessments using networks of amplicon sequence variants

Effective and precise grouping of highly similar sequences remains a major bottleneck in the evaluation of high-throughput sequencing datasets. Amplicon sequence variants (ASVs) offer a promising alternative that may supersede the widely used operational taxonomic units (OTUs) in environmental sequencing studies. We compared the performance of a recently developed pipeline based on the algorithm DADA2 for obtaining ASVs against a pipeline based on the algorithm SWARM for obtaining OTUs. Illumina-sequencing of 29 individual ciliate species resulted in up to 11 ASVs per species, while SWARM produced up to 19 OTUs per species. To improve the congruency between species diversity and molecular diversity, we applied sequence similarity networks (SSNs) for second-level sequence grouping into network sequence clusters (NSCs). At 100% sequence similarity in SWARM-SSNs, NSC numbers decreased from 7.9-fold overestimation without abundance filter, to 4.5-fold overestimation when an abundance filter was applied. For the DADA2-SSN approach, NSC numbers decreased from 3.5-fold to 3-fold overestimation. Rand index cluster analyses predicted best binning results between 97% and 94% sequence similarity for both DADA2-SSNs and SWARM-SSNs. Depending on the ecological questions addressed in an environmental sequencing study with protists we recommend ASVs as replacement for OTUs, best in combination with SSNs.