Importance of full-length amplicons
Although FishCARD typically increased taxonomic resolution of ASVs from our eDNA samples, in one specific case the inclusion of additional California Current-specific 12S barcodes led to substantially reduced taxonomic resolution, with 10 ASVs initially assigned toOxyjulis californica subsequently assigned only to family, Labridae. This result appears to suggest that inclusion of barcodes from local fauna decreases assignment accuracy. However, all 10 of these ASVs were forward sequences only. The Señorita (Oxyjulis californica ) and Rock wrasse (Halichoeres semicinctus ), both native to the California Current share a high degree of similarity in the forward12S sequence fragment, and the diagnostic SNPs occur in the reverse sequence fragment. This issue is resolved when using the full-length merged reads which capture diagnostic SNPs between these two species. Given genetic similarities between these two species, a recent phylogenetic study of new world Halichoeres wrasses has even argued for synonymizing Oxyjulis into Halichoeres(Wainwright et al., 2018).
Despite the above, single direction reads can provide accurate taxonomic assignment for many California Current fishes. For example, within Labridae, the same family as Señorita and Rock wrasse, forward-reads provided species-level resolution for the California sheephead,Semicossyphus pulcher . Such variation in taxonomic assignment based on single-direction sequences highlights the importance of full-length merged reads. To ensure the most accurate taxonomic classifications, researchers should maximize the generation of merged reads by choosing the appropriate sequencing platform (e.g. paired end 300 bp vs. paired end 75 bp).
Although we conducted 300 bp paired-end high throughput sequencing for a 176 bp amplicon, we still obtained a substantial number of orphaned single end reads. This result is largely due to the generation of low quality reverse sequence reads, a persistent issue found across Illumina sequencing platforms (Callahan et al., 2016; Kwon, Park, Lee, & Yoon, 2013). As such, it is important to think carefully about whether or not to include single end reads on a case-by-case and potentially species-by-species basis. In the present study, we were only able to make this determination because of the relative completeness of the FishCARD database, which included all labrids native to the California Current.