Because traditional taxonomic methods to survey diversity are time consuming and require high taxonomic expertise, DNA metabarcoding is becoming a popular tool to survey biodiversity. Metabarcoding techniques allow for quick, low-cost, whole community DNA sequencing, but performance can be influenced by the choice of genetic marker and the completeness of the reference database used for the taxonomic matching. We compared the taxonomic identification of zooplankton samples in the Gulf Stream with morphology, barcodes, and a metabarcoding approach using three genetic markers (COI, 18S V1-2, 18S V9) and two types of sequence clustering (OTUs and ZOTUs). We subsampled plankton tows, sorted, identified and barcoded one portion while processing another for metabarcoding. We used a gap analysis tool to determine how a short targeted effort by a group of taxonomists can add new sequences to reference databases (such as GenBank) and evaluated if the new barcodes database can help improve the outcome of taxonomic assignment using metabarcoding. Additionally, we compared the number and identity of taxa found with metabarcoding for the different genetic markers and types of clustering.
Results/Conclusions
The StreamCode barcodes database resulted in 1,067 COI sequences and 1,137 18S V1-2 sequences. The largest contribution of new sequences of barcode quality to GenBank (by Dec 2019) was for cnidarians, polychaetes, sipunculans, and gastropods. We found important differences in the metabarcoding results for the different markers. As expected given the variability of each marker, more taxa were detected using COI for most taxonomic groups. On the other hand, 18S V1-2 performed much better to assign taxa to phylum, while COI and 18S V9 had a higher percentage of unidentified taxa. Although both markers and clustering types agree on crustaceans being the dominant group, some combinations of marker and clustering types identified different taxa as the second more abundant group (cnidarians or molluscs). After matching the unidentified sequences with the StreamCode barcodes database, we improved the number of taxa assigned to Phylum for COI. The taxonomic differences in the taxa identified with each marker suggest that a multi-marker approach should be used when using metabarcoding methods for biodiversity surveys. The taxonomic experts identified unique families not identified by metabarcoding, which suggest that initial implementation of metabarcoding methods will benefit from collaboration with taxonomists to improve the resolution of taxonomic assignments.