Environmental metabarcodes for insects: <i>in silico</i><scp>PCR</scp> reveals potential for taxonomic bias

Studies of insect assemblages are suited to the simultaneous DNA-based identification of multiple taxa known as metabarcoding. To obtain accurate estimates of diversity, metabarcoding markers ideally possess appropriate taxonomic coverage to avoid PCR-amplification bias, as well as sufficient sequence divergence to resolve species. We used in silico PCR to compare the taxonomic coverage and resolution of newly designed insect metabarcodes (targeting 16S) with that of existing markers [16S and cytochrome oxidase c subunit I (COI)] and then compared their efficiency in vitro. Existing metabarcoding primers amplified in silico <75% of insect species with complete mitochondrial genomes available, whereas new primers targeting 16S provided >90% coverage. Furthermore, metabarcodes targeting COI appeared to introduce taxonomic PCR-amplification bias, typically amplifying a greater percentage of Lepidoptera and Diptera species, while failing to amplify certain orders in silico. To test whether bias predicted in silico was observed in vitro, we created an artificial DNA blend containing equal amounts of DNA from 14 species, representing 11 insect orders and one arachnid. We PCR-amplified the blend using five primer sets, targeting either COI or 16S, with high-throughput amplicon sequencing yielding more than 6 million reads. In vitro results typically corresponded to in silico PCR predictions, with newly designed 16S primers detecting 11 insect taxa present, thus providing equivalent or better taxonomic coverage than COI metabarcodes. Our results demonstrate that in silico PCR is a useful tool for predicting taxonomic bias in mixed template PCR and that researchers should be wary of potential bias when selecting metabarcoding markers.