Environmental metabarcodes for insects: in silicoPCR reveals potential for taxonomic bias

Abstract

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.

Effect of marker choice and thermal cycling protocol on zooplankton DNA metabarcoding studies

“…Although the use of COI for metabarcoding has been questioned due to lack of conserved primer‐binding sites (Clarke et al., 2014; Deagle et al., 2014), the taxonomic coverage and resolution provided by degenerate COI primers, combined with a comparatively well‐developed reference sequence database, make them valuable metabarcoding markers for biodiversity assessment. The potential for retrieving at least semiquantitative abundance data was confirmed for all markers, with 18S providing the strongest relationship between calanoid copepod biomass and number of HTS reads.…”

“…The detection of additional taxa supports the use of low annealing temperatures to maximize taxonomic coverage for any given marker (Clarke et al., 2014; Sipos et al., 2007). Amplification and sequencing of COI amplicons generated using the low annealing temperature protocol was highly repeatable in spite of using different sequencing chemistries (v2 and v3) and number of sequencing cycles (2 × 250 and 2 × 300), with OTUs representing >0.1% of reads in one replicate almost always detected in the corresponding replicate.…”

“…However, as COI is a protein‐coding gene, “third codon wobble” increases the chance of primer mismatches when targeting genetically diverse taxonomic groups. Indeed, the lack of conserved primer‐binding sites has been shown to cause taxonomic bias for many COI markers (Clarke, Soubrier, Weyrich, & Cooper, 2014; Piñol, Mir, Gomez‐Polo, & Agusti, 2015). Mitochondrial 12S and 16S rDNA has been proposed as an alternative source of metabarcoding markers (Clarke et al., 2014; Deagle, Jarman, Coissac, Pompanon, & Taberlet, 2014; Epp et al., 2012) to avoid taxonomic bias introduced by primer‐template mismatches but retain taxonomic resolution.…”

Protecting an Ecosystem Service

“…from 708 multiple pan traps), keeping costs low. However, this methodology is associated with various 709 problems such as biases in amplification success across different taxa, which may create false 710 negatives (Clarke et al, 2014;Tang et al, 2015), contamination risk and potential co-711 amplification of mitochondrial pseudogenes (Song et al, 2008), and the comparatively short 712 sequence achievable with the current NGS technology, which limits the analysis of the COI 713 gene to roughly a half-length 'minibarcode' and hence reduces discriminatory power (Tang et 714 al., 2015). 715 for each sample; these indexed primers can be used to sort the samples out bioinformatically 727 after sequencing.…”

Rare biosphere exploration using high-throughput sequencing: research progress and perspectives

“…Type I error has been widely reported from both laboratory-based and silicobased surveys for biodiversity assessment using HTS (e.g. Clarke et al 2014;Liu et al 2013;Ovaskainen et al 2013;van Velzen et al 2012;Toju et al 2012). This type of error is especially acute for those studies focusing on rare taxa (Bellemain et al 2010;Engelbrektson et al 2010).…”