Does phylogeny explain bias in quantitative DNA metabarcoding?

Abstract

Estimating species biomass or abundance from the number of high-throughput sequencing (HTS) reads is an aspirational goal for DNA metabarcoding, yet studies have found varied correlations. Performance varies depending on the gene marker and taxonomic group and, in part, may be related to primer-template mismatches, which are likely to exhibit phylogenetic signals. In this study, we compared commonly used fragments of two gene markers for beetles, the mitochondrial cytochrome c oxidase subunit I (COI) and 16S ribosomal RNA (16S), which have similar lengths, but different propensity for primer-template mismatches. We tested whether primer-template mismatches influence the relationship between species biomass and HTS read abundance and whether the effect of mismatches was explained by phylogeny. A significant correlation between species biomass and HTS read abundance existed for 16S, but not for COI, which had more primer-template mismatches. Models incorporating the effects of mismatch type or number improved the estimation of species biomass from HTS read abundance for COI and strong phylogenetic signals were identified. Researchers seeking to quantify biomass from metabarcoding studies should consider the effect of primer-template mismatches for the taxonomic group of interest and, for beetles, 16S appears a good candidate. Phylogenetic correction can also improve biomass estimation when using gene markers with higher primer mismatching.