Performance of amplicon and shotgun sequencing for accurate biomass estimation in invertebrate community samples

New applications of DNA and RNA sequencing are expanding the field of biodiversity discovery and ecological monitoring, yet questions remain regarding precision and efficiency. Due to primer bias, the ability of metabarcoding to accurately depict biomass of different taxa from bulk communities remains unclear, while PCR-free whole mitochondrial genome (mitogenome) sequencing may provide a more reliable alternative. Here we used a set of documented mock communities comprising 13 species of freshwater macroinvertebrates of estimated individual biomass, to compare the detection efficiency of COI metabarcoding (3 different amplicons) and shotgun mitogenome sequencing. Additionally, we used individual COI barcoding and de novo mitochondrial genome sequencing, to provide reference sequences for OTU assignment and metagenome mapping (mitogenome-skimming) respectively. We found that even though both methods occasionally failed to recover very low abundance species, metabarcoding was less consistent, by failing to recover some species with higher abundances, probably due to primer bias. Shotgun sequencing results provided highly significant correlations between read number and biomass in all but one species. Conversely, the read-biomass relationships obtained from metabarcoding varied across amplicons. Specifically, we found significant relationships for 8 out of 13 (amplicons B1FR-450bp, FF130R-130bp) or 4 out of 13 (amplicon FFFR, 658bp) species. Combining the results of all three COI amplicons (multi-amplicon approach) improved the read-biomass correlations for some of the species. Overall, mitogenomic sequencing yielded more informative predictions of biomass content from bulk macroinvertebrate communities than metabarcoding. However, for large scale ecological studies, metabarcoding currently remains the most commonly used approach for diversity assessment