Advancing aquatic biodiversity assessments of invertebrates using <scp>eDNA</scp> metabarcoding: A systematic evaluation of primers for marine and freshwater communities

Abstract Invertebrate richness and community composition are key indicators of aquatic ecosystem health. Compared with morphotaxonomy‐based survey methods, DNA metabarcoding is a highly efficient and scalable biomonitoring approach that has become increasingly used in invertebrate surveys. However, the accuracy and efficacy of this approach depend on the performance of universal primers; yet systematic comparative assessments of universal primer performance are lacking. Here, we evaluated the performance of 18 commonly used invertebrate metabarcoding primer pairs targeting the mitochondrial COI and nuclear 18S rRNA genes. We compared the taxonomic specificity, coverage and resolution of the primers via in silico and in vitro assessments using environmental DNA (eDNA) samples from both natural freshwater (rivers and lakes) and marine coastal habitats. We also assessed the effects of sequence reference databases (e.g. GenBank, BOLD, SILVA) on taxonomic assignments. The primers were highly variable in their specificity both in silico and in vitro. The in silico results showed that invertebrate sequences accounted for greater proportions of the total sequences amplified using the COI primers compared with the 18S primers; similar results were obtained in the in vitro assessments, but the proportions of invertebrate sequences were lower. Each primer pair detected 91–1524 invertebrate taxa from freshwater eDNA and 74–624 from marine eDNA and the taxonomic resolution was limited. The COI primers recovered substantially greater taxonomic richness than the 18S primers, but additional phyla were detected by certain 18S primers. The COI primers also preferentially recovered specific invertebrate orders such as bioindicator insects. We provide recommendations for primer and sequence database selection for the metabarcoding of invertebrate biodiversity. Overall, our findings have implications for improving the efficacy of (e)DNA‐based invertebrate surveillance and ecosystem bioassessments. Our results also highlight the gaps within public invertebrate sequence databases.