Influence of accuracy, repeatability and detection probability in the reliability of species-specific eDNA based approaches

Environmental DNA (eDNA) barcoding has a high potential to increase the cost-efficiency of species detection and monitoring in aquatic habitats. However, despite vast developments in the field, many published assays often lack detailed validation and there is little to no commonly (agreed upon) standardization of protocols. In this study, we evaluated the reliability of eDNA detection and quantification using published primers and assays targeting the Freshwater Pearl Mussel as a model organism. We first assessed limits of detection for two different target genes (COI and 16S) following the MIQE guidelines, and then tested the reliability of quantification in a double-blind mesocosm experiment. Our results reveal that different methodological indicators, namely accuracy, repeatability and detection probability affected the reliability of eDNA measurement at the different levels tested. The selection of the optimal analytical method was mainly determined by detection probability. Both the COI and 16S assays were highly specific for the targeted organism and showed similar accuracy and repeatability, whilst the limit of detection was clearly lower for the COI based approach. In contrast, the reliability of eDNA quantification hinged on repeatability, reflected by the scattering (r(2) = 0.87) around the relationship between eDNA and mussel density in mesocosms. A bootstrapping approach, which allowed for the assignment of measures associated with repeatability of samples, revealed that variability between natural replicates (i.e. accuracy) strongly influenced the number of replicates required for a reliable species detection and quantification in the field.