Detecting fish assemblages with environmental DNA: Does protocol matter? Testing eDNA metabarcoding method robustness

Environmental DNA (eDNA) metabarcoding has recently gain much attention to assess aquatic environment biodiversity. A great variety of protocols have been developed to collect, extract, and analyze eDNA, some of which are continuously evolving and optimized with technological improvements. Such technological shifts might deprecate the biological data produced with earlier protocols, leading to a loss of biological knowledge. Here, we investigated the robustness of an aquatic eDNA metabarcoding method through the comparison of two biodiversity datasets generated by two optimized protocols with different collection and extraction steps. To this end, we compared fish community richness and composition of 12 streams and 3 rivers from French Guiana, sampled with two distinct protocols commonly used in aquatic eDNA studies. Although sample collection with each protocol was not achieved the same year, our results show that species richness and species composition were only slightly affected by the protocol choice, both protocols producing similar fish assemblages at each sample site. Both protocols had a higher replicability in streams than in rivers, strengthening the importance to adapt sampling effort to waterbody type as rivers host a larger number of species than small streams. Despite the need for a standardized approach in eDNA metabarcoding studies, testing the robustness of datasets to protocol variations remains crucial to valorize old data, time series, or data collected in difficult to access locations.