Comparison of environmental DNA metabarcoding and conventional fish survey methods in a river system

Regular biological surveys are essential for informed management of freshwater ecosystems. However, current morphology-based biodiversity surveys can be invasive, time-consuming, and financially expensive. Recently, environmental DNA (eDNA) sequencing has been suggested as an alternative non-invasive, time- and cost-effective biological survey tool. However, eDNA sequencing tools require experimental validation in natural ecosystems before confidence in their use can be assumed. In this study, we compare fish community data obtained via eDNA metabarcoding to that of conventional fyke netting within two complex and drought-prone river systems. We also compare different eDNA sampling strategies and genetic markers for detecting rare and threatened fish species. We were able to detect 100% of the fyke net caught-species from eDNA when appropriate sampling strategies were used, including threatened and invasive species. Specifically, we found that two 1 L water samples per site were insufficient for detecting less abundant taxa; however, five 1 L samples per site enabled a 100% detection rate. Further, sampling eDNA from the water column appeared to be more effective for detecting fish communities than eDNA from sediments. However, on a per site basis, community discrepancies existed between the two methods, highlighting the benefits and limitations of both approaches. We demonstrate that careful interpretation of eDNA data is crucial as bioinformatic identification of sequences, without logical inference or local knowledge, can lead to erroneous conclusions. We discuss these discrepancies and provide recommendations for fish eDNA metabarcoding surveys.