Efficacy of environmental DNA sampling to detect the occurrence of endangered coho salmon (Oncorhynchus kisutch) in Mediterranean‐climate streams of California's central coast

Abstract

AbstractAnalysis of environmental DNA (eDNA) has emerged as an important tool for investigating the occurrence and distribution of fish and other species in aquatic environments. However, in lotic systems, uncertainty remains about how environmental factors influence the downstream transport, degradation, and dilution of eDNA, and hence how detection probability varies with distance from the eDNA source. We conducted cage experiments and paired eDNA and snorkel surveys to evaluate the potential for eDNA methods to detect endangered coho salmon in Mediterranean‐climate streams of the Santa Cruz Mountains, California. Juvenile coho salmon at two densities were placed in cages, and water samples were collected and analyzed (qPCR) at seven locations from 10 to 1,000 m downstream. Although we detected coho salmon eDNA up to 1,000 m downstream of the cage, the vast majority of detections were within 200 m, and detections at greater distances occurred only at the higher fish density. In field collections from 21 stream reaches across the Santa Cruz Mountains, eDNA methods and snorkeling produced identical outcomes in terms of reach‐level detection of coho salmon. Probability of occurrence in a water sample and detection in a qPCR replicate both increased with observed fish density; however, even at the lowest densities, a protocol of three water samples and two qPCR replicates resulted in a >90% cumulative probability of detection. Overall, our studies suggest that detection probability decreased rapidly with distance in Santa Cruz Mountain streams, likely because low stream discharges, slow transport velocities, and strong interaction between the water column and stream substrate result in rapid degradation or settling of eDNA. Our findings thus support the use of eDNA methods for detecting rare species, but also indicate that local conditions strongly influence transport dynamics and hence need to be considered when developing survey designs.