Establishing the Signal above the Noise: Accounting for an Environmental Background in the Detection and Quantification of Salmonid Environmental DNA

Abstract

A current challenge for environmental DNA (eDNA) applications is how to account for an environmental (or false-positive) background in surveys. We performed two controlled experiments in the Goldstream Hatchery in British Columbia using a validated coho salmon (Oncorhynchus kisutch) eDNA assay (eONKI4). In the density experiment at high copy number, eDNA in 2 L water samples was measured from four 10 kL tanks containing 1 to 65 juvenile coho salmon. At these densities, we obtained a strong positive 1:1 relationship between predicted copy number/L and coho salmon biomass (g/L). The dilution experiment simulated a situation where fish leave a pool environment, and water from upstream continues to flow through at rates of 141–159 L/min. Here, three coho salmon were placed in four 10 kL tanks, removed after nine days, and the amount of remaining eDNA was measured at times coinciding with dilutions of 20, 40, 80, 160, and 1000 kL. The dilution experiment demonstrates a novel method using Binomial–Poisson distributions to detect target species eDNA at low copy number in the presence of an environmental background. This includes determination of the limit of blank with background (LOB-B) with a controlled false positive rate, and limit of detection with background (LOD-B) with a controlled false negative rate, which provides a statistically robust “Detect” or “No Detect” assessment for eDNA surveys.