Sampling Designs for Landscape-level eDNA Monitoring Programs

Effective natural resources management requires accurate information about species distributions. Environmental DNA (eDNA) analysis is a commonly used method to determine species presence and distribution. However, when understanding eDNA-based distribution data, managers must contend with imperfect detection in collection samples and subsamples (i.e., molecular analyses) impacting their ability to detect species and estimate occurrence. Occurrence models can estimate 3 probabilities: occurrence, capture, and eDNA detection. However, most occurrence models do not. To quantify imperfect detection in rare versus common species, we examined multiple field capture and detection probabilities. We studied this with 3 objectives: Determine sample sizes required to detect eDNA given imperfect detection, determine sample sizes required to estimate eDNA capture parameters, and examine performance of a 3-level occurrence model. We found detecting eDNA in ≥1 sample at a site required ≤15 samples per site for common species, but detecting eDNA when looking for rare species required 45 to 90 samples per site. Our occurrence model recovered known parameters unless capture and detection probabilities were <0.2 where >100 samples per site and ≥8 molecular replicates were required. Our findings illustrate the importance of sample size and molecular replication for eDNA-based work. Integr Environ Assess Manag 2019;15:760-771. Published 2019. This article is a US Government work and is in the public domain in the USA.