Evaluating environmental DNA metabarcoding for describing freshwater fish and shellfish occurrence in an Eastern Cascades watershed
Skylar Wolf, Lisa Crosson, Marie WinkowskiAbstract
Objective
Environmental DNA (eDNA) metabarcoding offers a promising, cost-effective tool to collect data across a large spatial scale, but analyzing these data poses several critical decisions that are uncommon to traditional field sampling. The objective of our study was to evaluate data collection, filtering, and analytical components of occurrence data generated using an eDNA metabarcoding approach.
Methods
During 2024, we collected replicated water samples from 89 sites distributed across the Wenatchee River Hydrologic Unit Code 8 watershed located in eastern Washington. We processed samples using two metabarcoding markers that targeted the cytochrome c oxidase subunit 1 (CO1; Leray-XT) or 12S ribosomal RNA (MiFish-U) mitochondrial gene region, and we summarized differences in detection, taxonomic coverage, and data filtering requirements for each marker. Finalized metabarcoding data were used to fit occupancy models within a Bayesian framework. We evaluated model results with a focus on species-specific detection probability, the stability of predictions using the full data set and reduced numbers of visited sites, and comparisons between fitting models using a single-species or multispecies occupancy model approach.
Results
The metabarcoding approach generated detections for 26 unique taxonomic assignment groups that closely matched species-level assignments for fish and shellfish. Taxonomic coverage and resolution were greater using the CO1 marker, but the MiFish-U marker appeared to be more sensitive to detections. We found that setting data filtering criteria was an important component of finalizing results—for example, defining minimum read thresholds for detections generated using the MiFish-U marker. Occupancy modeling suggested a high detection probability for most species. Additionally, basinwide estimates of occurrence were comparable between single-species and multispecies models, with increased precision apparent for many species using the multispecies approach.
Conclusions
Collectively, we combined contemporary sampling and analytical approaches to investigate inland fish and shellfish occurrence at a watershed scale. We provide researchers and managers with important considerations for metabarcoding-based monitoring projects. Additionally, our results provide eDNA-inferred occupancy predictions for freshwater species in an eastern Washington watershed.