A cross-national network for environmental DNA monitoring of nearshore fish biodiversity across the Northeast Pacific

Technical Summary: A Cross-National Network for Environmental DNA Monitoring of Nearshore Fish Biodiversity Across the Northeast Pacific

Problem Statement
Monitoring marine biodiversity at large spatial scales is essential for understanding nature's responses to global change, yet conventional survey methods often lack the scalability and non-invasive nature required for broad inference. While environmental DNA (eDNA) metabarcoding offers a promising solution for detecting marine organisms across extensive spatial and temporal scales, a critical limitation remains: the lack of coordinated, networked approaches. Without standardized methods and synchronized sampling efforts across time and space, it is difficult to disentangle sampling biases from true biological responses or to accurately track biogeographical range shifts. Existing eDNA programs often focus on targeted species or isolated communities, lacking the interoperability necessary for comparative large-scale inference.

Methodology
The authors present the Pacific eDNA Coastal Observatory (PECO), a collaborative international network comprising 18 partners across 16 regional nodes spanning 26 degrees of latitude from San Diego, California, to Juneau, Alaska. The study covers four years of coordinated sampling (2021–2024) focused on eelgrass (Zostera marina) habitats.

• Sampling Design: The network aimed for a balanced design across biogeographic breaks (e.g., Point Conception, Cape Mendocino). Within each region, five sites were visited annually between July 15 and August 15. At each site, four 1.5 L seawater samples were collected along a shoreline transect, filtered through 0.22 μm Sterivex filters using a standardized positive-pressure pumping system. Field blanks were collected at every site.
• Laboratory Processing: DNA was extracted using a modified Qiagen Blood and Tissue Kit. Amplification targeted a fragment of the 12S rRNA gene using MiFish-U primers in a dual-PCR metabarcoding protocol. The study utilized Illumina DNA/RNA UD Indexes to mitigate index hopping and sequenced libraries on a NextSeq 1000 instrument.
• Bioinformatics: A rigorous pipeline was employed involving primer trimming (cutadapt), denoising (DADA2), and chimera removal. Amplicon Sequence Variants (ASVs) were clustered into Molecular Operational Taxonomic Units (MOTUs) using SWARM and curated with LULU. Taxonomic assignment was performed using the CRABS pipeline against a custom database derived from MitoFish and NCBI, with strict filtering to remove non-regional species and non-fish sequences.
• Validation: The study employed three validation approaches:
  1. Beach Seining: Direct comparison with beach seining surveys conducted in 12 regions.
  2. Database Coverage: Assessment of overlap between detected taxa and regional reference databases (FishBase, OBIS).
  3. Sampling Effort Analysis: Evaluation of species richness saturation relative to sequencing depth, number of water samples per site, and number of sites per region.
• Quality Control: Negative controls (field, extraction, and PCR blanks) were processed at every stage. A significant contamination event in 2021 was identified via field blanks with high ASV richness; samples clustering with these contaminated blanks were removed from the final dataset to ensure data integrity.

Key Results

• Dataset Scope: The final dataset comprises 1,041 samples (after quality control) collected across 94 sites and 16 regions.
• Biodiversity Detection: The network detected 64 fish families and 136 identified species, ranging from northern temperate to subtropical fauna. Of the 1,145 unique ASVs, 982 (86%) were resolved to the species level.
• Validation with Beach Seining: In paired surveys, eDNA detected 77% of the species captured by beach seines at the genus or species level (ranging from 49% to 100% across regions). eDNA also detected 41 additional species and 7 additional genera not captured by beach seines, suggesting higher sensitivity for certain taxa.
• Sampling Coverage: Rarefaction analysis indicated that sequencing depth was sufficient to saturate species detection within individual samples (Chao2 coverage = 1). However, coverage at the site and regional levels was incomplete (median coverage ~0.8), suggesting that increasing the number of sites rather than samples per site would be more efficient for capturing regional richness.
• Biogeographic Patterns: The data revealed distinct biogeographic patterns at the family level, with 14 families showing southerly ranges, 24 showing northerly ranges, and 14 spanning the entire study area.
• Non-Native Species: The study identified several non-native species, including Salmo salar (Atlantic salmon), though the authors note that detections of S. salar may reflect genetic material from open-net farms rather than wild populations.

Significance and Claims
The paper claims that the PECO network demonstrates the feasibility and value of a coordinated, networked approach to eDNA monitoring. By standardizing methods across a vast latitudinal gradient, the study provides a high-quality, quality-controlled dataset that serves as a foundation for a biomolecular time-series. The authors assert that this approach:

1. Augments Existing Programs: It allows for comparative large-scale inference that is difficult to achieve with disparate, uncoordinated monitoring efforts.
2. Validates eDNA Utility: The high overlap with beach seining data (77% detection of captured species) validates eDNA as a robust tool for nearshore fish biodiversity monitoring, while its ability to detect additional species highlights its potential for capturing rare or cryptic taxa.
3. Enables Future Inference: The dataset is designed to be interoperable with other datasets (adhering to FAIR principles) and is intended to track biodiversity change over time, specifically aiming for a 10-year data series to support the UN Ocean Decade.
4. Addresses Data Gaps: The study highlights that while sequencing depth is sufficient, spatial sampling effort (number of sites) remains the limiting factor for full biodiversity coverage, providing guidance for future monitoring design.

The authors maintain a modest tone regarding limitations, acknowledging that some species (particularly those with low biomass) are still missed by eDNA and that taxonomic resolution for certain groups (e.g., Sebastes rockfish) remains at the genus level, requiring future methodological refinements. The work emphasizes the importance of Indigenous data sovereignty, noting that data from Haida territory were collected with specific permissions and are subject to CARE principles.