Shotgun metagenomics for AIS surveillance — a 2026 landscape snapshot

This page summarises where the published literature stands on using shotgun metagenomic sequencing — as distinct from PCR-based metabarcoding — to surveil aquatic invasive species (AIS). It exists because we keep being asked "is there a standard reference panel for this?" and the honest answer requires some unpacking.

The short answer

There is no off-the-shelf shotgun-metagenomic AIS reference panel in published use. The dominant DNA-based surveillance approach is amplicon-based: COI, 12S MiFish, 16S rRNA, or species-specific qPCR assays, typically read on Illumina or — increasingly — Nanopore. Shotgun approaches are emerging but remain a research method, not an operational one, for invasive-species early detection.

What the operational programs distribute today:

• Curated taxonomic species lists (e.g. NOAA's GLANSIS for the Laurentian Great Lakes, USGS NAS nationally, state DNR watchlists). These are species-level inventories and watchlists, not genomic resources.
• qPCR / amplicon marker assays for high-priority targets (zebra/quagga, bigheaded carps, round goby, etc.).
• Genome assemblies in NCBI/ENA for individual species — never curated as a "panel".

What we are building under nanopore_live/--mapping_refs sits somewhere between (1) and (3) — a user-assembled panel of WGS references, used to map shotgun nanopore reads against species-by-species and stored per-barcode in a mapping DuckDB table. The concept exists in the literature; the operational pipeline does not.

Why shotgun metagenomics is rare for AIS, and changing

Metabarcoding has dominated because:

• Amplicons concentrate the signal — a target read after PCR is far more likely to be the species you care about than a random shotgun read.
• Marker references (BOLD, MIDORI, MitoFish) are vastly more complete than whole-genome references for invertebrates.
• Cost per sample is much lower for amplicons than for shotgun at any useful depth.

The arguments for moving toward shotgun:

• No primer bias. Whatever DNA you can sequence, you can in principle classify.
• Single library serves many taxa — invertebrates, vertebrates, microbes — without separate primer panels.
• Per-position alignment evidence: position-of-hit distributions, identity histograms, and CIGAR-level data let you distinguish clean signal from cross-mapping or repetitive artifacts in ways amplicons cannot.
• Nanopore long reads reduce ambiguity at the species level by spanning multiple polymorphic regions per read.

The current blockers are (1) reference-genome coverage and (2) the absence of curated panels — exactly the gap the recent literature is starting to name.

Key references (chronological)

Spens et al., bioRxiv 2020 amplicon / nanopore

Speeding up the detection of invasive aquatic species using environmental DNA and nanopore sequencing. The most-cited "nanopore + AIS" paper to date. Uses 12S MiFish amplicons; nanopore is the read-out platform. Demonstrates rapid in-field detection but does not depart from the metabarcoding paradigm.

Urban et al., eLife 2021 shotgun / nanopore

Freshwater monitoring by nanopore sequencing. Shotgun-style nanopore metagenomics applied to river water; identifies microbial and eukaryotic taxa from raw reads. Closer to what we are doing, but the framing is "biodiversity / pollution monitoring", not "AIS surveillance".

Rossi et al., Mol Ecol Resour 2022 methodology

Genomic data is missing for many highly invasive species, restricting our preparedness for escalating incursion rates. Audits genomic resources for known-invasive taxa. Reports that population-genetic data exists for ~82% of high-impact invasives but population-genomic data only for ~32%; reference genomes are unavailable for a large fraction. Methodologically useful: supplementary tables function as a per-species inventory of what is sequenceable today.

Buchner et al., Nat Comms 2022 shotgun / reference design

Shotgun metagenomics of soil invertebrate communities reflects taxonomy, biomass, and reference genome properties. Closest methodological analog to AIS shotgun panels. Uses ~270 invertebrate genomes as a reference database and characterises systematic biases: larger genomes and more contiguous assemblies attract disproportionately more reads at the same true biomass. Important caveat for any panel that mixes chromosome-scale and fragmented references.

McCartney et al., Front Environ Sci 2023 policy

Time to invest in the worst: a call for full genome sequencing of the 100 worst invasive species. Explicit call to close the reference-genome gap for the IUCN "100 worst" list, of which (as of writing) ~55% lack a reference genome. Frames panel-building as policy infrastructure, not just a research output.

Various, 2024–2026 surveys

The recent reviews — Carøe et al. (NAR Genom Bioinform 2024), the 2025 multi-species eDNA screening review, and the 2025 nanopore-vs-Illumina AIS-tracking comparison — consistently treat shotgun metagenomics as an emerging but not standard approach. None describe a published, named AIS shotgun panel.

The reference-genome gap

Empirically, when we tried to assemble a Lake Winnipeg–relevant panel by walking down the AIS list and querying NCBI Datasets:

That ratio — chromosome-scale for charismatic vertebrates and the dreissenid mussels, marker-pool or nothing for most invertebrates — is exactly the pattern Rossi et al. (2022) and McCartney et al. (2023) document at scale. It directly constrains which species can be detected by shotgun mapping at all.

Implementation considerations for a shotgun-metagenomic AIS pipeline

1. Reference-genome biases are real. Buchner et al. show that read assignment is biased by genome size and assembly contiguity. A panel mixing a 140 Mb CDS-only reference with a 1.8 Gb chromosome-scale assembly will report wildly different hit counts at the same true biomass. Standardising to whole-genome assemblies of comparable contiguity, where available, is worth the rebuild.
2. Close relatives cross-map. Sister species (zebra ↔ quagga; silver ↔ bighead carp; sea ↔ silver lamprey) will share reads at moderate identity. Per-sample identity histograms and a tunable HQ-identity cutoff (e.g. shifting from 90% to 96–98% for sister-taxon cases) are essential UI tools; baking a single cutoff into the data is a mistake.
3. Position-of-hit distributions disambiguate signal. A read pile-up clustered to one repetitive locus is an artifact; uniform spread across the genome is consistent with real coverage. Per-sample linear-genome histograms make this visually obvious and add discriminative power that amplicons cannot offer.
4. Marker-only "references" are still useful. For taxa with no WGS assembly (water flea, crayfish, many bryozoans/mollusks), a multifasta of the available GenBank marker records (COI / 18S / mitochondrion) gives a usable but lower-sensitivity reference. Label it as such in the pipeline so downstream users do not over-interpret hit counts.
5. Per-position evidence belongs in the database. Storing (reference, qname, rname, pos, mapq, identity_pct, cigar, aligned_len) rows in DuckDB — rather than pre-summarising to counts — preserves the ability to re-derive thresholds and visualise alignment quality after the fact.

Where danaSeq fits

The nanopore_live pipeline's --mapping_refs module (added 2026-05-28) implements the design above:

• Directory of reference subdirs, each holding <name>.idx + <name>.contigs.json + meta.json.
• Per-barcode minimap2 alignment, filtered to mapq ≥ 1 and aligned length ≥ 10 bp, streamed into a mapping DuckDB table keyed on reference name.
• SPA (the AIS view at /ais on each lab's deployed dashboard) provides per-species cycling, a live HQ-identity slider, per-sample identity and genome-position histograms, and a dynamic-range exponent for marker sizing.

The implementation is documented in detail in docs/mapping-references.md in the danaSeq repository.

References cited above. Where available, links go to the open-access PMC version.

• Spens J. et al., bioRxiv 2020 — Speeding up the detection of invasive aquatic species using environmental DNA and nanopore sequencing.
• Urban L. et al., eLife 2021 — Freshwater monitoring by nanopore sequencing.
• Rossi G. et al., Mol Ecol Resour 2022 — Genomic data is missing for many highly invasive species.
• Buchner D. et al., Nat Comms 2022 — Shotgun metagenomics of soil invertebrate communities reflects taxonomy, biomass, and reference genome properties.
• McCartney M. et al., Front Environ Sci 2023 — Time to invest in the worst: a call for full genome sequencing of the 100 worst invasive species.
• Carøe C. et al., NAR Genom Bioinform 2024 — Long-read nanopore shotgun metagenomic DNA sequencing for river biodiversity monitoring.
• NOAA GLERL, GLANSIS database — glerl.noaa.gov/glansis.
• IUCN/ISSG, 100 of the World's Worst Invasive Alien Species — iucngisd.org/gisd/100_worst.php.