Environmental DNA reads the shed cells of everything that passed through a litre of seawater. eDNAtlas turns that into a map you can click: one number and one plain sentence per site, with the full evidence a click deeper.
Every marker is a coastal sampling site scored by the Ecosystem Health Index. Pick one to read it three ways: a glance, a plain-language meaning, and the raw evidence underneath.
Real Southern California sites and coordinates; the site-level scores are modeled illustrations of the health index until wired to live sequence data (see the real dataset below). The interaction, the scoring model, and the layout are the product.
The Southern California sites above are modeled on the methodology. This one is not: these are actual Darwin Core records from a real marine environmental-DNA study, loaded straight into the same interface.
| Botrylloides sp. | Ascidiacea | 85 reads |
| Botryllus schlosseri | Ascidiacea | 56 reads |
| Ascidiella sp. | Ascidiacea | 6 reads |
| Molgula sp. | Ascidiacea | 2 reads |
Real records from a Darwin Core conversion (DaCA-utils, Flanders Marine Institute) of the study "Environmental DNA reveals seasonal shifts and potential interactions in a marine community" (Djurhuus et al., Nature Communications, 2020). Full sequence data at NCBI PRJNA433203.
A ranking turns a lone score into a story a community can act on. Sites are only compared against like-for-like habitat classes, so an enclosed harbour is never scored against open kelp forest.
Civic pride is a retention loop. People check back on a place they are attached to, and a rank gives them a reason to.
It makes trends legible. A score of 51 means little alone. "51, sixth of eight, down two places" is a story.
It creates pressure. Being last on a public list moves municipal budgets that a PDF report never will.
Ranking without habitat matching is just geography. eDNAtlas normalises every score against a habitat-matched reference site, so 74 means the same thing in two different bays.
Bars share one hue: this is a single measure across sites, not eight categories.
Every stage emits something a machine can consume. The map is the last step, not the whole product.
FASTQ, ASV tables, or a Darwin Core archive. Field metadata captured at submission, not reconstructed later.
Mapped to Darwin Core plus the DNA-derived data extension. Units, CRS, markers, and primers reconciled.
Sequences matched against reference libraries. Confidence and the library version stored with the call.
The Ecosystem Health Index computed from stored inputs. Recomputable, versioned, never hand-edited.
Open REST, bulk download, and the map. The same numbers through every door.
Recomputable. Stored inputs and versioned weights. Anyone can rerun the score and get ours.
Comparable. Normalised against a habitat-matched reference site.
Contestable. Every component links to the evidence behind it. Disagreement is a feature.
eDNA tells you what DNA was in the water, not how many individuals or whether they were alive. We label detections as detections and never print an abundance we did not measure.