Arctic Tracker / Methods
How the tracker is built
The Arctic Tracker brings together trade, catch, listing and conservation-status records for 43 CITES-listed Arctic species, so that the effects of a CITES listing and of the trade restrictions that follow it can be explored species by species. This page describes the tracker itself. Analyses built from it, starting with the first paper, are under Research.
What it is built from
| Source | Release held | Obtained |
|---|---|---|
| Arctic illegal-trade seizure records | compiled 2025-07-28 | 2025-07-28 |
| CITES Trade Database | v2025.1 | 2025-07-28 |
| CMS Appendices | download 2025-07-10 | 2025-07-10 |
| IUCN Red List | snapshot 2026-07-22 | 2026-07-22 |
| NAMMCO catch database | scrape 2025-06-11 | 2025-06-11 |
| Species+ API (CITES listing history) | queried 2026-08-12 | 2026-08-12 |
Each source is held as the release named above, and the basis for each date is recorded with it in the deposited table source_releases.csv. The species list and their names are curated: entered by a named person and versioned in the repository.
What a species page holds
- Trade: CITES trade records, per traded term, per unit and per reporting side, with the trend over time.
- CITES listings: the listing history, from the first listing to the current appendix, including split listings.
- IUCN Red List: the assessment history, with the scope of each assessment stated, since a regional assessment is not the species' status.
- Reported catch: catch statistics where they exist, as context for trade.
- Illegal trade seizures: seizure records whose reported taxon resolves to the species.
- Timeline: curated events, each marked when no table in the tracker corroborates it.
Terms used on these pages are defined in the Glossary.
Where it is kept
The tracker is frozen as 100 deposited files, 175.4 MiB, each with a SHA-256 checksum, so that anything built from it can be checked against the exact data it used. The files and their checksums are on the Dataset page.
How to read these pages
Each section declares its provenance before showing anything. Three classes appear throughout:
- derived
- Computed from a checksummed source file. Reproducible by replay.
- curated
- Entered by a named person and versioned in the repository.
- withheld
- Machine-drafted and withheld: no subject expert has signed it.
Three things these pages do differently
Each is stated where it applies rather than in an appendix. The numbers, reasons and finding documents are on What this rebuild shows differently.
- No total quantity. Trade is reported per unit dimension, per traded term and per reporting side. Kilograms and individual animals have no common unit; a live animal is not a carving; and an importer report and an exporter report are two accounts of one shipment.
- Only matched seizures. A seizure row is shown only when its reported taxon matches the attributed species; each shows the name the source reported.
- No unreviewed prose. Machine-drafted narrative sections exist and none is shown. They were withheld because no subject expert had checked them; they have since been checked mechanically, against these same tables. 35 sentences in them name a CITES appendix or an IUCN category. The deposit can adjudicate 26 of those, the ones naming an appendix, and 5 of the 26 name an appendix the listing table contradicts. The other 9 name only a category, which the deposit cannot settle either way: IUCN's terms do not permit redistributing assessments, so it carries their identifiers and not their categories. The sections are being rebuilt from the corpus rather than reviewed. The count withheld is disclosed on every affected page.