How we handle the data

The registry combines records from several upstream sources, each authoritative for a different slice of the picture:

• IUCN Red List — the source of record for conservation assessments (category, criteria, assessment date, assessors).
• FishBase — ecological and morphological data (habitat, maximum length, diet notes).
• GBIF — occurrence records (georeferenced observations and museum specimens) published under Darwin Core.
• CARES — the Conservation, Awareness, Recognition and Encouragement for Species priority list maintained by the fishkeeping-hobbyist community, distinct from the IUCN Red List and focused on species held in private breeding programs.
• SHOAL 1,000 Fishes Blueprint — global priority alignment for freshwater fish conservation.
• Expert contributions — named researchers and keepers who have published or shared records for Malagasy endemics. Individual contributors are credited on the records they informed.

The IUCN category shown on every species profile is a mirror of the most recent accepted assessment for that species. The platform does not hand-edit the badge. When a new IUCN assessment is ingested, the mirror updates; when no assessment exists, the profile reads "not yet assessed" rather than showing a stale category.

Manual expert input is supported, but only by creating a new assessment record with an assessor, date, and reasoning — the same path IUCN data takes. If a human-reviewed assessment disagrees with an incoming IUCN assessment, the conflict is flagged for review rather than silently overwritten. One source of truth, auditable back to the assessment record.

Occurrence records for non-sensitive species are shown with their original coordinates. For species assessed as threatened (Critically Endangered, Endangered, or Vulnerable), coordinates are generalized to roughly 0.1 degrees — about 11 km at the equator — following GBIF's sensitive-species best practices. This obscures exact localities while preserving the overall distribution pattern, which is what the public map is for.

Exact coordinates for sensitive species are available to conservation coordinator accounts (Tier 3) and above, for the organizations coordinating field work and transfers. The rationale is straightforward: small, range-restricted populations have been targeted by collectors in the past, and obscuring the precise pin is a cheap and effective protection.

The upstream record is imperfect, and the platform does not pretend otherwise.

• Transcription errors in source records. A small number of GBIF occurrences place Madagascar endemics in the Indian Ocean, on the wrong coast, or in other countries entirely. These are almost always data-entry or georeferencing errors in the source. The platform flags suspect records for admin review rather than silently dropping them; corrected records are published back where possible.
• Undescribed species. Madagascar's freshwater fish fauna includes morphospecies awaiting formal description. These appear in the registry under provisional names (for example, Ptychochromis sp. "Tsimembo") with a status of "not yet assessed." They are real species that keepers and researchers are working on; leaving them out would misrepresent the fauna.
• Introduced species. Introduced and invasive species in Malagasy waters (tilapia, gambusia, and others) are hidden from the directory by default to keep the focus on endemics. They can be surfaced with a filter toggle, because they are part of the ecological picture even when they are not the subject of conservation effort.
• Sparse husbandry data. For many species, published keeping and breeding guidance is thin or absent. Profiles say so plainly rather than inventing content.

Some data is deliberately behind tier-gated accounts:

• Exact coordinates for sensitive (threatened) species.
• Per-institution captive inventory detail from ZIMS and partner records (public view shows aggregates only).
• Breeding-program genetics, studbook-level data, and transfer recommendations.
• Contact information for individual keepers and researchers.

Access is granted by the platform operator on request to qualified individuals at partner organizations.

Occurrence data is modeled against the Darwin Core standard, which is the biodiversity community's shared vocabulary for species records. Publication of the platform's datasets to GBIF via an Integrated Publishing Toolkit (IPT) instance is planned after the MVP. IUCN categories, assessment criteria, and CARES priority status are carried through without re-interpretation.

If you see a record that looks wrong, or you have data that should be here and is not, the About page has contact and repository links. Corrections are welcomed and credited.