Dead names in a frozen archive: a name-by-name re-identification of the stranded Fonozoo frog-call pile, with a maintained-archive control
Sound archives assembled before the molecular era are filed under the taxonomy of their day. When such an archive is abandoned — its files stranded on the web, its catalogue frozen — every subsequent taxonomic revision silently falsifies a fraction of its labels, and no one re-counts them. We recovered the complete stranded web pile of the Fonoteca Zoológica (MNCN, Madrid): 1,013 frog-call audio files published as web companions to printed sound guides, present in no formal archive. Of the 629 recordings filed under firm species names, 146 (23.2%) no longer carry that name in the GBIF Backbone Taxonomy — predominantly genus recombinations (129 recordings), with 15 junior synonyms and 2 split elevations; a further 6 recordings (5 names) are split-ambiguous, their filed name now denoting a complex of two or more species. Joining the institution's own catalogue (3,618 Darwin Core records) resolved 4 of the 6 ambiguous recordings on geographic evidence. The burden is region-structured: Neotropical recordings changed at 30.7% (genus moves), Palearctic recordings carry a further 16% split-ambiguity (splits, not moves), and Malagasy recordings only 5.7%, their guide postdating the region's genus reorganisation. A second ruler — the Catalogue of Life (ASW lineage) — agrees with the GBIF Backbone on 87.4% of names and leaves the aggregate rate nearly unchanged (24.2% vs 22.8%), yet genuinely disagrees on 18 names (a whole genus, Platypelis, among them) and drops a further 47 of the pile's misspelt filename names that GBIF's fuzzy matcher rescues — so a re-identification is only as stable, and as complete, as the backbone and matcher it is run against. As a control, we ran the identical instrument over the maintained Fonoteca Neotropical Jacques Vielliard (5,793 firm-named amphibian recordings): 1.1% changed, its residue running in the opposite direction (recent moves and newly described species the backbone has not yet absorbed). The 23% figure is therefore not a property of an era but of a filing state: the measured dead-name burden of a frozen archive, against a living one. The method — recovery, two-ruler resolution, catalogue join — is archive-agnostic and is released with the per-recording tables.
Authors
- Atharva Chourey (ORCID: https://orcid.org/0009-0009-9423-1718)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-21
- DOI
- https://doi.org/10.5281/zenodo.24409229
- Primary Topic
- Research Data Management Practices
- Type
- preprint