Birds · Dupont's lark (Chersophilus duponti) / Dataset descriptor

Dupont's lark songs and calls: an annotated vocal-type dataset

Scientific Data 13, Article 770 (2026); version of record 22 May 2026 2026 · Note reviewed 2026-10-12

The article describes standardized recordings from 20 Spanish populations and reports 4,297 annotated songs from 191 singing males, representing 401 song types, plus 795 calls from 97 calling males, representing 80 call types. Experts categorized acoustic song and call types and provide event timing, frequency bounds and individual IDs. One ornithologist performed the primary annotation; uncertain classifications and types reported as shared among populations received additional expert review. Zenodo resolves the article's cited concept DOI 10.5281/zenodo.17264916 to current version record 10.5281/zenodo.17264917; the API identifies 17264916 as conceptdoi, 17264917 as the version doi, and declares record-level CC BY 4.0. Its eight-file metadata inventory includes provider-supplied sizes and MD5 values. The separate Digital.CSIC distribution also identifies version DOI 17264917 and declares CC BY 4.0. The CSIC archive bytes and their relationship to the Zenodo file set remain unverified.

What this does — and doesn’t — establish

The study covers 20 populations in Spain. Birds were not individually marked; measures reduced but do not prove that repeat recordings are independent, and song/call individuals could not be matched. One expert made the primary annotations; uncertain or population-shared types received additional review. Labels are acoustic type categories, not independently observed behaviors or word meanings. No published grouped ML split was reviewed. The authors advise against broad frequency-feature comparisons because of spatial dialect variation. Zenodo's per-file MD5 values are provider metadata and were not locally verified. The 3,711,242,291-byte CSIC archive was not fetched or checked against the eight-file Zenodo inventory; a size comparison alone would not establish matching contents.

What we can build on

This is a useful model for pairing event boundaries, vocal-type labels, individual metadata and geographic context, while keeping acoustic form separate from behavior and meaning. The DOI difference is resolved: the article cites concept DOI 17264916, which Zenodo redirects to version DOI 17264917. Both the Zenodo record and the CSIC catalog/README declare CC BY 4.0 for their respective dataset distributions. The 3.71 GB CSIC archive remains uninspected, and provider MD5 values are not locally verified. Any computational evaluation would need artifact-level integrity checks, qualified review and a specialist-selected split that groups likely repeated individuals and population structure. This resource does not supply the independent behavior labels needed for Talk2Nature's candidate context-prediction question and is not admitted to AutoResearch.

See the related tools and reading paths →

The next reading task

The DOI identity is now established through the official Zenodo API. Before any acquisition or analysis, obtain qualified avian-bioacoustics review of a bounded, non-semantic question; define a storage and integrity plan; and decide whether the Zenodo files or CSIC mirror are actually needed. Provider file sizes and MD5 values are recorded but not locally verified, and the CSIC archive's contents are not matched to the Zenodo inventory. Keep the R script unexecuted until its contents, dependencies and notices are reviewed. Do not treat acoustic type labels as behavior or meaning.

Go to the sources

1. Pérez-Granados et al. (2026), Dupont's lark songs and calls ↗

   A 2026 Scientific Data descriptor reports a curated acoustic dataset from 20 Spanish populations: 4,297 annotated songs from 191 males and 795 calls from 97 males, categorized into song and call types. Its cited Zenodo concept DOI resolves to a version record whose metadata declares CC BY 4.0; article rights remain separate.

2. Zenodo concept DOI and version record for Dupont's lark songs and calls ↗

   The official Zenodo API resolves concept DOI 10.5281/zenodo.17264916 to version record 10.5281/zenodo.17264917: its record ID is 17264917 and its conceptdoi is 17264916. The record-level license field is CC BY 4.0 and its eight-file inventory includes provider-supplied sizes and MD5 values. No file bytes were downloaded or locally verified.

   Official Zenodo API response for the concept DOI and current version record ↗

3. Digital.CSIC distribution metadata for the Dupont's lark dataset ↗

   The Spanish public-data catalog identifies CSIC as publisher and lists the public Digital.CSIC distribution under handle 10261/424783: 17264917.zip and README.txt. It declares CC BY 4.0 for both distributions. Directly reviewed README and archive-response provenance are recorded in source 212; the archive bytes and checksum remain unverified.

4. Digital.CSIC README and distribution headers for the Dupont's lark dataset ↗

   The full 4,169-byte README recommends citing the dataset as Zenodo DOI 10.5281/zenodo.17264917 and states CC BY 4.0. The hosted 17264917.zip response headers cite the same DOI and report 3,711,242,291 bytes. This identifies the CSIC distribution, not the uninspected archive contents or a source-published checksum.

Original editorial note by Talk2Nature contributors. AI-assisted, source-grounded synthesis; not a peer review or an independent replication. Linked works retain their own rights.