WeatherXBiodiversity: Soroye et al. (2020) Replication for Iberian Bombus

Pin Phase 3 occurrence data to a citable GBIF download DOI

Provenance-only release. The Phase 3 Iberian replication of Soroye et al. 2020 previously used the GBIF search API (a non-citable preview mechanism). This release replaces it with a download API snapshot:

• GBIF Occurrence Download: https://doi.org/10.15468/dl.3frmsq
• 36,560 records, Bombus in Spain + Portugal, georeferenced, no geospatial issues
• License: CC-BY-NC 4.0
• Citation: GBIF.org (2026-04-25) GBIF Occurrence Download https://doi.org/10.15468/dl.3frmsq

Numerical results are unchanged. The 24 extra records relative to the v0.2.0 search-API extract did not survive the Kerr species filter, so the cleaned analysis dataset (7,487 occurrence records, 31 species, 99 cells, 100 km cylindrical equal-area grid) is bit-for-bit identical, and Phase 3 still yields sc_TEI_delta = +0.48 [0.27, 0.69] (mixed-effects logistic GLMM). The change is purely provenance: the published Phase 3 numbers are now pinned to a permanent, citable GBIF snapshot suitable for FORRT replication chains.

What's in this release

• notebooks/01b_gbif_download_doi.py — submits a GBIF Occurrence Download request, polls until completion, saves the DOI and download metadata.
• soroye_port/01_clean_data_iberia.py — now reads the GBIF download SIMPLE_CSV (tab-separated) instead of the search-API CSV.
• README badge + CITATION.cff + codemeta.json updated with the GBIF DOI.

Citation If you use this software, please cite this repository (Zenodo concept DOI 10.5281/zenodo.19756173), the GBIF download (DOI 10.15468/dl.3frmsq) and Soroye et al. (2020, 10.1126/science.aax8591).