Reproducibility materials for upstream workflow uncertainty in global Unionida biodiversity inference

The archive covers taxonomic reconciliation, GBIF record-level quality control, adjudicated GBIF–IUCN comparisons, 54 taxonomy × QC × status × scale workflows, grouped spatial validation, PFAF2 and random-CV sensitivity, Region LOGO validation, and GAM, random-forest and XGBoost evaluation and interpretation outputs.

The Stage26K3 process archive was recovered to 151/151 registered assets without rerunning scientific analyses. Post-freeze Figure 1 and Table 1 reconciliation packages and the current frozen Supplementary release are also included. Original GBIF occurrence records, original IUCN Red List range geometries and original HydroBASINS source layers are not redistributed.