Skip to content

Source-bound taxa review package

The current private review projection contains 10,908 rows from the source authorities listed below. Its rows preserve source identities, native names and qualifiers, country observation counts, exact-unit values, source citations and separate proposed group, LRA taxon, PFT, PPE and fall-speed fields. Every row remains needs_discussion; all proposed mappings start unset. The package is for qualified review, with no classification admission or publication right. The completed projection has bundle SHA-256 1220483aa405b55aa08da56175d568ba7d3f31bed6f7f9095e804cb3551c5cfb; the completion reader verified all three payload members and their byte counts.

Source family Review rows Scope
Neotoma 1,351 Source taxon concepts and 370,936 source observations
SEAD 1,975 Source taxon records and 177,763 source observations
Rootsi private workbook quarantine 7,459 Columns in 107 sheets from 79 workbooks; headers unresolved
LandClim 31 Reported parameter and crosswalk labels
LRA reference resources 81 Author example labels and parameters
animal aDNA 10 Explicit non-pollen identities
AADR 1 Explicit non-pollen source identity

The Rootsi inventory is retained beside the 79 workbooks in the private source store as landclim/rootsi-workbooks/header-review-inventory.json (SHA-256 15a8a0d333c9b0a3928aa0f8e66efbf870860fdc27b8179465a0bee6d2059196). Its 7,459 column records carry each source workbook's SHA-256, and all 79 workbook identities agree with that store's custody manifest. The inventory retains the first six rows as header candidates; it does not select a taxon header or authorize analytical use. The workbook archive identity is f64d4dd411bc92362c9db2701c919063c2432e53fa523052e602da5f1a86183c.

Rebuild the private package

Set ROOTSI_SOURCE_ROOT to the private landclim/rootsi-workbooks directory. From the repository root, create a fresh artifact directory and copy the inventory into it before building:

mkdir -p artifacts/jobs/taxa-review/source-input
cp "$ROOTSI_SOURCE_ROOT/header-review-inventory.json" \
  artifacts/jobs/taxa-review/source-input/header-review-inventory.json
shasum -a 256 artifacts/jobs/taxa-review/source-input/header-review-inventory.json
PYTHONDONTWRITEBYTECODE=1 .venv/bin/python -m \
  bijux_pollenomics_dev.taxa_review build --root . \
  --rootsi-inventory artifacts/jobs/taxa-review/source-input/header-review-inventory.json \
  --output artifacts/jobs/taxa-review/complete-review

Use an absent output directory for each run. The builder records 60 consumed input files and writes taxa-review.json, a CSV projection, review instructions and a completion receipt with file hashes. The complete JSON and CSV remain private because the workbook material is restricted. The source inventory in the private store, the governed source files and this command suffice to regenerate the package; no prior artifact tree is an input authority.

The reviewer return has not been received. The return validator can check identity, vocabulary and documented intent, but it cannot grant source rights, accept taxonomic groups, resolve the LandClim 12-versus-13 PFT discrepancy, or admit an LRA/vegetation comparison. Neotoma and SEAD observation counts are source-row frequencies, not comparable pollen mass or coverage percentages. Current mapping statuses are 7,459 quarantined unclassified, 1,410 unmapped, 1,974 not accepted, 53 not applicable, 11 not applicable to pollen and one unresolved. These statuses are source projection labels, not reviewer decisions.

Neotoma priority and citation coverage

The current governed Neotoma source screen yields 10,637 prerequisite samples, 228,323 agreeing pollen count observations and 1,006 unresolved review concepts across 169 datasets. The country partition and exact source-unit sums are reported on the Neotoma source page. The replay binds the full private review bundle to the governed raw and relational Neotoma manifests. Its priority, dataset attribution and screen SHA-256 values are respectively:

  • 68cc019466e4e251e66bff0cdeb6627419b6e6444d844cf5cd01eed25ee2054c
  • 9bb1bebdf9155fa8ba2ebb7afb5b743258730aad967201fba6c0bdef13a5eb63
  • 7d1ec565e9eec4317c61414b553a05b43cfd476a6fd0c087f0a536da8f2554ee

All 1,006 priority rows and all 169 dataset-attribution rows equal the governed scoped review exactly. Their whole-file hashes differ because the private replay binds the 10,908-row full review bundle while the public table binds only its 1,351 Neotoma concepts.

The exact dataset-publication responses reside in the private source store at neotoma/publication-links/2026-09-29/. Its 169-response capture manifest has SHA-256 d6dee3065e6fcfae8a7e484406cf37e2b5aa0424bee66d3a6922bbbdc1e867da. The manifest retains each source URL, retrieval time, byte count and digest; 58 prior exact responses were reused and 111 additional datasets were queried. The replayed citation report has SHA-256 9f37bb406b30beef6d8a2f274cc27ec509441c6c5b0d4d3a31c87aca6e93e24d. It records 223 linked publication records across 169 datasets: 149 with one provider-primary publication, seven with multiple provider-primary links, two with links but no primary flag and eleven without a linked publication. Publication DOI is null on 172 records. All 169 datasets identify the European Pollen Database as their constituent database. The citation leads have not been independently checked against original publications.

Set NEOTOMA_PUBLICATION_ROOT to the private neotoma/publication-links/2026-09-29 directory. After building the private review package as above, reproduce the current source replay with an absent output directory:

PYTHONDONTWRITEBYTECODE=1 .venv/bin/python -m \
  bijux_pollenomics_dev.taxa_review.neotoma_priority \
  --repository . \
  --review-directory artifacts/jobs/taxa-review/complete-review \
  --capture-root "$NEOTOMA_PUBLICATION_ROOT" \
  --output artifacts/jobs/taxa-review/neotoma-replay

The replay verifies every capture member, source manifest and candidate dataset identity before writing its four reports and receipt. All outputs remain private review projections (publication_allowed=false, scientific_admission_allowed=false). The old 3,538-sample, 58-dataset projection used an earlier source screen and is not the current cohort.