Skip to content

Neotoma native hierarchy round trip

The retained report compares the eight relational surfaces' source_payload fields with their parent-scoped source payloads in the validated nine-part Neotoma download archive. The comparison includes sites, collection units, datasets, chronologies, controls, samples, ages and observations. For repeated parent records it compares distinct source payloads; for ages and observations it also compares multiplicity. Control payloads retained as conflict variants count as preserved evidence. The materialization reader separately validates surface hashes, row counts, ownership and referential integrity before this comparison.

The pinned source snapshot has 209 downloads and 200 distinct site payloads. The audit finds zero missing or unexpected parent-scoped payloads in all eight surfaces, including 23,281 age claims and 370,936 observations. The source has 2,867 control rows and 2,836 distinct parent-scoped control payloads; 2,798 control rows are accepted and 70 variant rows appear in conflict records. These counts describe preservation of source evidence. They do not approve age selection, count denominators, taxa, rights or scientific admission.

Replay from the repository root with the configured environment:

PYTHONDONTWRITEBYTECODE=1 .venv/bin/python \
  docs/report/source-relations/neotoma/native-roundtrip/replay.py \
  > artifacts/jobs/neotoma-native-roundtrip/report-replay.json

Compare the replay file with report.json. The source identity and materialization digest must match the retained report before reusing its conclusion. The audit does not independently verify the original acquisition query or pagination receipts, selected chronology semantics, normalized observation convenience fields, compact-site derivation or refresh review.