Animal Ancient DNA Evidence

Animal ancient-DNA publication normally begins with a source-backed sample, not a project title or species mention. Papers, archive projects, supplements, sample tables, sites, chronology statements, and coordinates remain distinct evidence units until their relationships are explicitly curated. The current point surface also contains one explicitly qualified project-anchored context feature; it must not be described as a recovered sample.

Evidence Chain

flowchart LR
    Paper["paper DOI"] --> Project["archive project"]
    Supplement["captured supplement"] --> Project
    Project --> Sample["stable sample identity"]
    Sample --> Locality["sample locality evidence"]
    Sample --> Chronology["sample chronology evidence"]
    Locality --> Coordinate["coordinate basis and precision"]
    Sample --> Species["species-normalized view"]
    Chronology --> Decision{"product admission"}
    Coordinate --> Decision
    Species --> Decision
    Decision -->|admit or qualify| Published["atlas and country evidence row"]
    Decision -->|exclude| Accountability["gap, conflict, or refusal surface"]

Evidence Units And Authorities

Evidence unit	Governing surface	Required distinction
project	source_library/project_registry.json	archive identity is not a sample identity
paper	source_library/paper_registry.json	publication identity is not project identity
captured artifact	project source bundle and supporting-material manifest	discovered URL is not recovered content
sample	project sample_master.json	source labels and stable repository identity remain linked
locality	project sample_locality_evidence.json and species site_evidence.json	verbatim place, resolved site, and publication precision differ
chronology	project sample_chronology_evidence.json and chronology review surfaces	source text, normalized interval, basis, and caveat remain linked
coordinate	species coordinate_provenance.json	supplied, resolved, approximate, substituted, and unresolved differ
publication	atlas evidence row and product manifest	visible membership is downstream of admission

Recovery States

stateDiagram-v2
    [*] --> Discovered
    Discovered --> Captured: paper and supporting material acquired
    Captured --> Extracted: stable sample rows recovered
    Extracted --> Reviewed: locality, chronology, taxonomy, and coordinates evaluated
    Reviewed --> Admitted: product requirements satisfied
    Reviewed --> Qualified: material precision limit remains visible
    Reviewed --> Excluded: required evidence is absent or conflicting
    Discovered --> Deferred: required source material unavailable
    Captured --> Deferred: sample-bearing content not recoverable

Tracked but deferred evidence is not equivalent to a negative scientific result. It records what is known about the source and what remains unavailable.

Current Evidence Depth

The governed foundation currently contains 894 preparation rows across 10 species and 40 archive projects. Of those rows, 502 are fully grounded, 256 are partly grounded, 29 are blocked by missing metadata, four by missing location detail, and 103 by weak chronology. This is the evidence-preparation population, not a sample or publication count.

The generated project intake review separately contains 868 recovered sample-master rows across the 40 tracked projects. Only four projects have an exact expected-sample denominator, 22 have a defensible minimum floor, and eight are flagged for implausibly low recovery. The repository therefore does not present 868 as a complete census of the tracked projects or as a row-for-row subset of the foundation.

The point-evidence review contains 234 admitted rows. Of these, 233 retain supplementary-table coordinates and one uses documented named-site geocoding at approximate confidence. These are admitted evidence rows, not proof that every project, species, locality, or chronology has reached the same maturity.

The same split applies to identity. The 233 supplementary-coordinate rows have final sample identity backed by directly extracted table rows. The remaining feature, the Wadi Halfa dromedary context for project SRP073444, has provisional project-anchored identity and not_yet_recoverable sample status. Its paper names Site 1040 near Wadi Halfa, and its published geometry is an approximate named-place resolution. The supported statement is therefore that the product carries a qualified dromedary context feature—not that a recovered sample row has exact excavation coordinates.

Point population	Rows	Identity and coordinate posture
directly extracted sample evidence	233	final sample identity and supplementary-table coordinates
project-anchored dromedary context	1	provisional identity and approximate Wadi Halfa named-place geocode
total admitted point-evidence rows	234	mixed evidence surface; preserve the class of each row

Three Ledgers Answer Three Questions

Ledger	Unit	Question answered
foundation truth	curated preparation row	how completely is the available identity, locality, chronology, and metadata evidence grounded?
project sample master	recovered sample identity	which source labels resolve to a stable sample within a project?
point publication	product member	which sample-backed or qualified context claims satisfy this map contract?

No universal completeness percentage spans all three. A preparation blocker is not necessarily an identity ambiguity; a final identity is not necessarily point-ready; and a qualified context point is not a recovered sample.

Evidence Depth Is Dimension-Specific

Dimension	What the repository can establish	Remaining boundary
source discovery	which projects, papers, and supplements are tracked	discovery does not prove sample-bearing material was recovered
sample recovery	which source rows became stable sample records	most projects do not have an exact expected-row denominator
locality	which verbatim localities resolve to sample- or group-owned sites	regional or conflicting assignments remain non-point evidence
coordinate	whether a pair is source-supplied or repository-resolved, with confidence	coordinate class does not imply uniform real-world precision
chronology	which source statement, interval, and basis belong to a sample	broad or contextual time does not authorize synthetic numeric bounds
publication	which rows satisfy a named product contract	admission does not establish collection completeness

There is therefore no single collection-wide maturity score. A project can be well documented at the paper and supplement level while its locality recovery is incomplete; a spatially admitted sample can still have chronology that is too broad for numeric comparison. The governing files preserve those different states instead of averaging them into one quality label.

flowchart LR
    Inventory["40 tracked projects"] --> Foundation["894 preparation rows"]
    Foundation -. "related, not one-to-one" .-> Recovery["868 recovered sample identities"]
    Recovery --> Review["identity, locality, chronology, coordinate review"]
    Review --> Samples["233 final sample-backed points"]
    Inventory --> Context["1 qualified project-anchored context point"]
    Samples --> Points["234 admitted point-evidence rows"]
    Context --> Points
    Inventory --> Gaps["blocked, under-recovered, and unresolved projects"]
    Gaps --> Accountability["recovery review and refusal surfaces"]

The generated reports make both branches visible. The published branch shows what satisfies the point contract; the accountability branch prevents the admitted subset from masquerading as collection completeness.

A Published Point Is A Typed Projection

The point row does not become a new authority for the facts it displays. It is a product-specific projection over separately governed evidence:

Point field	Governing owner	Projection rule
feature identity	publication manifest and evidence-row identity	stable within the named product and linked to the governed sample or qualified context member
sample label and accession	project sample master	display the admitted identity without replacing source-native aliases
species	species-normalized sample record and taxonomy decision	use the governed taxon posture, including qualification or conflict
locality label	sample locality and site evidence	display at the admitted resolution; broad text remains broad
geometry	coordinate-provenance decision	supplied, resolved, approximate, substituted, or withheld posture travels with the pair
time label or interval	sample chronology evidence	preserve source wording, normalized basis, ownership, and comparability caveat
admission posture	named product rule	distinguish sample-backed, qualified context, excluded, and deferred populations

flowchart LR
    Sample["sample identity"] --> Projection["product projection"]
    Locality["locality and site evidence"] --> Projection
    Coordinate["coordinate provenance"] --> Projection
    Chronology["chronology evidence"] --> Projection
    Taxonomy["species decision"] --> Projection
    Projection --> Member["typed publication member"]
    Projection --> Refusal["qualification or exclusion"]

This model permits a narrow supported point without declaring the whole project complete. It also permits curation to strengthen later without rewriting history: a new evidence decision creates a reviewable projection change and affected product diff rather than silently mutating the earlier source claim.

Audit A Published Animal Point

1. resolve its feature and evidence-row identifiers in the product traceability surface;
2. confirm the species record under data/adna/species/<latin_name>/normalized/sample_records.json;
3. inspect the species site_evidence.json, then follow its project linkage to sample_sites.json and sample_locality_evidence.json;
4. inspect project sample_chronology_evidence.json and the cross-project chronology review for temporal posture;
5. inspect coordinate_provenance.json for basis and precision;
6. follow the sample lineage to the project sample master, source bundle, paper, and captured supporting artifact;
7. confirm that the product manifest records the point's admission posture.

Publication Boundary

A species-level presence, archive project, or paper citation cannot substitute for a recoverable sample row. A broad locality cannot become an exact point. A cultural period cannot acquire a synthetic numeric interval. A visible point cannot outrank its governing evidence.

The point contract is also narrower than “animal evidence available.” A project can contribute paper, supplement, taxonomy, or broad locality evidence without contributing an atlas point. Conversely, a point admitted under its present traceability and precision does not certify complete recovery of its project. The release posture records the narrower claim that the admitted subset satisfies its point contract while project-level recovery denominators, blocked sources, and unresolved evidence remain visible outside that subset.

The Wadi Halfa context feature is the concrete reason publication type must travel with the row. It is spatially admitted under the current product but does not satisfy the stronger recovered-sample identity posture of the other 233 rows. Analyses requiring sample-level independence or recovered sample denominators must exclude or separately classify it and account for that decision.

Continue with animal source intake, sample records, locality evidence, chronology, and point publication rules.