Species Evidence Views

A species directory is a governed projection over source and project evidence. It groups records by the accepted taxon so readers can inspect curation, spatial support, recovery gaps, and product posture without treating the species folder as a new source of scientific facts.

flowchart LR
    Source["source release, paper, project, supplement"] --> Owner["governing record and fact owner"]
    Owner --> Species["species evidence view"]
    Species --> Review["taxon-level review and recovery posture"]
    Review --> Decision{"named product contract"}
    Decision -->|admitted| Member["publication member"]
    Decision -->|not admitted| Account["qualified, excluded, or recovery evidence"]

The projection direction matters. A species view may repeat a sample label, locality, chronology, or coordinate so the taxon can be inspected as a unit. That repeated value remains subordinate to the project-owned sample and claim identified by its lineage.

Two Lifecycle Models

Human and non-human ancient DNA share a species-owned directory shape, but they do not currently share one material lifecycle.

View	Current source route	Material posture	Supported use
homo_sapiens	governed link to the AADR v66 release capture	raw capture present; normalized, manifest, review, and species-report members absent	inspect retained AADR metadata and release identity at the captured panel boundary
ten non-human species	project and paper source library with generated species projections	raw inventories, normalized evidence, manifests, reports, and species review are materialized	inspect recovered evidence, curation posture, project deficits, and downstream fitness at each declared boundary

Directory symmetry therefore does not prove evidence symmetry. A query must inspect material artifacts and their contracts rather than infer lifecycle completion from the existence of a folder.

Non-Human Species Views

The tracked animal program currently exposes cattle, cat, dog, donkey, dromedary camel, goat, horse, pig, reindeer, and sheep views. Each README is generated from the same evidence code that materializes the species records, so the narrative and counts share an owner.

Each view separates four questions:

Question	Evidence surface
What material was tracked?	raw archive inventory and source snapshot
Which samples, sites, localities, and coordinates were normalized?	typed files under normalized/
Which project and curation revision produced the view?	manifests under manifests/
What remains blocked or incomplete?	recovery reports and species review

The README reports samples, projects, sites, coordinate classes, Nordic members, recovery deficits, pending projects, and rejected projects as different observation units. Those counts describe related evidence surfaces; they are not stages of one universal attrition funnel.

Roles Are Evidence Policy

product_role, dataset_bucket, and curation_class are repository policy fields. They describe how the current evidence may participate in governed products and comparisons. They are not taxonomic ranks, domestication facts, or permanent scientific labels.

Policy signal	Governs	Does not establish
product role	intended contribution to a named comparison or publication family	biological importance or collection completeness
dataset bucket	current grouping by evidence posture	a natural scientific category
curation class	rule used to evaluate project and paper support	that every tracked project passed review
species release gate	whether the taxon-level review satisfies its declared gate	that every member is sample-complete or publication-ready
supported-status eligibility	whether current evidence permits stronger support language	universal fitness for every analysis or product

A comparator can satisfy its comparator release contract while remaining ineligible for domesticated-core interpretation. A species can have supported curated projects while retaining blocked, pending, rejected, or under-recovered projects. Preserve both statements.

Taxonomic Reassignment Does Not Rewrite Sample History

The accepted taxon is a versioned claim about a governed sample, not the sample's identity. A revised taxonomy, synonym decision, or stronger molecular assignment can move a sample between species views while preserving its project namespace, source labels, locality, chronology, and evidence lineage.

flowchart LR
    Sample["stable project-owned sample"] --> TaxonClaim["taxonomic claim + authority"]
    TaxonClaim --> SpeciesView["accepted species projection"]
    TaxonClaim --> Alternative["synonym, ambiguity, or competing assignment"]
    SpeciesView --> Product["taxon-scoped product decision"]

Change	Stable authority	Required reassessment
display-name or synonym change	sample and source lineage	labels, taxonomy revision, and discovery views
accepted species changes	sample identity	species projection, taxon counts, role, and every taxon-scoped product
assignment becomes ambiguous	competing attributed claims	eligibility, aggregation, and public qualification
project species expectation disagrees with sample evidence	sample-owned evidence and conflict record	project summaries and affected taxonomic decisions

Species totals are therefore revision-dependent projections. They cannot be used as permanent specimen counts without the accepted taxonomy revision and the included sample identities.

Audit One Species Claim

1. Name the proposed claim and its observation unit: source project, sample, site, chronology, coordinate, species aggregate, or product member.
2. Open the species README and identify the current role, evidence bucket, release posture, and blocking reasons.
3. Resolve the stable sample and project identity in the normalized records and project-owned sample master.
4. Inspect locality, chronology, and coordinate evidence independently; one strong dimension does not repair another weak dimension.
5. Recover the paper, project, supplement, table, sheet, row, or archive locator that supports the disputed fact.
6. If the claim concerns a map or report, inspect the product-specific admission decision and manifest rather than inferring membership from the species view.
7. Carry unresolved projects and non-members into any coverage statement.

flowchart TB
    Claim["species claim"] --> Unit{"observation unit"}
    Unit --> View["species posture"]
    View --> Sample["stable sample and project identity"]
    Sample --> Dimensions["locality, chronology, coordinate evidence"]
    Dimensions --> Origin["captured source locator"]
    Dimensions --> Admission["product decision where applicable"]
    Origin --> Verdict["bounded interpretation"]
    Admission --> Verdict

Cross-Species Comparison Contract

Species views make taxon-level discovery easier, but comparison requires an explicit population contract. Record at least:

Contract member	Why it is required
included species and accepted taxonomy revision	fixes which views and names were compared
observation unit	prevents projects, recovered samples, sites, localities, and publication points from being counted together
eligible population	identifies which records could have entered the comparison
evidence requirements	fixes identity, locality, chronology, coordinate, and source-lineage thresholds
geography and temporal operation	prevents visual proximity from becoming an ungoverned association
exclusions and qualified members	keeps weak or out-of-scope evidence in the denominator account
source, curation, and product revisions	makes the comparison recoverable after a refresh

Do not compare raw species README counts as though every taxon had equal project recovery, site granularity, chronology depth, coordinate support, or publication rules. Normalize the question and evidence requirements—not the facts or their uncertainty.

Reuse Packet

A reusable species-level result carries the species identity, policy role, observation unit, selected record identities, source and curation revisions, eligibility rule, included and excluded populations, dimension-specific evidence posture, product identity where applicable, and unresolved blockers.

A screenshot, species total, or release-gate Boolean is an entry into that packet, not a substitute for it. The result remains defensible when another reader can recover the governing sample facts, reproduce the population, and see which stronger claims were refused.

Continue with animal source intake, sample records, locality evidence, chronology evidence, and record admission.