[jt55401]

Refs #631. Supersedes the closed draft #1310, which was closed without merge.

What changed since #1310

• Removed the invented US-wide absent phase (June–September). It contradicted ADW, which reports isolated breeding populations throughout the northern United States, and it was not a sourced window. Only the CA and US-AK breeding phases remain.
• Strategy is ActiveYearRound as a placeholder. This is the open gap (see below). The schema requires an absent phase for Migration. No consulted licensed source gives a defensible region-scoped absence window, so migration is not modelled.
• Activity curve described honestly. The US curve is an occurrence-frequency proxy, weighted to winter and to the coasts. Low June–September values are not evidence of absence, since Alaska and Canada breeders are largely unrecorded. The iNaturalist histogram total is exactly 10000, which matches the API result cap, so its shares come from a capped set. GBIF supplies 99.9% of the records. The re-run of fetch_activity.py reproduced the stored counts.
• Behaviour. The description now says plainly that dives are not depicted: eat_nearest with gait swim moves the bird below the surface in the simulation, but the model has no submerged pose. Feeding starts only in daylight. A feeding bird roosts on water after dark (sun elevation between -90 and -6). ADW says winter birds forage at night, and this program does not model that.
• Poses. Fly amplitude 1.3. Swim has paddle keyframes. Rest has a head tuck.
• ADW. Re-checked the lifespan figures (2.5 male, 2.3 female; 1969–1973 New York banding, 56–159 birds per sex, accuracy unknown) and the use conditions (CC BY-NC-SA 3.0). The citation is corrected to ADW's own form. NC coverage is narrowed to the datums that need it: range and strategy now rest on Wikipedia (CC BY-SA).

Missing value (1 factual gap)

• seasonal.strategy / seasonal absence. The species migrates: it breeds in Alaska and Canada and winters on the US coasts, the southern US and the Great Lakes (Wikipedia; ADW). A licensed, region-scoped absence window for the breeding populations is needed to use Migration. Checked: Wikipedia (lead, Description, Breeding, Diet, raw wikitext of the current revision) and ADW (Geographic Range, Breeding, Lifespan, Behavior). Neither gives a dated absence for CA or US-AK, and ADW reports breeding in the northern United States. The fix needs a licensed source for the absence window, or a decision to accept ActiveYearRound.

Validation (commit 397288e)

• uv run tools/validate.py species/bucephala-albeola --write-attribution --report: 0 errors, 1 warning (NonCommercialLicense, ADW CC BY-NC-SA 3.0, last resort). Provenance 34/34. Attribution regenerated; git diff -- species empty afterwards.
• uv run tools/validate.py --self-check: 629 files, 0 findings.
• uv run tools/validate.py --report: 192 species, 0 errors.
• uv run tools/validate.py --locale all --report: 33 locales, all pass.
• uv run --with pytest --with jsonschema pytest tools/tests -q: 347 passed, 8 skipped.

Preview (inspected)

Previewer at 127.0.0.1 on the built site, pose screenshots in headless Chromium (no console errors). Inspected: idle, swim, fly, rest.

• fly: wings spread and flap clearly; reads as a flying duck.
• swim: weak. The shared biped_winged renderer has no swim branch, so swim falls through to the walk/hop cycle. The paddle keyframes give a modest leg difference, and there is no waterline or submerged body. This is a shared renderer issue that also affects the five other biped_winged species with a swim pose. It is not fixed here.
• idle / rest: similar at preview scale; the head tuck is slight.

Not checked

• Native and introduced status beyond Wikipedia and ADW (no introduced range was found there). GBIF introduced flags were not queried.
• Alaska-specific breeding dates.
• The other lifespan sources (AnAge, USGS BBL, EOL) were not re-checked today.
• The simulation's handling of ActiveYearRound for a migrant (birds may stay present in winter in the breeding regions) cannot be observed from this repository.

Closes nothing: Refs #631.

🤖 Generated with Claude Code

[jt55401]

Coordinator review: the record explicitly uses ActiveYearRound as a placeholder for a migratory species because the required regional absence window is unresolved. The preview also reports an unrepresentative swim pose. This leaves seasonal behavior and pose review unfinished, so it does not meet the strict near-complete draft rule in AGENTS.md. Preserving pushed branch wave-20261101d-631-bucephala-albeola and the sourced evidence on issue #631; no merge.