[jt55401]

Closes #1081

Evidence

• Taxon: GBIF 2484254 (ACCEPTED, SPECIES, Sturnella magna (Linnaeus, 1758)) and iNat 9528 (species, active); both checked 2026-11-01. GBIF synonym: Alauda magna. No disagreement.
• Range (native, country list): derived from the Wikipedia subspecies list. It excludes the Chihuahuan meadowlark (S. lilianae/auropectoralis, split 2022). The US code is coarse and cannot exclude the split taxon's south-west US population, which is a known limitation.
• Morphology: mass 90–150 g and length 19–26 cm from ADW (midpoint derived). Wikipedia gives 76–150 g and 19–28 cm; the two disagree, and ADW was used. Wingspan 35–40 cm (both sources, midpoint derived).
• Diet: AUTHORED split (invertebrate 0.75, seed 0.2, fruit 0.05). Derived from Wikipedia ("about three-quarters" animal) and ADW's seasonal diet (spring insects, summer beetles and grasshoppers, winter weed seeds, grain, fruit).
• Lifespan: lifespan_y = 5, Low confidence. Source is ADW "expected lifespan of five years in the wild" (no cited source on ADW). ADW's own 151-month wild average (attributed to BBL) conflicts; AnAge gives a 12.6-year wild maximum, not used. The 9-year wild longest is a maximum, not used.
• NC source justification (ADW, CC BY-NC-SA 3.0): this is the only checked source with an explicit typical wild lifespan. The following were checked and do not supply one: Wikipedia (CC BY-SA 4.0; no lifespan); AnAge (maximum only; licence not used); COSEWIC (Canadian Government terms, not allowlisted). No open-access mark-recapture study with a typical lifespan or annual adult survival rate was found. ADW is also the source for mass, length, wingspan, breeding timing, diet, roosting and locomotion, all of which carry the NC tier in the warnings. Wikipedia covers song and the range split.
• Activity (US): fetch_activity.py sturnella-magna --region US --write. iNat tier A, total 1,997 (place 1, usable); GBIF tier A, total 1,926,953 (country US, usable). Monthly values are in the record; September is the thinnest month on iNat (30 observations).
• Seasonal: ActiveYearRound. Breeding 05-25..08-31 and territories 03-01..08-31 are from ADW. The northern migrant fraction (ADW: leave by end of November, return in March) is not modelled; no sourced fraction exists, and the migrant share is only described in provenance.

Behaviour (species-local bt-sturnella-magna-v1)

• Executed actions, audited in behavior.json:
  • Threat within 6 m: take off, fly to the nearest perch, perch.
  • Fear above 0.2: take off, flee in flight.
  • Night (sun not day): rest at shelter (ground grass roost).
  • Song: months 3–8, sun elevation −6°..12°, perch within 30 m, sing with a 20 s cooldown.
  • Nest: need above 0.5, months 5–8 only, build nest within 40 m.
  • Hunger above 0.4: eat invertebrate by walking.
  • Otherwise: walk, then idle.
• Month sets are explicit ([3,4,5,6,7,8] and [5,6,7,8]); no inherited April/May entries.
• Day/night exits: the top-level night check is re-evaluated each tick, so the animal rests at shelter at night and does not sing or forage then.
• Gait fit: ground walking for foraging, flight for flushing and perching. The shared songbird program was not used because it hops and hard-codes nesting months 4–5.
• Limitations: the program only forages invertebrate. The seed and fruit diet fractions have no reachable food action, so the program does not simulate seed-eating. The migrant fraction is not modelled. Defence (Flee) and wariness (0.5) are judgement, not sourced. Song cooldown (20 s) is judgement.

Look and previews

• Authored rendering block: biped_winged, judgement proportions and palette. The breast uses the belly palette slot (yellow); the black V is not modelled.
• Previewer (local _site build, Chromium via Playwright, preview.html?species=sturnella-magna&pose=…), all five declared poses viewed:
  • idle: reads as a bird, yellow breast visible after the slot fix.
  • walk: reads as a bird on the ground.
  • fly: the wings hang as flat slabs on the shared winged rig, so the flight reads poorly. This is a rendering limitation, not a record bug.
  • perched: reads correctly.
  • rest: reads correctly.
• No console errors; only WebGL performance messages.

Validation (commit 96ef232)

• validate.py species/sturnella-magna --report: 0 errors, 1 warning (the ADW NC source, justified above), provenance 35/35.
• validate.py --self-check: 629 files, 0 findings.
• validate.py --report (all species): exit 0, no failing species.
• validate.py --locale all --report: exit 0, no failing locales.
• pytest tools/tests -q: 347 passed, 8 skipped.
• validate.py --write-attribution: regenerated; attribution is fresh.

Checklist

• Issue claimed and then released after the first pass; the coordinator re-claimed.
• Record validates; ATTRIBUTION.md regenerated and committed.
• Activity fetched for US, with usable flags and counts reported.
• Identifiers resolve to the stated taxon and rank.
• Behaviour actions, month sets and day/night transitions audited against the record.
• Every declared pose previewed; limitations reported above.
• Licence: NC source justified above; no ND or unknown licences.
• No invented numbers: judgement values are marked derived or judgement in provenance.
• Title and Closes line as required.
• One species per PR.

🤖 Generated with Claude Code

[jt55401]

Coordinator review hold: the submitted behaviour reaches only invertebrate feeding while the record declares seed and fruit as 25% of the diet. The fly pose also needs visual correction. A normal species PR requires the feeding paths and poses to be reviewed as fitting. The branch is preserved and a follow-up worker is addressing these findings; this PR can be replaced when the current head passes the full gates.