Compare named specimens, morphology matrices, histology, taphonomy and functional models while preserving every observation–interpretation boundary and handing Avialae/flight evidence to the crocodylomorphs-birds package.
Each step names its evidence claims, with references below. Interactive Explorer states are available in the full application; the narrative remains an editorial synthesis.
Baron et al. recovered Ornithoscelida from 457 characters; Langer et al. recovered the traditional split after rescoring and added taxa. Both are topology tests.
The 233–228 Ma atlas interval is a display context for early dinosaur fossils, not an age inferred by either morphology matrix.
claim:event:dinosaur-radiationULBRA-PVT280 preserves recurved serrated teeth. Faunivory near the sauropodomorph root is an ancestral-state model, not a direct ancestor.
Formation correlation bounds the age; the dietary transition is a mapped hypothesis across sampled taxa, not a linear sequence observed in one lineage.
claim:event:buriolestes-holotype-feeding-boundarySAM-PK-K8025 is a bounded partial skeleton. Matrix placement and formation age do not establish a global ornithischian first appearance.
The broad display interval follows formation-level correlation and must not be read as a direct specimen date or a global ornithischian FAD.
claim:event:eocursor-holotype-ornithischian-testMussaurus digital models combine several growth stages. The bipedal-to-quadrupedal shift is ontogenetic and modelled, not an evolutionary ladder.
The fossils constrain anatomy; centre of mass and stance are sensitivity-tested reconstructions and cannot be generalized to every sauropodomorph.
claim:event:mussaurus-ontogenetic-stance-modelBP/1/7120 supports an adult giant with flexed limbs. Twelve tonnes and habitual quadrupedality are comparative estimates from a partial skeleton.
The specimen is incomplete and the earliest-Jurassic interval is formation-bounded; it is not a universal threshold or ancestor for sauropod gigantism.
claim:event:ledumahadi-body-mass-quadrupedality-modelNHMUK R1111 preserves associated osteoderms. Lost armour, inferred keratin and stem-ankylosaur placement remain separate from observation.
Missing armour and preparation history prevent a complete in-place map; placement is one analysis rather than an uncontested thyreophoran topology.
claim:event:scelidosaurus-r1111-dermal-skeletonIVPP V14530 preserves a nearly complete early ceratopsian skeleton. Its topology is a matrix result, not a direct ancestor.
The atlas interval is formation-level; topology and character polarity depend on outgroups, coding and the sampled matrix.
claim:event:yinlong-v14530-ceratopsian-mosaicNDGS 2000 preserves skin and feeding traces. Scavenging-assisted desiccation is a pathway inferred for this carcass, not a universal rule.
The evidence concerns one specimen and one depositional history; skin presence does not establish original colour, physiology or preservation frequency.
claim:event:edmontosaurus-ndgs2000-skin-taphonomyThree Yutyrannus skeletons preserve filamentous integument. Function, complete coverage and Avialae flight are separate questions.
Non-avian integument stays here; Avialae, flight models and living-bird COL26.8 names remain owned by crocodylomorphs-birds.
claim:event:yutyrannus-feathered-tyrannosauroidIGM 100/979 lies over a clutch in a bird-like posture. Brooding is well supported but remains behaviour inferred from association.
The original paper explicitly retains alternative explanations and limits testing of parental-system hypotheses in the fossil record.
claim:event:oviraptorid-igm100979-nest-associationTyrannosaurid growth marks are observations; reconstructed ages, masses and logistic growth rates are model outputs.
The 83–66 Ma atlas window spans sampled tyrannosaurids, not one population; the fitted curve is not a universal dinosaur growth schedule.
claim:event:tyrannosaurid-histology-growth-curvesMOR 799 traces and bone-bearing coprolite support osteophagy, while force and pressure derive from a seven-skull biomechanical model.
The models test capability rather than observe a feeding event; MOR 799 and the coprolite do not identify every individual or every feeding context.
claim:event:tyrannosaurus-osteophagy-biomechanics