Parallax: Divergence Between Umwelten as a First-Class Object in Ecological Knowledge Graphs

A shared ecological knowledge graph records, for many co-occurring species, the environmental values each can tolerate, use, and read. Prior work in this programme scaled Jakob von Uexküll's Umwelt upward — composing organism-level perceptual worlds into ecosystem-scale entities and translating between them. Parallax proposes the complementary and, I argue, more revealing move: to hold the constituent Umwelten apart and treat the divergence between them — where one and the same environmental value carries opposite meaning for two species sharing the same place — as a first-class, computable object over the graph. The claim advanced here is that these divergences are the locus of ecological insight that a single (human, measurement-based) standpoint structurally cannot see: a value that reads as safe from one standpoint and as lethal from another is invisible to any observer who reads the value only once. The readings are grounded in the documented physiology and behaviour of the actual organisms and traceable to primary literature, not narrated by a language model — a distinction the paper argues is essential. A first grounded demonstration (an estuarine dissolved-oxygen split between diadromous fish and tolerant benthic species, at oxygen levels the estuary actually reaches each summer) shows the operation surfaces true ecological structure with its sources attached. This is a conceptual and priority record within an evolving programme. It states the concept, its lineage, and evidence that it works, at a level sufficient to establish authorship; it deliberately does not specify the controlled vocabulary, the value-to-meaning resolution procedure, the commensurability-enforcement mechanism, the serving interface, or the sourcing pipeline in full.