A Cross-Substrate Biological Spectrum Using the Consciousness Gradient Index v2: Proof of Concept Across Four Anchor Species

No existing consciousness measurement framework places neural and non-neural biological systems on a single quantitative axis without substrate-specific normalisation assumptions. Existing frameworks either make unverifiable metaphysical identity claims (Integrated Information Theory), are restricted by design to neural architectures (Global Workspace Theory, Higher-Order Theories), or remain at the stage of theoretical extension without empirical demonstration across substrate types. We present CGI_v2, a revised operationalisation of the Consciousness Gradient Index (CGI = √(φ × ρ)), in which structural integration φ is measured using network modularity Q and adaptive response richness ρ is measured using a unified behavioural repertoire ceiling. Both components are defined identically across substrate types, requiring no tier-specific corrections. CGI makes no identity or reduction claim about consciousness: it measures functional correlates that co-vary with consciousness-like states, without asserting that these properties constitute consciousness. Applied to four anchor species — Escherichia coli K-12, Arabidopsis thaliana, Caenorhabditis elegans, and Drosophila melanogaster — using published network modularity values (E. coli: Q = 0.650; Arabidopsis: Q = 0.630; C. elegans: Q = 0.420) and a directly computed value for Drosophila (Q = 0.702, FlyWire v783, Zenodo 10676866), the formula produces: E. coli 3.804, Drosophila 4.576, Arabidopsis 4.701, C. elegans 5.150. The principal finding is counterintuitive: C. elegans (302 neurons) scores above Drosophila (134,181 neurons). This is a biologically grounded prediction: Drosophila's high modularity (Q = 0.702, 264 communities) reflects architectural specialisation that reduces global integration; C. elegans has a proportionally more globally integrated connectome (Q = 0.420). The prediction that integration matters more than neuron count is falsifiable via functional connectivity analysis of existing calcium imaging datasets. This paper constitutes a proof of concept for cross-substrate CGI computation, not a claim to a validated universal law. Three conditions for universality are identified: vertebrate extension, ecosystem validation, and independent replication. None is currently met. Project DOI 10.17605/OSF.IO/TFBM3