Born Free, Bound by Function: The Topological Lifecycle of the C. elegans Connectome

How does a nervous system acquire its topology? We introduce the Spearman rank correlation between node degree k(v) and local clustering coefficient σ(v)—denoted corr(k,σ)—as a scalar order parameter for network development and apply it to the eight serial-section electron microscopy connectomes of Caenorhabditis elegans reconstructed by Witvliet et al. (2021), spanning from hatching to adulthood. We report five principal findings. First, the newborn connectome (dataset 1) exhibits corr(k,σ) ≈ 0, indicating dimensional independence between local clustering and global connectivity—a state we term the membrane phase. Development proceeds away from this state toward increasingly negative correlation, as hub neurons progressively lose triangles while peripheral neurons gain them. Second, a covariance decomposition of the multiplex network into chemical-synapse and gap-junction layers reveals that both layers contribute negative covariance in the same direction, with chemical synapses dominating the total signal at approximately 9:1 ratio—demonstrating topological dominance, not interference. Third, edges present in both layers simultaneously (overlap dyads, growing from 18 to 135 across development) exhibit 5–8× enrichment for triangle participation, identifying them as crystallization nuclei for local clustering. Fourth, simulated degradation of the adult connectome reveals that approximately 50% edge loss returns corr(k,σ) to zero—a ghost membrane that is topologically isomorphic to the birth state but structurally impoverished, with depleted triangle counts distinguishing ruin from potential. Fifth, combining empirical data with simulation and backward extrapolation yields a complete lifecycle trajectory: Embryo(≈+) → Birth(≈0) → Adult(≈−) → Ghost(≈0) → Death(≈+), crossing the membrane phase twice—once with developmental potential, once as structural ruin. These findings reframe neural development as ontogenesis from freedom rather than toward it, and suggest that corr(k,σ) may serve as a clinically translatable biomarker for neurodegenerative disease, where the approach toward zero could mark a pre-symptomatic transition window.