Evolutionary Selection on Transient Dynamical Retention Time in Energy-Coupled Chemical Oscillators

Fourth paper in a series on dynamical rigidity in autocatalytic chemical networks. Papers 1–3 established that network growth dilutes the activation ratio, that feedback-aligned growth preserves viability but not attractor dimension, and that timescale separation between fast oscillatory cores and a slow energy reservoir transiently inflates the correlation dimension D₂ before phase-locking restores rigidity. Here we test whether evolutionary selection can prolong this transient exploratory phase. Using the enzyme-complex coupled Brusselator model from Paper 3 with fixed network topology, we subject the energy dissipation rate γ to tournament selection for transient retention time τ (number of sliding windows with D₂ > 1.2). With population 20, 40 generations, and 10 selection + 10 neutral replicates, selection drives mean τ from 1.22 to 3.39 while neutral drift produces 1.08 (Cohen's d = 3.7, Mann–Whitney p < 0.0001, Wilcoxon p = 0.001). The mechanism is convergent reduction of γ by 5.6×, extending slow–fast timescale separation. Complete separation between conditions is observed (minimum selection τ = 2.65, maximum neutral τ = 2.28). These results demonstrate that simple evolutionary pressure acting on a dynamical phenotype can prolong high-dimensional state-space exploration in a prebiotic chemical network without genetic encoding or structural change. V2 confirmed run (n = 10); V1 pilot (n = 3, d = 4.9) published separately.  

All simulation scripts, raw data (JSONL), and publication figures are available in the paper4/ directory of the accompanying GitHub repository.