(A) Example stack of bright-field images of the recorded network. Pixel intensities encode the contraction state (tube dilation) at each point of the network. Principal Component Analysis is performed on a stack of post-processed bright-field frames. (B) Ranked spectrum of relative eigenvalues in percent (orange), plotted against the mode rank on a log-log graph. The eigenvalue spectrum is continuous, without a natural cutoff. Spectrum of randomised data (gray) shown for comparison. The cutoff for the continuous spectrum is defined by the largest eigenvalue of the spectrum from randomised data (black line). (C) (i) Structure of the four highest-ranking modes with their respective coefficients shown in (ii). The red-blue colour spectrum indicates the contraction state. The modes are eigenvectors of the covariance matrix. The coefficient of the first mode captures the organism’s characteristic oscillation period of , while the coefficients show considerable variation in amplitude and frequency over time. The PCA was performed on a data segment with 1500 frames, at the rate of 3 sec per frame.