Correlative coherence analysis: variation from intrinsic and extrinsic sources in competing populations

The concept of the correlation between two signals is generalized to the correlative coherence of a set of n signals by introducing a Shannon-Weaver-type measure of the entropy of the normalized eigenvalues of the n-dimensional correlation matrix associated with the set of signals. Properties of this measure are stated for canonical cases. The measure is then used to evaluate which subsets of a particular set of n signals are more or less coherent. This set of signals comprises extrinsic, stochastic resource inputs and the population trajectories obtained from simulations of a discrete time model of competing biological populations driven by these resource inputs. The analysis reveals that, at low levels of competition, the correlative coherence of the combined system of intrinsic population and extrinsic resource variables is relatively low, but increases with increasing variation in the resources. Further, at intermediate and high competition levels, the correlative coherence depends more strongly on competition than entrainment of stochasticity in the extrinsic resource variables. Density dependence has the effect of amplifying variation in noise only when this variation is relatively large. Also, chaotic systems appear to be entrained by sufficiently noisy environmental inputs.