Anisotropic probabilistic cellular automaton for a predator-prey system

We consider a probabilistic cellular automaton to analyze the stochastic dynamics of a predator-prey system. The local rules are Markovian and are based in the Lotka-Volterra model. The individuals of each species reside on the sites of a lattice and interact with an unsymmetrical neighborhood. We look for the effect of the space anisotropy in the characterization of the oscillations of the species population densities. Our study of the probabilistic cellular automaton is based on simple and pair mean-field approximations and explicitly takes into account spatial anisotropy.

💡 Research Summary

The paper presents a stochastic, lattice‑based model of predator‑prey dynamics that builds directly on the classical Lotka‑Volterra equations but implements them through a probabilistic cellular automaton (PCA). Each site of a two‑dimensional square lattice can be in one of three states: predator (P), prey (H), or empty (E). The local update rules are Markovian and occur synchronously across the lattice. Three elementary processes are encoded probabilistically: (i) prey reproduction into a neighboring empty site with probability b, (ii) predation, where a predator consumes an adjacent prey and becomes a predator itself with probability p, and (iii) predator death with probability d.

A distinctive feature of the model is the choice of an anisotropic (unsymmetrical) neighborhood: a site interacts only with its right‑hand and upper neighbors. Consequently, information and individuals are preferentially transmitted in a single diagonal direction, mimicking real‑world asymmetries such as wind, water flow, or terrain slope. The authors ask how this spatial anisotropy influences the collective oscillations of the two populations.

The analysis proceeds in two stages. First, a simple mean‑field (MF) approximation is applied. By defining global densities ρ_P(t) and ρ_H(t) and averaging over all sites, the authors derive a pair of ordinary differential equations that are formally identical to the well‑known Lotka‑Volterra equations. In this coarse‑grained picture the anisotropy disappears; the system exhibits a single Hopf‑type limit cycle with a frequency and amplitude that depend only on the reaction probabilities (b, p, d). Thus, the MF level cannot capture any directional effects.

To incorporate spatial correlations, the authors develop a pair mean‑field (pair‑MF) approximation. They introduce joint probabilities P_{XY} for finding a directed pair of neighboring sites in states X and Y (X,Y ∈ {P,H,E}). Because the neighborhood is directed, nine independent pair variables are required. The evolution equations for these pairs retain the asymmetry of the interaction rules, leading to a closed set of nonlinear difference equations. Numerical integration of the pair‑MF system reveals several key phenomena absent from the simple MF description:

1. Traveling density waves – The predator and prey densities do not oscillate uniformly in space; instead, a wave of high density propagates along the direction defined by the anisotropic neighborhood (right‑up diagonal).
2. Direction‑dependent damping – Behind the wave front, oscillations are strongly damped, while ahead of the front they persist, indicating a directional stability.
3. Velocity proportional to anisotropy – The speed of the traveling wave scales with the degree of asymmetry (i.e., the relative weight of the right versus upper neighbor). When the neighborhood becomes symmetric, the wave velocity tends to zero and the system reverts to the homogeneous MF limit cycle.

These results demonstrate that spatial anisotropy can fundamentally alter the macroscopic dynamics of predator‑prey systems. By biasing the flow of individuals, the lattice model generates self‑organized spatiotemporal patterns that may correspond to ecological phenomena such as seasonal migrations, river‑driven fish movements, or wind‑driven insect swarms. Moreover, the anisotropic PCA provides a mechanistic explanation for how local directional interactions can stabilize populations: the wave‑like transport of predators prevents localized over‑exploitation of prey, while the trailing damping region reduces the risk of predator over‑population.

The paper concludes by emphasizing the broader implications of the work. First, it highlights the limitations of traditional homogeneous mean‑field approaches for spatially extended ecological models. Second, it showcases the pair‑MF framework as a tractable yet sufficiently detailed analytical tool to capture direction‑dependent effects without resorting to full Monte‑Carlo simulations. Finally, the authors suggest several extensions: incorporating heterogeneous landscapes (varying site capacities), adding more species (e.g., a second predator or a top‑down control), and coupling the PCA to external periodic forcing (seasonality) to explore resonance phenomena.

In summary, the study provides a clear demonstration that anisotropic interaction neighborhoods in a probabilistic cellular automaton generate traveling population waves and direction‑dependent damping, fundamentally reshaping the collective oscillations predicted by classical Lotka‑Volterra dynamics. This insight bridges stochastic cellular automata, spatial ecology, and nonequilibrium statistical physics, opening avenues for more realistic modeling of ecosystems where environmental anisotropies play a crucial role.