Delayed Pattern Formation in Two-Dimensional Domains

This study investigates how the interaction between gene expression time delay and domain size governs spatio-temporal pattern formation in a reaction-diffusion system. To investigate these phenomena, we utilize a modified version of the Schnakenberg model called the ligand internalisation (LI) model. In a one-dimensional domain, a linear relationship has been observed between the gene expression time delay and the time it takes for patterns to form. We extend the model to the two-dimensional domain and confirm that a similar relationship holds there as well. However, our exploration reveals a non-monotonic correlation between domain size and the time required for pattern emergence. To unravel these dynamics, we consider a range of initial conditions, including random perturbations of the spatially homogeneous steady state and initial conditions from its unstable manifold. We compute a two-parameter chart of patterns with respect to time delay and domain size.

💡 Research Summary

This paper investigates how gene‑expression time delay and the size of the spatial domain jointly influence the emergence of Turing patterns in a reaction‑diffusion system. The authors employ a modified Schnakenberg model, termed the ligand internalisation (LI) model, in which the activator dynamics contain a discrete time delay τ representing the intracellular processing time of a ligand. While previous work demonstrated a linear relationship between τ and the time required for pattern formation in one‑dimensional domains, the present study extends the analysis to two‑dimensional rectangular domains and explores the additional role of domain dimensions (Lx, Ly).

The model equations consist of two coupled partial differential equations for the activator u and inhibitor v, each diffusing with coefficients du = 0.01 and dv = 0.2. Neumann boundary conditions are imposed to reflect impermeable biological boundaries. The homogeneous steady state E* = (a + b, b/(a + b)²) is identified, and linear stability analysis is performed by perturbing around this equilibrium. Assuming modal solutions of the form e^{λt}cos(kxπx/Lx)cos(kyπy/Ly), the authors derive a characteristic equation that includes an exponential factor e^{−λτ}, which encapsulates the effect of the delay on the eigenvalues.

From the characteristic equation, the maximal real part of the spectrum, α(τ, Lx, Ly), is defined. Positive α indicates a Turing instability and thus the onset of spatial patterning. The analysis reveals several key findings:

1. Delay‑induced slowing – As τ increases, the real part of the leading eigenvalue decreases roughly linearly, extending the time before a pattern becomes observable. A critical τ exists beyond which the homogeneous state remains stable for the chosen kinetic parameters (a, b).

2. Non‑monotonic domain‑size effect – Varying Lx (or Ly) does not produce a simple monotonic change in α. Instead, α exhibits a bell‑shaped dependence on domain size: for small domains the most unstable modes are suppressed; at intermediate sizes a resonance between the domain’s natural wavelengths and the most unstable wave numbers maximises α; for larger domains α declines again as higher‑order modes become dominant but grow more slowly. This non‑monotonicity explains why pattern formation time can both increase and decrease with domain enlargement.

3. Initial‑condition sensitivity – Random perturbations of the homogeneous steady state typically excite the most unstable mode, leading to canonical stripe, spot, or lattice patterns. In contrast, initial conditions taken from the unstable manifold of the steady state can preferentially select subdominant modes, producing alternative morphologies (e.g., circular islands) even under identical parameter sets. This highlights the importance of early‑stage fluctuations in determining final pattern geometry.

4. Two‑parameter bifurcation maps – By scanning the kinetic parameters (a, b) while varying τ and domain size, the authors construct heat‑maps of α. With τ = 0 the Turing region occupies a broad swath of the (a, b) plane. Introducing a delay shrinks this region and can eliminate it entirely for sufficiently large τ. However, for certain domain sizes the Turing region re‑expands, illustrating a compensatory interaction between temporal delay and spatial scale.

The numerical simulations corroborate the analytical predictions, showing that the time required for pattern emergence scales linearly with τ and varies non‑monotonically with Lx, Ly. The paper concludes that both temporal delays and geometric constraints must be considered jointly when predicting or engineering pattern formation in biological or synthetic systems. The presented framework can be extended to more complex geometries, heterogeneous diffusion, and multiple delays, offering a versatile tool for future studies in developmental biology, tissue engineering, and synthetic morphogenesis.