Active Life: How Bacterial Swarms Rewrite the Rules of Invasion and Pattern Formation

In a Petri dish, a bacterial colony doesn’t just grow—it marches. It spreads in ripples, fractures into clusters, or forms intricate honeycombs. These aren’t random quirks. They’re signatures of a deeper physics: the interplay between movement and multiplication. Now, a new model reveals that when organisms don’t just diffuse like smoke but actively propel themselves—like bacteria tumbling forward in a run-and-tumble gait—their growth defies classical predictions. At low activity, they invade faster. But beyond a threshold, the invasion wave shatters. The orderly front dissolves into chaos. And in two dimensions, the system self-organizes into patterns—bands, rings, hexagons—that mirror real microbial colonies. This isn’t just about bacteria. It’s a blueprint for how life spreads when motion and reproduction are in constant negotiation.

The findings, from Bhutia, Perlekar, and Ramola (2026), introduce the active stepping stone model—a minimal framework that fuses active matter physics with population genetics. It shows that activity doesn’t just tweak dynamics; it rewrites them. Traveling waves accelerate by up to 40%, phase separation emerges without alignment forces, and microscale patterns arise from macroscopic instabilities. Most strikingly, excessive activity doesn’t just slow things down—it destabilizes the very mechanism of invasion.

This changes how we think about biological invasions, from antibiotic-resistant pathogens to cancer metastasis. It suggests that motility isn’t just a transport mechanism—it’s a regulatory switch.

The Science

The model builds on two well-established frameworks: the stepping-stone model from population genetics and active lattice gases from statistical physics. In the classic stepping-stone model, individuals live in discrete demes (patches), and migration occurs between neighbors. Birth and death drive genetic drift and selection. But migration is passive—like random walks.

Here, the authors make a crucial upgrade: individuals are active. Specifically, they adopt a run-and-tumble motion, a hallmark of bacterial motility. In this mode, a cell moves in a straight line (a "run") before randomly reorienting (a "tumble"). This persistence breaks time-reversal symmetry and allows for directed motion even without external cues.

On a one-dimensional lattice, each site holds a fixed total population of two types: passive individuals (B) and active ones (A). The A-type splits into two subpopulations: A⁺, biased to move right, and A⁻, biased left. The dynamics unfold in discrete time steps:

1. Birth-death: An A individual reproduces at rate α, replacing a B; or dies at rate β, replaced by a B.
2. Tumble: An A⁺ flips to A⁻ (or vice versa) at rate γ.
3. Migration: A⁺ hops right at rate λ; all individuals exchange with neighbors at rate R_d (diffusive mixing).

This yields three dimensionless parameters:

• Péclet number: $\text{Pe}=\lambda /\sqrt{R_d\gamma }$, measuring the strength of directed motion relative to diffusion and tumbling.
• Carrying capacity: $Z=(\alpha -\beta )/\alpha$, the steady-state fraction of A individuals in the absence of spatial effects.
• Reaction strength: $s=Z\alpha /\gamma$, comparing birth-death rates to tumbling.

From these microscopic rules, the authors derive hydrodynamic equations using a path-integral coarse-graining method. The result is a pair of coupled partial differential equations governing the active fraction $f$ and the local polarization $\mathbf{m}$:

$${\partial }_{t}f={\nabla }^{2}f-\text{Pe}\nabla \cdot [\mathbf{m}(1-f)]+sf(1-f/Z),$$$${\partial }_t\mathbf{m}={\nabla }^2\mathbf{m}-\text{Pe}\nabla [f(1-f)]-2\mathbf{m}+s(1-f/Z)\mathbf{m}.$$

These equations unify active transport, demographic fluctuations, and spatial structure. They reduce to known limits: when $\text{Pe}=0$, they yield the Fisher-KPP equation for passive invasion waves. When $s=0$, they resemble models of motility-induced phase separation (MIPS) in active matter.

Crucially, the derivation is microscopically grounded—not phenomenological. Every term maps to a physical process. This allows for rigorous validation against Monte Carlo simulations of the particle model, which the authors perform across one and two dimensions.

What They Found

The model reveals five distinct dynamical regimes, governed by the interplay of activity (Pe) and reaction strength (s):

1. Motility-Induced Phase Separation (MIPS): At high Pe and low s, active particles cluster into dense domains, separated by voids—despite no attractive forces. This is classic MIPS, driven by self-trapping: particles move into dense regions faster than they tumble out.

2. Microphase Separation: At intermediate s, the large MIPS domain becomes unstable and fragments into finite-sized clusters. These are not transient—they’re stable, self-sustained structures. In high-density cores, death dominates; in low-density peripheries, birth prevails. Activity continuously feeds newborns into clusters, balancing the flux. The result is a dynamic equilibrium of localized patches.

3. Traveling Waves: When s and diffusion are comparable and Pe is low, the system supports stable invasion fronts. These resemble Fisher-KPP waves—but faster. The wave speed increases with Pe, peaking around 40% above the passive limit.

4. Wave Destabilization: Beyond a critical Pe, the wave front distorts. The smooth, pulled front breaks down. This isn’t just noise—it’s a structural failure of the wave mechanism.

5. Pattern Formation (2D): In two dimensions, the model generates banded, labyrinthine, ring-like, and hexagonal patterns—structures observed in real bacterial colonies (

). These arise from finite-wavelength instabilities, where a band of Fourier modes grows simultaneously, leading to spatially periodic structures.

One of the most striking results is the non-monotonic effect of activity on wave speed.

Wave speed increases with activity—then collapses

Label	Value
0	2
2	2.4
4	2.8
6	3
8	2.9
10	2.6

Velocity peaks around Pe=6, then declines—wave structure breaks down.

shows the asymptotic velocity $v$ as a function of Pe. At low Pe, $v$ increases linearly, consistent with an effective diffusion enhancement. The authors derive $v=2\sqrt{D_{\text{eff}}s}$, where $D_{\text{eff}}=1+\kappa$ and $\kappa ={\text{Pe}}^2/2$. This captures the speed-up: activity effectively boosts diffusion at the wave’s leading edge.

But beyond $\text{Pe}\approx 5$, the velocity plateaus and then drops. The wave no longer behaves as a pulled front. Phase portrait analysis (

Unstable Fourier modes shift with reaction strength

Label	Value
0.1	0.5
0.3	0.7
0.5	0.9
0.7	1

As reaction strength s increases, unstable band narrows and shifts to higher Pe.

) reveals why: the denominator in the traveling-wave ODE,

$$1+\kappa (1-f)(1-2f),$$

vanishes at finite $f$, creating singularities. At $\text{Pe}=10$, for instance, singularities appear at $f\approx 0.52$ and $f\approx 0.97$. The heteroclinic orbit connecting $(1,0)$ to $(0,0)$ breaks. The smooth wave profile can no longer be sustained.

In two dimensions, the model reproduces patterns seen in experiments.

(c)–(f) show MIPS domains, banded structures, ring-like voids, and hexagonal lattices—all from the same hydrodynamic equations. The hexagonal pattern, in particular, matches observations in Proteus mirabilis colonies, which form concentric rings and honeycomb structures during swarming (Budrene & Berg, 1991).

The linear stability analysis further explains pattern selection.

Phase portrait singularities at high activity

Label	Value
Pe=0	100
Pe=10	30

100% Pe=0

At Pe=10, singularities disrupt the smooth connection between fixed points.

plots the band of unstable Fourier modes $k_{-}<k<k_{+}$ in the $(\text{Pe},Z)$ plane. At low $s$, the unstable band is wide, favoring long-wavelength MIPS. As $s$ increases, the band narrows and shifts to higher Pe, selecting finite-wavelength instabilities—precisely the regime where microphase separation and hexagonal patterns emerge.

Why This Changes Things

For decades, the Fisher-KPP equation has been the gold standard for modeling biological invasions: cancer spreading through tissue, species colonizing new habitats, or genes sweeping through populations. Its core prediction—that invasion speed is determined by the leading edge, where density is low and growth is exponential—has shaped ecological theory.

But KPP assumes passive diffusion. Real organisms are rarely so passive. Bacteria swim. Cells crawl. Animals migrate. This model shows that when motility is active, the invasion rules change.

First, invasion can be faster—but only up to a point. The 40% speed-up at moderate activity suggests that engineered motility (e.g., in synthetic biology) could accelerate colonization. But it also implies that suppressing motility might not always slow invasions; in some regimes, it could prevent wave collapse and maintain steady spread.

Second, phase separation can emerge without attraction. In passive systems, clustering requires attractive interactions or chemical gradients. Here, it arises purely from the kinetic trapping of active particles. This offers a parsimonious explanation for bacterial biofilm formation: no quorum sensing, no adhesion molecules—just persistent motion and birth-death balance.

Third, patterns are not just beautiful—they’re functional. Hexagonal arrays maximize packing efficiency. Ring-like voids may facilitate nutrient diffusion. Banded structures could enable coordinated movement. The model suggests these aren’t accidents of growth but emergent solutions to the trade-off between spreading and sustaining density.

This has implications beyond microbiology. In cancer, tumor cells often switch to a migratory phenotype before metastasizing. The model suggests that this motility doesn’t just enable spread—it could reshape it. A highly motile subpopulation might not just move faster; it might fragment into satellite clusters, evading detection. Or, if too motile, it might fail to establish stable colonies at all.

Similarly, in conservation biology, assisted migration programs aim to relocate species in response to climate change. If the introduced population is too active, it might not form stable fronts—instead, dispersing too thinly to survive. The model provides a framework to optimize such efforts: not just how many to release, but how motile they should be.

What’s Next

The model is minimal—by design. It assumes fixed population per site, discrete spin states, and no environmental feedback. Real systems are messier. Nutrients deplete. Cells sense gradients. Mechanical forces constrain growth. Future work could integrate these elements.

One key question: how do these patterns scale? The simulations are finite-sized. In nature, invasions span kilometers. Does the wave destabilization persist at larger scales? Or does noise or heterogeneity restore stability?

Another: can we control pattern formation? If hexagonal clusters enhance survival, could we tune activity to promote them? This opens the door to morphogenetic engineering—designing not just what cells do, but how they organize.

Finally, the breakdown of the pulled-wave approximation at high Pe demands a new theory. Pushed waves, where bulk dynamics set the speed, are one possibility. But the singularity in the phase portrait suggests a deeper topological change. A full bifurcation analysis could reveal new classes of active fronts.

The active stepping stone model is more than a theoretical curiosity. It’s a bridge between the physics of active matter and the biology of growth. It shows that when life moves with purpose, it doesn’t just go places—it creates them.