Explosive dispersal of non-motile microbes through metabolic buoyancy

TL;DR

Non-motile microbes in quiescent fluids are shown to escape diffusion limits via metabolism-driven buoyancy that launches a living Rayleigh-Bénard convection. The authors combine viscous-medium experiments with a minimal scaling framework to show a parameter-free relation $\beta = (\alpha + 2)/4$ between front and area growth, and extend the theory to fractal biomass distributions, predicting $D_{eff} \approx 2.3$–$2.5$ to account for observed deviations. The resulting Circulation-Driven Aggregation produces fractal colony morphologies with $D_f \approx 1.71$ and an algebraic seed-size distribution with $\xi \approx 1.15$, and the mechanism generalizes across yeast and bacteria, indicating a universal active-matter dispersal strategy. This work introduces a physics-based, autocatalytic transport engine for proliferating active matter with potential implications for biofilms, sediment ecology, and niche construction across diverse habitats.

Abstract

For non-motile microorganisms, spatial expansion in quiescent fluids is presumed to be limited by diffusion. We report that microbial colonies can explosively circumvent this constraint through a self-amplifying physical process. As non-motile yeast and bacteria metabolize dense nutrients into lighter waste within their fluid environment, they generate buoyancy-driven Rayleigh-Bénard convection, an ubiquitous fluid-dynamical phenomenon that organizes material on scales from chemical reactors to planetary atmospheres. This robust, self-generated flow fragments and disperses cellular aggregates, which seed new growth sites, enhancing total metabolic activity and further strengthening the convective flow in an autocatalytic cycle. The resulting expansion follows accelerating power-law kinetics, quantitatively captured by a physical theory linking metabolic flux to flow velocity, and produces fractal patterns through a flow-focusing instability we term Circulation-Driven Aggregation, the hydrodynamic analogue of Diffusion-Limited Aggregation. This `metabolic fireworks' mechanism establishes a canonical instance of proliferating active matter, where cellular metabolic activity self-organizes a physical transport engine--a living Rayleigh-Bénard convection--providing a fundamental, physics-based dispersal strategy.

Figures (15)

• Figure 1: The Metabolic Fireworks Phenomenon. (a) Side-view time series showing the vertical stretching and subsequent dispersal of the initial yeast colony seeded at the bottom ($\eta/\eta_{w}\sim10^{4}$, movie S1). Experimental flow fields (PIV) from the (b) side view, (c) top plane (showing outward flow), and (d) bottom plane (showing inward flow), revealing a toroidal circulation. The colour bars show the flows speeds in units of mm/h. (e) Time sequence (side view) showing flow-induced fragmentation and detachment of a colony segment (red arrow indicates a representative fragmentation event), movie S2. (f) The resulting 2-dimensional morphology of the colony, movie S3. H = 8 mm in a-d. Scale bars, 10 mm (a, b, c, d), 2 mm (e), 20 mm (f).
• Figure 2: Explosive expansion kinetics. (a) Schematic of the experimental setup showing key parameters and observed scaling behaviors. (b) Time series (bottom view) showing the fractal, firework-like expansion of the colony over 4 days. The kinetics of expansion exhibit power law behaviors (H = 8 mm, $\eta/\eta_{w}\sim10^{4}$). (c) The total colonized area $A(t)$ grows explosively, $A(t) \propto t^{\alpha}$ ($\alpha \approx 3.5$). (d) The radius of the expanding front $R(t)$ shows super-linear growth, $R(t) \propto t^{\beta}$ ($\beta \approx 1.3$), indicating acceleration. (e) Local growth of individual seeds $S(t)$ follows a slower power law, $S(t) \propto t^{\gamma}$ ($\gamma \approx 1.0-1.9$, fig. \ref{['figS:Controls-extras']}b). The power law exponents for (f) area expansion ($\alpha$) and (g) radial expansion ($\beta$) increase with the fluid height H, for different container diameters. (h) Experimental validation of the theoretical prediction $\beta = (\alpha+2)/4$ across all tested conditions (including variations in height H, diameter D, and viscosity $\eta$, movies S3-S6). Scale bars, 20 mm (b).
• Figure 3: Fractal Morphogenesis and the Transport Cycle. (A.) Diversity of fractal morphologies observed under varying conditions (D: baseline diameter; 2D: double diameter; 2H: double height; 2$\eta$: double viscosity, movies S3-S6). Insets show magnified views. (b) Box-counting analysis reveals a robust fractal dimension $D_{f} \approx 1.71$ across conditions, characteristic of DLA/CDA processes. (c) The probability distribution function (PDF) of satellite colony sizes (S) follows a power law $P(S) \sim S^{-\xi}$, consistent with the hyper-scaling prediction $\xi = 2 - D_f/2$. (d) Close-up image showing characteristic "horseshoe" shapes of dispersed seeds. (e) Time sequence (side view) capturing flow-induced stretching, fragmentation, and folding of colony filaments during transport (red circles highlight the formation and detachment process, movie S7, S8). Scale bars, 10 mm (a-left column), 20 mm (a-right column), 5 mm (d), 2.5 mm (e).
• Figure 4: Generality of the mechanism. (a) Robustness to initial conditions. The self-organized dispersal pattern emerges regardless of the initial inoculum geometry (point, streak, circle, spiral) in high viscosity media ($\eta/\eta_{w}\sim10^{4}$). (b) Dispersal persists at medium to low viscosities ($\eta/\eta_{w} \sim 10^2, 10^1$), resulting in "Swiss-cheese"or "tafoni"-like morphology due to increased erosion (movie S9). (c) Dispersal even in nutritious water (YPD, $\eta/\eta_{w} \sim 1$) Bottom: Corresponding PIV flow field (color bar units: mm/h). (d) Fractal colony morphologies observed in a phylogenetically distinct, non-motile coccus microbe at intermediate viscosity ($\eta/\eta_{w} \sim 10^{2}$). Dispersal dynamics (top) and PIV flow field (bottom; color bar units: mm/h) for the coccus microbe, demonstrating the generality of the physics-based strategy. Scale bars, 10 mm (a), 5 mm (b), 20 mm (c-top), 10 mm (c-bottom), 10 mm (d), 10 $\mu m$ (d-inset), 10 mm (d-bottom).
• Figure S1: Controls.a. Simultaneous bulk metabolic firework and range expansion at the air-liquid substrate interface atis2019microbial (white circles) observed for $\eta/\eta_w \sim 10^{4}$, $H = 8$ mm. When yeast is inoculated on the top surface of the liquid substrate (white arrow), part of the inoculum detaches and sinks to the bottom of the dish, initiating the firework, while the remaining portion forms a donut-like top colony (white arrow), leading to a parallel range expansion at the interface. b. Slow radial growth in 2% agar. c. Very slow dynamics under low-sugar conditions (0.2%), affirming the role of metabolism driving the flows. d,e. Top and bottom plane PIV flow fields at 24 h. f. Flow-field control in side view without yeast. Scale bars, 10 mm (a-f).