REVIEW 3 major objections 5 minor 2 references
Spontaneous spatial sorting by cell shape in growing colonies of rod-like bacteria
T0 review · 3 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash
Pith's one-line read In growing colonies of rod-shaped bacteria, cells with longer aspect ratio increasingly dominate the periphery through purely mechanical interactions, even with equal division times.
desk verdict Periphery enrichment of long rods is new and plausible, but the length-dependent periphery metric means the headline effect may be partly a detection artifact. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The central object is the bidisperse colony of rod-like agents modeled as lengthening, dividing circo-rectangles (rectangles with semicircular caps) of fixed width and two heritable maximum aspect ratios. Dynamics are overdamped Brownian motion with Hertzian repulsion between overlapping rods, so all ordering arises from steric forces during growth. The quantitative measures that carry the argument are the periphery fraction (cells within 5 cell widths of an $\alpha$-complex boundary), the local radial alignment parameter $\bar{s}_r = \frac{1}{N}\sum_i \cos(2(\theta_i - \phi_i))$ that distinguishes radially versus tangentially oriented cells, and local heterozygosity for mixing. Together they link nematic order in radial microdomains to transport of long cells outward and tangential anchoring to their retention at the edge.
What would settle it
A decisive test would be to repeat the growth protocol from an initially well-mixed but radially unbiased, or deliberately tangentially biased, dense droplet and check whether the longer cells still accumulate at the periphery; if the enrichment disappears or reverses, the effect is an artifact of the initial radial bias rather than an emergent mechanical sorting.
Extended reading notes
Core claim
The central discovery is a mechanical shape-sorting effect: in a bidisperse colony with equal division times and no growth-rate or nutrient differences, the fraction of the periphery occupied by the longer-aspect-ratio population rises steadily over time, while shorter cells cluster into monoallelic pockets in the bulk. The enrichment at the boundary scales linearly with aspect-ratio difference and survives changes in the width used to define the periphery. The paper traces the effect to two mechanisms: radial expansion pressure organizes longer cells into radial nematic microdomains that help them break through to the boundary, and tangential active anchoring at the periphery makes long cells harder to pull back into the bulk, while active mixing flows carry shorter cells inward. Even under an equal-elongation-rate protocol where short cells dominate by number, long cells remain overrepresented at the periphery relative to their abundance.
Load-bearing premise
The main load-bearing premise is that the sorting is emergent and not inherited from how the colony is started: the initialization protocol packs a small isotropic droplet, which already gives a slight radial orientation bias and puts shorter cells slightly closer to the center, and the paper does not test a control with the opposite or zero initial bias.
Editorial extensions
If this is right
- Periphery composition is a function of cell shape alone: with equal division times, the longer cell type's periphery fraction rises over time and the effect grows linearly with the aspect-ratio gap.
- Bidispersity suppresses intermixing: when one cell type is below the nematic-ordering threshold, the colony develops large monoallelic clusters and lower local heterozygosity than a monodisperse colony of the same average shape.
- A growing colony acts as an active analogue of granular convection: longer rods rise to the periphery as shorter rods flow inward, analogous to the Brazil nut effect.
- Under nutrient-limited conditions, this mechanical sorting implies a selective pressure favoring high aspect ratio, because the periphery is where growth would occur.
- Even when shorter cells divide faster, longer cells are still overrepresented at the edge relative to their share of the population.
Reading between the lines
- A direct experimental test would be a microfluidic 2D colony of two isogenic strains engineered to have different lengths but identical growth rates; tracking edge composition over roughly nine doublings should reproduce the linear periphery-fraction slope if the mechanism is purely mechanical.
- The same sorting logic may apply to mixtures differing in stiffness or adhesion rather than aspect ratio, since the mechanism only requires shape-dependent nematic order and boundary anchoring.
- Because the initial radial bias is a possible confound, the authors' claim would be sharpened by showing that a tangentially biased or fully random well-mixed initial state still produces long-cell enrichment; absent that control, the early-time sorting signal could partly stem from initialization.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper uses a two-dimensional Brownian dynamics model of growing, dividing circo-rectangle "cells" to study colonies containing two populations with different heritable aspect ratios. Under an equal-division-time protocol, the authors report that the longer-aspect-ratio population becomes progressively overrepresented at the colony periphery, and they propose two mechanisms: radial nematic microdomains that drive longer cells outward, and tangential anchoring plus active nematic mixing that retain them at the boundary. They also report that bidispersity suppresses local genetic mixing when one population is below the nematic-ordering threshold, and that under an equal-elongation-rate protocol longer cells are still overrepresented at the periphery relative to their abundance in the bulk. The manuscript includes statistical averaging over 5-10 independent runs, standard errors, robustness checks on the periphery-thickness parameter, lineage tracking, and tracer-particle control simulations.
Significance. If the central sorting claim holds, the result is interesting and potentially important: it would show that purely mechanical interactions in a growing colony can sort microbial populations by cell shape, with possible evolutionary consequences under nutrient-limited conditions where the periphery is the growth zone. The paper's strengths are its careful statistical reporting (multiple independent runs, standard errors), the use of two different growth-rate protocols, the explicit robustness scan over periphery thickness in SI Fig. 8, and the lineage-history analysis that attempts to connect sorting to radial order. However, the central claim currently rests on a periphery definition that is biased by rod length: longer rods are more likely to be counted as peripheral even in a perfectly mixed colony. Because this confound directly affects the paper's main quantitative evidence, the result is not yet established to the standard required for publication.
major comments (3)
- [Methods, 'Periphery Definition'; Fig. 3; SI Fig. 8] The periphery definition counts a cell if any of five points along its rod lies within wp = 5d of the alpha complex. For a rod of length L at angle θ to the boundary normal, the center can be up to (L/2)|cos θ| farther from the boundary and still be counted, so longer rods have a higher detection probability in a perfectly mixed colony. The positive slope of ϕB versus aB in Fig. 3B and the decrease of that slope with increasing wp in SI Fig. 8 are exactly the signatures of this length-dependent cross-section effect. The caption of Fig. 3A says the periphery comprises cells whose centers lie within 5d, which conflicts with the Methods description of using five points along each rod; if centers were used the bias would largely disappear, but the reported analysis appears to be body-based. Please provide a null-model correction (e.g., random label shuffling or a center-based periphery metric) or otherwise demonstrate that the sorting signal survives removal of this geometric detection bias.
- [Methods, 'Simulation Initialization and Stopping Conditions'; Fig. 3A] The initialization protocol is acknowledged to introduce a radial orientation bias and to place shorter cells slightly more centrally on average. Because the paper's key claim is that the periphery enrichment emerges over time, the analysis must show that this initial bias does not seed or amplify the observed effect. No control simulation with an alternative initial condition, nor a quantitative bound on the initial bias, is provided. The abstract's statement that the findings are robust across initial conditions is not supported by the presented experiments, which vary the growth-rate protocol but not the initialization protocol.
- [Fig. 7D-E (equal elongation rate)] The equal-elongation-rate comparison uses the same body-based periphery metric as the equal-division-time analysis. In this protocol the longer cells are also the rarer population, so the overrepresentation measure ϕA/NA − ϕB/NB is inflated by the same length-dependent detection probability: a small number of long rods contributes a disproportionately large periphery count. A center-based or otherwise length-unbiased periphery measure should be applied before concluding that longer cells are overrepresented at the periphery relative to the bulk in this protocol.
minor comments (5)
- [Fig. 3A caption vs. Methods] The Fig. 3A caption states that periphery cells are those whose centers lie within 5d of the alpha complex, while Methods states that five equally spaced points along each rod are used. Please reconcile this discrepancy, as it is directly relevant to the geometric-bias concern.
- [Fig. 4 and surrounding text] The paragraph after Fig. 4 says 'Replacing the circle of passive tracers with a circle of actively growing Population B in this scenario further confirms...', but Fig. 4A-B show passive tracers and Fig. 4C-D show actively growing rings; the cross-references to panels are confusing and should be corrected.
- [References 40-41] The Introduction cites Refs. 40-41 for the Brazil nut effect, but those references are about phase transitions in mixtures of rods and spheres, not granular convection. The actual Brazil nut references (48-49) are cited later; please fix the citation placement.
- [Model description] The term 'circo-rectangles' is used without a definition or standard reference; please define it at first use or use a more common term such as spherocylinders or rounded rectangles.
- [Fig. 5D discussion] The text states that at early times the more radially aligned cell type is that with the higher aspect ratio, and at late times larger-aspect-ratio cells have higher tangential alignment; it would help to state explicitly that the crossover time is not a fitted parameter and to quantify its uncertainty if possible.
Circularity Check
Reported periphery-vs-aspect-ratio correlation is partly generated by a rod-length-dependent periphery definition.
-
self definitional
[Methods and Materials, 'Periphery Definition'; SI Appendix, Fig. 8]
"To determine which cells belonged to the periphery, a KD-tree was constructed from the points along the alpha complex, and the distance from each rod to the boundary was determined by calculating the nearest-neighbor distance for five equally spaced points along each rod. Cells located within a distance of 5d from the complex were classified as being on the periphery."
The binary 'periphery' label depends on whether any sampled point on the rod lies within wp=5d of the boundary. For a rod of length L and orientation θ relative to the boundary normal, the center may lie up to wp + (L/2)|cosθ| away and still be counted, so longer rods have a larger acceptance window. Under a uniform-placement null, the periphery fraction is therefore an increasing function of aspect ratio by construction, which is exactly the positive correlation reported in Fig. 3B. The SI observation that the slope decreases as wp grows (Fig. 8) is the same geometric scaling predicted by this length-dependent cross-section, not independent confirmation that the effect is a property of the periphery.
full rationale
One measurement step is self-definitional: the periphery label is applied to rods whose body intersects a fixed boundary layer, so longer rods are geometrically more likely to qualify. This injects a length-dependent term into the headline ϕB-vs-aB correlation and into the SI slope-vs-thickness trend, so the score is 6 rather than 0. The circularity is partial: the time evolution of ϕA, the backward lineage statistics, and the passive-tracer/ring experiments do not reduce to the metric and provide independent evidence of active rearrangement. The self-citations (Refs. 32, 37, 44) set parameter values and prior phenomenology but are not load-bearing for the sorting claim. The acknowledged initialization bias is a robustness concern, not a circular construction; however, the Methods-based rod-body periphery criterion is a genuine definitional confound.
Assumptions & free parameters
free parameters (3)
- alpha complex parameter alpha =
1.3
- periphery width w_p =
5d
- radial order threshold =
s_r > 0.8 for two division times
assumptions (5)
- domain assumption Overdamped Brownian dynamics with Hertzian contact forces repels overlapping circo-rectangles.
- domain assumption Nutrients are uniformly distributed, so growth rate is position-independent.
- domain assumption 2D modeling of colony growth captures the sorting mechanism.
- domain assumption Elongation-rate variability is uniformly drawn from [g_X/2, 3g_X/2] to avoid division synchronization.
- domain assumption The dimensionless friction parameter g zeta / F0 L = 5e-7 is taken from prior work (Refs 32, 37, 44).
Cite this review
Pith. "Pith review of Spontaneous spatial sorting by cell shape in growing colonies of rod-like bacteria." pith.science (2026). https://pith.science/paper/AE6ZOHL5
@misc{pith2026250111177,
author = {Pith},
title = {Pith review of: Spontaneous spatial sorting by cell shape in growing colonies of rod-like bacteria},
year = {2026},
howpublished = {\url{https://pith.science/paper/AE6ZOHL5}},
note = {Machine review of arXiv:2501.11177}
}
read the original abstract
Mechanical interactions among cells in a growing microbial colony can significantly influence the colony's spatial genetic structure and, thus, evolutionary outcomes such as the fates of rare mutations. Here, we computationally investigate how this spatial genetic structure changes as a result of heritable phenotypic variations in cell shape. By modeling rod-like bacterial cells as lengthening and dividing circo-rectangles in a 2D Brownian dynamics framework, we simulate the growth of a colony containing two populations with different aspect ratios. Compared to monodisperse colonies, such bidisperse colonies exhibit diminished intermixing between sub-populations when the less elongated cells are too short to nematically order, instead forming large clusters. We find that the cells with longer aspect ratio gradually segregate to the colony periphery. We present evidence that this demixing is related to nematic order in the bulk and to active nematic mixing dynamics near the periphery. These findings are qualitatively robust across different growth rate protocols and initial conditions. Because the periphery is often an advantageous position when nutrients are limited, our results suggest a possible evolutionary selective pressure of mechanical origin that favors large cell aspect ratio.
Figures
Figures from the paper (6 more)
Reference graph
Works this paper leans on
-
[1]
Genetic drift at expanding frontiers pro- motes gene segregation
1O. Hallatschek, P. Hersen, S. Ramanathan, and D. R. Nelson, “Genetic drift at expanding frontiers pro- motes gene segregation”, Proceedings of the National Academy of Sciences 104, 19926–19930 (2007). 13 DRAFT - JANUARY 22, 2025 2L. Excoffier and N. Ray, “Surfing during population expansions promotes genetic revolutions and structura- tion”, Trends in Ec...
arXiv 2007
-
[2023]
51N. J. Mottram and C. J. P. Newton, Introduction to Q-tensor theory, version 2, (2014) https://arxiv. org/abs/1409.3542. 15 DRAFT - JANUARY 22, 2025 SI Appendix Videos of the Growth Supplementary movie S1: Growth of a colony composed of cells with maximum aspect ratio aA = 2 (blue) and aB = 2 (brown), with equal division times TA = TB, from 50 to 25000 c...
arXiv 2014
Reviewed August 10, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.