{"id":"16ce5879-b1d5-41ff-a604-4f8dc8548bcd","arxiv_id":"2505.05437","paper_version":4,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Cell size polydispersity, not shear flow, controls the disorder-to-crystal transition in the developing fly wing epithelium.","lead":"This paper finds that the developing fruit fly wing turns from a disorderly jumble of cells into a regular honeycomb pattern because the cells become more uniform in size, not because the tissue is stretched by flows. The result offers a new physical control knob for how biological tissues order themselves during development.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Measured polydispersity is an outcome variable in experiments; the paper does not establish that it tracks the model's quenched preferred-area polydispersity.","rationale":"The paper has real strengths: the vertex-model phase diagram with finite-size scaling (SI Figs. S6, S9), defect-density analysis (SI Fig. S7), the distal-ablation and cdc2 experiments showing crystallization without shear or division, and the shear/quench simulations that reproduce the local vs tissue hexatic decoupling. These establish that quenched preferred-area polydispersity can control melting in the model and that shear accelerates alignment but is not required. The central claim, however, is that the experimental crystallization is controlled by cell size heterogeneity. That requires the measured Δ_ex to be a faithful, causal proxy for the model's Δ. The manuscript explicitly defines Δ_ex on smoothed actual areas (SI Eq. S27) and Δ on preferred areas (Sec. III.A) but never tests their relationship. Because actual areas are mechanical outputs, the observed narrowing of the area distribution (Fig. 4G, S10A) may simply accompany crystallization. The cdc2 result—where crystallization occurs despite Δ_ex staying higher than in WT—already hints that Δ_ex is not the sole determinant; the authors invoke finite-size shifts (SI Sec. 5), which further loosens the quantitative mapping. The concrete simulation proposed would settle whether Δ_ex computed from the model behaves like the experimental observable; if it does not, the experimental support for the central claim reduces to a correlation that could be reversed. This is an addressable gap, so the conditional verdict is appropriate; the current evidence does not warrant rejection, but the central claim should not be accepted without the mapping test.","tokens_in":34781,"tokens_out":6432,"duration_ms":76375,"concrete_test":"Run overdamped vertex-model simulations (same parameters as SI Sec. 6, Λ_F=0.35Λ0) with fixed quenched preferred-area polydispersities Δ=0.1, 0.2, 0.3. From the simulated actual cell areas, compute Δ_ex exactly as in SI Sec. 5: 1 h moving average, exclusion of dividing cells and vein-adjacent rows, and the same L4–L5-like subregion. Plot Δ_ex versus time and versus ⟨|ψ6|⟩. If Δ_ex decreases or correlates with order while input Δ is held constant, then the experimental Δ_ex trajectory is not evidence that the model's control parameter decreases; the causal arrow in the comparison of Fig. 4E–G is confounded. If instead Δ_ex remains near input Δ across the ordering transition, the proxy assumption is supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing link between experiment and model is the identification of the measured cell-size polydispersity Δ_ex (SI Eq. S27) with the model's quenched preferred-area polydispersity Δ (Sec. III.A). Δ_ex is computed from actual cell areas after a 1 h moving average (SI Sec. 5); Δ is an input energy-landscape parameter. Actual cell areas in the vertex model are not equal to preferred areas: they are shifted and narrowed by mechanical interactions and by the same bond-tension noise that drives ordering. Hence a decrease in Δ_ex during pupal development (Fig. 4G) could be a consequence of the tissue becoming ordered, rather than an independent decrease in the proposed control parameter. The proximal-ablation experiment (Fig. 5) is consistent with the claim, but it is a single, pleiotropic perturbation that changes tissue attachment and mechanics, so it does not resolve directionality. The paper does not provide a calibration relating the variance of measured/smoothed areas to the variance of preferred areas, nor a simulation showing that Δ_ex computed from vertex-model cell areas tracks input Δ in the same way. Without that mapping, the apparent coincidence of the experimental onset around Δ_ex≈0.2 and the model's transition near Δ≈0.12–0.2 is not decisive.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies ordering of the Drosophila pupal wing epithelium and argues that a disorder-to-crystal transition is controlled by cell size polydispersity rather than by tissue shear flow. The authors quantify local and tissue-scale hexatic order in wild-type wings and in distal ablation, cdc2, dumpy, and PCP mutant conditions, finding that local crystalline order emerges even when shear flow is largely suppressed, whereas tissue-scale alignment of crystallites is weak without shear. They introduce a vertex model with quenched preferred-area polydispersity and bond-tension noise, and show, with finite-size scaling and defect analyses, a crystal-to-hexatic-to-liquid transition as a function of polydispersity Δ, with a critical value around Δ≈0.12. Kinetic simulations with and without imposed shear reproduce the decoupling of local and tissue hexatic order observed in distal ablation experiments. Experimentally, a smoothed measure of cell-area polydispersity Δ_ex decreases over pupal development, and a proximal ablation condition with persistently high Δ_ex fails to crystallize. The paper concludes that reduction of cell size heterogeneity controls the transition, while shear accelerates crystallite alignment but does not itself induce order.","tokens_in":35056,"tokens_out":5243,"duration_ms":65355,"significance":"If the central claim holds, the paper identifies cell size polydispersity as a control parameter for tissue-scale crystalline ordering, connecting the physics of polydisperse two-dimensional melting to epithelial morphogenesis. The work is careful and unusually broad in its perturbations: distal ablation, cdc2, dumpy, and several PCP mutants are analyzed, and the crystallite-size distributions and alignment strengths are quantified. The vertex-model phase diagram is supported by finite-size scaling, susceptibility-like variances, and topological-defect densities, and the kinetic simulations make a concrete, falsifiable prediction that shear accelerates alignment but does not generate order for high polydispersity. A notable strength is that the order parameters used in experiments and simulations are defined consistently, including the patch-wise removal of average cell elongation. The main weakness is the quantitative link between the experimentally measured Δ_ex and the model's preferred-area polydispersity Δ, which is not directly validated; this affects the strength of the claim that the observed decrease in Δ_ex causes the ordering rather than being a consequence of it.","major_comments":[{"comment":"The identification of the measured cell-size polydispersity Δ_ex with the model's quenched preferred-area polydispersity Δ is not established. Δ_ex is computed from smoothed actual cell areas (SI Eq. S27), whereas Δ is an input energy-landscape parameter. In the vertex model, actual cell areas are not equal to preferred areas: they are shifted and narrowed by mechanical interactions and by the bond-tension noise that drives ordering. A decrease in Δ_ex during development could therefore be a consequence of the tissue becoming ordered rather than an independent decrease of the proposed control parameter. The proximal ablation experiment is consistent with the claim but is a single, pleiotropic perturbation. The paper should provide a calibration: compute Δ_ex from vertex-model cell areas using the same moving-average and exclusion protocol, and show that it tracks the input Δ across the phase diagram, and further test whether Δ_ex changes as order develops at fixed input Δ. Without this mapping, the apparent crossing of the experimental onset near Δ_ex≈0.2 and the model transition near Δ≈0.12–0.2 (Figs. 4G and 5G) is not decisive.","section":"Sec. III.A and SI Sec. 5, Eq. S27"},{"comment":"The proximal ablation experiment is presented as the critical causal test, but it does not isolate polydispersity as the manipulated variable. Ablation at the hinge–blade junction changes tissue attachment, stress distribution, and potentially signaling, and Δ_ex remains an outcome variable in that experiment. The observed failure to crystallize and the elevated Δ_ex are equally consistent with the alternative that the ablation prevents ordering through a mechanical or signaling route which in turn keeps Δ_ex high. To support the causal direction, the authors should either identify a perturbation that specifically changes cell-size heterogeneity without altering tissue mechanics or explicitly model the PA condition and show that the elevated Δ_ex alone, with the same mechanics as WT, reproduces the lack of crystallization.","section":"Sec. IV, Fig. 5"},{"comment":"The quantitative comparison between model kinetics and experiments depends on the chosen noise magnitude Λ_F=0.35Λ0, which is selected because it reproduces the experimentally observed decoupling of local and tissue hexatic order (SI Fig. S11). Since Λ_F is not measured independently, the placement of the experimental trajectory relative to the model's critical polydispersity is partly selected by this choice. The authors should quantify how sensitive the predicted ordering threshold and the kinetics are to Λ_F, and should test whether the conclusion that polydispersity controls the transition remains robust for a range of noise levels. A direct estimate of Λ_F from cell-area or junction-length fluctuations in the imaging data would strengthen the comparison.","section":"Sec. III.C, SI Sec. 6.3"},{"comment":"The quoted critical polydispersities Δ_t=0.120±0.001 and Δ_6=0.125±0.001 appear to come from finite-size simulations, while SI Fig. S9 shows that susceptibility peak positions shift with system size. If the quoted values are not extrapolated to the thermodynamic limit, the comparison with the experimental onset around Δ_ex≈0.2 is quantitatively compromised. Please state explicitly at which system size the main-text transition points are evaluated and provide the N→∞ extrapolation if available. In addition, the kinetics simulations reduce Δ(t) over time (SI Eq. S30), but the implementation of this reduction is not described: if preferred areas are dynamically reassigned or rescaled, the quenched disorder that defines the steady-state phase diagram is effectively annealed, which could alter the transition. Clarify the implementation and discuss its effect on the kinetic conclusions.","section":"Sec. III.B and SI Sec. 4.3, 4.6"}],"minor_comments":[{"comment":"The phrase 'even if tissue shear have been inhibited' should be 'even if tissue shear flow has been inhibited'; similar grammar corrections are needed in several places (e.g., 'the dumpy mutant wing the average cell hexatic' in SI Sec. 2.2).","section":"Abstract"},{"comment":"The main text uses cumulative shear strain ν=1/2, while the SI describes a simple shear implemented via Lees–Edwards boundary conditions and then states that a total pure shear strain γ=1/2 was imposed. Please make the notation and the shear protocol consistent throughout.","section":"Fig. 4 and SI Sec. 6.1"},{"comment":"The robustness of Δ_ex to the smoothing window T is reassuring, but the authors should also state whether the exclusion of dividing cells affects the late-time Δ_ex values in WT, since division rates decline over the same period; this would clarify whether the Δ_ex decrease could partly reflect the division-rate decrease.","section":"SI Sec. 5, Fig. S10H"},{"comment":"In the distal ablation experiment, 'largely reduces' shear flow is not the same as the zero-shear condition simulated in the vertex model. A sentence quantifying the residual shear rate in the DA wing (or referencing its measurement from the original study) would help the reader judge how directly the no-shear simulations map onto the experiment.","section":"Sec. II, Fig. 2"},{"comment":"The radial hexatic correlation function in the crystal phase is described as remaining constant, but the plotted data appear to decay near the box-size scale; the text should acknowledge that system-size effects limit the ability to distinguish algebraic from exponential decay, which is already mentioned later but should be stated at the first presentation of the plot.","section":"SI Sec. 4.4, Fig. S7E"}],"recommendation":"major_revision","confidential_remarks":"The paper is potentially impactful for epithelial tissue physics, and the experimental breadth is commendable. The main risk is that the quantitative claim rests on an unvalidated identification between the measured cell-area polydispersity and the model's quenched preferred-area polydispersity. This is fixable within the manuscript's scope by adding vertex-model calibration of Δ_ex and by strengthening the causal interpretation of the proximal ablation experiment. I would not support rejection, but the current version does not yet establish the central causal claim at the level the abstract states."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The short version: this paper makes a credible, well-supported case that cell size polydispersity — not tissue shear — controls the disorder-to-crystal transition in the fly wing, and that shear only accelerates alignment of crystallites. It deserves a serious referee. The main soft spot is a calibration gap between the measured and modeled polydispersity; it's addressable and not fatal.\n\nWhat's new: the polydispersity-as-control-parameter claim for a developing epithelium is a real departure from the shear-centric picture in Sugimura & Ishihara 2013. The authors back it with a broad perturbation set — distal ablation (local order persists without shear), cdc2 (no divisions), dumpy, and PCP mutants — and they show that tissue-scale order specifically requires shear while local order does not. The vertex model side is careful: finite-size susceptibility analysis, a hexatic phase with sequential dislocation/disclination unbinding, and kinetic simulations where shear aligns crystallites but cannot induce order in the liquid phase. The finite-size prediction that smaller tissues crystallize at higher polydispersity, matching the cdc2 wing, is a nice touch.\n\nWhere I'd push back: the link between the measured Δ_ex and the model's quenched preferred-area polydispersity Δ is the load-bearing joint, and it is not calibrated. Δ_ex is computed from smoothed actual cell areas; in the vertex model, actual areas are narrowed by mechanics and noise, so a decrease in Δ_ex during development could partly be a consequence of ordering rather than an independent decrease in the proposed control parameter. A simulation that computes Δ_ex from vertex-model output and shows it tracks input Δ would close this. Until then, the coincidence of the experimental onset near Δ_ex ≈ 0.2 and the model transition near Δ ≈ 0.12–0.2 is suggestive, not decisive. The proximal ablation is the right kind of test — high polydispersity, no crystallization — but it's a pleiotropic perturbation, so it does not fully resolve directionality on its own.\n\nMinor points: Λ_F = 0.35Λ_0 is chosen to reproduce the local/tissue hexatic decoupling, so the kinetic match is a fit, not an independent prediction, though the paper states this openly. No code or processed data is released, which is a reproducibility gap but not a scientific flaw. The authors are also honest that what drives the polydispersity decrease is unknown.\n\nOverall: the central claim holds up; the directionality worry is real but does not break the paper, especially with the proximal ablation and dumpy data in view. This is for developmental biophysicists and soft-matter people. Send it to review — with a request for the calibration check.","headline":"Solid experiment-plus-model case that cell size polydispersity controls fly wing crystallization, with shear demoted to an accelerator; the measured-to-model polydispersity mapping needs calibration but the central claim holds.","tokens_in":35603,"tokens_out":7176,"would_cite":true,"duration_ms":73928,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Cell size spread, not shear, gates fly-wing crystallization","keywords":["cell size polydispersity","fly wing epithelium","crystallization","vertex model","hexatic order","tissue shear flow","KTHNY melting","morphogenesis"],"falsifier":"Track a wing in which cell size polydispersity is held above the inferred threshold (for example by perturbing cell size homeostasis without blocking division) while tissue shear flow remains intact: if local hexatic order still rises strongly, the polydispersity-control claim fails. Conversely, if polydispersity is reduced early in development while shear is absent and tissue-scale hexatic order still grows substantially, the shear-enhancement claim would need revision.","tokens_in":34620,"feed_emoji":"🪰","tokens_out":3641,"duration_ms":38901,"temperature":0.7,"pith_summary":"This paper argues that the fruit fly wing epithelium crystallizes during pupal development because its cells become more uniform in size, not because tissue shear flow orders them. Analyzing time-lapse movies of wild-type and perturbed wings, the authors find local hexagonal order appears even when shear flow is suppressed, while tissue-scale alignment requires shear. A vertex model with randomly assigned preferred cell areas shows a sharp crystal-to-disordered transition at a critical polydispersity around $\\Delta \\approx 0.12$, with a narrow hexatic window. Experimental polydispersity starts above this threshold and falls below it during development, matching the slow coarsening kinetics of the model. If right, cell size heterogeneity is a control parameter for tissue ordering, with consequences for how regular epithelia form and how tissue fluidity is regulated.","feed_headline":"Cell size spread, not shear, gates fly-wing crystallization","feed_subtitle":"As fly wing cells even out in size they lock into a crystal; shear only aligns the patches.","key_machinery":"The central object is a vertex model of epithelial tissue with cell size polydispersity: each cell is assigned a preferred area $A_{0,c}$ uniformly drawn from $[(1-\\sqrt{3}\\Delta)A_0, (1+\\sqrt{3}\\Delta)A_0]$, with $\\Delta$ the normalized standard deviation of preferred areas, and bond tensions fluctuate via an Ornstein-Uhlenbeck process with magnitude $\\Lambda_F$. The model is simulated both quasistatically and with overdamped vertex dynamics, including T1 transitions and Lees-Edwards shear. Order is quantified by the cell hexatic $\\Psi_6$, the tissue hexatic $\\langle \\Psi_6 \\rangle$, and a translational order parameter, with susceptibility peaks and defect densities locating the transition. The machinery works by showing that increasing $\\Delta$ melts the crystal through sequential unbinding of dislocations and disclinations, and that reducing $\\Delta$ across the transition produces the slow coarsening kinetics observed in the wing.","core_discovery":"The central claim is that the disorder-to-crystal transition in the developing fly wing is controlled by cell size polydispersity, defined as the normalized standard deviation of the cell area distribution, and not by tissue shear flow. In a vertex model where cells are assigned preferred areas spread over an interval of width $\\Delta$, the authors find a sharp transition near $\\Delta \\simeq 0.12$: below this critical value the packing is crystalline, above it the packing is disordered, with the translational and orientational order parameters peaking at slightly different polydispersities ($\\Delta_t = 0.120 \\pm 0.001$, $\\Delta_6 = 0.125 \\pm 0.001$), hinting at a hexatic phase. Experimentally, the authors measure cell area polydispersity in wild-type wings and in distal ablation, proximal ablation, $cdc2$, $dumpy$, and planar cell polarity mutants. They find that polydispersity decreases during pupal development in wild-type and distal ablation wings, that local hexatic order emerges when $\\Delta_{\\mathrm{ex}}$ falls below roughly $0.2$, and that proximal ablation keeps polydispersity high and blocks crystallization. Shear flow, they conclude, does not drive the transition itself but significantly accelerates the alignment of locally ordered crystallites, which is why tissue-scale hexatic order fails to develop when shear is suppressed.","pith_inferences":["A direct, testable prediction of this picture is that artificially maintaining a broad cell size distribution should block crystallization even when tissue shear flows are intact; this could be probed with perturbations that disrupt cell size homeostasis without arresting divisions.","Because polydispersity also controls relaxation times in the model, the paper implies that cell size heterogeneity could tune tissue fluidity and solidity during morphogenesis, connecting to jamming and glass-transition ideas in other epithelia.","The proximity of the translational and orientational transition points suggests a KTHNY-like two-step melting scenario in a biological tissue, but the finite system sizes leave open whether the hexatic phase is genuine or a finite-size effect.","The mechanism that reduces polydispersity during development is left open; the authors note it may reflect cell-cycle arrest and uniform mechanical state, or a proximal coordinating cue, and distinguishing these would sharpen the causal story."],"forward_implications":["Crystallization can proceed without tissue shear: distal ablation and $dumpy$ mutant wings, where shear flow is largely suppressed, still show a strong increase in local hexatic order.","Shear's role is kinetic, not thermodynamic: it aligns already-formed crystallites and raises tissue-scale hexatic order, but it cannot induce ordering when polydispersity is high.","Cell divisions are not required for crystallization: the $cdc2$ mutant, with division arrested, crystallizes similarly to wild-type.","The transition is sharp in the model and the experimental onset of ordering near $\\Delta_{\\mathrm{ex}} \\simeq 0.2$ is consistent with a quench across a phase transition.","Reducing polydispersity over time in the model recapitulates the experimental rise of local hexatic order and the shear dependence of tissue hexatic order."],"supporting_citations":[{"why":"Previous proposal that tissue shear flow and mechanical anisotropy promote hexagonal packing ordering; the paper's experimental perturbations are designed to test this claim.","marker":"[29]"},{"why":"Source of the wing time-lapse data and the distal ablation, proximal ablation, and cdc2 perturbation experiments reanalyzed here.","marker":"[32]"},{"why":"Basis of the vertex model for epithelial packing, including the work function and parameter regime used in the simulations.","marker":"[20, 21]"},{"why":"Colloidal and hard-sphere literature establishing polydispersity as a control parameter for melting, motivating the analogy drawn for cellular packings.","marker":"[17-19]"},{"why":"Report of a hexatic phase in an active Voronoi model, used as comparison for the two-step melting interpretation.","marker":"[24]"},{"why":"Framework of coarsening dynamics of systems quenched across a phase transition, used to interpret the slow ordering kinetics and the accelerating role of shear.","marker":"[35]"}],"fun_headline_variants":["Cell size spread, not shear, decides fly wing crystal order","Fly wing crystallization driven by cell size diversity, not shear","Polydispersity gates fly wing crystal packing, not shear","Cell size heterogeneity, not shear, controls fly wing crystal transition"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The measured polydispersity $\\Delta_{\\mathrm{ex}}$, computed from cell areas after a one-hour moving average and excluding cells adjacent to veins and dividing cells, is assumed to faithfully represent the model's preferred-area spread $\\Delta$, and the noise magnitude $\\Lambda_F = 0.35\\Lambda_0$ is chosen because it reproduces the experimental decoupling of local and tissue hexatic order; if either choice is off, the apparent crossing of the model's critical polydispersity with the experimental ordering onset could be coincidental.","fun_headline_variants_meta":{"raw":{"variants":["Cell size spread, not shear, decides fly wing crystal order","Fly wing crystallization driven by cell size diversity, not shear","Polydispersity gates fly wing crystal packing, not shear","Cell size heterogeneity, not shear, controls fly wing crystal transition"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000214,"raw_usage":{"total_tokens":1515,"prompt_tokens":1126,"completion_tokens":389,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":742,"completion_tokens_details":{"reasoning_tokens":319}},"tokens_in":742,"tokens_out":389,"duration_ms":4019,"temperature":1.0,"reasoning_tokens":319,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T23:04:04.573503+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Track a wing in which cell size polydispersity is held above the inferred threshold (for example by perturbing cell size homeostasis without blocking division) while tissue shear flow remains intact: if local hexatic order still rises strongly, the polydispersity-control claim fails. Conversely, if polydispersity is reduced early in development while shear is absent and tissue-scale hexatic order still grows substantially, the shear-enhancement claim would need revision.","supporting_citations":[{"cited_title":"The mechanical anisotropy in a tissue promotes ordering in hexagonal cell packing,","cited_arxiv_id":null,"evidence_quote":"Previous proposal that tissue shear flow and mechanical anisotropy promote hexagonal packing ordering; the paper's experimental perturbations are designed to test this claim."},{"cited_title":"Hexatic phase in a model of active biological tissues,","cited_arxiv_id":null,"evidence_quote":"Report of a hexatic phase in an active Voronoi model, used as comparison for the two-step melting interpretation."},{"cited_title":"Coarsening dynamics of phase-separating systems,","cited_arxiv_id":null,"evidence_quote":"Framework of coarsening dynamics of systems quenched across a phase transition, used to interpret the slow ordering kinetics and the accelerating role of shear."}],"review_version":1}