{"id":"3a722af9-a285-4091-b1f5-41b11f654a84","arxiv_id":"2501.03328","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Star-forming clumps in z~1.5 galaxies span 0.1-1 kpc, follow a mass-size relation and a mass function consistent with hierarchical star formation, and bridge lensed and unlensed observations.","lead":"Using sharp JWST images, researchers split 32 distant galaxies into bulges, disks, and bright clumps, measuring star-forming clumps about 0.1 to 1 kiloparsec across in galaxies at cosmic noon. The work connects small clumps seen in gravitationally lensed galaxies with larger ones seen in ordinary surveys, supporting the idea that clumps form a continuous hierarchy of star-forming structures.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The mass–size and mass-function slopes rely on sub-PSF F150W sizes that are not validated for systematic bias; the Sec. 2.4 completeness test does not re-fit recovered clumps, so a size–flux degeneracy could shift Eq. 1 and α.","rationale":"The paper's central quantitative claims are Eq. 1 and the cSMF slope α=−1.85±0.19. Both depend on sizes measured in F150W, which has PSF FWHM ≈0.4 kpc at the sample redshift. The analysis is careful in many respects: forward modeling with GALIGHT/LENSTRONOMY, BIC-based model selection, PSFEx PSFs, and a detection-completeness test as a function of flux and size (Fig. 7). It also checks that excluding poor SED fits does not significantly change the mass–size slope and that the final clump sample lies within the 90% completeness region. These are real strengths. However, completeness is not accuracy: a clump can be detected yet have its flux and size systematically mis-measured, particularly when its size is below the PSF FWHM. The Sec. 2.4 test does not re-fit recovered artificial clumps, and the Sec. 2.2 uncertainty injections use sources at the assumed real size, so they estimate random scatter rather than bias at sub-PSF scales. A bias at r_e ≈ 0.2–0.4 kpc would directly change the slope of Eq. 1 and the inferred mass-function cutoff, weakening the hierarchical interpretation. The authors' statement that results are unchanged when sub-PSF clumps are excluded mitigates the concern, but the mass–size sample still includes r_e = 0.2–0.4 kpc objects, so the issue is not fully settled. This is a data-in-hand, testable problem, which is exactly why the CONDITIONAL verdict is appropriate; I do not see grounds to move to ACCEPT or REJECT without running the recovery test.","tokens_in":30865,"tokens_out":5300,"duration_ms":49367,"concrete_test":"Re-run the Sec. 2.4 injection on a subset of real F150W images with artificial clumps at known r_e = 0.1, 0.2, 0.3, 0.4, 0.6, and 1.0 kpc, with fluxes spanning the observed range; run the full bulge+disk+clump pipeline and re-fit each recovered clump rather than only counting detections. Compare recovered vs. input r_e and flux as a function of r_e/PSF(FWHM) and local disk contrast, and re-fit Eq. 1 using only clumps with r_e > 0.4 kpc; if the slope moves outside the quoted ±0.07 error or the recovery bias exceeds ~30% for r_e < 0.4 kpc, the sub-PSF size claim is not yet supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Eq. 1 (Sec. 3.3) reports r_e ∝ M_*^0.52±0.07 over 0.1–1 kpc, and Sec. 3.4 uses Eq. 1 to set the completeness-based exclusion before fitting the mass-function slope α=−1.85±0.19. These quantitative results are driven by Gaussian sizes measured in F150W, where the PSF FWHM is ~0.4 kpc at z~1.5 (Sec. 2.1); a 0.1 kpc clump is a quarter of the PSF. The Sec. 2.4 injection test (Fig. 7) only determines whether injected clumps are detected; it does not re-fit recovered sources and compare output flux and size to the injected values. The uncertainty estimates in Sec. 2.2 inject sources with the same size as the real clump, so they characterize scatter at that assumed size rather than systematic bias for truly sub-PSF sources. Sub-PSF Gaussian fitting is degenerate with the disk model and with PSF-model errors; if recovered sizes at r_e < 0.4 kpc are biased low or high, the slope of Eq. 1 and the low-mass cutoff of the cSMF shift together, and the hierarchical conclusion in Sec. 4.1 loses its quantitative anchor.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a two-stage spatial deconstruction of 32 massive (log M* > 10.5) main-sequence galaxies at z_spec ~ 1.5 using JWST/NIRCam (F115W, F150W, F277W, F444W) and ALMA 870 μm data. The authors fit a bulge+disk model in F444W, detect clumps in F150W contrast images, and then deblend all bands with a fixed-shape bulge+disk+clump model. They report clump stellar masses of 10^8.0–10^9.5 M_sun and half-light radii of 0.1–1 kpc, a mass–size relation r_e ∝ M_*^0.52±0.07 (Eq. 1), and a clump stellar mass function slope α = −1.85±0.19 (Fig. 15). They interpret these as evidence that high-redshift clumps are part of a hierarchical star-forming structure spanning from individual star clusters to kpc-scale complexes, and they find that clumps preferentially lie on residual spiral features. They also report that clumps contribute 1–30% of the host SFR and have elevated sSFR relative to the main sequence.","tokens_in":31184,"tokens_out":7986,"duration_ms":65546,"significance":"If the sub-PSF size measurements are reliable, this is an important step: it is among the first un-lensed studies to resolve clumps down to ~0.1 kpc at z ~ 1.5, bridging the gap between lensed and unlensed samples. The method of simultaneous bulge/disk/clump modeling in native-resolution filters is a genuine advance over aperture-based approaches, and the careful use of BIC for model selection and injection-based completeness is commendable. The mass–size relation and cSMF slope provide quantitative anchors for simulations of disk fragmentation and clump survival. However, the central quantitative claims rest on sizes measured well below the F150W PSF FWHM, and the current validation does not directly test size/flux recovery accuracy; this is the main load-bearing concern.","major_comments":[{"comment":"The completeness test in Sec. 2.4 (Fig. 7) determines detection completeness only. The text explicitly states that the artificial clumps are not re-fit after detection, so no information is provided on whether the fitted size and flux of recovered sub-PSF sources are accurate. Yet the paper claims in Sec. 2.4 that 'we can characterize sub-PSF scales' and that results are robust to inclusion/exclusion of sub-PSF clumps. Since Eq. (1) and the cSMF analysis (Sec. 3.4) use sizes in the range 0.2–1 kpc with F150W FWHM ≈ 0.4 kpc, a size–flux degeneracy or PSF mismatch could bias the slope 0.52±0.07 and the completeness-based mass cut of 10^8.7 M_sun. Please add an injection-recovery test that re-fits recovered clumps and reports bias and scatter in size and flux as a function of injected size, flux, and local background, or restrict the quantitative analysis to r_e ≳ 0.4 kpc and demonstrate that the conclusions are unchanged.","section":"Sec. 2.4, Fig. 7"},{"comment":"The uncertainty procedure in Sec. 2.2 and Appendix A injects sources 'of the same size but varied flux' and then 'fixing the flux and varying the sizes.' This measures random uncertainty at the fitted parameter values but cannot detect a systematic bias in sub-PSF Gaussian sizes, which are degenerate with the local disk/spiral residual background and with errors in the PSF model. Because the validity of sizes below the PSF FWHM is a load-bearing premise for Eq. (1) and for the cSMF completeness cut, a direct validation of size recovery accuracy is required.","section":"Sec. 2.2, Appendix A"},{"comment":"The mass–size relation is fitted after discarding clumps with r_e < 0.2 kpc, but the range 0.2–0.4 kpc is still below the F150W PSF FWHM. The detection threshold in Fig. 7 is a strong function of size and flux, so the correlation between M* and r_e among detected clumps may be shaped by incompleteness even within the adopted cuts. The paper should test the stability of the fitted slope when using a size cut at the PSF FWHM (≈0.4 kpc), and/or apply an explicit completeness correction to the mass–size fit.","section":"Sec. 3.3, Eq. (1)"},{"comment":"The clump SEDs are fit with a constant star-formation history using only four JWST bands. In this setup, the SFR and sSFR are largely determined by the assumed SFH and the fitted age, and the reported Δlog sSFR offsets of 0–0.4 (Sec. 3.2, Fig. 11) may be partly an artifact of the SFH prior rather than an empirical measurement. The authors should test the sensitivity of the sSFR offsets to alternative SFH parameterizations (e.g., delayed-τ or two-component SFH), or explicitly state that the sSFR offsets are model-dependent.","section":"Appendix C, Sec. 3.2"}],"minor_comments":[{"comment":"The cSMF fitting method is not fully specified; the text gives only 'log(dN/dM★)=α log(M★)−const'. Please state whether the fit is binned or a maximum-likelihood fit, the binning scheme, and how the uncertainty on α is obtained.","section":"Sec. 3.4"},{"comment":"The PSFs are described as created with PSFEx on the mosaic; please state the number of stars used and the estimated PSF model uncertainty, since PSF errors propagate directly into sub-PSF clump sizes.","section":"Sec. 2.1"},{"comment":"The phrase 'we use the major-axis value in order for projection effects to not influence our results' is unclear; if the clumps are assumed intrinsically round, the major axis is the deprojected size, but the text should spell this out.","section":"Sec. 2.2"},{"comment":"The top x-axis relates mass to size via the average mass-to-light ratio; the caption should note that this conversion carries the 0.5 dex uncertainty discussed in Sec. 2.4.","section":"Fig. 7"},{"comment":"The reduced-χ² cut of 4 is motivated by the behavior of the SED uncertainties; please show the distribution of reduced-χ² values or the uncertainty behavior in a figure, since this cut removes 65 of 167 detected clumps.","section":"Sec. 3.2"},{"comment":"There is a typo: 'contributution' should be 'contribution'.","section":"Sec. 4.2"}],"recommendation":"major_revision","confidential_remarks":"The paper presents a valuable observational dataset and a novel modeling approach. The main concern is that the headline quantitative results (Eq. 1 and the cSMF slope) depend on sub-PSF size measurements whose systematic accuracy is not yet demonstrated. This is fixable with an injection-recovery test or by restricting the analysis to resolved sizes. I see no issues with citation practice or novelty. The length and structure are appropriate for MNRAS."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Genuinely new measurement set. JWST native-resolution F150W images allowed bulge+disk+clump decomposition in 32 massive z~1.5 main-sequence galaxies, yielding clump sizes down to ~0.1 kpc and masses 10^8–10^9.5 M_sun. The two headline results—mass-size slope 0.52±0.07 and mass-function slope −1.85±0.19—bridge the gap between lensed and unlensed clump studies, and the hierarchical interpretation is reasonable given the data.\n\nThe analysis is careful: BIC-based model selection, per-clump uncertainty estimates, a detection-completeness injection test, and robustness to including high-χ² clumps (slope shifts to 0.49). The data are public. That is real work, and the paper is a solid observational contribution.\n\nThe stress test is right. The Sec. 2.4 completeness test only records whether an artificial clump is detected; it does not re-fit the recovered source and compare output size and flux to the injected values. The per-clump uncertainties are derived by injecting sources with the same size as the real clump, which measures scatter at that size, not systematic bias for sources at or below half the PSF FWHM. The mass-size fit uses clumps down to 0.2 kpc (F150W FWHM ~0.4 kpc). The paper claims the results hold regardless of sub-PSF clumps, but it does not show the mass-size fit with a strict r_e > 0.4 kpc cut, nor does it quantify how a size bias would propagate into Eq. 1 and the cSMF completeness limit. That is the soft spot, and it is addressable.\n\nThe SFR-mass correlation and Δlog sSFR offsets are partially imposed by the constant-SFH CIGALE setup in Appendix C; the clump masses used in Eq. 1 come from the same SED fits, so there is a second-order systematic from the assumed SFH. The mass-function fitting procedure is described too briefly; a referee will want the fitting details and the treatment of the completeness boundary.\n\nMy take: send to referees. The novelty is real, the weaknesses are specific and fixable, and the paper will be a useful reference whether or not the slopes survive the extra tests. Ask the authors for (1) a re-fitting injection test that recovers sizes and fluxes, and (2) a mass-size fit restricted to r_e > 0.4 kpc, plus a cSMF with a mass limit that does not rely on Eq. 1. That would settle the load-bearing question.","headline":"A genuinely new unlensed sub-kpc clump census at z~1.5 whose headline mass-size and mass-function slopes need one more validation pass—size recovery and a stricter size cut—before fully convincing.","tokens_in":31893,"tokens_out":5396,"would_cite":true,"duration_ms":50590,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper establishes that star-forming clumps in massive main-sequence galaxies at z~1.5 form a single hierarchical population spanning 0.1–1 kpc, with mass–size and mass-function scalings that connect unlensed and lensed observations.","keywords":["star-forming clumps","galaxy deconstruction","mass-size relation","clump stellar mass function","cosmic noon","JWST/NIRCam","disk instabilities","hierarchical star formation"],"falsifier":"An injection–recovery test that re-fits artificial clumps through the full pipeline and checks whether the recovered half-light radii and fluxes match the injected values at 0.1–0.4 kpc, or an independent high-resolution observation of the same galaxies that resolves the clumps and directly measures their sizes.","tokens_in":30690,"feed_emoji":"🔭","tokens_out":11597,"duration_ms":92687,"temperature":0.7,"pith_summary":"The paper tries to establish that the bright clumps seen in galaxies at cosmic noon are not a separate class of objects but a single hierarchy of star-forming structures, spanning sizes $0.1$–$1$ kpc and stellar masses $10^8$–$10^{9.5}\\,M_\\odot$. It reaches this by decomposing 32 massive main-sequence galaxies at $z\\approx1.5$ into bulge, disk, and clumps using JWST/NIRCam images with roughly $0.3$ kpc resolution. The measured mass–size relation ($r_e \\propto M_\\star^{0.52\\pm0.07}$) and clump stellar mass function (slope $\\alpha=-1.85\\pm0.19$) both match expectations for turbulence- and gravity-regulated fragmentation. If correct, the results close the observational gap between lensed studies, which see small clumps, and unlensed studies, which previously resolved only kpc-scale clumps, and imply that still smaller sub-clumps await higher resolution.","feed_headline":"Cosmic noon clumps are one hierarchy from 0.1 to 1 kpc","feed_subtitle":"JWST-based galaxy deconstruction links kpc-scale clumps to lensed and local star-forming regions.","key_machinery":"The central mechanism is the two-stage forward-model 'deconstruction' of each galaxy: a bulge+disk model is fit to the F444W image, clumps are detected on contrast images from F150W and added as elliptical Gaussians until the Bayesian information criterion stops improving, and the fixed composite model is then used to deblend photometry in all four NIRCam bands and the ALMA 870 $\\mu$m image. This yields the fluxes and sizes that feed the SED fits, which in turn produce the stellar masses and star formation rates. The argument for hierarchy is carried by Eq. (1), the mass–size relation $r_e \\propto M_\\star^{0.52\\pm0.07}$, combined with the completeness limit of about $10^{8.7}\\,M_\\odot$ below which the mass function is not fitted.","core_discovery":"The paper claims that star-forming clumps in massive main-sequence galaxies at $z\\sim1.5$ are coherent, multiscale structures: a single population with half-light radii $r_e \\sim 0.1$–$1$ kpc and stellar masses $\\sim10^{8.0}$–$10^{9.5}\\,M_\\odot$. It derives an empirical mass–size relation $\\log(r_e/\\mathrm{kpc}) = 0.52(\\pm0.07)\\log(M_\\star/M_\\odot) - 4.98$ and a clump stellar mass function with slope $\\alpha=-1.85\\pm0.19$, both consistent with the hierarchical, instability-driven picture of star-forming regions in the local universe. The paper interprets the agreement as evidence that clumps at all scales are part of a cascade bounded by the Toomre length (the largest scale on which a disk can fragment, $\\sim1$–$5$ kpc) and the Jeans length (the smallest self-gravitating scale, $\\sim10$–$500$ pc), and that the apparent divide between lensed and unlensed clump samples is a resolution effect rather than a physical difference. It also reports that over 70% of clumps lie along spiral features seen in the residual near-IR images.","pith_inferences":["If the hierarchy is real, lower-resolution HST-era samples likely blended several sub-clumps into one, overestimating clump masses and sizes; degrading these JWST images to HST resolution and rerunning the pipeline should reproduce the older distributions.","The similar mass–size scaling to giant molecular clouds suggests each massive clump may have a gas-cloud progenitor of comparable mass; ALMA CO or [CI] maps at matched resolution could test this link directly.","The inferred hierarchy implies the clump stellar mass function should continue with the same slope below about 10^8 solar masses; deep lensed-field JWST observations can test that continuity.","The preferential location on spiral arms raises the possibility that spiral-arm compression triggers or concentrates clump formation; resolved H-alpha kinematics could check whether clumps sit at predicted compression sites."],"forward_implications":["Clump masses and sizes measured here overlap and extend the lensed clump population, so lensed and unlensed studies can be combined into one mass–size plane.","The mass–size slope above 0.5 means clump density decreases with size, consistent with gravity/turbulence-regulated structures rather than simple scaled-up star clusters.","The mass function slope near -2 matches the prediction for disk fragmentation, supporting an in-situ, instability-driven origin for the clump population.","With over 70% of clumps on spiral features, clump formation and spiral structure appear linked in these massive disks.","Clumps contribute only 1–20% of stellar mass and 1–30% of SFR per galaxy, bounding their direct role in bulge growth."],"supporting_citations":[{"why":"Supplies the contrast-image clump detection method and the near-IR-selected comparison sample whose mass distribution is contrasted with this sample.","marker":"Kalita et al. 2024a"},{"why":"Provides the HST-based clump stellar mass distribution in similar-mass galaxies that this work extends to sub-kpc scales.","marker":"Guo et al. 2018"},{"why":"Supplies the lensed clump stellar mass function slope (about -1.7) used to compare with the measured -1.85 slope.","marker":"Dessauges-Zavadsky & Adamo 2018"},{"why":"Gives the theoretical prediction of about -2 for the slope of clumps formed by turbulent disk fragmentation.","marker":"Elmegreen et al. 2006"},{"why":"Provides the lensed clump size–mass compilation and scaling relation that this sample overlaps and extends.","marker":"Claeyssens et al. 2023"},{"why":"Gives the local star cluster size–mass relation used to argue that clumps are not a simple extrapolation of star clusters.","marker":"Brown & Gnedin 2021"},{"why":"Provides the 3D instability framework with the Toomre and Jeans scales used to bound the hierarchical range.","marker":"Meidt 2022"},{"why":"Defines the star-forming main sequence used for sample selection and for the clump sSFR offsets.","marker":"Schreiber et al. 2015"},{"why":"Supplies the UV-plane fitting method used to deblend ALMA 870 micron flux into bulge and disk components.","marker":"Tan et al. 2024b"}],"fun_headline_variants":["Cosmic noon clumps: a single hierarchical family","Clumps from 0.1 to 1 kpc: one population at z≈1.5","JWST+ALMA reveal multiscale clumps in star-forming galaxies","Mass–size relation shows clumps are hierarchical everywhere","Sub-kpc clumps bridge lensed and unlensed galaxy studies"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that clump sizes measured down to about 0.1 kpc in F150W, where the point-spread function is about 0.4 kpc, are true sizes and not artifacts of the Gaussian model or of blending with disk and spiral residuals.","fun_headline_variants_meta":{"raw":{"variants":["Cosmic noon clumps: a single hierarchical family","Clumps from 0.1 to 1 kpc: one population at z≈1.5","JWST+ALMA reveal multiscale clumps in star-forming galaxies","Mass–size relation shows clumps are hierarchical everywhere","Sub-kpc clumps bridge lensed and unlensed galaxy studies"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00051,"raw_usage":{"total_tokens":2647,"prompt_tokens":1278,"completion_tokens":1369,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":894,"completion_tokens_details":{"reasoning_tokens":1284}},"tokens_in":894,"tokens_out":1369,"duration_ms":9456,"temperature":1.0,"reasoning_tokens":1284,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T21:53:05.885059+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"An injection–recovery test that re-fits artificial clumps through the full pipeline and checks whether the recovered half-light radii and fluxes match the injected values at 0.1–0.4 kpc, or an independent high-resolution observation of the same galaxies that resolves the clumps and directly measures their sizes.","supporting_citations":[],"review_version":1}