{"id":"32e2982d-7115-434c-b2f1-9253098e8606","arxiv_id":"2505.21622","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"A uniform JWST catalog of Sérsic and bulge+disk morphology fits for over 340,000 galaxies across four major fields, with initial results linking morphology to star formation and redshift.","lead":"This paper presents a catalog of galaxy shapes for over 340,000 galaxies seen by JWST, with two different model fits for each galaxy. It is a large, uniform public data release that could support many future studies of how galaxies grow and stop forming stars.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The shared-shape constraint across bands (Sect. 3.2), with different NIRCam filter sets per field, can make the published R_e and n_S wavelength-averaged in a field-dependent way, and this is not tested.","rationale":"The catalog's central value is uniformity across four major fields, enabling community statistical studies. For that to hold, the morphological parameters must be on the same physical scale everywhere. Section 3.2's shared-shape fit is a deliberate simplification; if morphology is genuinely wavelength-dependent, the fit returns a S/N-weighted compromise. The problem is that the weights and the band sets differ across fields, so the compromise is not the same in EGS as in GOODS/PRIMER. This can create offsets comparable to known size-wavelength trends (~0.1-0.2 dex), which is large relative to the quoted size-evolution slopes (power-law indices of -0.6 to -0.9, roughly 0.1-0.2 dex per unit redshift). Thus even a moderate wavelength-dependent bias could mimic or suppress real evolution. The paper acknowledges the averaging in Sect. 4.2 but does not test it. The van der Wel comparison checks only early-type galaxies and uses a redshift- and mass-independent scaling relation that the original authors themselves refined; it is not a test of the cross-band constraint. The GitHub code is a strength, as it allows the proposed ablation test to be run directly. My recommendation is to keep the CONDITIONAL verdict: the catalog can be accepted if the authors add a band-set ablation test and a wavelength-gradient diagnostic, or explicitly caveat catalog entries with the effective average wavelength. This is compatible with the reader's verdict, so I do not change it.","tokens_in":18394,"tokens_out":9242,"duration_ms":94772,"concrete_test":"Run an ablation on the EGS field: rerun the published SourceXtractor++ configuration using only the six bands common to all four fields (F115W, F150W, F200W, F277W, F356W, F444W) instead of the full EGS set, keeping identical priors and PSFs. Cross-match the resulting catalog to the released catalog and compare matched sources with flag=2; if the median absolute change exceeds 0.05 dex in log R_e or 0.1 in n_S, the field-dependent band sets create systematic offsets that break the claimed cross-field uniformity.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 3.2 constrains R_e, n_S, (a/b), and θ to be identical across all NIRCam bands for each source, so each galaxy ends up with a single morphology that is effectively averaged over the fitted bands. The four fields use different filter sets (Table 1): EGS includes F182M, F210M, and F410M, while GOODS and PRIMER do not. Because the average is weighted by band S/N, which depends on the source SED and filter set, the rest-frame wavelength being measured varies across fields and with redshift. Galaxies have color gradients (bulge-disk differences, dust), and R_e and n_S are known to change by roughly 0.1-0.2 dex across the rest-UV to rest-optical range. The external validation (Sect. 3.5) is limited to 3,263 early-type galaxies matched to HST F160W and rescaled with a fixed slope of -0.25 (Eq. 1); it does not test the wavelength-independence assumption for star-forming galaxies or cross-field uniformity. If a galaxy with a red bulge and blue disk is forced to a single size across F115W-F444W, the recovered parameters depend on which bands are available and on the SED, producing systematic offsets that differ between EGS and the other fields. The paper flags the averaging caveat in Sect. 4.2 but does not quantify it, leaving the central 'uniform catalog' claim insufficiently supported.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript presents a JWST/NIRCam morphological catalog for over 340,000 galaxies spanning 0<z<12 in the CEERS/EGS, GOODS, and PRIMER fields, built from DJA v7 mosaics. SourceXtractor++ is used to fit both a single-Sérsic model and a bulge+disk (B+D) model, with morphological parameters constrained to be identical across all NIRCam bands. Completeness is quantified relative to the DJA catalogs (80.4% of S/N>3, magnitude-limited sources have both fits), and the Sérsic effective radii are validated against van der Wel et al. (2012) for 3,263 early-type galaxies. The science demonstrations link Sérsic index and B/T to UVJ colors, present size-redshift trends for bulge/disk and quiescent/star-forming galaxies, and show that quiescent galaxies have higher stellar mass surface densities at high redshift.","tokens_in":18798,"tokens_out":8934,"duration_ms":102663,"significance":"If the catalog is as uniform as claimed, it is a valuable community resource: it is one of the largest JWST morphology catalogs to date, with public code and catalog release, dual model fits, quantified completeness, and an external HST-based validation. The paper also makes testable claims about the morphology-star formation bimodality out to z>3 and about the high compactness of early quiescent galaxies. However, the headline uniformity claim is not yet fully supported: the band-averaging construction and the limited external validation leave open the possibility of field-dependent and wavelength-dependent systematics in the published sizes and Sérsic indices.","major_comments":[{"comment":"The shared-shape constraint (R_eff, n_S, (a/b), and theta identical across all NIRCam bands) turns the published morphology into a weighted average over 0.8-5 micron whose effective rest-frame wavelength depends on the field's filter set and on each source's SED. EGS includes F182M, F210M, and F410M while GOODS and PRIMER do not, so a given galaxy type is not measured at the same rest-frame wavelength in all fields; this can create field-dependent offsets in R_eff and n_S that propagate into every comparison in Sect. 4. The external check in Sect. 3.5 is restricted to 3,263 early-type galaxies and rescales with a fixed slope (Eq. 1), so it does not test the wavelength-independence assumption for star-forming galaxies, for B+D parameters, or across fields. The paper should quantify this, for example by fitting short- and long-wavelength bands as separate groups and showing parameter stability, or by reporting a per-source effective wavelength and demonstrating that the science results are unchanged when the filter-dependent averaging is accounted for.","section":"Section 3.2, Table 1"},{"comment":"The Bulge+Disk results are used as a primary morphology indicator (B/T) for the bulge/disk classification and for the UVJ bimodality claims, but they are not validated against any external two-component measurements. The B/T prior is described only as a 'bell curve' with a wavelength-dependent mean and spread, which is not enough to judge whether the recovered B/T values are prior-dominated for faint or low-surface-brightness galaxies. The internal consistency between n_S and B/T (Figs. 4-6) is expected from the same imaging and is not an independent check. Please add an external or simulated validation of B/T, or at minimum a prior-sensitivity analysis demonstrating that the classifications and the z>3 star-forming bulge population do not depend on the adopted prior.","section":"Section 3.2 and Sect. 4.1"},{"comment":"The redshift-independent UVJ cut is applied out to z=6 although the authors note that the criteria change with redshift. Because this selection defines the quiescent and star-forming samples used for the size evolution (Sect. 4.2) and Sigma_e trends (Sect. 4.3), the paper should show how many sources move across the boundary when a redshift-dependent UVJ criterion is used, particularly at z>3 where the quiescent sample is small. If the conclusions are robust to this choice, state that explicitly; if not, soften the wording of the claims.","section":"Section 4.1, Eq. (2)"}],"minor_comments":[{"comment":"The comparison with van der Wel et al. (2012) is reported only qualitatively; please provide median offset, scatter, and outlier fraction for R_eff and n_S, and state whether the comparison changes if the more complex redshift- and mass-dependent scaling from van der Wel et al. is used.","section":"Section 3.5, Fig. 3"},{"comment":"The 0.3 arcsec cross-match radius and the flag thresholds are set manually; a brief test showing that nearby alternative choices do not change the completeness or parameter distributions would strengthen the catalog description.","section":"Sections 3.3.1 and 3.4"},{"comment":"The phrase 'There evolution is modeled' should read 'Their evolution is modeled'.","section":"Fig. 7 caption"},{"comment":"The priors should be described numerically in the text or with explicit axis labels and values, since the current description does not allow the reader to reproduce the adopted priors.","section":"Appendix B, Fig. B.1"}],"recommendation":"major_revision","confidential_remarks":"The catalog release and code are strengths, and the external HST comparison for Sérsic radii is a useful check. The main blocking issue is the unquantified field-dependent wavelength averaging from the shared-shape constraint; this is fixable within the manuscript's scope by adding a split-band or per-source-wavelength test. The UVJ and B/T validation concerns are also addressable in revision and do not by themselves warrant rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: the genuinely new thing here is scale. A uniform Sersic and bulge+disk catalog for ~340k galaxies across CEERS, GOODS-S/N, and PRIMER fields, with catalog and code released, is a real community resource. The validation against van der Wel et al. (2012), with R_e rescaled to 2.5 micron, shows reasonable agreement for early-type galaxies, and the paper is honest about the wavelength caveat in Sect. 4.2. The science demonstrations are mostly confirmatory, but that is fine for a catalog paper; the catalog is the deliverable.\n\nThe soft spot that matters is the cross-band constraint in Sect. 3.2. Fitting one R_e, n_S, axis ratio, and PA across all NIRCam bands makes each galaxy's morphology a S/N-weighted average over the fitted bands. EGS has a different filter set (F182M, F210M, F410M; the others do not), and the weights depend on source SED and redshift, so the rest-frame wavelength baseline is field-dependent. The paper flags this caveat but does not quantify it. Given that R_e and n_S change by roughly 0.1-0.2 dex from rest-UV to rest-optical, this can introduce field-dependent systematics at the level of the science results in Sect. 4, and it does weaken the 'uniform catalog' claim. The validation against van der Wel only covers 3,263 early types in F160W; it does not test star-forming galaxies or cross-field uniformity. This is fixable—show per-band or per-field consistency, or provide the effective wavelength of each fit—but it needs attention before the catalog is used blindly.\n\nOther issues are minor. The UVJ selection is fixed at the 0.7<z<1.3 criteria for all redshifts, which the authors admit is qualitative. The B/T prior is wavelength-dependent, which is a choice. The completeness analysis is adequate, and the claim that flag=0/1 sources are not biased by magnitude, redshift, or mass is plausible but only marginally checked.\n\nNet: the catalog is worth having and the paper deserves refereeing as a resource paper. I would recommend acceptance after the authors either quantify the cross-field wavelength effect or relax the uniform-shape assumption, and ideally add a cross-field comparison where fields overlap. Who is it for: anyone doing JWST morphology evolution, especially studies of quiescent galaxies at z>3. I would cite it once the catalog is properly validated.","headline":"A large and useful JWST morphology catalog, but the field-dependent wavelength baseline from the shared-shape constraint needs quantifying before I would call it uniform.","tokens_in":19283,"tokens_out":2758,"would_cite":true,"duration_ms":30680,"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":"The paper builds a uniform JWST morphology catalog for over 340,000 galaxies and shows that bulge-dominated galaxies are predominantly quiescent and increasingly compact at high redshift.","keywords":["galaxies: structure","galaxies: evolution","catalogs","techniques: image processing","Sérsic profile","bulge-disk decomposition","JWST NIRCam","galaxy quenching"],"falsifier":"Inject synthetic galaxies with known Sérsic parameters and realistic fluxes into the DJA mosaics, run the same SourceXtractor++ pipeline, and compare recovered $R_e$ and $n_S$ with the inputs; a systematic offset larger than the quoted uncertainties at S/N>10 would disprove the catalog's stated accuracy.","tokens_in":18252,"feed_emoji":"🔭","tokens_out":10519,"duration_ms":103936,"temperature":0.7,"pith_summary":"This paper builds the kind of dataset that was missing: a single, uniformly processed morphological catalog for more than 340,000 galaxies detected by JWST across four major extragalactic fields, spanning $0<z<12$. Every source is fit twice, once with a Sérsic profile and once with a bulge-plus-disk model, so a user gets effective radius, Sérsic index, axis ratio, and bulge-to-total ratio from the same pipeline. The authors then show that these measurements reproduce and extend the known link between structure and star formation: bulge-dominated galaxies preferentially sit in the quiescent region of the $UVJ$ diagram, while disk-dominated galaxies are mostly star-forming, out to $z>3$. Quiescent galaxies are more compact, with stellar mass surface densities nearly an order of magnitude higher at $z\\sim4$ than at $z\\sim1$. If the catalog is as accurate as claimed, it turns the JWST archive from a set of images into a statistical resource for studying how galaxy structure evolves and how quenching connects to bulge growth.","feed_headline":"340,000 JWST galaxies get one uniform morphology catalog","feed_subtitle":"Bulge and disk fits across four deep fields tie galaxy shape to star formation and quenching from z<1 to z~5.","key_machinery":"The carrying mechanism is dual-model surface brightness fitting inside SourceXtractor++: every galaxy is modeled both as a single Sérsic profile and as a bulge-plus-disk composite, with all shape parameters forced to be identical across NIRCam bands so that the fitted values are a weighted average over wavelength. Accurate fits depend on empirical point-spread functions built with PSFEx from point sources selected along the 'starline' in the magnitude-versus-peak-surface-brightness plane, and the resulting catalogs are cross-matched to the archive's SExtractor and eazy catalogs to attach photometric redshifts, stellar masses, and rest-frame colors. The consistency between the independently fitted Sérsic index and bulge-to-total ratio is what lets the paper classify galaxies as bulge- or disk-dominated and connect morphology to the $UVJ$ diagram.","core_discovery":"The central claim is that a uniform, flux-limited sample of over 340,000 JWST-detected galaxies in CEERS, PRIMER-UDS, PRIMER-COSMOS, and GOODS can be reliably characterized by two-dimensional profile fitting, yielding consistent structural parameters across two independent models. For each source, SourceXtractor++ fits both a single Sérsic profile and a bulge-plus-disk decomposition in which the bulge follows a de Vaucouleurs profile and the disk is exponential; the effective radii, Sérsic index, axis ratio, and position angle are constrained to be identical across all NIRCam bands, so the reported values represent a wavelength-averaged morphology over roughly 0.8--5\\,µm. Using these measurements together with the archive's photometric redshifts and physical parameters, the paper reports that bulge-dominated galaxies (high $n_S$ and $B/T$) occupy the quiescent $UVJ$ region, that a minority population of quiescent disks and star-forming bulge-dominated galaxies persists beyond $z\\sim3$, and that quiescent galaxies are markedly more compact, with $\\Sigma_e$ roughly an order of magnitude higher at $z\\sim4$ than at $z\\sim1$. The paper interprets these trends as evidence that bulge growth and quenching are closely connected, with compact star-forming phases preceding quiescence.","pith_inferences":["Editorial inference: forcing morphology to be identical across all NIRCam bands suppresses genuine color gradients; a band-by-band release of the same fits would show how much of the reported size evolution is a wavelength-averaging artifact.","Editorial inference: the pipeline could be run on mock galaxies with known input profiles to map the bias and scatter of $R_e$ and $n_S$ as a function of signal-to-noise, providing corrections the current catalog does not include.","Editorial inference: applying the same fitting to separate bands, or to rest-frame-optical-selected samples, could test whether the claimed quiescent disk population at $z>3$ is genuinely gas-poor or merely dust-reddened."],"forward_implications":["Users of the archive can now select galaxies by $R_e$, $n_S$, $B/T$, or axis ratio across four major fields without reprocessing the images, enabling statistical studies of rare populations such as quiescent galaxies at $z>3$.","The observed bimodality, with quiescent disks and compact star-forming bulge-dominated galaxies coexisting beyond $z\\sim3$, implies that morphology alone does not determine star formation state at early times.","The order-of-magnitude increase in stellar mass surface density of quiescent galaxies from $z\\sim1$ to $z\\sim4$ constrains models in which high-redshift quenching happens in very dense stellar cores.","The size evolution fits $r_e = R(1+z)^a$ for disk-, bulge-, and all-dominated samples provide direct quantitative targets for simulations of galaxy growth.","Because the catalog supplies both total and component sizes, it allows separate evolution of bulges and disks to be tracked, showing bulge sizes are established early while disks keep growing."],"supporting_citations":[{"why":"These papers present SourceXtractor++, the fitting engine that carries out the catalog-wide multiband profile fits.","marker":"Bertin et al. 2020; Kümmel et al. 2020"},{"why":"PSFEx builds the empirical point-spread-function models from selected point sources, and the fits are convolved with these models.","marker":"Bertin 2011"},{"why":"This HST-based catalog supplies the comparison sample and the wavelength scaling relation used to validate the measured effective radii and Sérsic indices.","marker":"van der Wel et al. 2012"},{"why":"This paper describes the DAWN JWST Archive mosaics and catalogs that provide the imaging, photometric redshifts, and physical parameters.","marker":"Valentino et al. 2023"},{"why":"The eazy code produces the photometric redshifts and rest-frame colors used to classify galaxies and convert angular sizes to kiloparsecs.","marker":"Brammer et al. 2008"},{"why":"SExtractor produces the archive source catalogs that the morphology measurements are cross-matched against.","marker":"Bertin & Arnouts 1996"},{"why":"This paper defines the Sérsic profile used as the single-component model and as the bulge component of the two-component fits.","marker":"Sérsic 1963"},{"why":"The starline method for selecting point sources in the magnitude versus peak-surface-brightness plane is adapted to build the PSF sample.","marker":"Leauthaud et al. 2007"}],"fun_headline_variants":["JWST traces galaxy shape to star formation over 12 billion years","340k galaxies' morphology from JWST ties bulge to quenching","Bulge and disk fits for 340k galaxies map cosmic quenching","Galaxy structure from JWST: 340k sources reveal form-quench link","Uniform JWST morphology catalog: 340k galaxies, one method"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the empirical PSF models built from starline-selected point sources are accurate and that each galaxy's morphology is the same in every NIRCam band; if either fails, the fitted effective radii and Sérsic indices shift systematically and every Section 4 trend inherits that shift.","fun_headline_variants_meta":{"raw":{"variants":["JWST traces galaxy shape to star formation over 12 billion years","340k galaxies' morphology from JWST ties bulge to quenching","Bulge and disk fits for 340k galaxies map cosmic quenching","Galaxy structure from JWST: 340k sources reveal form-quench link","Uniform JWST morphology catalog: 340k galaxies, one method"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000329,"raw_usage":{"total_tokens":1951,"prompt_tokens":1177,"completion_tokens":774,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":793,"completion_tokens_details":{"reasoning_tokens":681}},"tokens_in":793,"tokens_out":774,"duration_ms":9413,"temperature":1.0,"reasoning_tokens":681,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T13:25:47.698897+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Inject synthetic galaxies with known Sérsic parameters and realistic fluxes into the DJA mosaics, run the same SourceXtractor++ pipeline, and compare recovered $R_e$ and $n_S$ with the inputs; a systematic offset larger than the quoted uncertainties at S/N>10 would disprove the catalog's stated accuracy.","supporting_citations":[{"cited_title":"V., & Coppi, P","cited_arxiv_id":null,"evidence_quote":"The eazy code produces the photometric redshifts and rest-frame colors used to classify galaxies and convert angular sizes to kiloparsecs."},{"cited_title":"& Arnouts , S","cited_arxiv_id":null,"evidence_quote":"SExtractor produces the archive source catalogs that the morphology measurements are cross-matched against."},{"cited_title":"2007, The Astrophysical Journal Supplement Series, 172, 219","cited_arxiv_id":null,"evidence_quote":"The starline method for selecting point sources in the magnitude versus peak-surface-brightness plane is adapted to build the PSF sample."}],"review_version":1}