{"id":"50a87186-419d-44b7-8b19-f9844d0e53b8","arxiv_id":"2507.06406","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Across quenching stages in 643 CALIFA galaxies, molecular gas decreases, but the decisive step beyond the green valley is a drop in star formation efficiency, especially in the inner regions.","lead":"This paper maps how gas, stars, and star formation relate in 643 nearby galaxies at six different 'quenching stages', from actively star-forming to fully retired. It finds that once galaxies leave the green valley, their central regions stop forming stars efficiently even when molecular gas remains.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Retired-stage SFE drop may be an artifact of CO detection bias: fR has only 27% detections, and flaring plots use detections only, so low central SFE could reflect selection of gas-rich, low-SFR objects.","rationale":"The paper is a transparent, well-structured observational study; it publishes detection rates, uses linmix for censored regression, and explicitly flags the non-detection problem in Section 4.1. The most load-bearing step in the argument is the translation of the measured SFE medians into the physical conclusion that molecular gas remains but star formation efficiency drops. For the fR stage, 73% of the galaxies are CO non-detections, so the median SFE is determined by the 27% detected subset. The detection criterion selects for high CO luminosity; because the QueStNA classification already fixes the Hα-based SFR to be low in retired galaxies, the detected subset preferentially contains high-Mmol, low-SFR objects, producing artificially low SFE values. The non-detected galaxies could have Mmol just below the detection threshold, in which case their SFE could be closer to that of star-forming galaxies, meaning quenching could proceed through gas depletion alone. This is not an internal inconsistency in the paper; it is a limitation that the authors partly acknowledge. However, the abstract's headline claim goes beyond what the current data can establish, and a feasible stacking analysis of the existing non-detected spectra would settle whether the SFE drop is real. Because the reader already recommended a conditional verdict with additional data needed, this concern does not alter the recommended verdict.","tokens_in":35152,"tokens_out":8071,"duration_ms":98081,"concrete_test":"Stack the APEX CO(2-1) spectra of the CO-non-detected nR and fR galaxies (S/N<3) in the 26.3 arcsec beam aperture, measure the mean CO luminosity and mean Mmol of these subsets, and combine with the mean beam SFR from Hα to compute the mean SFE of the full retired samples. If the stacked SFE lies within 0.5 dex of the SF-group median, the claim that retired centres require a significant SFE drop is not supported by current data. As an alternative or complement, refit the beam SFR versus Mmol relation for nR and fR with a censored Bayesian model that treats all non-detections as left-censored, and compare the resulting median SFE with the detections-only values; a shift upward of more than 0.5 dex would indicate that the reported drop is dominated by detection bias.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that a significant SFE decrease is necessary to quench galactic centres rests chiefly on the beam/global flaring plots (§3.3, Fig. 7) and the SFE medians in Table 2. For the retired stages these numbers are computed from CO detections only: the nR detection rate is 54% and the fR rate is 27% (Table 3.1). Because fR and nR are defined by low WHα, their SFRs are uniformly low; the CO-detected subset is therefore biased toward the highest Mmol at a given SFR, which mechanically lowers SFE = SFR/Mmol. The non-detected 73% of fR galaxies have Mmol upper limits, and if their true Mmol lies near the APEX sensitivity limit their SFE could be much higher, possibly close to the SF value. The paper itself acknowledges in §4.1 that the retired-group relations 'can be driven by upper limits' and that 'we cannot fully establish with our sample that retired KS laws actually exist'; the same caveat applies to the SFE drop that carries the abstract's conclusion. The violin plots in §3.2 that include upper limits do not resolve the issue because the flaring plots and Table 2 medians for the retired groups use detections only, and the linmix fits for the full samples are not used to derive the quoted median SFE values.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper uses the iEDGE database (643 CALIFA galaxies with CO data from APEX, CARMA, and ACA) to classify galaxies into six QueStNA quenching stages on the basis of WHα maps and BPT diagnostics, then compares sSFR, Mmol, SFE, and fmol across stages, both globally and in the central ('beam') regions. The main results are that SFE is roughly constant for the SF, QnR, and cQ stages and declines for MX, nR, and fR; that this decline is more pronounced in the centre than globally; that the Mmol-M* relations become shallower and the SFR-Mmol relations steeper with quenching; and that a 3D SFR-M*-Mmol relation exists only for the SF group. The paper concludes that quenching beyond the green valley cannot be explained by molecular gas depletion alone and requires a significant SFE decrease, particularly in galaxy centres.","tokens_in":35311,"tokens_out":7625,"duration_ms":82760,"significance":"If established, the result that molecular gas remains present but is used with reduced efficiency in the centres of retired galaxies would be an important constraint on quenching mechanisms, favoring efficiency-drop scenarios over pure gas exhaustion and supporting inside-out quenching. The paper has real strengths: a large homogenized multi-telescope sample, explicit automatic classification quality flags, linmix regressions that treat upper limits, 2D KS tests, and stacked CO spectra that do not depend on SFR calibration assumptions. However, the central claim rests on the retired-stage SFE measurements, which are the least secure part of the dataset; the paper itself concedes in §4.1 that the retired-stage scaling relations may be driven by upper limits, and that same caveat extends to the SFE drop that carries the abstract's conclusion.","major_comments":[{"comment":"The central claim that a significant SFE decrease is necessary to quench galaxy centres is carried by the flaring plots, which use only CO-detected galaxies (S/N>3). In the retired stages the detection fraction is low: 54% for nR and 27% for fR (Table 1). Because nR and fR are selected to have uniformly low Hα-based SFR, the detected subsample is biased toward the highest Mmol at a given low SFR; if the true SFE were constant, this selection would mechanically lower the measured median SFE = SFR/Mmol. The paper itself states in §4.1 that the retired-group relations can be driven by upper limits and that it cannot fully establish with this sample that retired KS laws actually exist, and this caveat applies equally to the SFE drop that supports the abstract's conclusion. The authors should add a quantitative upper-limit-aware analysis—for example survival-analysis medians for Table 2, linmix full-sample predictions evaluated at the group medians, or an injection test that assigns non-detections SF-like SFE and re-derives the flaring ratios—and show that the beam SFE decline survives.","section":"§3.3 / Fig. 7 and §4.2 / Fig. 13"},{"comment":"The 'centre' used in the inside-out quenching claim is not the same aperture as the QueStNA central classification. The cQ and QnR stages are defined by WHα patterns within 0.5 Re (K21; Appendix A), and §4.2 discusses quenching within 0.5 Reff. The beam measurements, however, are the APEX 26.3'' FWHM CO aperture, which the authors themselves state corresponds to a median of 1 Re (Section 2). For a typical galaxy, a 1-Re aperture includes a substantial part of the disc, so the beam SFE decrement may be diluted relative to the true central value or may reflect processes at a different radius than the 0.5-Re 'centre' invoked in the abstract. The quantitative inside-out comparison should either use an aperture-matched central definition or explicitly quantify the Re distribution and the fraction of the beam that lies within 0.5 Re for the galaxies in each stage.","section":"§3.3 / §4.2 and Section 2"},{"comment":"The reported steepening of the SFR-Mmol relation with quenching rests on linmix fits for groups in which non-detections dominate: the nR slope is 2.11 with a fitted range of roughly 1.94-2.32 and the fR slope is 1.58 with a fitted range of roughly 1.47-1.71, but the Spearman coefficients for detections-only samples are 0.49 and 0.43, and the fitted relations are explicitly acknowledged in §4.1 to be possibly driven by upper limits. As the paper says, retired KS laws cannot be established. Since the abstract presents the steepening as one of the main findings, the authors should either soften the claim to 'consistent with, but not conclusive of, steepening' or perform a sensitivity test that fits all stages using a common S/N or Mmol completeness threshold, so that the slope comparison is not confounded by the very different detection fractions across stages.","section":"§3.4 / Table 3 and §4.1"}],"minor_comments":[{"comment":"There are numerous typographical and formatting issues, including 'K21]kalinova2021' (several occurrences), 'e fficient' and 'efficiency' inconsistencies, 'redshit' for 'redshift', and 'Quantities (expect rP and rS)' in the Table 3 notes, which should be 'except'.","section":"Throughout"},{"comment":"The text refers to 'Fig. 7 left and middle panels' but the figure as presented has two panels (global and beam); also the Fig. 7 caption labels the star formation efficiency line as 'SFR, cyan' instead of 'SFE, cyan'.","section":"§3.3 and Fig. 7"},{"comment":"The APEX beam size is given as 26.2'' in the Fig. 1 caption and 26.3'' in Section 2; these values should be harmonized.","section":"Fig. 1 and Section 2"},{"comment":"The abstract states that a significant SFE decrease is 'necessary' to retire galaxy centres, while §4.1 explicitly says the retired-stage relations may be driven by upper limits and that retired KS laws cannot be fully established; the abstract should be tempered to match this caveat.","section":"Abstract and §4.1"},{"comment":"The sentence claiming that excluding active galaxies allows the authors to 'rule out AGN-driven gas depletion as the primary mechanism for quenching in our sample' is too strong, since BPT-based AGN selection can miss diluted or previously active nuclei; the subsequent qualifying discussion is appropriate, but the strong wording should be revised.","section":"§4.3"},{"comment":"The stacked CO spectra are a valuable non-model-dependent check, but no error bars or quantitative significance tests are provided for the amplitude differences between stages or between active and non-active subsamples; adding these would strengthen the AGN gas-retention point.","section":"Fig. 4"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of A&A and is likely to be of interest to the galaxy-evolution community. The main issue is not the quality of the dataset or the classification effort, but the gap between the strong abstract-level claim and the paper's own admission that the retired-stage measurements are dominated by non-detections. This gap is fixable with a focused sensitivity analysis and a more carefully hedged conclusion."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThis is worth a serious look. The paper extends the QueStNA quenching-stage framework from 238 to 643 galaxies with homogenized CO data from iEDGE, and derives stage-resolved scaling relations for SFR, M*, and Mmol. Two genuinely new findings: only star-forming galaxies define a tight 3D relation in that space, and the SFE is flat across SF/QnR/cQ then drops sharply for MX/nR/fR, especially in the center. That latter point is the abstract's headline.\n\nThe paper does several things well. The sample is large and homogenized; they use linmix to handle upper limits, 2D KS tests for distribution differences, and stacked CO spectra as a model-independent check. They are also honest about their main limitation, stating in §4.1 that they cannot fully establish that retired Kennicutt-Schmidt laws exist.\n\nThe soft spot is that the central SFE-drop claim rests heavily on the flaring plots and Table 2 medians, which for the retired groups use CO detections only. nR has a 54% detection rate, fR 27%. Selecting gas-rich, low-SFR objects can mechanically lower SFE. This is exactly the concern in the stress-test note, and it is not a straw man. However, it is not fully decisive: the full-sample medians in Table 2, which include upper limits, show a similar drop, and for fR the all-galaxy and detected-only SFE medians are nearly identical. So the bias is probably not as extreme as feared, but the authors' own caveat about upper limits applies to the SFE drop as much as to the scaling relations, and they do not say so explicitly. The claim would be more robust with a survival-analysis treatment of SFE or deeper CO observations.\n\nTwo smaller issues: the 26.3 arcsec APEX beam corresponds to about 1 Re, while the QueStNA central classification uses 0.5 Re, so the inside-out claim is based on a looser definition of 'center.' And the 3D PCA eigenvalue ratios have no attached uncertainties.\n\nOverall, this is a solid, careful paper that deserves referee time. I would recommend conditional acceptance, asking the authors to extend the detection-bias caveat to the SFE claim, add PCA uncertainties, and address the aperture mismatch.\n\nBest,","headline":"Solid stage-resolved study with a strong central claim that is undercut by low CO detection rates in the retired groups.","tokens_in":36009,"tokens_out":6869,"would_cite":true,"duration_ms":73525,"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":"This paper argues that star formation quenching requires a collapse in star formation efficiency in galactic centres, not molecular gas depletion alone, based on 643 CALIFA galaxies with CO data.","keywords":["galaxy evolution","star formation quenching","molecular gas","star formation efficiency","scaling relations","CALIFA survey","CO observations","emission-line classification"],"falsifier":"Resolve the central half effective radius of nearly and fully retired galaxies in sensitive CO observations: if their central molecular gas is detected and forms stars at normal efficiencies, the claimed central efficiency collapse would disappear; if the central SFE deficit persists at sub-kiloparsec resolution, the inside-out quenching claim would be confirmed.","tokens_in":34865,"feed_emoji":"🌌","tokens_out":5735,"duration_ms":60462,"temperature":0.7,"pith_summary":"This paper asks why galaxies stop forming stars, and specifically whether quenching is a problem of running out of molecular gas or of failing to use the gas that remains. Using 643 nearby galaxies from the CALIFA survey combined with CO observations in the iEDGE database, the authors sort galaxies into six quenching stages by the spatial pattern of ionised gas. They find that the molecular gas fraction falls steadily from star-forming to retired galaxies, but the star formation efficiency stays roughly constant through the first stages and then drops sharply for nearly and fully retired galaxies. The efficiency drop is much larger in galactic centres than in the discs, meaning quenching proceeds inside-out. The paper concludes that depleting the molecular gas alone cannot retire a galaxy; beyond the green valley, the centres must also lose the ability to convert their remaining gas into stars.","feed_headline":"Galaxies quit star formation by losing efficiency, not just gas","feed_subtitle":"Nearly and fully retired galaxies still hold molecular gas but form stars far less efficiently, especially in their centers.","key_machinery":"The central machinery is the QueStNA emission-line classification, which assigns each galaxy a quenching stage from the spatial pattern of H-$\\alpha$ equivalent width ($W_{\\mathrm{H}\\alpha} > 6$ \\AA{} marking star-forming regions, $W_{\\mathrm{H}\\alpha} < 3$ \\AA{} marking retired regions, and the range between them marking mixed ionisation), together with the iEDGE homogenised database of CO(1-0) and CO(2-1) fluxes and CALIFA optical maps. Beam quantities measured within the 26.3-arcsec APEX beam, which has a median size of about one effective radius, are compared with global quantities to isolate central behaviour. The quantitative analysis is carried by Bayesian linear regressions that treat CO non-detections as upper limits, flaring plots that compare each stage's median to the star-forming median, two-dimensional Kolmogorov-Smirnov tests, and principal component analysis eigenvalue ratios that diagnose whether a three-dimensional relation is line-like.","core_discovery":"The central claim is that star formation quenching is a two-stage process: a steady loss of molecular gas relative to stellar mass brings galaxies from the star-forming sequence into the green valley, and then a collapse in star formation efficiency, most pronounced in the central regions, carries them into retirement. The evidence includes median star formation efficiencies that are nearly constant for the star-forming, quiescent-nuclear-ring, and centrally quenched stages, then fall by factors of 20 to 30 for the nearly retired and fully retired stages. The molecular gas--stellar mass relation flattens from a slope near 1.08 for star-forming galaxies to about 0.57 for fully retired galaxies, while the star formation rate--molecular gas relation steepens from about 0.79 to values above 2 as the intercept drops, so retired galaxies sit at much lower efficiency for a given gas mass. Principal component analysis shows a line-like three-dimensional relation among star formation rate, stellar mass, and molecular gas mass only for the star-forming group, with the other stages scattered in that space. Taken together, the paper argues that a significant decrease in star formation efficiency is necessary to retire the centres of galaxies beyond the star formation green valley.","pith_inferences":["If the central claim is confirmed, galaxy evolution models should treat quenching as requiring an efficiency-reduction channel rather than gas exhaustion alone; prescriptions that only remove gas will struggle to reproduce the red sequence.","The result implies that CO surveys of green-valley and retired galaxies should prioritise sensitivity to low-surface-brightness molecular gas, since non-detections currently dominate the fully retired stage and the fitted relations depend on how those upper limits are treated.","The inside-out efficiency drop is consistent with dynamical suppression or loss of the feedback support that sustains star formation in dense centres, and a direct test would compare central turbulence, shear, and molecular-cloud properties in retired galaxies at sub-kiloparsec resolution.","Because active galactic nuclei were excluded from the analysis, the central efficiency collapse in this sample cannot be attributed to instantaneous AGN feedback, pointing instead to long-term preventative or dynamical mechanisms operating after gas depletion begins."],"forward_implications":["The molecular gas fraction $f_{\\rm mol}$ declines continuously across quenching stages, so gas removal is real and matters, but it is not the whole story.","Star formation efficiency is flat from the star-forming stage through the centrally quenched and quiescent-nuclear-ring stages, with the onset of the efficiency loss occurring at the green-valley (mixed) stage.","Central star formation efficiency in nearly and fully retired galaxies drops by more than two orders of magnitude relative to the star-forming stage, while the global drop is about one order of magnitude, so the centre leads the quenching.","The scaling relations change systematically: the SFR--$M_{\\rm mol}$ slope steepens from about 0.79 to 2.11, while the $M_{\\rm mol}$--$M_*$ slope flattens from about 1.08 to 0.57, implying retired galaxies are gas-poor relative to their stellar mass and inefficient at using the gas they retain.","Only star-forming galaxies show a line-like three-dimensional relation among SFR, $M_*$, and $M_{\\rm mol}$; the other quenching stages scatter in that parameter space, so stage-specific scaling relations are needed."],"supporting_citations":[{"why":"Defines the QueStNA emission-line quenching stages and the H-alpha equivalent-width thresholds that this paper automates and applies to 643 galaxies.","marker":"K21"},{"why":"Supplies the APEX-EDGE CO data and the green-valley boundary used here, and earlier evidence for central SFE-driven quenching that this paper extends.","marker":"Colombo et al. 2020"},{"why":"Supplies the CARMA-EDGE CO survey whose measurements enter the iEDGE database, predominantly for star-forming galaxies.","marker":"Bolatto et al. 2017"},{"why":"Supplies the ACA-EDGE CO observations and the finding that both SFE and fmol decline in the centres of green-valley galaxies.","marker":"Villanueva et al. 2024"},{"why":"Provides the Bayesian linear-regression method that treats CO non-detections as upper limits in the scaling-relation fits.","marker":"Kelly 2007"},{"why":"Defines the star-forming main-sequence fit used to order the quenching stages and to locate the green-valley boundary.","marker":"Cano-Díaz et al. 2016"},{"why":"Establishes the molecular-gas main sequence and the three-dimensional SFR--M*--Mmol relation that this paper tests across quenching stages.","marker":"Lin et al. 2019"},{"why":"Gives the resolved and integrated three-dimensional star-formation relation and the hidden-parameter interpretation that the PCA analysis extends to retired galaxies.","marker":"Sánchez et al. 2021"}],"fun_headline_variants":["Quenching needs efficiency collapse, not just gas loss","Star formation quits when efficiency falls, not gas runs out","Galaxy retirement hinges on efficiency drop, not gas drain","Central star formation dies from efficiency plunge, not gas loss","Quenching: gas remains, but efficiency collapses in centers"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The retired-stage conclusions depend on treating CO non-detections as upper limits and on taking the 26.3-arcsec beam, roughly one effective radius, as the galaxy centre while the quenching classification defines the centre within half an effective radius.","fun_headline_variants_meta":{"raw":{"variants":["Quenching needs efficiency collapse, not just gas loss","Star formation quits when efficiency falls, not gas runs out","Galaxy retirement hinges on efficiency drop, not gas drain","Central star formation dies from efficiency plunge, not gas loss","Quenching: gas remains, but efficiency collapses in centers"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000173,"raw_usage":{"total_tokens":1370,"prompt_tokens":1126,"completion_tokens":244,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":742,"completion_tokens_details":{"reasoning_tokens":163}},"tokens_in":742,"tokens_out":244,"duration_ms":2928,"temperature":1.0,"reasoning_tokens":163,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T19:05:40.419953+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Resolve the central half effective radius of nearly and fully retired galaxies in sensitive CO observations: if their central molecular gas is detected and forms stars at normal efficiencies, the claimed central efficiency collapse would disappear; if the central SFE deficit persists at sub-kiloparsec resolution, the inside-out quenching claim would be confirmed.","supporting_citations":[],"review_version":1}