{"id":"22819682-1167-4314-acc2-d2ce67e23201","arxiv_id":"2608.10165","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"A 577 arcmin^2 ALMA Band 4 spectral search yields 52 CO/[CI]-detected galaxies at 0.5<z<5, mostly massive main-sequence systems with large gas reservoirs, plus five Green Valley galaxies with a gas-fraction peak at cosmic noon.","lead":"This paper searches ALMA Band 4 data cubes from the Ex-MORA survey for molecular gas emission lines and finds 52 galaxies in CO or [CI] across redshifts 0.5 to 5. It shows that gas-selected samples recover massive, gas-rich galaxies in many evolutionary stages, including some transitioning galaxies that optical surveys miss.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The abstract's GV mu_gas peak at cosmic noon is not established in the body: with N=5 GV galaxies and a gas-rich z~2.4 protocluster in the same bin, environment is a direct alternative to the feedback interpretation.","rationale":"I agree with the reader's conditional verdict but think the most load-bearing vulnerability is not the metallicity-dependent conversion chain per se; even with a perfect conversion factor, the GV peak at cosmic noon is threatened by small-N and the z=2.4 protocluster that Section 8.4 itself identifies as biasing that bin. The conversion factor concern is real and should be part of the robustness section, but it is a systematic offset common to all bins unless metallicity evolution is strong; the environmental confound is specific to the headline claim. The body's own language ('might not be significant') and the abstract's confident phrasing create an internal inconsistency, which is more damaging than relying on literature alpha_CO values. Hence no change to the reader's conditional verdict is needed; the requested re-scoping should target the GV peak.","tokens_in":23500,"tokens_out":6012,"duration_ms":61666,"concrete_test":"Using the protocluster membership definition in Section 8.4, remove all GV galaxies associated with the z~2.4 overdensity and recompute the GV mu_gas curve in Fig. 10; then run a bootstrap or t-test on the z~2-3 GV bin against the z<2 GV bin. If the peak is no longer significant at >=2 sigma or vanishes, the abstract's feedback interpretation is unsupported and must be re-scoped to note that the GV peak is driven by the protocluster.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Catalog-level claims (52 line detections, dual search) are well supported, but the central physical claim — that five Green Valley galaxies exhibit a peak in molecular gas fraction at cosmic noon, implying feedback heats rather than depletes gas — is the least secure part of the argument. Section 8.2, which presents Fig. 10, stops at a 'trend' and says error bars are non-negligible; it does not report a significance test for any GV peak. The only quantitative statement about the z~2.3-2.5 epoch is in Section 8.4, where the protocluster at z~2.4 is found to be gas-rich and to bias the redshift bin, with three sub-MS (i.e. GV-class) galaxies inside it. Thus the elevated GV point at cosmic noon is plausibly an environmental effect of the overdensity, not an intrinsic evolutionary stage. The abstract upgrades this to a result ('exhibit a mu_gas peak at cosmic noon') without the supporting test, and the suggested feedback heating is one of several interpretations. Because this is the main new astrophysical conclusion beyond the catalog, it is load-bearing.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents the Ex-MORA molecular line search: 577 arcmin^2 of ALMA Band 4 data in the COSMOS field are searched for CO and [CI] emission using a dual strategy of an unbiased line-search algorithm and a targeted search with spectroscopic redshift priors. The final sample contains 52 galaxies (54 line detections) at 0.5<z<5. The authors derive stellar masses and SFRs with CIGALE using non-parametric SFHs, convert CO and [CI] luminosities to molecular gas masses with metallicity-dependent conversion factors, and analyze gas fractions, depletion times, morphology, kinematics, and environmental effects. The main claims are that the sample is dominated by massive main-sequence galaxies with large gas reservoirs, that gas fractions increase with redshift for active populations, and that five Green Valley galaxies show a peak in molecular gas fraction at cosmic noon, possibly indicating stellar feedback that heats gas rather than depleting it.","tokens_in":23708,"tokens_out":2476,"duration_ms":26357,"significance":"If the results hold, this would be one of the largest molecular-gas-selected samples at 0.5<z<5, demonstrating the value of repurposing wide-area ALMA continuum surveys as line searches. The methodology is a real strength: the unbiased search uses SNR>=6 plus PPessimistic<0.5, the targeted search is calibrated by a random-frequency control test limiting contamination to 1%, and every unbiased line has a COSMOS-Web counterpart. These choices are transparent and justify confidence in the catalog itself. The physical interpretation, however, is much more uncertain because the gas masses inherit a chain of model assumptions (Sanders et al. 2021 metallicities, Tacconi et al. 2018 alpha_CO, Heintz & Watson 2020 alpha_[CI]) with no direct metallicity measurements, and the central astrophysical claim about Green Valley galaxies rests on only five sources with no significance test and a plausible environmental alternative. The paper would be a valuable contribution with a revised, more cautious framing of the GV peak.","major_comments":[{"comment":"The abstract states that 'Five Green Valley galaxies exhibit a μ_gas peak at cosmic noon' and links this to stellar feedback, but this is not established in the body. Section 8.2 says only that 'there is a trend for GV ... to have lower gas fractions' and that 'error bars are non-negligible'; no significance test is reported for the cosmic-noon GV point. Section 8.4 shows that the z~2.4 bin is biased by a gas-rich protocluster containing three sub-MS (GV-class) galaxies, so the elevated GV point is plausibly an environmental effect rather than an intrinsic evolutionary stage. This is the main new astrophysical conclusion beyond the catalog, so the claim must either be supported by a quantitative test (e.g., a bootstrap or permutation test comparing the GV μ_gas at z~2.3-2.5 with other bins or with field-only GV sources) or be explicitly downgraded to a tentative interpretation in the abstract and conclusions.","section":"Section 8.2 and Fig. 10; Abstract"},{"comment":"The molecular gas fractions are compared with the Tacconi et al. (2018) scaling relations, but the same Tacconi et al. (2018) relation is used to set alpha_CO, which enters M_mol and hence μ_gas. Since the metallicities used to derive alpha_CO come from the Sanders et al. (2021) mass-metallicity relation applied to the CIGALE stellar masses, the comparison is partially circular: any systematic offset in M_* propagates directly into M_mol through alpha_CO and then is compared against a relation that was calibrated with the same alpha_CO. The paper should quantify this coupling (e.g., by recomputing μ_gas with a fixed alpha_CO or an independent alpha_CO prescription) or explicitly discuss the direction and magnitude of the resulting bias.","section":"Section 6.1 and Section 7"},{"comment":"The discussion of the z~2.4 protocluster says that field galaxies at that redshift have 'significantly lower gas fractions' than overdensity members, but no significance test or uncertainty is reported for this comparison. Given that only four field galaxies are in that bin, the statement needs a quantitative measure (e.g., a two-sample test or resampling confidence interval) before it can support the conclusion that the protocluster 'significantly biases the results.'","section":"Section 8.4 and Fig. 11"}],"minor_comments":[{"comment":"Typo: 'posibility' should be 'possibility'.","section":"Section 9"},{"comment":"The error bars in Figures 10 and 11 are large, and the figures would be more informative if individual galaxy points were overlaid, so that readers can see the sample sizes and the scatter behind each averaged bin.","section":"Figure 10 and 11"},{"comment":"The paper says CIGALE v2025.1 is used and the SFH grid follows Arango-Toro et al. (2025) but with seven bins instead of ten; this change is not justified. A brief comment on why the default of seven was chosen would be useful.","section":"Section 5"},{"comment":"The final catalog is described as '54 high-fidelity emission detections' but elsewhere the paper says '52 galaxies (54 emission lines)'. This is consistent, but the wording in Section 4.3 could be clarified to state that two galaxies each contribute two lines, so the catalog contains 52 unique galaxies.","section":"Section 4.3"},{"comment":"In Table D.1, some redshift entries have no quoted uncertainty (e.g., z = 2.29(1) vs z = 2.32(2)), and the meaning of the parenthetical digits should be stated; for spectroscopic redshifts the uncertainty appears to be in the last digits, but for some entries it is omitted.","section":"Appendix D"}],"recommendation":"major_revision","confidential_remarks":"The catalog construction and validation are robust and will be of value to the community. The main risk is that the headline physical conclusion (GV μ_gas peak at cosmic noon) is not supported by the analysis as written; I would advise the editor that this is fixable with a significance test and a more cautious framing, and the paper should be acceptable after major revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read the Ex-MORA line search paper. The catalog and the search methodology are the real news; the headline astrophysical claim about Green Valley galaxies is not supported by the body.\n\nWhat's genuinely new: a 577 arcmin^2 ALMA Band 4 line search in COSMOS, producing 52 gas-selected galaxies with 54 line detections. The dual search (unbiased Lineseeker plus targeted spec-z priors) is carefully calibrated: SNR>=6 with PPessimistic<0.5 for the unbiased sample, and a random-frequency control test limiting contamination to 1% for the targeted sample. Every unbiased line has a COSMOS-Web/spectroscopic counterpart. That is a clean, reproducible catalog and a useful reference for future wide-area line surveys. The authors also deserve credit for explicitly analyzing the two overdensities and flagging that the z~2.4 protocluster biases the high-redshift gas fractions.\n\nThe soft spots are real but localized. The abstract states that five GV galaxies 'exhibit a mu_gas peak at cosmic noon,' implying a result. Section 8.2 merely describes a 'trend' with non-negligible error bars; there is no significance test for any GV peak. With N=5 GV galaxies total, and three of them inside the gas-rich z~2.4 protocluster, the elevated cosmic-noon point is as plausibly environmental as it is evolutionary. The protocluster discussion in Section 8.4 actually makes that case. So the abstract overstates the body. The fix is either to soften the abstract and conclusions or to add the missing robustness tests (remove protocluster members, try a starburst alpha_CO, check sensitivity to the MS definition). Also worth noting: gas masses are derived with the Tacconi+2018 alpha_CO and then compared to the Tacconi+2018 scaling relation; that's a common practice but it does create a degree of circularity in the gas-fraction trends.\n\nNone of this is fatal. The survey is a solid observational contribution and the catalog will be cited. It deserves a serious referee, but it needs a revision to align the claims with the evidence.","headline":"Solid survey paper; the catalog is the contribution, but the Green Valley peak in the abstract overreaches the evidence in the body.","tokens_in":24380,"tokens_out":3195,"would_cite":true,"duration_ms":31252,"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":"A 577-square-arcminute ALMA Band 4 line search recovers 52 gas-rich galaxies between redshifts 0.5 and 5, dominated by massive main-sequence systems and including five transitional Green Valley galaxies.","keywords":["molecular gas","CO line emission","[CI] fine-structure lines","galaxy evolution","main sequence","green valley galaxies","ALMA surveys","gas fraction"],"falsifier":"Measure rest-frame optical gas-phase metallicities for the 52 galaxies, or at least the five Green Valley systems, and recompute the CO and [CI] conversion factors; if the Green Valley gas fractions no longer peak above the field near $z\\approx2.4$, the feedback-heating interpretation loses its empirical support.","tokens_in":23249,"feed_emoji":"🌌","tokens_out":15343,"duration_ms":124624,"temperature":0.7,"pith_summary":"Using the spectral axis of a 577 arcmin$^2$ ALMA Band 4 blank-field survey, this paper tries to build a census of cold molecular gas selected by gas emission rather than by starlight. Combining a blind line search with a targeted search at known spectroscopic redshifts, it detects 52 galaxies in CO or [CI] transitions across $0.5<z<5$. The resulting sample is dominated by massive main-sequence galaxies (median $\\log M_*/M_\\odot = 10.83$) with large molecular reservoirs (median $\\log M_{\\rm mol}/M_\\odot = 10.69$), and it includes five Green Valley galaxies (transitional systems between the star-forming main sequence and quiescence) whose molecular gas fraction peaks around cosmic noon. If the analysis holds up, it would show that wide-area line searches recover gas-massive galaxies at all evolutionary stages, including transitional systems that optical selection misses, and that some Green Valley galaxies may have star formation suppressed by feedback that heats gas rather than depleting it.","feed_headline":"Mining ALMA cubes finds 52 gas-rich galaxies across cosmic time","feed_subtitle":"A Band 4 spectral search catches massive gas reservoirs and five transitional Green Valley galaxies.","key_machinery":"The load-bearing mechanism is the dual search over the full spectral axis of the MORA and Ex-MORA ALMA Band 4 mosaics. A blind line-search algorithm convolves each cube with Gaussian kernels of various widths and assigns a conservative fidelity value by comparing positive detections with negative noise spikes; a targeted search then fits Gaussian profiles at the expected frequencies of CO or [CI] lines at positions with prior spectroscopic redshifts. Line luminosities are converted to molecular gas masses through metallicity-dependent conversion factors for CO and [CI], with metallicities taken from a stellar mass-metallicity relation and stellar masses from SED fitting with non-parametric star-formation histories. This conversion chain carries every gas fraction and depletion time reported in the paper.","core_discovery":"The paper's central claim is that a molecular-gas-selected sample built from ALMA Band 4 spectral cubes is both feasible and scientifically productive: 18 high-fidelity emitters from a blind search plus 34 galaxies recovered by targeting known spectroscopic redshifts give 52 galaxies (54 line detections) in CO and [CI] transitions between $z=0.5$ and $z=5$. These galaxies are mostly main-sequence systems with stellar masses near or above $10^{10.8}\\,M_\\odot$ and molecular gas masses typically above $10^{10}\\,M_\\odot$. The gas fraction rises with redshift among active galaxies, consistent with existing scaling relations, but the paper shows that the high-redshift trend is inflated by a gas-rich protocluster at $z\\approx2.4$, while the overdensity at $z\\approx0.73$ matches the field population. The five Green Valley galaxies show a peak in gas fraction at cosmic noon, which the authors interpret as evidence that stellar feedback can heat gas and suppress star formation without immediately consuming the reservoir, while explicitly cautioning that this need not hold for all Green Valley galaxies.","pith_inferences":["A direct test of the heating interpretation would be to measure CO excitation ladders or [CI]/CO ratios in the five Green Valley galaxies: if feedback is heating the gas, higher-$J$ CO lines should be relatively bright compared with CO(2-1).","The targeted component's yield depends on the completeness of the spectroscopic catalog, so the true line-emitter population is probably richer; deeper blind searches over the same area would quantify how many gas-rich galaxies lack prior redshifts.","If this selection is as effective as claimed, future wide-band, large-area line surveys could construct near-complete gas-mass samples at fixed stellar mass, letting the molecular-gas scaling relations be measured without any luminosity or color preselection.","The use of [CI] alongside CO hints that some of the gas may be CO-dark; per-galaxy joint CO and [CI] detections could be used to map CO-dark fractions and check the metallicity dependence of the conversion factors."],"forward_implications":["If the method works, existing wide-area ALMA continuum surveys can be re-mined as line searches, expanding the cosmic volume probed for cold gas without new observations.","Gas-selected samples can detect galaxies in transitional phases, meaning optical preselection may miss a population of gas-rich quasi-quiescent systems whose quenching is not caused by gas exhaustion.","Environmental overdensities such as the $z\\approx2.4$ protocluster can dominate a line-selected sample's high-redshift gas-fraction trend, so environment must be accounted for when interpreting such trends.","The prevalence of rotating disks among resolved sources implies that massive gas-rich galaxies at these redshifts are largely rotationally supported, consistent with disk formation before cosmic noon.","Compared with the deeper pencil-beam ASPECS survey, the wide but shallower Ex-MORA search recovers more massive gas reservoirs, demonstrating a complementary selection function."],"supporting_citations":[{"why":"Defines the Ex-MORA observations and continuum catalog whose spectral cubes are searched here.","marker":"Long et al. 2024"},{"why":"Presents the MORA survey data that form the base of the combined mosaic.","marker":"Casey et al. 2021"},{"why":"Provides the metallicity-dependent CO-to-H2 conversion factor and the main-sequence gas-fraction scaling relations used for comparison.","marker":"Tacconi et al. 2018"},{"why":"Supplies the mass-metallicity relation used to assign metallicities, hence conversion factors, to every galaxy.","marker":"Sanders et al. 2021"},{"why":"Supplies the metallicity-dependent [CI]-to-H2 conversion factor used for the neutral-carbon lines.","marker":"Heintz & Watson 2020"},{"why":"Gives the redshift-dependent CO excitation ratios used to normalize all CO transitions to CO(1-0).","marker":"Boogaard et al. 2020"},{"why":"Supplies the spectroscopic redshift compilation that defines the targeted search priors and line identifications.","marker":"Khostovan et al. 2025"},{"why":"Provides the COSMOS-Web photometric catalog, photometric redshifts, and SED inputs used for cross-matching and physical properties.","marker":"Shuntov et al. 2025"},{"why":"Describes the blind line-search algorithm that produces the unbiased detections and fidelity estimates.","marker":"González-López et al. 2017"},{"why":"Provides the SED-fitting code used to derive stellar masses and star-formation rates with non-parametric star-formation histories.","marker":"Boquien et al. 2019"}],"fun_headline_variants":["Blind ALMA search nets 52 gas-rich galaxies up to z=5","Gas-selected ALMA survey maps massive reservoirs in 52 galaxies","Green Valley galaxies peak in gas fraction at cosmic noon","ALMA cubes reveal gas-rich galaxies dominate high-z sample"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"All gas masses and gas fractions rest on conversion factors whose metallicities are inferred from a stellar mass-metallicity relation, with no direct metallicity measurements for any of the 52 galaxies.","fun_headline_variants_meta":{"raw":{"variants":["Blind ALMA search nets 52 gas-rich galaxies up to z=5","Gas-selected ALMA survey maps massive reservoirs in 52 galaxies","Green Valley galaxies peak in gas fraction at cosmic noon","ALMA cubes reveal gas-rich galaxies dominate high-z sample"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001355,"raw_usage":{"total_tokens":5585,"prompt_tokens":1112,"completion_tokens":4473,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":728,"completion_tokens_details":{"reasoning_tokens":4400}},"tokens_in":728,"tokens_out":4473,"duration_ms":29877,"temperature":1.0,"reasoning_tokens":4400,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T04:10:54.285606+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure rest-frame optical gas-phase metallicities for the 52 galaxies, or at least the five Green Valley systems, and recompute the CO and [CI] conversion factors; if the Green Valley gas fractions no longer peak above the field near $z\\approx2.4$, the feedback-heating interpretation loses its empirical support.","supporting_citations":[{"cited_title":"S., Casey, C","cited_arxiv_id":null,"evidence_quote":"Defines the Ex-MORA observations and continuum catalog whose spectral cubes are searched here."},{"cited_title":"M., Zavala, J","cited_arxiv_id":null,"evidence_quote":"Presents the MORA survey data that form the base of the combined mosaic."},{"cited_title":"J., Genzel, R., Saintonge, A., et al","cited_arxiv_id":null,"evidence_quote":"Provides the metallicity-dependent CO-to-H2 conversion factor and the main-sequence gas-fraction scaling relations used for comparison."},{"cited_title":"L., Shapley, A","cited_arxiv_id":null,"evidence_quote":"Supplies the mass-metallicity relation used to assign metallicities, hence conversion factors, to every galaxy."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the metallicity-dependent [CI]-to-H2 conversion factor used for the neutral-carbon lines."},{"cited_title":"A., van der Werf, P., Weiss, A., et al","cited_arxiv_id":null,"evidence_quote":"Gives the redshift-dependent CO excitation ratios used to normalize all CO transitions to CO(1-0)."},{"cited_title":"A., Kartaltepe, J","cited_arxiv_id":null,"evidence_quote":"Supplies the spectroscopic redshift compilation that defines the targeted search priors and line identifications."},{"cited_title":"B., Paquereau, L., et al","cited_arxiv_id":null,"evidence_quote":"Provides the COSMOS-Web photometric catalog, photometric redshifts, and SED inputs used for cross-matching and physical properties."}],"review_version":1}