{"id":"1957cf9b-20d5-4010-b356-bb6b78b7249a","arxiv_id":"2412.03790","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Galaxies in the SPT2349-56 protocluster core show an elevated [CI](2-1)/[CI](1-0) line ratio, implying warmer or denser gas than field galaxies at the same epoch.","lead":"Using ALMA, this paper measures two atomic carbon lines in 25 dusty galaxies inside a massive protocluster at redshift 4.3, finding the line ratio is nearly twice as high as in comparable field galaxies. The result suggests the crowded protocluster environment is heating or concentrating the galaxies' cold gas, an early sign of environmental effects on the first galaxy clusters.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The reported mean/median ratio statistics appear to drop the many 3-sigma upper limits in Table 1; if faint protocluster members have lower [CI](2-1)/(1-0), the claimed offset is a selection artifact. A survival-analysis reanalysis is needed before the environmental interpretation is accepted.","rationale":"I read the paper as a measurement paper whose central claim is the ratio offset and its environmental interpretation. The data are new and the line-deblending checks are careful (e.g., applying the algorithm to the field sample), which supports the measurements themselves. The weak point is not the ALMA reduction but the population statistics. The reader flagged the lensed field sample, and I partially agree: the equal-magnification assumption is untested, and the 11 Gururajan sources without lens models could bias the field comparison. However, that issue may not be decisive because differential magnification of the more compact [CI](2-1) line would tend to raise the field ratio, making the correction reduce the field value and strengthen the offset. The more immediately load-bearing issue is internal: the Gaussian Monte Carlo described in Section 3 is only well-defined for sources with both line detections, while Table 1 is full of 3-sigma upper limits. If those upper limits are simply omitted, the protocluster mean/median is a selected-subset statistic, and the comparison with a fully detected lensed sample is apples-to-oranges. The paper itself does not specify the inclusion criterion or provide a censored-data treatment. A survival-analysis reanalysis using the published table can settle this, so the correct verdict remains conditional pending that check. I therefore leave the reader's CONDITIONAL verdict unchanged.","tokens_in":18277,"tokens_out":11353,"duration_ms":121032,"concrete_test":"Recompute the protocluster and field mean/median [CI] line-ratio statistics from Table 1 with a censored-likelihood estimator that treats all 3-sigma upper limits as censored data (e.g., ASURV or a Bayesian tobit model), and also refit the L([CI](2-1))-S850 power law with the same censoring. If the protocluster-field offset drops below ~2 sigma or the mean ratio shifts by more than ~0.2, the central claim fails; if the offset survives, the upper-limit concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 3 computes the headline statistics by drawing 10^5 Gaussian realizations of the [CI](1-0) and [CI](2-1) measurements. That procedure can only use sources with detections in both lines. Table 1, however, reports 3-sigma upper limits for a large fraction of the 25 protocluster galaxies (e.g., C11, C12, C14, C16, C17-C23, and one line in C19/C21/C22). The text never states how these censored measurements enter the mean, median, or power-law fits, and it gives no survival-analysis or limit-inclusion test. If the excluded galaxies preferentially have low excitation (C16 and C21 have [CI](2-1) upper limits that imply L' ratios below the reported median), the mean ratio 1.094 and median 0.94 would be biased high. The field sample, being strongly lensed, is nearly all detected, so the two samples are not censored symmetrically. Because the claimed environmental offset is only a factor ~1.6 in the mean ratio, selection on detections could plausibly produce it. The reader's equal-magnification worry is real but secondary; differential lensing may even act to increase the field ratio, so the censoring issue is the more load-bearing threat.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents ALMA Band 3 and Band 4 observations of the [C I] 492 GHz and 806 GHz fine-structure lines toward 25 dusty star-forming galaxies (DSFGs) in the core of the z = 4.3 protocluster SPT2349-56. The authors measure [C I](1-0), [C I](2-1), and CO(7-6) line strengths, deblending the latter two using [C II] profiles as templates, and compare the ratio L'[CI](2-1)/L'[CI](1-0) with field DSFGs from Gururajan et al. (2023). They report a higher mean ratio in the protocluster (mu = 1.094 +/- 0.090 versus 0.671 +/- 0.052), a higher median (0.94 versus 0.55), an excess of [C I](2-1) at fixed 850 micron flux density (power-law fits give r21/r10 = 1.54 +/- 0.39), and LTE excitation temperatures of 59.1 K versus 33.9 K. The interpretation is that interactions in the dense protocluster environment drive cold gas to galaxy cores, raising the average gas density and excitation temperature. The central ratio measurement is direct and does not rely on fitted parameters; the main interpretive steps are the LTE assumption and the representativeness of the lensed field comparison sample.","tokens_in":18707,"tokens_out":32689,"duration_ms":284799,"significance":"The dataset is unique: [C I](1-0) and [C I](2-1) measurements for a statistical sample of protocluster DSFGs at z ~ 4.3, along with a field sample at comparable redshift and 850 micron flux density. If the offset is real, it is one of the clearest pieces of evidence for environmental effects on the ISM excitation of galaxies during protocluster assembly, and it speaks directly to the gas-concentration and gas-stripping scenarios debated in the literature. The paper's strengths include the careful deblending procedure validated on the comparison sample, the spatial-coincidence checks in Appendix B, the explicit robustness test for the bright source N1 in Appendix A, and the transparent statement of the equal-magnification assumption for lensed sources without lens models. The central ratio measurement is independent of fitted calibrations, so circularity is not a concern; the only shared calibration (X_CI x alpha_CO from Gururajan et al. 2023) is used in the secondary gas-mass comparison. However, the treatment of censored measurements is incomplete and the quoted significance of the headline offset is likely overstated, so the quantitative claims need additional work.","major_comments":[{"comment":"The statistics in Section 3 silently exclude the censored measurements in Table 1: only sources with both lines detected above an (apparently) 3.5 sigma threshold enter the mean, median, and power-law fits, while 11 of the 25 protocluster galaxies (C11, C12, C14, C16, C17, C18, C19, C20, C21, C22, C23) have at least one 3 sigma upper limit and 7 have neither line detected. The text never states this threshold or how upper limits enter the analysis; the 3.5 sigma criterion appears only in the caption of Fig. 2. Because the strongly lensed field sample is nearly fully detected, the two samples are censored asymmetrically, and the comparison requires either a survival-analysis treatment of the ratios or, at minimum, a sensitivity test that assigns upper-limit values to the censored sources and recomputes the median, mean, and fits. Concretely, C16 and C21 have [C I](2-1) upper limits that bound their L' ratios to about 0.45 and 0.60, below the reported protocluster median of 0.94. The paper should also state explicitly in the abstract and text that the quoted median and mean are based on roughly 10 detected sources, not all 25 galaxies.","section":"Section 3, Table 1"},{"comment":"The quoted uncertainty on the protocluster mean, mu = 1.094 +/- 0.090, appears to be the standard deviation of the per-realization means obtained by drawing only from the Gaussian measurement errors; it does not include the sampling variance of the mean, sigma/sqrt(N) ~ 0.42/sqrt(10) ~ 0.13. Including both contributions gives a total uncertainty of about 0.16 on the protocluster mean, and the offset of 0.423 relative to the field mean is then significant at roughly 2.5-3 sigma rather than the roughly 4 sigma implied by a naive combination of the quoted errors. The paper should report a significance that accounts for both measurement noise and population variance (for example, a Welch t-test or a permutation test over sources), or explicitly state that the quoted errors are measurement-error-only bounds. This is load-bearing because the abstract's claim that the ratios are 'markedly different' rests on this significance.","section":"Section 3"},{"comment":"The headline statistics are sensitive to the S/N threshold used for inclusion, and that threshold is not justified. With the 3.5 sigma criterion the mean ratio is about 1.09, but with the 2 sigma convention of Table 1 the weak-[C I](1-0) sources C8, C10, and C13 enter and raise the mean to about 1.3; these are the same sources whose individual L' ratios (about 2.1-2.5) sit at or above the optically-thin LTE limit R = 2.11 where Eq. (2) diverges. The paper should state the threshold in the text, show how the mean, median, and Tex vary with the threshold, and test whether the ratio correlates with S850 or with [C I](1-0) S/N within the protocluster, since any such correlation would make the threshold choice a selection effect in the comparison with the fully detected field sample.","section":"Section 3, Eq. (2)"},{"comment":"The power-law fits that support the 'excess of [C I](2-1) at fixed S850' claim appear to use different source subsets for the two lines: [C I](2-1) is detected in more protocluster galaxies than [C I](1-0) at the 3.5 sigma level (for example, C10 and C13 enter the [C I](2-1) fit but not the [C I](1-0) fit), and the text never specifies which sources enter the ODR fits. The authors should restrict both fits to the same source list and to the same treatment of upper limits, including the censored galaxies discussed above, and they should tabulate the field fit parameters alpha and gamma, which currently appear only in Fig. 2, so that the quoted r10 = 1.41 +/- 0.26 and r21 = 2.17 +/- 0.38 can be verified.","section":"Section 3, Table 2"},{"comment":"The field comparison rests on the inclusion of 11 Gururajan et al. (2023) sources without lens models under the explicit assumption that both [C I] lines are magnified equally; the paper should test the sensitivity of the mean field ratio and of r10 and r21 to excluding these sources, since differential magnification of a more compact [C I](2-1) region would bias the field ratio upward and shrink the reported offset, while the opposite geometry would enlarge it. In addition, the field sample uses S870 while the protocluster uses S850; at these frequencies the S850/S870 ratio is about 1.1 for a modified blackbody with beta ~ 1.8 and Td ~ 35-40 K, and the paper should state how this offset is handled in the fits, even though it largely cancels in the line-ratio comparison.","section":"Section 3, field comparison"}],"minor_comments":[{"comment":"The abstract states that 'the protocluster galaxies exhibit a median ratio of 0.94,' but this median is computed from only the roughly 10 sources with both lines detected; the statement should make clear that the median refers to the detected subset, or better, to a properly censored analysis.","section":"Abstract and Section 3"},{"comment":"The 3.5 sigma inclusion criterion for the statistics in Section 3 appears only in the caption of Fig. 2; it should be moved to the text and applied consistently throughout.","section":"Fig. 2 caption"},{"comment":"The name 'Canameras' is rendered as 'Ca~nameras' in several places in the text and in the reference list.","section":"Throughout"},{"comment":"The sentence beginning '(Cortzen et al. 2020) also measured...' should read 'Cortzen et al. (2020) also measured...'.","section":"Section 4"},{"comment":"The phrase 'we averaged over the channels within 2 sigma of the expected position' should specify whether the 2 sigma refers to the line width or to the per-channel noise.","section":"Section 2.2"},{"comment":"The definition of the integration variable u and the limits are given in the text, but the notation in Eq. (1) is dense; a short verbal description of rbar as the mean predicted ratio over the 2-20 mJy interval would improve readability.","section":"Eq. (1)"}],"recommendation":"major_revision","confidential_remarks":"The central claim depends on a single field dataset (Gururajan et al. 2023), with some CO(4-3) measurements coming from private communication; the authors should consider whether including the GEMS and Walter et al. (2011) samples in a robustness figure, or at least quantifying the effect of excluding the no-lens-model sources, would harden the comparison. The statistical issues in Section 3 (censoring and significance) are fixable with additional analysis, and the qualitative direction of the result appears likely to survive, so I do not recommend rejection. The paper is a good fit for the journal's scope, but the headline numbers should be placed on firmer statistical footing before acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The short version: this is the first comparison of [CI](2-1)/(1-0) ratios between a protocluster core and a field DSFG sample at z~4.3, and the ALMA data are real and carefully reduced. But I think the stress-test note is right: the mean and median statistics in Section 3 quietly ignore most of the protocluster upper limits, and that makes the central offset look fragile.\n\nWhat the paper does well: new Band 3 and Band 4 observations of 25 protocluster galaxies, a sensible deblending scheme for CO(7-6) and [CI](2-1) using [CII]-profile templates, explicit checks against beam effects and aperture choices, and spatial coincidence tests with [CII] cutouts. The power-law comparison of L'[CI] versus S850 is a clean way to show the excess—it does not depend on fitted parameters in any circular way. The line ratio measurement itself is model-independent, and the LTE Tex inference is secondary.\n\nThe load-bearing problem is censoring. Table 1 lists 3-sigma upper limits for a large fraction of the sample (C11, C12, C14, C16, C17-C23 in at least one line). The Gaussian-realization method described in Section 3 can only use sources with detections in both lines. The paper never states how many sources enter the statistics or how limits are handled, and there is no survival analysis. If the faint protocluster members have lower [CI](2-1)/(1-0) ratios—which is exactly what you'd expect if they are less excited—the reported mean of 1.094 and the excess over the field would be selection artifacts. The field sample, being strongly lensed, is nearly all detections, so the two samples are not censored symmetrically. The equal-magnification assumption for Gururajan sources without lens models is a real secondary worry, but it is at least stated; differential lensing could plausibly go in either direction, so I agree with the stress-test note that censoring is the more serious threat.\n\nMinor points: the Tex values assume LTE and optically thin emission, which is fine given the stated optical depths, but they inherit whatever bias is in the mean ratio. Removing N1 does not change the picture, and the CO(7-6)/CO(4-3) ratio comparison also shows a hint of excess, which is independent support and should be kept in the paper.\n\nWho gets value: anyone working on protocluster gas physics, cold gas excitation, or environmental effects at high redshift. The measurement table and the deblending method are useful even if the interpretation does not survive. I would not desk-reject this. But the censoring issue needs to be addressed before the environmental claim is accepted—either with a survival-analysis reanalysis or at minimum with a clear statement of which sources enter the statistics and tests including the limits. If the offset survives that, it's a solid result; if not, the paper becomes a data paper with an interesting null.\n\nRecommendation: send to peer review, but ask for a major revision focused on the censoring treatment.","headline":"First [CI] ratio measurements in a z~4 protocluster core, but the headline statistics quietly drop most of the upper limits—so the environmental offset is less secure than the abstract claims.","tokens_in":19281,"tokens_out":2716,"would_cite":true,"duration_ms":29136,"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":"Protocluster galaxies at z=4.3 have hotter, denser gas than field galaxies","keywords":["protocluster","dusty star-forming galaxies","neutral carbon fine-structure lines","line luminosity ratio","gas excitation temperature","galaxy environment","high-redshift galaxy evolution","submillimeter observations"],"falsifier":"Measure both [CI] lines in a sample of unlensed, non-interacting dusty star-forming galaxies at z approximately 4.3 matched in 850 micron flux density; if their median line ratio matches the protocluster value rather than the lensed-field value, the claimed environmental excess disappears. Alternatively, a non-LTE excitation model that reproduces the field ratios at higher gas densities would remove the need for an environmental temperature difference.","tokens_in":18138,"feed_emoji":"🌌","tokens_out":4864,"duration_ms":43008,"temperature":0.7,"pith_summary":"The paper tries to establish that living in a dense protocluster core changes the interstellar gas inside galaxies, not just their surroundings. Using millimeter observations of the two neutral carbon fine-structure lines in 25 dusty star-forming galaxies in the z=4.3 protocluster SPT2349-56, it measures a median line-luminosity ratio of 0.94, well above the 0.55 seen in comparable field galaxies. The excess comes from the higher-excitation [CI](2-1) line at a given 850-micron flux, and under local thermal equilibrium it translates to a gas excitation temperature of about 59 K versus 34 K in the field. If right, this is direct evidence that the dense environment heats and compresses the gas in these galaxies, likely through interactions.","feed_headline":"Protocluster gas runs hotter than field gas at z = 4.3","feed_subtitle":"Neutral carbon line ratios reveal denser, more excited gas in the SPT2349-56 core.","key_machinery":"The machinery is the neutral carbon atom's two fine-structure transitions, $[CI](1-0)$ at 492 GHz and $[CI](2-1)$ at 809 GHz, which together form a simple three-level system. Their luminosity ratio $R_{CI} = L'_{[CI](2-1)}/L'_{[CI](1-0)}$ is converted to an excitation temperature through the LTE formula $T_{ex} = 38.8 \\, \\mathrm{K} / \\ln(2.11 / R_{CI})$. Because both lines are observed in the same galaxies at the same redshift, the ratio is a self-calibrating thermometer that does not depend on distance or overall luminosity. The paper also uses power-law fits of each line luminosity against the 850 micron flux density to isolate which line drives the ratio difference.","core_discovery":"The central discovery is an environmental difference in the neutral carbon line ratio. Protocluster galaxies in SPT2349-56 have a mean $L'_{[CI](2-1)}/L'_{[CI](1-0)}$ ratio of $1.094 \\pm 0.090$, compared with $0.671 \\pm 0.052$ for field dusty star-forming galaxies drawn from a lensed sample at comparable redshift and 850 micron flux density. Power-law fits of line luminosity versus $S_{850}$ show that while $[CI](1-0)$ is similar between the two populations, $[CI](2-1)$ is elevated by a factor of about two in the protocluster. Assuming the lines are optically thin and in local thermal equilibrium, this yields an excitation temperature of $59.1^{+8.1}_{-6.8}$ K for the protocluster and $33.9^{+2.4}_{-2.2}$ K for the field. The authors interpret this as cold gas being driven to the galactic cores by interactions in the dense environment, raising gas density and excitation.","pith_inferences":["If the line-ratio excess survives a comparison against unlensed field galaxies, it would strengthen the case that environment, not lensing selection, drives the difference.","A testable extension is to look for a correlation between the $[CI]$ ratio and merger stage or projected separation within the protocluster; the paper reports no trend with distance from the center, but merger state is a more direct probe.","The same ratio method could be applied to other protocluster cores to ask whether the heating is universal or specific to SPT2349-56's extreme over-density.","The implied higher excitation temperature should also be visible in other tracers, such as the $CO(7-6)/CO(4-3)$ ratio, which the paper finds is also elevated."],"forward_implications":["Gas in protocluster galaxies is on average hotter and likely denser than in field galaxies at the same epoch, implying the environment alters the interstellar medium before quenching.","The elevated $[CI](2-1)/S_{850}$ ratio provides a new observational marker for environmentally affected galaxies at high redshift.","Gas mass estimates from $[CI]$ agree with CO-based estimates within a factor of two, so the higher excitation does not invalidate $[CI]$ as a gas tracer.","The shorter gas depletion timescales previously reported for SPT2349-56 are consistent with the higher radiation intensity per unit gas mass implied by the hotter gas.","Future resolved observations should test whether the extra $[CI](2-1)$ is concentrated in galaxy cores, as predicted by the interaction scenario."],"supporting_citations":[{"why":"Supplies the primary field comparison sample of strongly lensed DSFGs with both [CI] line measurements.","marker":"Gururajan et al. (2023)"},{"why":"Provides positions, [CII] line profiles, and 850 micron flux densities used for line extraction and scaling.","marker":"Hill et al. (2020)"},{"why":"Reports shorter gas depletion timescales in the protocluster, supporting the higher excitation interpretation.","marker":"Hill et al. (2022)"},{"why":"Shows extended CO(4-3) emission in SPT2349-56, giving context for how gas is distributed in the core.","marker":"Zhou et al. (2025)"},{"why":"Describes the deblending method used to separate CO(7-6) from [CI](2-1).","marker":"Chapman et al. (2024)"},{"why":"Derives the LTE excitation temperature formula for the [CI] line ratio.","marker":"Schneider et al. (2003)"},{"why":"Provides the SPT lensed source catalog from which the field comparison sample is selected.","marker":"Reuter et al. (2020)"},{"why":"Establishes [CI] as a molecular gas tracer, supporting part of the analysis.","marker":"Papadopoulos & Greve (2004)"}],"fun_headline_variants":["Protocluster gas runs hotter than field at z=4.3","Dense protocluster boosts [CI](2-1) emission","Protocluster gas excitation 59 K vs 34 K in field","Interactions in protocluster heat gas to 59 K","Environmental effect: elevated [CI] ratio"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The field galaxies are strongly lensed, and for about a third of them the two carbon lines are assumed to be magnified equally even without a lens model; if that magnification ratio is wrong, the field line ratios could be biased low and the environmental difference could be an artifact.","fun_headline_variants_meta":{"raw":{"variants":["Protocluster gas runs hotter than field at z=4.3","Dense protocluster boosts [CI](2-1) emission","Protocluster gas excitation 59 K vs 34 K in field","Interactions in protocluster heat gas to 59 K","Environmental effect: elevated [CI] ratio"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000932,"raw_usage":{"total_tokens":4073,"prompt_tokens":1116,"completion_tokens":2957,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":732,"completion_tokens_details":{"reasoning_tokens":2869}},"tokens_in":732,"tokens_out":2957,"duration_ms":20865,"temperature":1.0,"reasoning_tokens":2869,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T22:05:58.268370+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure both [CI] lines in a sample of unlensed, non-interacting dusty star-forming galaxies at z approximately 4.3 matched in 850 micron flux density; if their median line ratio matches the protocluster value rather than the lensed-field value, the claimed environmental excess disappears. Alternatively, a non-LTE excitation model that reproduces the field ratios at higher gas densities would remove the need for an environmental temperature difference.","supporting_citations":[{"cited_title":"2003, , 406, 915, 10.1051/0004-6361:20030726","cited_arxiv_id":null,"evidence_quote":"Derives the LTE excitation temperature formula for the [CI] line ratio."}],"review_version":1}