{"id":"10962152-ba00-4776-9816-998ff6b677a7","arxiv_id":"2501.14893","paper_version":4,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"In the starburst galaxy M 82, CO-to-PAH scaling relations measured on 306 molecular clouds have shallower slopes than in main-sequence spiral galaxies.","lead":"JWST mid-infrared images of the starburst galaxy M 82 are matched to archived carbon monoxide radio maps, and 306 molecular clouds are used to test how well CO traces gas in extreme outflows. The measured CO-to-PAH scaling slopes are shallower than in normal spirals, suggesting CO undercounts molecular gas in small clouds; this matters for interpreting gas flows and star formation in starbursts.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claimed shallower CO-PAH slopes in M 82 rest on a CO-detection-selected sample with a hard SNR threshold; without a censoring correction, the offset versus PHANGS is not yet established.","rationale":"I read the paper as a careful, transparent study whose headline result is explicitly hedged. The central claim is the slope offset relative to PHANGS/C24, and the weakest link in that claim is the CO-selected sample: because clouds are defined only where CO is detected above a high SNR threshold, the regression is subject to y-truncation that flattens slopes. This is exactly the assumption the reader identified, and the paper's own caveats in Sections 4.1 and 5 confirm it. The proposed censored-regression test is feasible with the public NOEMA cube and MIRI maps and would settle whether the offset is physical or an artifact. The R21 assumption is secondary: a constant R21 shifts normalization, not slope, and no evidence is presented for a systematic R21 gradient correlated with PAH brightness. The paper deserves credit for publishing the comparison with explicit caveats, but the central physical interpretation should remain conditional until the censoring test is done. Therefore the reader's CONDITIONAL verdict stands unchanged.","tokens_in":17261,"tokens_out":5196,"duration_ms":62745,"concrete_test":"Re-fit the CO-PAH relations on a MIRI-selected sample (all pixels or MIRI clumps above the F770W/F1130W noise within the NOEMA field) with CO non-detections included as upper limits: set I_CO,lim = 3 sigma x Delta v using the Krieger et al. (2021) rms and line width, and run a censored regression (e.g., Bayesian survival analysis with upper limits). If the recovered slopes move to within about 1 sigma of the PHANGS values (0.9-1.0), the claimed shallower slope is a selection artifact. As a cross-check, apply the identical SNR threshold and clump-finding to the PHANGS/C24 data and see whether their slopes also drop below unity; if they do, the M 82 versus PHANGS offset is not physical.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that the CO(1-0)-PAH power-law slopes in M 82 (m = 0.74 +/- 0.19 at F770W and 0.81 +/- 0.25 at F1130W) are intrinsically shallower than the PHANGS/C24 values (0.93 +/- 0.05 and 1.00 +/- 0.08). The entire sample is 306 clouds identified in the NOEMA CO(1-0) moment-0 map, which was blanked below SNR = 5 (Krieger et al. 2021; Section 2.2) and clump-found at >= 3 sigma (Appendix B). Because I_CO is effectively truncated from below, low-CO clouds at low MIRI intensity are absent, and an OLS fit of log I_CO on log I_MIRI is biased toward a shallower slope. The authors themselves flag this in Sections 4.1 and 5: missing fainter clouds could produce artificially shallower slopes. The PHANGS comparison uses CO(2-1) with different sensitivity, angular resolution, and selection, so the slope difference may reflect different truncation rather than the M 82 ISM. The R21 = 1.0 assumption affects normalization more than slope unless R21 varies with MIRI brightness; the censoring/selection effect is the more load-bearing threat.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents new JWST/MIRI F770W and F1130W images of M 82 and combines them with the archival NOEMA CO(1-0) moment-0 map to measure, for 306 CO-selected clouds, the power-law relations between CO(1-0) integrated intensity and the 7.7 and 11.3 micron PAH band intensities. The authors report global slopes m = 0.74 +/- 0.19 (F770W) and m = 0.81 +/- 0.25 (F1130W), compare these with the PHANGS/C24 slopes of 0.93 +/- 0.05 and 1.00 +/- 0.08, and interpret the lower values as evidence that CO does not trace the full molecular gas budget in smaller clouds exposed to hard radiation. They also examine the CO versus F770W/F1130W ratio and find moderate correlations for some regional and size-selected subsets. The paper explicitly acknowledges that the CO data sensitivity and cloud-selection threshold could bias the slopes toward shallower values.","tokens_in":17587,"tokens_out":4859,"duration_ms":48257,"significance":"If the claimed slope difference were robust, it would be an interesting constraint on the CO-to-H2 calibration in starburst outflows and on PAH emission in extreme environments. The new MIRI data are valuable, and the paper is commendably transparent about its sensitivity limitations and about the R21 = 1.0 assumption used to compare with PHANGS. However, the central quantitative claim is currently not established: the reported slope differences are within about one combined standard deviation, and the CO-selected sample is truncated from below in exactly the quantity used as the dependent variable, a bias that the authors themselves identify but do not quantify. The result is therefore best treated as a promising but unproven hint rather than a measured physical difference.","major_comments":[{"comment":"The central claim that the M 82 slopes are lower than in PHANGS/C24 is not supported by the quoted uncertainties. For F770W, the difference is 0.93 - 0.74 = 0.19 with a combined uncertainty of sqrt(0.19^2 + 0.05^2) = 0.20, i.e., less than one standard deviation. For F1130W, 1.00 - 0.81 = 0.19 with a combined uncertainty of about 0.26, also less than one standard deviation. The text states these slopes are 'significantly below' the C24 values, but no significance test is presented. The authors should either provide a proper test of the slope difference or soften the claim to a marginal difference.","section":"Section 4.1, Eqs. (2)-(3)"},{"comment":"The CO intensities are truncated from below: the moment-0 map was blanked below SNR = 5 and QUICKCLUMP was run at a 3-sigma threshold, so low-CO clouds are missing from the sample. An OLS fit of log I_CO against log I_MIRI on such a censored sample is expected to produce a shallower slope, and the authors acknowledge in Section 5 that this 'could drive biases in the derivations of the power-law parameters, e.g. producing artificially shallower slopes.' Because this is the exact quantity on which the comparison to PHANGS rests, the caveat needs to be made quantitative: for example, by adding a censored-regression fit, by including upper limits from the noise map, or by simulating the effect of the CO threshold on the recovered slope. Without such an analysis, the offset from the PHANGS slopes is not yet established.","section":"Sections 2.2, 4.1, 5; Appendix B"},{"comment":"The continuum subtraction is a first-order correction in which the continuum-to-total fraction is modeled by a four-parameter sigmoid fitted to MRS/IRS spectra and then extrapolated to all clouds. The continuum contributes 35-50% of the F770W band and about 35% of the F1130W band, so errors in this correction could alter the derived slopes and normalizations. The agreement with CAFE at a few locations is encouraging, but no uncertainty on the sigmoid parameters is propagated through Equation (1), and no test of the sensitivity of the final slopes to alternative continuum prescriptions is given. The authors should show that the main conclusions are robust to the continuum-correction choice.","section":"Appendix A and Section 3.1"},{"comment":"The note in Table 1 that 'all rp have p-values << 0.01' is not correct for several entries. For example, the streamer-east CO-F770W correlation has rp = 0.14 with only 46 clouds, which is consistent with the null hypothesis (p of order 0.3), and the outflow-south CO-F770W/F1130W correlation has rp = -0.22 with 104 clouds, which is not significant at p < 0.01. The blanket statement overstates the significance of the correlations and should be replaced by individually computed p-values.","section":"Table 1"}],"minor_comments":[{"comment":"The abstract states cloud sizes range from about 21 to 270 pc, while Section 3.2 and Figure 2 give a minimum size around 31 pc. Please make the numbers consistent.","section":"Abstract vs. Section 3.2"},{"comment":"The R21 = 1.0 assumption is clearly stated, but the paper should note explicitly that if R21 varies with PAH brightness or with position in M 82, the comparison slope could be affected; the current statement only addresses the normalization.","section":"Section 3.3"},{"comment":"The fitting procedure is described as 'binned in the x-axis,' but the number of bins, the binning rule, and whether the fit uses bin means or individual points are not specified. Please provide these details so the reader can assess the fit.","section":"Section 4.1"},{"comment":"The choice to identify clouds in the 2D moment-0 map rather than the 3D datacube is explained, but the resulting line-of-sight blending bias is mentioned only in passing. A sentence quantifying the expected effect on the derived slopes would be helpful.","section":"Section 3.2"}],"recommendation":"major_revision","confidential_remarks":"The paper is honest and the data are valuable, but the central comparison is currently under-supported: the slope differences are within the reported uncertainties, and the CO censoring bias is acknowledged but not quantified. I do not see circularity in the use of external comparison samples. The manuscript could become publishable after a serious sensitivity/censoring analysis and a corrected significance assessment."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Vivian,\n\nQuick take: this is a genuinely useful observational paper. The new JWST MIRI F770W/F1130W imaging of M 82, matched to the NOEMA CO(1-0) map, gives the first cloud-scale CO–PAH scaling relations in a starburst outflow, with 306 clouds. The slopes they measure (0.74 ± 0.19 and 0.81 ± 0.25) are lower than the PHANGS reference values (0.93 ± 0.05 and 1.00 ± 0.08), and the paper is careful about many systematics. But the central comparison is not yet convincing. The CO sample comes from a moment-0 map blanked at SNR = 5 and a clump-finder run at 3σ, which truncates the sample at low I_CO. An OLS fit on a sample truncated that way biases slopes shallow. The authors themselves flag exactly this in Sections 4.1 and 5. The PHANGS comparison uses CO(2-1) with different sensitivity, resolution, and selection, so the offset may be a measurement artifact rather than something about the M 82 ISM. The R21 = 1.0 assumption affects normalization more than slope; the truncation issue is the load-bearing concern. The continuum subtraction is first-order and the MIRI background pointings failed, so they rely on a model background; worth noting, probably not fatal.\n\nWhat the paper does well: the measurements are transparent, the caveats are explicit, and the size-split analysis (stronger correlations for clouds larger than 100 pc) actually supports the sensitivity interpretation rather than an intrinsic physics story. The physical discussion is balanced—they even mention a Kennicutt–Schmidt alternative to CO suppression.\n\nNet: this is an honest, incremental observational datapoint. The M 82 measurements are worth having, and the paper deserves a serious referee. But the headline claim of intrinsically shallower slopes should be softened, and a censoring-aware analysis (or at least a simulation of the truncation bias) would be needed to establish it. I'd cite it for the M 82 measurements, not for the slope comparison.","headline":"New JWST data give the first cloud-scale CO–PAH comparison in M 82, but the claimed shallower slopes are plausibly a sensitivity artifact, so the headline comparison to PHANGS is not yet established.","tokens_in":18190,"tokens_out":2707,"would_cite":true,"duration_ms":24159,"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":"M 82's starburst makes CO a poorer tracer of molecular gas in small clouds.","keywords":["starburst galaxy M 82","galactic outflows","CO(1-0) emission","PAH bands 7.7 and 11.3 microns","molecular cloud scaling relations","JWST MIRI imaging","CO-dark molecular gas"],"falsifier":"Re-derive the CO-PAH slopes after adding the faint clouds that fall below the current CO detection threshold, using a deeper, higher-resolution CO(1-0) map or a higher-J CO map matched to the MIRI resolution. If the recovered slopes rise to $\\approx 0.9$–$1.0$, the claimed intrinsic shallowness in M 82 is an artifact of missing faint CO clouds; if they stay near $0.7$–$0.8$, the physical suppression interpretation is supported.","tokens_in":17133,"feed_emoji":"🌌","tokens_out":8508,"duration_ms":69558,"temperature":0.7,"pith_summary":"This paper asks whether carbon monoxide emission still reliably traces molecular gas when a galaxy is being torn apart by an intense starburst. The authors combine new mid-infrared images of M 82 with an archival CO(1-0) map, identify 306 molecular clouds, and measure how each cloud's CO brightness scales with the 7.7 and 11.3 micron emission from polycyclic aromatic hydrocarbons (PAHs). They find power-law slopes of $0.74\\pm0.19$ and $0.81\\pm0.25$, below the near-unity slopes measured in local main-sequence spiral galaxies. They interpret this as evidence that in the hardest radiation environments CO no longer traces the full molecular-gas budget, especially in the smallest clouds, likely because photoionization or emission suppression weakens CO. If correct, CO-only measurements of starburst outflows systematically underestimate the molecular gas available to be ejected or to form stars.","feed_headline":"M 82's starburst weakens CO as a molecular gas tracer","feed_subtitle":"JWST mid-infrared images show PAH emission outshining CO in small clouds, hinting at gas CO cannot see.","key_machinery":"The load-bearing object is the per-cloud scaling relation between integrated CO(1-0) intensity and the intensity of the 7.7 and 11.3 micron PAH bands, measured on 306 cloud footprints extracted from the CO moment-0 map. PAHs are large carbon molecules that emit characteristic mid-infrared bands after absorbing ultraviolet photons, so they trace gas heated by young stars. The analysis sums each map inside the cloud footprints, fits log-log power laws, and compares the fitted slopes to control spirals, using a first-order continuum subtraction, MIRI filter-band corrections, and a fixed CO(2-1)-to-CO(1-0) line ratio to place the comparison on the same footing.","core_discovery":"The central claim is that in the nearby starburst galaxy M 82, the relations between CO(1-0) line emission and the mid-infrared PAH features at 7.7 and 11.3 microns are genuinely shallower than the same relations in local main-sequence spiral galaxies. Using 306 molecular cloud footprints identified in the inner 2 kpc of an archival CO moment-0 map and matched-resolution JWST MIRI images, the authors fit power laws with slopes $m=0.74\\pm0.19$ for F770W and $m=0.81\\pm0.25$ for F1130W, against $m=0.93\\pm0.05$ and $m=1.00\\pm0.08$ for spirals after converting CO(2-1) to CO(1-0) with a fixed line ratio $R_{21}=1$. They also find moderate correlations between CO intensity and the F770W/F1130W ratio in most regions, with the scatter increasing sharply above $\\log(I_{770}/I_{1130})\\approx0.56$, concentrated in small clouds at large projected distances. The authors conclude that the hard starburst radiation field suppresses CO emission in the smallest clouds, so CO does not trace the full molecular-gas budget and PAH emission becomes relatively brighter.","pith_inferences":["Editorial inference: If CO is suppressed rather than absent, high-excitation CO lines or direct H2 rotational lines should recover some of the missing gas, and a multi-J CO study of the same 306 clouds would test this directly.","Editorial inference: The same size-dependent bias would apply to high-redshift starbursts, where CO is often the only cold-gas tracer, so PAH-to-CO ratios could serve as a diagnostic of CO-dark gas.","Editorial inference: The near-zero CO-F770W correlation in the streamer-east region suggests the local radiation geometry, not just the global starburst intensity, sets the CO-to-PAH ratio."],"forward_implications":["CO-based molecular gas masses in M 82's outflow and streamer regions are likely lower limits, with the missing fraction greatest for clouds smaller than about 100 pc.","The 7.7 and 11.3 micron PAH bands become relatively brighter per unit CO as the starburst radiation field hardens, so PAH intensity can serve as a complementary gas tracer in extreme environments.","The CO-PAH scaling relations in starbursts are not universal; they depend on cloud size and location, so calibrations from normal spiral disks should not be applied to starburst outflows.","The sharp increase in scatter of the CO-F770W/F1130W relation above a band ratio of about 0.56 marks a transition to a regime where PAH ionization state and CO emission decouple."],"supporting_citations":[{"why":"Supplies the archival NOEMA CO(1-0) moment-0 map and the region classification that the cloud identification and intensity measurements are built on.","marker":"Krieger et al. (2021)"},{"why":"Provides the CO(2-1)-to-PAH scaling relations for local main-sequence spirals that the M 82 slopes are compared against.","marker":"Chown et al. (2024)"},{"why":"Supplies the assumed CO(2-1)-to-CO(1-0) line ratio R21=1 used to convert the comparison sample to CO(1-0).","marker":"Weiß et al. (2005)"},{"why":"Provides the MIRI filter band and PAH wing corrections used to turn raw F770W and F1130W intensities into PAH feature intensities.","marker":"Donnelly et al. (2025)"},{"why":"Supplies Spitzer IRS spectra used for the first-order continuum subtraction in the outflow regions.","marker":"Beirão et al. (2008)"},{"why":"Supports the point that CO(1-0)-PAH slopes differ from CO(2-1)-PAH slopes by about 0.2-0.3, contextualizing the observed offset.","marker":"Leroy et al. (2023)"},{"why":"Describes the JWST M 82 observations whose F770W and F1130W images are used in this work.","marker":"Bolatto et al. (2024)"}],"fun_headline_variants":["M82's starburst suppresses CO emission in small clouds, JWST finds","CO-PAH slopes shallower in M82's starburst than in spirals","JWST shows CO undercounts molecular gas in M82's small clouds","Starburst M82's hard radiation dims CO, making PAH relatively bright"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The claim depends on the assumption that the CO map's limited sensitivity and cloud selection do not remove faint clouds; if they do, the fitted slopes are artificially shallow and the comparison to normal spirals does not measure a real physical difference.","fun_headline_variants_meta":{"raw":{"variants":["M82's starburst suppresses CO emission in small clouds, JWST finds","CO-PAH slopes shallower in M82's starburst than in spirals","JWST shows CO undercounts molecular gas in M82's small clouds","Starburst M82's hard radiation dims CO, making PAH relatively bright"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000641,"raw_usage":{"total_tokens":3044,"prompt_tokens":1133,"completion_tokens":1911,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":749,"completion_tokens_details":{"reasoning_tokens":1823}},"tokens_in":749,"tokens_out":1911,"duration_ms":12577,"temperature":1.0,"reasoning_tokens":1823,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T14:48:22.017174+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-derive the CO-PAH slopes after adding the faint clouds that fall below the current CO detection threshold, using a deeper, higher-resolution CO(1-0) map or a higher-J CO map matched to the MIRI resolution. If the recovered slopes rise to $\\approx 0.9$–$1.0$, the claimed intrinsic shallowness in M 82 is an artifact of missing faint CO clouds; if they stay near $0.7$–$0.8$, the physical suppression interpretation is supported.","supporting_citations":[{"cited_title":"D., et al","cited_arxiv_id":null,"evidence_quote":"Supplies the archival NOEMA CO(1-0) moment-0 map and the region classification that the cloud identification and intensity measurements are built on."},{"cited_title":"K., Bolatto , A","cited_arxiv_id":null,"evidence_quote":"Supports the point that CO(1-0)-PAH slopes differ from CO(2-1)-PAH slopes by about 0.2-0.3, contextualizing the observed offset."},{"cited_title":"D., Levy , R","cited_arxiv_id":null,"evidence_quote":"Describes the JWST M 82 observations whose F770W and F1130W images are used in this work."}],"review_version":1}