{"id":"42b235de-10e8-4a7a-8670-45d612c03b2a","arxiv_id":"2506.10735","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"In ten 5-10 Myr old Upper Sco disks, dust masses are about an order of magnitude below Lupus values, yet the C18O-N2H+ flux correlation matches Lupus, indicating stable CO abundance over the Class II stage.","lead":"New ALMA observations of ten evolved planet-forming disks in Upper Scorpius find dust masses about ten times lower than in younger regions, while the gas chemistry traced by CO and N2H+ behaves as it does in younger disks. This matters because it tests how planet-forming material disappears over a few million years and when planets can form.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 'constant CO abundance' claim hinges on a fragile, internally inconsistent N2H+ detection set and a model-dependent flux–flux interpretation; a formal censored comparison with Lupus is missing.","rationale":"The reader's beam-smearing objection to the size comparison is valid and is explicitly conceded in Appendix D ('radii ... are strongly affected by beam smearing, and hence are overestimated'); it properly conditions the dust-trap conclusion. I do not think it is the most load-bearing issue for the strongest claim singled out in the verdict. The CO-abundance claim has a more direct weakness: the statistical and interpretive scaffolding for 'nearly constant CO' is thinner than the summary suggests. The paper is transparent about p≈4% and non-detections, and the companion modeling papers may well resolve this, but within this manuscript the evidence is insufficient. I therefore retain the CONDITIONAL verdict; no move to accept or reject is warranted. The requested check is a censored, model-based comparison that would settle whether the apparent trend is robust and whether XCO is actually constrained. Agreement with the reader is partial: they noted the modest correlation but selected beam smearing as the weakest assumption.","tokens_in":30436,"tokens_out":10147,"duration_ms":123632,"concrete_test":"Fit the AGE-PRO Lupus and Upper Sco F_C18O vs F_N2H+ data, with upper limits as defined in §4.3, to the thermochemical grid of Trapman et al. (2025b) with a free region-dependent CO-abundance offset, and report the posterior on Δlog10 XCO. If the 68% interval excludes a ≥0.5 dex offset, the constant-abundance reading survives; if it spans such offsets, or if the censored correlation disappears after excluding USco 1 and treating USco 6/8 per §4.3, the claim should be downgraded to an empirical trend requiring further modeling.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Summary point 5 ('CO abundance remains nearly constant') is the load-bearing claim. The evidence is weaker than the wording: (i) §4.3 says N2H+ is detected in only four disks, but Table 4 lists USco 8 at ~6.8σ and USco 6 at ~2.6σ, and the Figure 11 caption implies only USco 3–5 are nondetections; the censored Kendall-τ analysis is not reproducible from the stated counting. (ii) The p≈4% correlation, even if accepted, is not a test that the Lupus and Upper Sco relations are consistent: no formal slope/intercept comparison is presented. With roughly five detections, the trend may be driven by the bright, structured disk USco 1; the remaining detections (USco 7, 8, 9, 10) are not visibly monotonic. (iii) The interpretive step from an unchanged C18O–N2H+ flux trend to constant XCO is not demonstrated. As §5.3 itself notes, a positive correlation is the expected signature of a spread in gas mass at fixed XCO; an age-dependent XCO could be partially masked by correlated changes in gas surface density, disk size, or excitation. No thermochemical fit marginalizing over these nuisance parameters is reported, so the correlation is consistent with, but does not establish, constant CO abundance.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper presents ALMA Band 6 and 7 observations of ten Class II disks in Upper Scorpius from the AGE-PRO large program. The authors measure dust continuum fluxes and spectral indices, dust masses under standard optically thin assumptions, image-plane COG radii for gas and dust, and CO isotopologue and N2H+ line fluxes. The main empirical results are that Upper Sco dust masses are roughly an order of magnitude lower than in Lupus/Ophiuchus, gas and dust radii appear comparable to or larger than Lupus, and the C18O–N2H+ flux correlation looks similar to the younger Lupus sample. The paper interprets these findings as evidence for dust evolution, radial drift halting or dust trapping, and a nearly constant CO abundance across the Class II phase. Much of the detailed modeling and deconvolved size work is deferred to companion AGE-PRO papers.","tokens_in":30655,"tokens_out":5121,"duration_ms":64108,"significance":"If the conclusions hold, this is a valuable contribution: it extends a standardized, multi-tracer ALMA survey to an intermediate-age region, provides homogeneous flux and size measurements for a sparse but important sample, and places empirical constraints on gas and dust evolution between ~1–3 Myr and ~5–10 Myr. The paper's strength is its careful presentation of the observational material, the explicit comparison with the Lupus AGE-PRO sample, and the candid statement of resolution limitations in Appendix D. The most novel claim, constant CO abundance over the Class II stage, is interesting but is not fully established by the in-paper analysis; the supporting statistical and modeling steps need to be made reproducible and quantitative before the claim can be regarded as load-bearing.","major_comments":[{"comment":"The N2H+ detection set is internally inconsistent. The text in §4.3 states that N2H+(3–2) is detected in four disks (USco 1, 7, 9, 10) with SNR > 5σ and that the rest are nondetections, but Table 4 lists USco 8 at 27.3 ± 4 mJy km s−1, which is about 6.8σ, and USco 6 at 7.8 ± 3 mJy km s−1, about 2.6σ. The Figure 11 caption, in contrast, says that only USco 3, 4, and 5 are nondetections. This ambiguity makes the censored Kendall-τ analysis impossible to reproduce and directly affects the claimed C18O–N2H+ correlation, since including or excluding USco 8 and USco 6 changes the censoring pattern and the correlation strength. The authors should state explicitly which sources are detections, which are upper limits, and which are excluded, and rerun the correlation with that census.","section":"§4.3, Table 4, Figure 11 caption"},{"comment":"The claim that the Upper Sco and Lupus C18O–N2H+ correlations are indistinguishable rests on a visual comparison of two regression lines, not on a formal test. No slope or intercept uncertainties are quoted for the Upper Sco fit, no joint censored regression is presented, and no test of whether the two data sets are drawn from the same relation is performed. With p ≈ 4% for the Upper Sco correlation and a small number of detections, the evidence is weak; a bootstrap or likelihood-ratio comparison between the Lupus and Upper Sco relations is needed before the word \"indistinguishable\" is used.","section":"§5.4, Figure 11, bottom-right panel"},{"comment":"The interpretive step from an unchanged C18O–N2H+ flux–flux trend to a constant CO abundance is not demonstrated. As the paper itself notes in §5.3, a positive C18O–N2H+ correlation is the expected signature of a spread in gas mass at fixed XCO, and correlated changes in gas surface density, disk size, excitation, or dust temperature could mask an age-dependent CO abundance. No thermochemical fit that marginalizes over these nuisance parameters is reported in this paper; the correlation is therefore consistent with, but does not establish, constant CO abundance. Summary point 5 and the abstract should either be weakened or explicitly cite the companion modeling paper (L. Trapman et al. 2025b) as the source of the quantitative inference.","section":"§5.3, §5.4, Summary point 5"},{"comment":"The size comparison that supports the dust-trapping interpretation is based on image-plane COG radii that Appendix D admits are overestimated for all but USco 1. Figure 10 places all Upper Sco disks outside the drift-dominated region using these beam-limited R68 and R90 values, and Summary point 3 repeats the claim that dust disk sizes are comparable to or larger than Lupus. For the seven or eight unresolved or marginally resolved disks, the image-plane radii are dominated by beam smearing, so the empirical conclusion does not follow from this paper's own measurements. The main text should either plot the deconvolved/visibility radii from M. Vioque et al. (2025) and L. Trapman et al. (2025a) in Figure 10, or add an explicit caveat in the abstract and summary that the size comparison is only secure for USco 1 and otherwise relies on companion visibility fitting.","section":"§5.1, Figure 10, Table 5, Appendix D"}],"minor_comments":[{"comment":"The dust masses are listed without uncertainties or error bars; given that the dust mass formula in Eq. (1) depends on the assumed Tdust and κν, at minimum the adopted fixed values and an estimate of the systematic uncertainty should be stated in the table notes or text.","section":"Table 6"},{"comment":"The text \"3σ < flux < σ\" appears to be a typo; it should read \"3σ < flux < 5σ\".","section":"Appendix E"},{"comment":"The sentence \"As seen in the bottom-right panel of Figure 7\" refers to the flux–flux comparison, which is actually shown in Figure 11; the figure number appears to be a typo.","section":"§5.3"},{"comment":"The abstract's statement that disk radii are comparable to those in Lupus should be qualified as image-plane, beam-affected measurements, consistent with the acknowledgment in Appendix D that these values are overestimated for most sources.","section":"Abstract and Summary"},{"comment":"The caption's statement that only USco 3, 4, and 5 are N2H+ nondetections conflicts with the text in §4.3; the caption and the text should be made consistent after the detection census is clarified.","section":"Figure 11 caption"}],"recommendation":"major_revision","confidential_remarks":"The paper is part of a coordinated series, and several of the points I raise are addressed in companion papers that are cited but not included in this manuscript. The central in-paper claim about constant CO abundance, however, is not yet quantitatively supported in the presented text: the N2H+ detection census is inconsistent, the Lupus comparison is not formally tested, and the beam-limited radii cannot carry the size conclusion alone. These are fixable in revision, so I recommend major revision rather than rejection. I would encourage the editor to ask for a revised version that reconciles Table 4 with §4.3 and Figure 11, adds a formal test of the Lupus/Upper Sco relation, and moves the Appendix D caveat into the main text."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Worth reading if you work on disk evolution. The new data are a solid addition: simultaneous 12CO, 13CO, C18O, and N2H+ measurements for 10 Upper Sco disks, more sensitive than the earlier Barenfeld/Carpenter surveys. The dust mass deficit is consistent with Barenfeld et al. (2016), so that part is confirmation. The genuinely new piece is the C18O–N2H+ flux comparison between Lupus and Upper Sco, which is a reasonable empirical test of whether the CO abundance evolves across the Class II stage.\n\nThe paper is careful about calibration and literature cross-checks, and Appendix D is honest that the image-plane radii are beam-smearing limited for all but USco 1. Data products are available, which is good.\n\nWhere it strains: the abstract and summary point 5 claim a stable CO abundance from an unchanged C18O–N2H+ correlation. That is too strong. The correlation has p≈4% with roughly five detections, there is no formal slope or intercept comparison with Lupus, and the N2H+ detection counting is internally inconsistent: the text says four detections, but Table 4 lists USco 8 at about 6.8σ and USco 6 at about 2.6σ. The censored Kendall-τ test cannot be reproduced from the stated numbers. Also, as the paper's own §5.3 notes, a positive correlation is what you expect from a spread in gas mass at fixed CO abundance, so the flux–flux trend does not uniquely imply constant XCO.\n\nThe size comparison is a softer spot. The abstract says disk radii are comparable to Lupus, but Appendix D admits these image-plane radii are overestimated for most targets. The dust-trap conclusion relies on companion visibility-fitting papers. That is fine for a survey paper, but the abstract should carry the caveat.\n\nBottom line: this deserves a serious referee and, after revision, publication. The data are worth having, the analysis is mostly careful, and the overstatements are fixable with a tempered abstract and a corrected detection census. I would bring it to reading group and cite it.","headline":"Valuable new Upper Sco ALMA data, but the abstract's claims of stable CO abundance and comparable disk sizes outrun what the paper's own statistics and image-plane radii support.","tokens_in":31385,"tokens_out":3593,"would_cite":true,"duration_ms":39617,"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":"CO gas abundance stays nearly constant through the Class II stage of disk evolution, while dust masses drop by an order of magnitude in 5–10 Myr old disks.","keywords":["protoplanetary disks","disk evolution","CO isotopologues","N2H+ chemistry","dust radial drift","dust traps","Upper Scorpius","ALMA survey"],"falsifier":"If visibility fitting of the Band 6 continuum (as in Vioque et al. 2025) yields true dust radii that are substantially smaller than the image-plane R68 values, the radial-drift-halted interpretation fails. Alternatively, a thermochemical model of these specific disks that reproduces the observed C18O and N2H+ fluxes with a strongly varying CO abundance would falsify the constant-CO claim.","tokens_in":30185,"feed_emoji":"🔭","tokens_out":4716,"duration_ms":50134,"temperature":0.7,"pith_summary":"This paper presents ALMA observations of ten 5–10 Myr old protoplanetary disks in Upper Scorpius and argues that the CO gas abundance in the warm molecular layer remains nearly constant across the Class II stage of disk evolution. The key evidence is that the empirical correlation between C18O and N2H+ line fluxes is indistinguishable between these older disks and the younger Lupus disks at 1–3 Myr, even though dust masses are about an order of magnitude lower. If correct, the decline in CO brightness with age traces real gas loss rather than CO depletion, and the comparable gas and dust radii imply that millimeter grains must be trapped rather than freely drifting inward. The paper also finds more scatter in gas-versus-dust correlations for Upper Sco than for Lupus, possibly reflecting age spread or per-disk differences in CO abundance or gas-to-dust ratio.","feed_headline":"CO gas in planet-forming disks survives to 5–10 Myr","feed_subtitle":"C18O–N2H+ correlation matches young Lupus disks, so shrinking gas masses are real, not CO depletion.","key_machinery":"The load-bearing mechanism is the C18O (2–1) versus N2H+ (3–2) flux–flux correlation. N2H+ forms where CO is absent and is destroyed by CO, so a fixed relation between these fluxes across regions of different age implies a fixed CO abundance in the warm molecular layer. The paper measures this correlation using a censored Kendall's τ test and compares the best-fit linear regression with the Lupus sample from Deng et al. (2025b). For disk sizes, the curve-of-growth (COG) method provides the R68 and R90 radii from the image plane, which are then compared with the deconvolved radii from visibility fitting.","core_discovery":"The central claim is that the CO gas abundance does not evolve significantly between roughly 1–3 Myr and 5–10 Myr in Class II disks. Because N2H+ is destroyed by reactions with gaseous CO, the ratio of C18O to N2H+ emission constrains the CO abundance, and the C18O–N2H+ flux correlation in Upper Sco falls on the same line as in Lupus. The paper therefore concludes that the lower dust masses and lower CO fluxes in the older region reflect genuine depletion of the disk material, not a chemical masking of gas. This is paired with the finding that dust disk radii in the image plane are comparable to or larger than in Lupus, which the authors interpret as evidence that dust trapping halts the expected inward drift of grains.","pith_inferences":["If CO abundance is truly constant across the Class II phase, C18O-based gas mass estimates at intermediate ages could be calibrated against N2H+ without invoking evolution of CO depletion, a generalization the paper itself does not fully claim beyond its two regions.","The paper's dust-radius conclusion rests on image-plane R68 values that are beam-dominated for seven of ten disks, so the companion visibility-fitting analysis is the natural decisive test of the radial-drift-halted interpretation.","A testable extension is to apply the same C18O–N2H+ analysis to a region of intermediate age, to see whether the correlation line shifts gradually or abruptly and thereby distinguish smooth CO abundance evolution from a step change."],"forward_implications":["Gas disk masses inferred from CO in 5–10 Myr old disks are not systematically biased by varying CO depletion, so the observed gas deficit is real.","The combination of declining dust mass and non-shrinking dust radius points to dust traps that hold millimeter grains at large radii, preserving pebble reservoirs for late planet formation.","The increased scatter in the CO-versus-continuum correlations in Upper Sco compared with Lupus suggests either an age spread within the region or disk-to-disk variations in CO abundance or gas-to-dust ratio.","Higher-resolution imaging should directly test the dust-trap scenario; confirmation would establish dust trapping as a primary mechanism of disk evolution in older regions."],"supporting_citations":[{"why":"Supplies the Lupus flux–flux correlations and best-fit regression lines that the Upper Sco data are compared against.","marker":"D. Deng et al. 2025b"},{"why":"Provides the chemical model linking C18O and N2H+ flux trends to CO abundance, the interpretive basis of the central claim.","marker":"L. Trapman et al. 2022"},{"why":"Visibility fitting of the Band 6 continuum yields deconvolved dust radii against which the image-plane radii are checked.","marker":"M. Vioque et al. 2025"},{"why":"Companion modeling of gas masses confirms the evolutionary trend found empirically in this paper.","marker":"L. Trapman et al. 2025b"},{"why":"Previous Upper Sco ALMA survey whose millimeter fluxes and source sample define the context for this work.","marker":"S. A. Barenfeld et al. 2016"},{"why":"The 340 GHz survey from which the Upper Sco sample was selected, providing the disk population context.","marker":"J. M. Carpenter et al. 2025"},{"why":"Earlier finding that Upper Sco disks are compact, which this paper revisits with higher sensitivity.","marker":"N. Hendler et al. 2020"}],"fun_headline_variants":["CO abundance holds steady in old disks, gas loss is real","Old disks show real gas depletion, not CO masking","C18O-N2H+ link holds: CO stable in disks to 10 Myr","Dust traps stall drift: disk radii don't shrink with age","Gas signs persist in 10 Myr disks, contradicting quick loss"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The conclusion that dust radii in Upper Sco are comparable to or larger than in Lupus, and hence that dust is trapped, assumes that the image-plane R68 radii are not systematically inflated by the beam for the many unresolved or marginally resolved disks.","fun_headline_variants_meta":{"raw":{"variants":["CO abundance holds steady in old disks, gas loss is real","Old disks show real gas depletion, not CO masking","C18O-N2H+ link holds: CO stable in disks to 10 Myr","Dust traps stall drift: disk radii don't shrink with age","Gas signs persist in 10 Myr disks, contradicting quick loss"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000693,"raw_usage":{"total_tokens":3173,"prompt_tokens":1021,"completion_tokens":2152,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":637,"completion_tokens_details":{"reasoning_tokens":2058}},"tokens_in":637,"tokens_out":2152,"duration_ms":18752,"temperature":1.0,"reasoning_tokens":2058,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T04:19:28.348885+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"If visibility fitting of the Band 6 continuum (as in Vioque et al. 2025) yields true dust radii that are substantially smaller than the image-plane R68 values, the radial-drift-halted interpretation fails. Alternatively, a thermochemical model of these specific disks that reproduces the observed C18O and N2H+ fluxes with a strongly varying CO abundance would falsify the constant-CO claim.","supporting_citations":[{"cited_title":"T., Trapman, L., et al","cited_arxiv_id":null,"evidence_quote":"Visibility fitting of the Band 6 continuum yields deconvolved dust radii against which the image-plane radii are checked."},{"cited_title":"A., Carpenter, J","cited_arxiv_id":null,"evidence_quote":"Previous Upper Sco ALMA survey whose millimeter fluxes and source sample define the context for this work."},{"cited_title":"2020, ApJ, 895, 126","cited_arxiv_id":null,"evidence_quote":"Earlier finding that Upper Sco disks are compact, which this paper revisits with higher sensitivity."}],"review_version":1}