{"id":"b371ef63-1ad0-426b-984b-7cdf3fb081ea","arxiv_id":"2412.05650","paper_version":3,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"A JWST NIRSpec catalog of 42 accreting pre-main-sequence stars in NGC 3603 shows accretion rates higher, and ages older, than typical low-mass star-forming regions.","lead":"Using JWST spectroscopy, researchers classified 42 young stars in the massive cluster NGC 3603 as actively accreting gas from their disks. These stars appear to keep accreting for 10 to 20 million years, longer and at higher rates than stars in quieter regions, suggesting the dense UV-bright environment alters planet formation.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 'higher accretion at fixed age and mass' claim is not demonstrated: raw Mdot-age and Mdot-mass comparisons do not control for both variables simultaneously, and NGC 3603's mass-rich sample can mimic the reported offset.","rationale":"The reader's weakest_assumption targeted isochronal age systematics, but as the paper itself notes (Secs. 7.6 and 7.7), the known systematic shifts move the old sources to even older ages, so the existence of >=10 Myr accreting stars is a conservative lower limit and is not weakened by those effects. The more serious threat to the headline claim is the joint mass-age comparison: the paper compares Mdot versus mass (Fig. 11) and Mdot versus age (Fig. 12) separately, while Eq. 7 shows a steep mass dependence. Because NGC 3603 is more massive-rich than typical low-mass SFR samples, the reported offset in the Mdot-age plane could arise from the mass imbalance alone. The paper does not present a mass-normalized comparison or a simultaneous fit including the comparison sample, so the claim 'systematically higher for a given stellar age and mass' is not directly evidenced. This is a well-defined, fixable gap in the analysis, and the paper's otherwise transparent pipeline and careful nebular subtraction deserve credit. The appropriate verdict remains conditional pending this quantitative check, matching the reader's overall assessment.","tokens_in":39478,"tokens_out":9599,"duration_ms":94368,"concrete_test":"Obtain the individual stellar mass, age, and Mdot measurements for a low-mass SFR comparison sample (e.g., Lupus or Chamaeleon data, or the Trumpler 37 sources of Sicilia-Aguilar et al. 2005). Compute the mass-normalized accretion rate Mdot / M*^1.68 using the exponent from Eq. 7 and plot it against age for both samples. If the NGC 3603 points remain offset in this mass-normalized plane, the 'systematically higher for a given age and mass' claim is supported; if the samples overlap, the offset seen in Fig. 12 is a mass-distribution artifact rather than evidence of environmental enhancement.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The strongest claim includes both old accreting stars (>=10 Myr) and accretion rates 'systematically higher for a given stellar age and mass' than in low-mass SFRs. The age part is actually robust to the paper's own documented systematics: MIST age underestimation (Sec. 7.7) and the B10 extinction alternative (Sec. 7.6) both move the old sources to older, not younger, ages. The load-bearing weakness is the 'given age and mass' comparison. Figures 11 and 12 compare raw Mdot versus M* (against the D11 relation) and raw Mdot versus age (against the Sicilia-Aguilar relation), yet Mdot depends steeply on both variables: Eq. 7 gives log Mdot = 1.682 log M* - 0.795 log t - 2.731. NGC 3603's sample contains many 2-7 solar-mass sources (the largest symbols in Fig. 12), whereas the low-mass SFR comparison samples are predominantly below 2 solar masses. A mass-rich sample will lie above a low-mass sample in the Mdot-age plane even if the underlying Mdot(M*, t) law is identical. The abstract's 'for a given stellar age and mass' is therefore not directly demonstrated: no figure or statistical test controls for both variables at once, and the literature comparisons also use different isochrones, extinction laws, and Lacc calibrations. This is a testable gap rather than an internal inconsistency, but it is the weakest load-bearing point in the central claim.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper presents JWST NIRSpec MSA spectroscopy of 100 stars in NGC 3603, of which 42 are classified as pre-main-sequence stars with active accretion based on hydrogen recombination lines. The authors use an MCMC spectral fitting procedure with Phoenix models and HST photometry to derive effective temperatures, extinctions, veiling, and stellar parameters; ages and masses are obtained from MIST isochrones. Accretion luminosities are derived from Brγ (or Paα) line fluxes using literature calibrations, and an updated Lacc–LPaα relation is presented. The headline results are: twelve accreting sources with ages consistent with ≥10 Myr (four with ≥15 Myr), mass accretion rates spanning five orders of magnitude and 'systematically higher for a given stellar age and mass' than in low-mass star-forming regions, and a positive correlation between residual accretion rate and ambient nebular H2 emission.","tokens_in":39876,"tokens_out":9283,"duration_ms":78632,"significance":"If the central claims hold, the paper provides one of the first spectroscopic demonstrations that active disk accretion can persist for longer than ~10 Myr in a massive, UV-hostile starburst cluster, with elevated accretion rates and an environmental correlation with molecular gas density. The work also delivers a valuable catalogue: 42 spectroscopically classified PMS stars with MCMC-derived stellar parameters, extinction, veiling, and Monte Carlo line-flux uncertainties. The data reduction is notably careful: the scaled nebular subtraction is tested against He I residuals, the residual nebular contamination is quantified (≈2% to −8% in Paα), and optimal extraction is validated. The updated Lacc–LPaα relation with a factor ~2.5 reduction in coefficient uncertainties is a useful community resource. These strengths, however, do not by themselves establish the headline comparison claims, which rest on unquantified selection effects, heterogeneous literature calibrations, and a bivariate claim that is not directly tested.","major_comments":[{"comment":"The abstract and Section 6.3 state that twelve accreting sources have ages consistent with ≥10 Myr and four with ≥15 Myr. However, the best-fit ages listed in Table 2 give only nine sources with ages ≥10 Myr (1339, 1717, 1876, 4104, 1981, 1852, 1854, 1497, 1813) and only two with ages ≥15 Myr (1497, 1813). If 'consistent with' instead means that the 1σ uncertainty interval overlaps the threshold, the criterion is not stated and the counts change (e.g., 1436, 1530, and 2880 have intervals containing 10 Myr). Because the population of old accretors is a central claim, please state the exact criterion used, report the correct counts, and show the individual age posteriors or a table of the relevant sources.","section":"§6.3 and Table 2"},{"comment":"The claim that accretion rates are 'systematically higher for a given stellar age and mass' is not demonstrated. Figure 11 compares log Mdot vs log M* with the Donehew & Brittain (2011) relation, and Figure 12 compares log Mdot vs age with the Sicilia-Aguilar et al. (2005) relation, but neither controls for both variables simultaneously. Since Equation (7) gives log Mdot = 1.682 log M* − 0.795 log t − 2.731, a sample skewed toward higher masses (NGC 3603 contains many 2–7 M⊙ sources, while the comparison samples are predominantly below 2 M⊙) will appear above a low-mass sample in the Mdot–age plane even if the underlying Mdot(M*, t) law is identical. Please test the offset directly, for example by computing residuals from the multivariate relation or by comparing a mass-matched subsample, and report the significance of any residual offset.","section":"§6.5, Figures 11–12"},{"comment":"The comparison samples are not homogeneous with respect to the quantities being compared. The D11 relation combines Herbig AeBe stars, intermediate-mass T Tauri stars, and CTTSs, while the Sicilia-Aguilar sample includes the massive SFR Cepheus OB2; both use different accretion tracers (Brγ, Hα), different Lacc calibrations, different extinction laws, and different isochrones. The paper does not quantify how these systematic differences propagate into the claimed offset in Mdot at fixed age and mass. Without an assessment of these systematics, the environmental interpretation remains vulnerable to calibration-driven shifts.","section":"§6.5 and §7.6"},{"comment":"The claimed correlation between the residual accretion rate and nebular H2 emission is presented visually, but no statistical test is reported. Given that the sample size is 42 and multiple environmental variables (Hα, H2, radial distance) were examined, a Spearman rank correlation with a p-value (or an equivalent test) is needed to establish the significance of the H2 trend and to rule out a chance finding. The normalization of the axes in Figure 13 is also not defined in the text.","section":"§6.6 and Figure 13"}],"minor_comments":[{"comment":"The wording of the central comparison claim is inconsistent: the abstract says 'systematically higher for a given stellar age and mass', §6.5 says 'for a given mass' (Figure 11), and conclusion 10 says 'for a given age'. Please harmonize the wording with what is actually tested.","section":"Abstract and §6.5"},{"comment":"The fit in Equation (6) uses all sources including the low-mass end where the text states there is no apparent trend; consider fitting separate relations for M* < 1 M⊙ and M* ≥ 1 M⊙, or stating why a single power law is appropriate.","section":"§6.5, Equation (6)"},{"comment":"For the five continuum sources whose properties were derived from SED fitting alone, the comparison with MUSE spectra (Section 5.2.3) shows Teff underestimates of order 1000 K for source 152; please state explicitly how this systematic affects ages and masses of the SED-fitted sources and whether it influences the old-age population.","section":"§5.2 and Table 2"},{"comment":"The note says 'a number of sources shows net absorption line profiles for Br7' and that these used Equation (8) and Paα, but in Table C.1 only source 152 is flagged with '−−' for Brγ. Please clarify which sources actually used the Paα-based calibration.","section":"Appendix C, Table C.1"},{"comment":"The spectral index α is measured over 1.7–3.0 μm, which is a narrower range than the 1–10 μm normally used for Class I/II/III classification; the caveat in the text is appropriate, but consider also presenting the distribution of α for the restricted range to avoid over-interpretation of the Class III fraction.","section":"§6.2"},{"comment":"The discussion of the lack of circumstellar H2 would benefit from a quantitative upper limit on H2 line flux or EW for a representative source, rather than the purely qualitative statement that no H2 is detected.","section":"§7.4"}],"recommendation":"major_revision","confidential_remarks":"The paper is technically solid in its reduction and catalogue, but the headline claims need substantial tightening before acceptance. The age-count inconsistency in §6.3/Table 2 is a straightforward numerical issue that must be fixed. The more important gap is that the 'given age and mass' comparison is not actually performed; this is the pillar of the abstract's environmental claim. The authors should be asked either to provide a bivariate comparison or to soften the claim accordingly. I would also encourage the editor to seek a referee with specific expertise in accretion-rate calibration systematics, since the comparison samples are heterogeneous in ways that are not trivial to correct."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis is a solid observational catalog paper: first NIRSpec MSA spectra of PMS stars in NGC 3603, 42 accretors classified and characterized, with transparent MCMC fitting, Monte Carlo line fluxes, and a quantified nebular subtraction residual (2–8%). The expanded sample over their own previous work and the re-fit Lacc–LPaα relation are incremental but legitimate. The paper is honest about its systematics: Sections 7.6 and 7.7 explicitly concede that isochronal ages may be underestimated by a factor of ~2 and that the B10 extinction correction pushes the oldest sources to ~50 Myr. Both corrections make the old ages older, not younger, so the core claim of a ≥10 Myr accreting population actually survives the paper's own caveats. That is worth emphasizing because it is the most interesting result.\n\nThe load-bearing weakness is the 'systematically higher for a given stellar age and mass' claim. Figures 11 and 12 plot raw Mdot versus M* and versus age separately, but Mdot depends steeply on both (Eq. 7: log Mdot = 1.682 log M* − 0.795 log t − 2.731). The NGC 3603 sample is mass-rich (many 2–7 Msun sources) compared to the low-mass SFR comparison samples, so a mass-rich sample can sit above a low-mass sample in the Mdot–age plane even with an identical Mdot(M*, t) law. No figure or test controls for both variables simultaneously. This is a testable gap, not an internal contradiction, but it is exactly the part of the abstract that is not demonstrated.\n\nThe H2 environmental correlation is intriguing but driven by a few sources in the pillars; the authors themselves note the direction of causality is unclear. Minor point: the circularity score is low because nearly all Mdot uses external calibrations; the self-derived relation is used for one source, so that is not a real weakness.\n\nWho is this for? Anyone working on disk evolution in massive clusters, proplyd survival, or accretion diagnostics. It deserves a serious referee—the data are valuable and the claims are important enough to scrutinize. I would accept it for review, but the referee should push for a joint Mdot(M*, age) analysis or an equivalent statistical control before the elevated-accretion claim is accepted.\n\nVerdict: conditional acceptance with requested revisions; the catalog itself is publishable regardless.","headline":"A careful, honest NIRSpec catalog of 42 accreting PMS stars in NGC 3603, with a plausible but not fully demonstrated claim of elevated accretion rates at fixed age and mass.","tokens_in":40431,"tokens_out":656,"would_cite":true,"duration_ms":8276,"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":"JWST NIRSpec spectra show 42 stars in NGC 3603 are actively accreting, including twelve with ages of at least 10 Myr and four at 15 Myr or more.","keywords":["NGC 3603","pre-main-sequence stars","mass accretion rates","protoplanetary disks","external photoevaporation","JWST NIRSpec","star formation","H2 emission"],"falsifier":"Take the four sources with nominal ages of at least 15 Myr and date them with an independent clock, such as lithium depletion, kinematics, or binary orbit. If they turn out to be significantly younger than 10 Myr, the long-lived accretion claim fails. As a second test, repeat the H2–$\\dot{M}_{\\rm acc}$ correlation on a larger sample with a molecular gas tracer that does not depend on the same nebular subtraction; if the correlation disappears, the environmental claim fails.","tokens_in":39272,"feed_emoji":"⭐","tokens_out":9971,"duration_ms":88515,"temperature":0.7,"pith_summary":"This paper uses JWST NIRSpec multi-object spectroscopy to build the first detailed catalogue of 100 stars in NGC 3603, the Milky Way's nearest analogue to a starburst cluster, and identifies 42 as pre-main-sequence stars that are still actively accreting gas from their disks. Its central claim is that accretion is not shut off by the cluster's extreme ultraviolet radiation: twelve of these stars have ages consistent with at least 10 Myr, four with at least 15 Myr, well beyond the usual 5–10 Myr disk dispersal timescales. The paper further claims that their mass accretion rates span five orders of magnitude and are systematically higher for a given stellar mass and age than in nearby low-mass star-forming regions, and that the excess accretion tracks the density of ambient molecular gas traced by nebular H2 emission. If correct, this means the environments where most stars actually form can keep disks alive and accreting far longer than the standard picture allows, with the surrounding gas supply playing a central role in how planets grow.","feed_headline":"JWST finds 10-20 Myr old stars still accreting in NGC 3603","feed_subtitle":"Spectra of 42 pre-main-sequence stars show elevated accretion rates that track the surrounding molecular gas density.","key_machinery":"The argument is carried by four linked pieces. (1) A scaled nebular-subtraction scheme uses the He I 1.869 $\\mu$m doublet as a brightness reference, scaling the subtraction until He I vanishes and thereby recovering the stellar hydrogen lines from a bright, spatially variable nebula. (2) An MCMC spectral-fitting routine matches each source against Phoenix stellar atmosphere models reddened with the G23 extinction curve and veiled by a single blackbody component, yielding effective temperature, extinction, and veiling; five line-poor continuum sources are instead fitted to their optical SEDs. (3) Stars are placed on the Hertzsprung-Russell diagram with MIST isochrones to get masses, ages, and radii. (4) Br$\\gamma$ (or Pa$\\alpha$ for sources where Br$\\gamma$ is in absorption) line luminosity is converted to accretion luminosity with the A17 and D11 calibrations, then to $\\dot{M}_{\\rm acc} \\sim 1.25\\,L_{\\rm acc} R_*/(G M_*)$; a multivariate fit removes the mass and age dependence before the residuals are compared with nebular H$_2$, H$\\alpha$, and distance.","core_discovery":"The discovery is a population of old, actively accreting pre-main-sequence stars in a massive, UV-dominated cluster, plus a spatial correlation between how fast a star accretes and how much molecular gas surrounds it. Of 100 NIRSpec spectra, 42 show hydrogen recombination lines (Paα, Brγ, Brβ) above the chromospheric level after nebular subtraction, confirming ongoing magnetospheric accretion. Placing these stars on the Hertzsprung-Russell diagram with MIST isochrones yields masses from 0.5 to 7 $M_\\odot$ and a tail of ages at 10–20 Myr; accretion rates inferred from Brγ and Paα line luminosities span five orders of magnitude and, at fixed stellar mass and age, exceed those of comparison samples in low-mass star-forming regions. After fitting and removing the joint dependence of $\\dot{M}_{\\rm acc}$ on mass and age, the residuals correlate positively with nebular H2 brightness, suggesting late infall or shielding by dense molecular gas sustains accretion; no such correlation appears with ionized Hα or with radial distance from the cluster centre.","pith_inferences":["If the 10–20 Myr ages survive independent checks, then disk-lifetime estimates based on near-infrared continuum excess systematically underestimate disk longevity in massive clusters, since most of the accretors in this sample show no JHK excess.","The H2–$\\dot{M}_{\\rm acc}$ correlation, if causal, points to dense gas replenishing or shielding disks; a direct test would compare disk gas masses and radii, for example with submillimetre interferometry, between sources in the dense pillars and those in the low-density central bubble.","The steep $\\dot{M}_{\\rm acc}$–mass relation implies intermediate-mass stars in such clusters assemble their disks faster, which would shorten the time window for forming giant planet cores before disk dispersal.","The updated $L_{\\rm acc}$–Pa$\\alpha$ relation offers a route to measure accretion in even more crowded or more distant clusters using Pa$\\alpha$ alone, where Br$\\gamma$ is weaker or absorbed."],"forward_implications":["Active gas accretion persists in NGC 3603 for at least 10 Myr, with four sources at 15 Myr or more, under an ultraviolet field orders of magnitude stronger than in Orion.","At fixed stellar mass and age, $\\dot{M}_{\\rm acc}$ values are systematically higher than in the comparison low-mass star-forming regions, and the decline of $\\dot{M}_{\\rm acc}$ with age is shallower.","After removing the mass and age dependence, the residual accretion rate correlates positively with nebular H$_2$ brightness; no such correlation appears with nebular H$\\alpha$ or with distance from the cluster centre.","Most accreting sources (26 of 42) are Class III, with no detectable JHK excess, so surveys that identify disks by near-infrared continuum excess alone will miss most active accretors at these ages.","The older accreting sources lie at larger projected distances from the cluster centre than the younger ones, a separation supported by a K-S test and bootstrap confidence intervals."],"supporting_citations":[{"why":"Supplies the MIST isochrones and evolutionary tracks used to convert stellar temperature and luminosity into masses and ages.","marker":"Choi et al. (2016)"},{"why":"Provides the calibration converting Br$\\gamma$ line luminosity to accretion luminosity for classical T Tauri stars.","marker":"Alcalá et al. (2017)"},{"why":"Provides the analogous Br$\\gamma$-to-accretion-luminosity calibration for Herbig AeBe stars and the comparison accretion-rate–mass relation.","marker":"Donehew & Brittain (2011)"},{"why":"Supplies the formula $\\dot{M}_{\\rm acc} \\sim 1.25\\,L_{\\rm acc} R_*/(G M_*)$ that converts accretion luminosity to mass accretion rate.","marker":"Gullbring et al. (1998)"},{"why":"The HST photometric survey from which the 100 targets were drawn and which first suggested an old accreting population in NGC 3603.","marker":"Beccari et al. (2010)"},{"why":"Provides the comparison $\\dot{M}_{\\rm acc}$–age relation from several star-forming regions and the precedent of old, strongly accreting stars in Cepheus OB2.","marker":"Sicilia-Aguilar et al. (2005)"},{"why":"Reports the correlation between accretion rate and ambient gas density in Lupus that the H2 correlation in NGC 3603 is compared with.","marker":"Winter et al. (2024)"},{"why":"Supplies the G23 extinction curve used to deredden the spectra and photometry before fitting.","marker":"Gordon et al. (2023)"}],"fun_headline_variants":["Old stars in NGC 3603 accrete faster in denser gas","JWST: ancient stars still accreting in NGC 3603","NGC 3603: 10-20 Myr stars still munching on gas","Gas-rich surroundings keep old stars accreting in NGC 3603"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The ages of the stars, and therefore the claims of 10–20 Myr accretion and of slower accretion decline, rest entirely on the isochronal age scale and the adopted extinction curve; the paper itself states that a different extinction correction moves the nominal 15 Myr sources to about 50 Myr.","fun_headline_variants_meta":{"raw":{"variants":["Old stars in NGC 3603 accrete faster in denser gas","JWST: ancient stars still accreting in NGC 3603","NGC 3603: 10-20 Myr stars still munching on gas","Gas-rich surroundings keep old stars accreting in NGC 3603"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000538,"raw_usage":{"total_tokens":2656,"prompt_tokens":1094,"completion_tokens":1562,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":710,"completion_tokens_details":{"reasoning_tokens":1478}},"tokens_in":710,"tokens_out":1562,"duration_ms":13250,"temperature":1.0,"reasoning_tokens":1478,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T20:30:21.808793+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the four sources with nominal ages of at least 15 Myr and date them with an independent clock, such as lithium depletion, kinematics, or binary orbit. If they turn out to be significantly younger than 10 Myr, the long-lived accretion claim fails. As a second test, repeat the H2–$\\dot{M}_{\\rm acc}$ correlation on a larger sample with a molecular gas tracer that does not depend on the same nebular subtraction; if the correlation disappears, the environmental claim fails.","supporting_citations":[{"cited_title":"W., Hern \\'a ndez, J., Briceno, C., & Calvet, N","cited_arxiv_id":null,"evidence_quote":"Provides the comparison $\\dot{M}_{\\rm acc}$–age relation from several star-forming regions and the precedent of old, strongly accreting stars in Cepheus OB2."},{"cited_title":"J., Benisty, M., Manara, C","cited_arxiv_id":null,"evidence_quote":"Reports the correlation between accretion rate and ambient gas density in Lupus that the H2 correlation in NGC 3603 is compared with."}],"review_version":1}