{"id":"8b6f9d05-a8bb-4f61-91a6-afd18c4440c0","arxiv_id":"2412.02255","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"In IllustrisTNG, bars in massive disk galaxies are destroyed by mergers, while in lower-mass galaxies bars fail to form because the disks are too extended and dynamically hot.","lead":"Using the IllustrisTNG galaxy simulations, this study finds that massive disk galaxies lose their bars mainly because of galaxy mergers, while lower-mass disk galaxies remain unbarred because their disks are too diffuse and dynamically hot to form bars. The result helps explain why only some local disk galaxies have bars, and it highlights what the simulations get wrong about the centers of galaxies.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The low-mass branch of the central claim rests on a z=1 Toomre-Q offset that may be caused by bars already present at z=1, not by pre-existing disk responsiveness.","rationale":"The paper is a careful statistical study with real strengths: it uses two large public simulations, two independent bar measurement methods, KS tests over many parameters, and explicit merger histories. The massive-galaxy branch (mergers suppress bars) is supported by convergent evidence from fhalo, fex situ, and major-merger fractions, so I do not recommend rejection. The load-bearing weakness is in the lower-mass branch, which is half of the central claim. The z=1 Q2comp comparison in Fig. 11 is presented as evidence for a causal role of disk responsiveness, but the companion panels in Figs. 8 and 10 show that bar fractions at z=1 are already about 0.5 for all final classes. Since bars modify velocity dispersion and hence Q, a snapshot at z=1 cannot separate 'high Q prevents bar formation' from 'bar formation, or its absence, changes Q.' The compactness argument is also weakened because Re at z=0 is measured after the evolution being explained. This does not mean the conclusion is wrong; it means the causal arrow is unproven. The proposed test at z=2 or on the unbarred-at-z=1 subsample would settle it. This is an added condition, not a rejection, so I keep the reader's CONDITIONAL verdict.","tokens_in":33722,"tokens_out":8679,"duration_ms":98684,"concrete_test":"Using the public TNG merger trees, recompute Q2comp at z=2 for the same z=0-selected disk galaxies in the log(M*/M_sun)=10.6-10.8 TNG100 bin and the 10.0-10.6 TNG50 bins, before most bars have formed (downsizing: Fig. 14 bottom). Also, for galaxies that are unbarred at z=1, compare Q2comp at z=1 between those that become barred by z=0 and those that remain unbarred. If the final-barred/final-unbarred Q offset persists in the pre-bar or unbarred-at-z=1 subsample, the causal interpretation survives; if it disappears or reverses, the z=1 Q offset is a consequence of bar formation and the nature branch must be reworked.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Section 6 uses Eq. (3) (Romeo-Wiegert two-component Q) and Fig. 11 to argue that z=0 unbarred galaxies below log(M*/M_sun)<10.8 have higher Q2comp at z=1, interpreting this as a lack of responsiveness that prevents bar formation. The problem is that Figs. 8 and 10 show the bar fraction is already ~0.5 at t_LB=8 Gyr for barred, short-bar, and unbarred final samples; many of the final-barred galaxies already have bars before z=1. Bars heat the stellar disk, raising sigma_R and redistributing angular momentum, so Q* = sigma_R*kappa/(pi G Sigma) measured at z=1 is not a pristine initial condition. The lower Q2comp of the final barred sample could be a selection effect: the low-Q tail formed bars early and then modified Q, while the high-Q tail remained unbarred. The compactness measure (Re at z=0) is also an endpoint of the same evolutionary path. Thus the causal statement that a larger Toomre-Q 'generates' unbarred low-mass galaxies is not established by the z=1 snapshot; the data are consistent with Q being a consequence of the divergence the paper seeks to explain.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper uses ellipse fitting and Fourier decomposition on TNG100 and TNG50 disk galaxies to compare barred, short-bar, and unbarred galaxies at z=0, tracing their evolution back to z=1. The central claims are that all z=0 disk galaxy progenitors show similar bar fractions and strengths at z=1; that for log(M*/M_sun)>10.8, mergers suppress or destroy bars; and that for log(M*/M_sun)<10.8, a larger two-component Toomre-Q parameter associated with a less compact nature prevents bar formation. The analysis also addresses the properties and evolution of short bars, which are found to resemble unbarred galaxies more than barred ones. The methodology combines KS tests of galaxy parameters, evolutionary tracks of individual galaxies, and comparisons between TNG100 and TNG50.","tokens_in":34011,"tokens_out":5754,"duration_ms":68564,"significance":"If correct, the paper offers a mass-dependent physical explanation for which disk galaxies end up barred, with mergers dominating in massive systems and internal disk responsiveness dominating in lower-mass systems. The authors are careful in several respects: they cross-check ellipse-fitting and Fourier bar measurements, they make the bar evolution catalog publicly available, they validate TNG100 results with TNG50, and they openly flag the TNG50 central ellipticity problem and the discrepancy with observed galaxy sizes. I do not see a definitional circularity: the bar properties and the galaxy parameters are measured independently. The main risks are the causal interpretation of the Toomre-Q comparison at z=1, when a large fraction of the final barred galaxies already have bars, and the reliance on an ad hoc TNG50 bar-size correction for the low-mass branch of the argument.","major_comments":[{"comment":"The claim that a larger Toomre-Q at z=1 'generates' unbarred low-mass galaxies is not established by the present analysis. At z=1 the bar fraction is already about 0.5 for all final samples (Figs. 8 and 10), so the Q2comp profiles in Fig. 11 are measured after a sizable fraction of the final barred galaxies have already formed bars. Bars modify the stellar velocity dispersion and surface density, so Q2comp at z=1 is not a pristine initial condition; the lower Q2comp of the final barred sample may be a selection effect or a consequence of early bar formation rather than a pre-existing lack of responsiveness. The paper should measure Q2comp before bar formation, for example at z=2 or for galaxies that have no bar at z=1, or should explicitly show that the offset survives when all galaxies with bars at z=1 are removed. This is load-bearing for the low-mass branch of the central claim.","section":"§6, Eq. (3), Fig. 11"},{"comment":"The TNG50 bar measurements rely on an ad hoc 1.4 kpc inner cutoff for long bars, while short bars are measured using all radii, and the underlying central ellipticity enhancement in TNG50 is left unexplained. Because TNG50 is the only simulation used for the log(M*/M_sun)<10.6 mass range and for the left-hand panels of Fig. 11, a systematic misclassification of short bars or spurious detections caused by central ellipticity could bias the low-mass conclusions. The authors should quantify the sensitivity of their bar fractions and size measurements in TNG50 to this cutoff, for example by using Fourier-based bar sizes as an alternative, and should investigate the physical origin of the central ellipticity enhancement rather than treating it only as a fitting artifact.","section":"§2.2.2, Figs. 1-2"},{"comment":"The premise that all z=0 disk galaxy progenitors have 'similar bar features' at z=1 (abstract and Section 8, item 1) is asserted from overlapping medians without statistical quantification. No KS test, bootstrap confidence interval, or other uncertainty estimate is given for fbar, A2max, or Rbar at tLB=8 Gyr, and the sample is defined by z=0 populations traced through merger trees, whose completeness is not reported. Since the subsequent nature/nurture interpretation assumes equal initial bar conditions, the authors should provide a quantitative comparison of the z=1 bar properties and report the progenitor identification completeness. This is needed to make the 'similar at z=1' claim falsifiable.","section":"§4.2, §5.2, Figs. 8 and 10"}],"minor_comments":[{"comment":"There is a typo in 'Other subgrid physical processes may alse influence the size of bars'; 'alse' should be 'also'.","section":"§7.2"},{"comment":"'Enterpreneurship' should be 'Entrepreneurship' in the acknowledgement of the XMU Training Program.","section":"Acknowledgements"},{"comment":"The sentence 'Additionally, we also study the evolution of both the Toomre-Q' is incomplete; it should read 'both the stellar and gaseous Toomre-Q parameters' or similar.","section":"§8"},{"comment":"The text in §3.2 refers to p-values while the caption says the bracketed number gives the log p-value; please clarify consistently so readers do not misinterpret the reported significance levels.","section":"Fig. 4 caption and §3.2"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is squarely within the scope of A&A and I see no citation or attribution problems. The main issue is not circularity but causal overreach in the Toomre-Q interpretation and the fragility of the TNG50 bar measurements. Both are fixable within the manuscript's scope, hence major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a careful, mostly convincing analysis of bar formation in TNG100 and TNG50, and it will be useful to anyone who works on bar statistics in simulations. The main contribution is the explicit comparison of evolutionary histories—bar fraction, strength, size, merger rates, and two-component Toomre-Q—for z=0 barred, short-bar, and unbarred galaxies split by stellar mass. The result that all three groups have statistically indistinguishable bar properties at z=1 is clean and striking, and the merger-driven destruction story for high-mass galaxies is well supported by major-merger histories and f_halo/ex-situ fractions. The short-bar analysis is a useful addition, and the public data release is a real asset.\n\nThe soft spots are mostly about framing rather than the underlying measurements. The stress-test concern about Toomre-Q at z=1 is real but not fatal. Because roughly half of each final sample already has a bar at z=1, the Q measured there is not a pristine initial condition; bars heat the disk and redistribute angular momentum, so Q can be altered by the very process the paper seeks to predict. That said, the final-barred and final-unbarred samples have similar bar fractions at z=1, so the Q offset is not simply a consequence of one group being more barred at that lookback time. The causal direction remains genuinely unclear, and the abstract's phrase 'generating unbarred disk galaxies' overstates what a z=1 snapshot can show. The paper would be stronger if it acknowledged this explicitly and either measured Q at higher redshift or restricted the causal claim to a subsample that is bar-free at z=1.\n\nOther issues: the TNG50 1.4 kpc inner cutoff for bar sizes is ad hoc, though the agreement with Fourier decomposition is reassuring. The abstract's '60% in the local Universe' and 'all conclusions validated with TNG50' overreach—the statistics are from TNG100, and TNG50 has fewer, smaller galaxies. The 'similar at z=1' claim also lacks quantified uncertainty. And the many KS tests without multiple-testing corrections are fine for exploratory work, but the bolded p-values should be read with caution.\n\nWho is this for? People working on bar fractions, the short-bar overproduction problem, or the physics of bar destruction in cosmological simulations. It deserves a serious referee: the analysis is sound enough to publish after revisions that tighten the causal language and the TNG50 correction. Send it to review.","headline":"A careful statistical study of TNG bars with a plausible mass-dependent nature/nurture story, but the low-mass Toomre-Q claim is more correlational than the text admits.","tokens_in":34534,"tokens_out":4797,"would_cite":true,"duration_ms":52108,"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":"Mergers suppress or destroy bars in massive disk galaxies, while extended, dynamically hot disks prevent bars in lighter ones.","keywords":["galactic bars","disk galaxies","IllustrisTNG","galaxy mergers","Toomre Q parameter","bar fraction","secular evolution","cosmological hydrodynamical simulations"],"falsifier":"Use a higher-resolution simulation (or a successor TNG run with smaller softening, plus an inner-cutoff-free bar finder) and measure bar fraction and strength at $z=1$ for the same $z=0$-selected barred and unbarred populations; if the progenitors already differ in bar properties at $z=1$, the similar-at-$z=1$ claim fails. Separately, if a substantial population of massive unbarred galaxies with merger histories as quiet as those of barred galaxies turns up in independent simulations or in observed close-pair and tidal-debris counts, the claim that mergers dominate bar destruction above $\\log(M_*/M_\\odot)>10.8$ would be falsified.","tokens_in":33501,"feed_emoji":"🌌","tokens_out":7501,"duration_ms":74250,"temperature":0.7,"pith_summary":"The paper asks why some disk galaxies in the local Universe have stellar bars and others do not, using the IllustrisTNG cosmological simulations to trace the same galaxies back in time. It claims the answer is mass-dependent: above $\\log(M_*/M_\\odot)\\sim 10.8$, mergers are the dominant factor, heating or destroying bars, while below that mass, a more extended, dynamically hotter disk (higher Toomre-Q parameter) simply fails to respond to bar instabilities. A striking intermediate result is that the $z=0$ barred, short-bar, and unbarred disk galaxies all look similar in bar fraction and strength at $z=1$, so their different fates are decided by what happens after that epoch. This matters because reproducing the observed bar fraction is a demanding test of whether cosmological simulations get internal galaxy dynamics right.","feed_headline":"Mergers destroy bars in massive galaxies, hot disks in smaller ones","feed_subtitle":"Every local disk looks alike at z=1 in TNG; afterwards mergers and disk temperature set the bar fate.","key_machinery":"The argument is carried by a diagnostic toolkit rather than a single identity. Bars are identified by isophotal ellipse fitting (ellipticity $\\epsilon$ peak, position-angle constancy, outer drop, with bar size $R_{85}$) and cross-checked by Fourier $A_2$ profiles; galaxies are followed through time with subhalo merger trees; and the difference between barred and unbarred populations is assessed by Kolmogorov-Smirnov tests on parameters that include ex-situ stellar fraction, kinematically derived stellar-halo fraction, effective radius, and rotation. The physical probe of disk responsiveness is the two-component Toomre parameter $Q_{\\rm 2comp}$ from the Romeo-Wiegert approximation, which combines stellar and gaseous velocity dispersions and surface densities; the paper uses its radial profile at $z=1$, $z=0.5$, $z=0.2$, and $z=0$ to show that unbarred low-mass disks are dynamically hotter than their barred counterparts.","core_discovery":"Using ellipse fitting and Fourier $A_2$ measurements on TNG100 and TNG50 disk galaxies, the paper finds that bar presence evolves after $z\\sim1$ and is governed by different physics on either side of $\\log(M_*/M_\\odot)\\approx10.8$. In massive systems, unbarred galaxies have experienced far more major mergers since $z\\sim1$ and carry larger stellar halos and ex-situ mass fractions; about 60% of the local unbarred population once hosted a bar that was destroyed. In lower-mass systems, mergers are minor, and unbarred galaxies are instead the extended, less compact disks with higher two-component Toomre-Q values, where bar instability is suppressed even when rotation is high; barred galaxies follow a denser, earlier-assembling compact pathway. Short bars form at about the same epoch as normal bars but either contract or barely grow, and the simulations overproduce them, indicating that the central regions of TNG galaxies are insufficiently affected by mergers and gas inflows.","pith_inferences":["If bar survival in TNG tracks merger rate, the measured high-mass bar fraction could be turned around as an observational constraint on merger rates, by comparing TNG's bar fractions against merger-rate measurements from close pairs or tidal features.","The claim that about 60% of unbarred galaxies once hosted a bar implies that unbarred samples at $z=0$ are a mix of destroyed-bar and never-barred objects; future observational work could try to split them using kinematic or metallicity relics of a past bar.","The resolution-dependent difference in short-bar abundance suggests a testable numerical prediction: a higher-resolution rerun of TNG50 or zoom-in simulations with smaller softening should produce longer bars and a bar fraction closer to S4G.","The correlation between compactness and bar presence links bar formation to angular-momentum-driven mass-size relations, suggesting bar fraction could be predicted from halo spin and assembly time rather than from local disk conditions alone."],"forward_implications":["At the massive end, the $z=0$ bar fraction becomes a probe of recent merger history: disks with quiet merger histories stay barred, so a deficit of bars implies a violent assembly.","Since all $z=0$ disk populations look alike at $z=1$, the early formation stage does not preordain bar presence in this mass range; the discriminating physics happens between $z=1$ and $z=0$.","Low-mass unbarred galaxies need not have lost a bar; many are simply disks too extended and dynamically hot to go unstable, so environment and halo spin, not just mergers, set the unbarred population.","Short bars are a distinct population closer to unbarred galaxies in properties; measurements of the bar fraction in simulations should either exclude them or classify them separately to match observed S4G fractions.","The overproduction of short bars and the TNG50/TNG100 differences in central density point to resolution and subgrid effects in galaxy centers as limiting the fidelity of simulated bar sizes."],"supporting_citations":[{"why":"Supplies the TNG100 disk galaxy catalog and the ellipse-fitting bar identification criteria and R85 size definition the analysis builds on.","marker":"Zhao et al. (2020)"},{"why":"Provides the ex-situ stellar mass fractions used to quantify merger influence on bar survival.","marker":"Rodriguez-Gomez et al. (2016)"},{"why":"Provides the kinematic decomposition (auto-GMM) giving the stellar halo fraction fhalo used as a merger/nurture indicator.","marker":"Du et al. (2020)"},{"why":"Supplies the S4G bar fraction versus mass relation used to compare TNG bar fractions and identify excess short bars.","marker":"Erwin (2018)"},{"why":"Supplies the bar size-stellar mass relation and 20th percentile used to classify short bars.","marker":"Erwin (2019)"},{"why":"Defines the ellipticity and position-angle criteria adopted for bar identification.","marker":"Martinez-Valpuesta et al. (2006)"},{"why":"Gives the critical Toomre-Q around 2.2 for bar formation cited in interpreting disk responsiveness.","marker":"Athanassoula & Sellwood (1986)"},{"why":"Provides the two-component Toomre-Q approximation used to measure disk dynamical temperature.","marker":"Romeo & Wiegert (2011)"},{"why":"Documents that TNG50 bars are about 35% shorter than MaNGA bars, the comparison behind the short-bar overproduction claim.","marker":"Frankel et al. (2022)"},{"why":"Provides the A2/phase criteria and the downsizing trend in bar formation used for comparison.","marker":"Anderson et al. (2024)"}],"fun_headline_variants":["Mergers kill bars in big galaxies; hot disks in small ones","Bar formation in galaxies: mergers vs disk temperature","TNG overproduces short bars: central regions too calm","Mergers destroy bars in massive disks; hot disks suppress them"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The central assumption is that bar detection at $z=1$ and in TNG50, including the ad hoc 1.4 kpc inner cutoff motivated by an unexplained central ellipticity enhancement, does not systematically misclassify bars, and that the merger trees correctly recover the $z=0$ galaxies' progenitors.","fun_headline_variants_meta":{"raw":{"variants":["Mergers kill bars in big galaxies; hot disks in small ones","Bar formation in galaxies: mergers vs disk temperature","TNG overproduces short bars: central regions too calm","Mergers destroy bars in massive disks; hot disks suppress them"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000647,"raw_usage":{"total_tokens":3044,"prompt_tokens":1089,"completion_tokens":1955,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":705,"completion_tokens_details":{"reasoning_tokens":1884}},"tokens_in":705,"tokens_out":1955,"duration_ms":15683,"temperature":1.0,"reasoning_tokens":1884,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T23:39:54.263965+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Use a higher-resolution simulation (or a successor TNG run with smaller softening, plus an inner-cutoff-free bar finder) and measure bar fraction and strength at $z=1$ for the same $z=0$-selected barred and unbarred populations; if the progenitors already differ in bar properties at $z=1$, the similar-at-$z=1$ claim fails. Separately, if a substantial population of massive unbarred galaxies with merger histories as quiet as those of barred galaxies turns up in independent simulations or in observed close-pair and tidal-debris counts, the claim that mergers dominate bar destruction above $\\log(M_*/M_\\odot)>10.8$ would be falsified.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the two-component Toomre-Q approximation used to measure disk dynamical temperature."}],"review_version":1}