{"id":"b27780b1-fc5f-4c3b-811d-83d9279faf7d","arxiv_id":"2607.16467","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"Compact-group galaxies have HI disks at least ~71% smaller than isolated galaxies of the same optical size, with truncation increasing along the group evolutionary sequence.","lead":"This paper measures how far atomic hydrogen gas extends in isolated galaxies and in crowded compact groups, and finds that compact-group galaxies are missing at least ~71% of their expected HI disk at a given optical size. The result puts a quantitative, bias-free number on how dense environments strip gas from galaxy outskirts.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 71% HCG deficit rests on a D_HI–D25 baseline in which 372/407 AMIGA points are inferred from the size–mass relation, not measured; the claim that 407 galaxies make the fit statistically secure is therefore misleading.","rationale":"The reader's weakest_assumption correctly identifies that the AMIGA baseline is built almost entirely from inferred D_HI values. This is the most load-bearing concern because every residual—and hence the 71% deficit—is defined relative to this baseline. If the baseline is biased, the headline number changes directly. The concern is not merely hypothetical: the resolved AMIGA sample (N=35) yields a size–mass fit that is offset from the adopted combined fit by about 1.4σ in slope and 1.6σ in intercept, and the paper's assertion that 407 galaxies make the fit statistically secure is misleading because 372 of those D_HI values are not independent measurements. The proposed test would settle whether the resolved-only baseline or a calibration without HCGs changes the median HCG residual by a meaningful amount. If it does not, the central claim is robust; if it does, the headline needs to be rephrased as conditional on the adopted size–mass relation. The reader's CONDITIONAL verdict is therefore appropriate; no adjustment is needed.","tokens_in":34552,"tokens_out":12971,"duration_ms":116055,"concrete_test":"Re-fit the AMIGA D_HI–D25 baseline two ways: (a) using only the 35 resolved AMIGA galaxies; (b) using 407 AMIGA galaxies but with the 372 inferred D_HI values recomputed from the size–mass relation fitted to MIGHTEE+Wang2016 only (excluding HCGs). For each baseline, recompute the Kaplan–Meier median residual for the 124 HCG members (73 upper limits). If the median residual shifts by more than 0.1 dex (i.e., the 71% deficit moves outside 60–75%), the inferred-D_HI baseline is load-bearing and the headline needs qualification.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The headline 71% deficit is measured relative to an AMIGA D_HI–D25 baseline (Eq. 12) in which 372 of 407 galaxies have D_HI not measured but inferred from the size–mass relation of Sect. 3.2. The text states 'With 407 galaxies, the AMIGA fit is not limited by statistics' (Sect. 4.1), but only 35 D_HI values are direct measurements; the remaining 372 are model predictions that inherit the slope/intercept of a relation fitted to a combined sample that includes HCGs (Table 3, N=727). If the size–mass relation is not universal, or if the resolved AMIGA subset is unrepresentative, the baseline's slope (0.70) and zero-point shift, and the median HCG residual (≲−0.535 dex) changes. The paper's own Table 3 shows the resolved AMIGA size–mass fit is offset from the adopted combined fit by ~1.4σ in slope and ~1.6σ in intercept; the claim that this is due to small-number statistics is plausible but not tested by a resolved-only baseline.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper measures HI disk diameters at the 1 M_sun/pc2 isophote for 35 resolved AMIGA galaxies and 51 HCG detections, adds 73 HCG nondetections as beam-size upper limits, and constructs an AMIGA D_HI-D25 baseline from 407 galaxies (35 measured plus 372 D_HI values inferred from a size-mass relation). Residuals from this baseline are used to claim that HCG galaxies have HI disks at least ~71% smaller than isolated galaxies of the same optical diameter, with truncation increasing monotonically along the HCG phase sequence, and that HCGs are statistically indistinguishable from the Virgo cluster (VIVA). The paper also compares loose groups, Hydra I subsets, and galaxy pairs, and provides a fully reproducible Snakemake/Python workflow.","tokens_in":34946,"tokens_out":5845,"duration_ms":55195,"significance":"The question is timely and the measurement strategy is mostly sound: the residual-based metric avoids the size-dependent D_HI/D25 bias, the Monte Carlo error treatment accounts for beam-correlated noise, and the survival analysis handles nondetections in a principled way. The reproducibility framework (installable package plus Snakemake workflow) is a clear strength and sets a good standard for the field. If the baseline-construction concern is resolved, the paper would provide a valuable quantitative environmental sequence from isolated galaxies to compact groups and clusters. The main scientific claim, however, depends on an AMIGA baseline that is largely built from inferred rather than directly measured D_HI values, so the headline deficit is not yet established at the level the text claims.","major_comments":[{"comment":"The AMIGA baseline is not an independent isolated-galaxy reference: 372 of the 407 D_HI values entering Eq. (12) are inferred from the size-mass relation fitted to a combined sample that includes the HCGs being compared (Table 3, N=727). The statement 'With 407 galaxies, the AMIGA fit is not limited by statistics' overstates the effective information; only 35 of those points are direct measurements. Table 3 itself shows the resolved AMIGA size-mass fit is offset from the literature by ~1.4sigma in slope and ~1.6sigma in intercept, an offset the text attributes to small-number statistics. This assumption should be tested directly: (i) fit the D_HI-D25 baseline using only the 35 resolved AMIGA galaxies; (ii) infer the 372 single-dish D_HI values using a size-mass relation fitted to MIGHTEE+Wang16 only (excluding AMIGA and HCGs); then recompute the HCG median residual. If the headline defic","section":"Sect. 3.2 / 4.1, Eq. (12), Table 3"},{"comment":"The statistical tests treat the 124 HCG members as independent, but these galaxies are clustered in 30 groups and share a common environment; the reported p-values (e.g., Gehan p=1.1e-32, Mann-Whitney p=2.3e-5) therefore overstate significance. A group-level bootstrap or cluster-robust test should be used for the HCG-vs-AMIGA comparison. In addition, the claimed monotonic phase trend is not present in the detections-only statistics: Table 7 gives median residuals of -0.116 dex (Phase 1), -0.099 dex (Phase 2), and +0.011 dex (Phase 3c), so the monotonic increase appears only after assigning upper limits (Table 8). The text in Sect. 5 partially acknowledges the Phase 3c behavior, but the abstract's unqualified 'increases monotonically' overstates what the censored medians establish, especially given that the Phase 3c/3a Kaplan-Meier medians are unconstrained upper bounds.","section":"Sect. 4.1, Tables 5-8"},{"comment":"For the pairs sample, D_HI is inferred from ALFALFA M_HI using the same size-mass relation adopted for the AMIGA single-dish galaxies. This makes the 'least truncated' ranking of Bok+20 partly tautological relative to the AMIGA baseline. The text acknowledges that this comparison probes an HI-mass deficiency expressed in diameter units, but the abstract/conclusions place pairs at the least-truncated end without this caveat. This should be either removed from the central environmental ranking or presented only as a mass-based consistency check.","section":"Sect. 4.2, Table 9, Bok+20 rows"}],"minor_comments":[{"comment":"The caption text 'HCG KM median: 0.54' should read '-0.54'; the red dashed line and the stated upper bound -0.535 indicate a missing minus sign.","section":"Fig. 6 caption"},{"comment":"The table formatting is garbled: the entries under alpha, beta, sigma_obs, rho_S, p_S, <Delta>low, <Delta>high are not cleanly separated, making some values unreadable. Please reformat for clarity.","section":"Table 4"},{"comment":"The Phase 3c statement that detected members are statistically indistinguishable from isolated galaxies (Mann-Whitney p=0.69) is based on only six detections; this low-N caveat should appear in the abstract or conclusions, not only in Sect. 5.","section":"Sect. 5 / Abstract"},{"comment":"The AMIGA sample excludes nondetections in the single-dish sample (only 'measurements' are used). If any of the 372 single-dish AMIGA galaxies are nondetections, the baseline is biased toward gas-rich systems; please clarify whether the Jones et al. (2018) values are detections only or include upper limits, and discuss the effect on the baseline.","section":"Sect. 2.1 / 4.1"},{"comment":"Minor typographical issues: 'Hinondetections' and 'Hidiameter' spacing appears inconsistently; 'phase' capitalization in Table 8 captions is uneven.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The central issue is the construction of the isolated baseline. It is fixable: a resolved-only AMIGA baseline or a literature-calibrated size-mass inference, plus a group-level resampling test, would settle whether the ~0.5 dex HCG offset is robust. If those tests confirm the offset, the paper is a strong A&A contribution; if they shift it substantially, the headline claim needs re-scoping. I would request these tests before acceptance rather than require an entirely new sample."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Your instinct on this one is about right, with one caveat: the central result is probably true, but the headline number is shakier than the prose admits.\n\nWhat the paper does well: it measures D_HI at a fixed column density for a strict isolated sample (AMIGA) and a compact-group sample, uses residuals from a D_HI-D25 baseline instead of the size-biased D_HI/D25 ratio, and handles nondetections with survival analysis. The error treatment—beam-correlated Monte Carlo for diameter uncertainties, and careful choice of fitting estimator—is solid. The detections-only offset (median -0.1 dex, p~2e-5) is statistically robust and already shows HCGs are truncated. The supplied Snakemake package is a real plus.\n\nThe soft spot is exactly where you put it: the baseline that defines 'expected' D_HI is built from 407 AMIGA galaxies, but 372 of those D_HI values are not measured; they are predicted from a size-mass relation fitted to a sample that includes the HCGs. The claim that 'with 407 galaxies the fit is not limited by statistics' papers over the fact that 91% of the baseline is model prediction. Their own Table 3 shows the resolved AMIGA fit is offset from the combined fit by 1.4-1.6 sigma; they wave this off as small-N, but they never show a resolved-only D_HI-D25 baseline. If the size-mass relation is not universal, or the resolved AMIGA subset is unrepresentative, the 71% deficit shifts. The direction won't flip—the detections-only residuals and the censored analysis both point the same way—but the accuracy of the headline is genuinely uncertain.\n\nTwo smaller issues. First, there is no morphology control. If HCGs have a different morphological mix than AMIGA at fixed D25, part of the truncation could be type-driven rather than environment-driven. Second, the 'monotonic increase' in truncation along the phase sequence is partly definitional: Phase 3a is by construction severely stripped. The detections-only median for Phase 3c is actually +0.01 dex, so the trend is less clean than the abstract implies; the paper discusses this, but the summary overstates it.\n\nOverall: a useful, reproducible measurement and a fair comparison. It deserves a serious referee. The main request should be a resolved-only baseline as a robustness test, plus a morphology-matched comparison if feasible. I would not cite the 71% number without that check, but I would cite the paper for the method and the HCG-Virgo comparison.","headline":"Central result likely right, but the headline deficit rests on a baseline that is 91% inferred, not measured.","tokens_in":35472,"tokens_out":4070,"would_cite":true,"duration_ms":35793,"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":"The paper establishes that galaxies in compact groups have atomic-hydrogen disks at least ~71% smaller than isolated galaxies of the same optical size, placing compact groups alongside the Virgo cluster as the most truncating environments.","keywords":["galaxy evolution","HI disk truncation","compact groups","isolated galaxies","HI 21 cm line","disk scaling relations","environmental processing","galaxy groups"],"falsifier":"Recompute the AMIGA baseline using only the 35 resolved isolated galaxies (or new resolved observations of the 372 single-dish galaxies) and re-measure the compact-group residuals. If the median residual moves above −0.535 dex or the baseline slope changes by more than the quoted uncertainty, the ≥71% deficit as stated is not robust.","tokens_in":34495,"feed_emoji":"📡","tokens_out":5669,"duration_ms":52154,"temperature":0.7,"pith_summary":"The paper asks how strongly the dense, interacting environment of a compact group shrinks a galaxy's atomic-hydrogen disk relative to an otherwise similar isolated galaxy. It measures HI diameters at the 1 M_sun pc^-2 contour and builds a baseline relation between HI and optical diameter from isolated galaxies, then expresses each galaxy's offset as a residual. It finds compact-group members sit far below the baseline, with a median residual implying disks at least ~71% smaller, and that the deficit grows along the group evolutionary sequence. The same baseline places compact groups at the most-truncated end, statistically indistinguishable from the Virgo cluster sample and more truncated than loose groups, cluster infall, or pairs. The analysis avoids the traditional size ratio, which is biased because HI and optical diameters scale nonlinearly.","feed_headline":"Compact groups truncate HI disks by ~71%","feed_subtitle":"Dense group members fall far below the isolated-galaxy baseline, matching Virgo cluster truncation.","key_machinery":"The load-bearing device is the residual from the isolated-galaxy baseline, Δlog D_HI = log D_HI,obs − log D_HI,exp(D_25), with the baseline fitted by Bayesian regression to 407 isolated galaxies. Because D_HI grows as D_25^0.70, a fixed ratio D_HI/D_25 would systematically vary with galaxy size; the residual-at-fixed-D_25 removes that bias. The measurement chain also includes direct ellipse fitting to the 1 M_sun pc^-2 HI isodensity contour, beam deconvolution, and Kaplan–Meier/Gehan survival statistics to incorporate nondetected members as upper limits.","core_discovery":"Using the 1 M_sun pc^-2 isophote, the authors measure D_HI by ellipse fitting and establish an isolated-galaxy baseline, log D_HI = 0.704 log D_25 + 0.691 (scatter 0.153 dex). Compact-group members sit far below this baseline, with Kaplan–Meier median residual ≲ −0.535 dex — HI disks at least ~71% smaller than isolated galaxies of the same optical diameter — and ~90% below baseline once nondetections are included as upper limits. The deficit grows monotonically along the group evolutionary sequence, and the compact-group distribution is statistically indistinguishable from the Virgo cluster sample. The HI size–mass relation is shown to be tight and environment-invariant, so the truncation re","pith_inferences":["A testable extension: apply the same baseline to resolved HI observations of the 372 single-dish isolated galaxies; if their measured diameters confirm the inferred baseline, the 71% deficit is robust.","The Phase 3c 'recently accreted' interpretation predicts that the gas-rich member's HI disk should be kinematically undisturbed and symmetrically placed; a resolved kinematic study could discriminate between accretion and survival.","The same residual framework could rank simulated galaxies from cosmological runs, providing a direct observational benchmark for tidal-stripping prescriptions in group-scale potentials.","Because more than half of compact-group members are HI nondetections, the quoted deficit is a lower bound; deeper HI observations could reveal true disks even smaller than the beam-size upper limits, pushing the deficit above 71%."],"forward_implications":["If the claim holds, compact groups are as effective as galaxy clusters at truncating HI disks, despite lacking a hot intracluster medium.","The monotonic phase 1→3 trend implies HI truncation is a progressive process tied to the group's evolutionary stage, not a single starburst or merger event.","Because the HI size–mass relation is unaffected, gas removal moves galaxies along the relation, so HI-deficiency surveys can be translated into disk-size deficits at fixed optical size.","The D_25-dependent bias in the classic D_HI/D_25 ratio means previous comparisons mixing samples of different size ranges may need re-evaluation.","Pairs of galaxies selected to be HI-bright show no truncation relative to isolated galaxies, so interaction alone is not sufficient; the group potential matters."],"fun_headline_variants":["Compact groups: HI disks 71% smaller than isolated","HI truncation in compact groups reaches ~71%","Group encounters shrink HI disks by ~71%","Compact group galaxies have HI disks ~71% smaller"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"That the HI diameter of an isolated galaxy can be reliably inferred from its HI mass through a universal size–mass relation, so that the baseline built mostly from 372 inferred diameters is unbiased; if including compact groups in the calibration or environment-dependent variations in the relation shifts those expected diameters, the 71% deficit changes.","fun_headline_variants_meta":{"raw":{"variants":["Compact groups: HI disks 71% smaller than isolated","HI truncation in compact groups reaches ~71%","Group encounters shrink HI disks by ~71%","Compact group galaxies have HI disks ~71% smaller"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000794,"raw_usage":{"total_tokens":3429,"prompt_tokens":933,"completion_tokens":2496,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":677,"completion_tokens_details":{"reasoning_tokens":2445}},"tokens_in":677,"tokens_out":2496,"duration_ms":19636,"temperature":1.0,"reasoning_tokens":2445,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T20:53:03.095103+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the AMIGA baseline using only the 35 resolved isolated galaxies (or new resolved observations of the 372 single-dish galaxies) and re-measure the compact-group residuals. If the median residual moves above −0.535 dex or the baseline slope changes by more than the quoted uncertainty, the ≥71% deficit as stated is not robust.","supporting_citations":[],"review_version":1}