{"id":"46955913-221c-4bfc-b76e-dcbd91071ea2","arxiv_id":"2504.13103","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Galaxy groups with active nuclei show a broad trough in cold hydrogen gas at halo masses around 10^11.8 solar masses, below where theoretical models expect an AGN-driven dip.","lead":"Using archived X-ray images stacked on thousands of nearby galaxy groups, this paper traces how hydrogen gas and hot X-ray gas change with group size, assembly stage, and active black holes. It reports a shallow dip in cold gas content at lower group masses than theory predicts, tied to groups containing active galactic nuclei.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"AGN-HI valley claim rests on unpublished AGN catalog; inverse-causality/selection bias (Sec. 5.3.1) is untested because the analysis does not control for FSMGR at fixed Mhalo.","rationale":"The reader identified the same core vulnerability I would: the AGN versus non-AGN comparison assumes the AGN catalog is not biased by the very quantity (HI content / FSMGR) being compared. The paper is unusually careful, presenting the inverse-causality scenario explicitly (Sec. 5.3.1) and noting they have not controlled for FSMGR (Sec. 5.3.3), so this is not an oversight the authors are hiding. Still, it is the load-bearing point: the abstract's 'AGN-hosting halos exhibit a valley' becomes evidence for AGN feedback only if the valley is not a selection artifact. The unpublished status of the AGN catalog and the reliance on the novel SF-AGN class mean no independent check of the selection function is possible. I agree with the reader's conditional verdict; my proposed fixed-FSMGR comparison is exactly the experiment that would settle the interpretation. I do not think the concern rises to rejection because the measurement of reduced HI in AGN-hosting halos is likely real in a statistical sense; the question is whether it reflects AGN physics or detection bias. Since the reader's verdict already conditions on this, I recommend UNCHANGED.","tokens_in":32535,"tokens_out":5659,"duration_ms":52172,"concrete_test":"Re-run the Figure 8 analysis in narrow bins of both Mhalo and FSMGRgrp (e.g., 0.3 dex in Mhalo and ~0.2 dex in FSMGR), comparing median MHI,grp/Mhalo for AGN-hosting vs non-AGN-hosting halos. If the HI offset vanishes at fixed FSMGR, the valley is consistent with the detectability-bias/inverse-causality scenario of Sec. 5.3.1 and cannot be claimed as AGN-driven. If a 0.2-0.25 dex offset persists, the concern is mitigated. As a second check, inject the Polimera et al. AGN selection function into mock group catalogs with no AGN-HI coupling and verify whether the observed valley depth and mass location can be reproduced by selection alone.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that AGN-hosting halos carve a ~0.25 dex valley in MHI,grp/Mhalo below Mhalo ~ 10^12.1 Msun depends on the AGN inventory being complete and unbiased with respect to HI content and FSMGR at fixed Mhalo. The paper itself flags the inverse-causality alternative in Sec. 5.3.1: reduced star formation may make AGN easier to detect, so the valley could be a selection effect rather than AGN-driven gas removal. Because the AGN catalog (M.S. Polimera et al. 2025, in prep.) is unpublished and the AGN selection includes the new SF-AGN class identified via optical line diagnostics, there is no external way to assess the selection function. The >5-sigma KS difference between AGN-hosting and non-AGN-hosting halos is consistent with either physical interpretation; it does not validate the valley as intrinsic. Figure 8 also shows a parallel valley in FSMGRgrp, and since FSMGR and HI content are tightly correlated (Kannappan et al. 2013), the HI valley may simply track the FSMGR deficit. The paper states in Sec. 5.3.3 that it has not replicated Figure 8 in fixed bins of Mhalo and FSMGRgrp together. That missing control is the load-bearing gap: without it, the claim that AGN presence itself, rather than the SF state that enables AGN detection, is associated with depressed HI is not established.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper combines the volume-limited RESOLVE and ECO surveys with the G3 group catalog to study drivers of group-integrated HI-to-halo mass ratio and X-ray emission. The authors stack archival ROSAT All-Sky Survey images in bins of halo mass, crossing time, AGN presence, and group-integrated fractional stellar mass growth rate, using random sky stacks and multiple X-ray binary estimators as controls. They report that hot gas is confidently detected for halos of 10^12.6–10^14 Msun and unambiguously above 10^14 Msun, that lower-crossing-time groups show lower HI content at fixed halo mass, and that AGN-hosting halos below ~10^12.1 Msun exhibit a broad ~0.25 dex valley in M_HI,grp/M_halo relative to non-AGN-hosting halos. Diluted by non-AGN halos, this produces a shallower valley in the full sample near 10^11.8 Msun, below the location of the theoretically predicted dip; a possible additional valley near 10^13 Msun is also discussed. The paper interprets the low-mass valley as evidence for AGN-associated HI suppression at lower masses than current semi-analytic models predict.","tokens_in":32822,"tokens_out":2989,"duration_ms":30325,"significance":"If the central claim holds, the paper provides a new observational constraint on AGN feedback in the dwarf-to-group regime, a mass range where such constraints are rare and where the paper's SF-AGN inventory is unusually complete. The analysis is careful in several respects: the X-ray stacking includes random sky-position control stacks processed identically to the data, X-ray binary contamination is assessed with 13 estimators, errors on medians are bootstrapped, and the crossing-time split is explicitly checked against alternative boundary values. The paper also gives a clear, falsifiable prediction: the valley in the M_HI,grp/M_halo relation should be traceable to the AGN-hosting subpopulation, with a deeper valley than in the full sample. The main weakness is that the central AGN comparison rests on an unpublished AGN catalog and on a control (fixed M_halo and FSMGR_grp jointly) that the authors state they have not performed; until that control is supplied, the AGN valley could be a selection effect rather than a physical AGN-driven depletion.","major_comments":[{"comment":"The paper explicitly states in §5.3.3 that the Figure 8 analysis has not been replicated in fixed bins of M_halo and FSMGR_grp together. This is a load-bearing missing control: because M_HI,grp/M_halo and FSMGR_grp are tightly correlated (as the paper itself notes from Kannappan et al. 2013), the observed HI valley for AGN-hosting halos may simply track the reduced FSMGR_grp of those halos, which in turn may be what makes AGN detectable in the first place. Without this joint control, the claim that AGN presence itself, rather than the star-formation state that enables AGN detection, is associated with depressed HI is not established. The authors should add the fixed-M_halo, fixed-FSMGR_grp comparison, or explicitly quantify the residual HI offset after matching on FSMGR_grp.","section":"§5.3.3, Fig. 8"},{"comment":"The reported >5 sigma two-sample Kolmogorov-Smirnov significance for the difference between AGN-hosting and non-AGN-hosting halos is not computed at fixed M_halo. The KS test is sensitive to differences in the overall M_halo distributions of the two samples; given the strong dependence of M_HI,grp/M_halo on M_halo, this significance does not by itself demonstrate a mass-independent or mass-matched HI offset. The valley 'a' significance should be reassessed with a stratified or matched test (e.g., fixed M_halo bins or propensity-score matching) to support the claim that the valley is intrinsic to AGN-hosting halos.","section":"§4.3, Fig. 8"},{"comment":"The 2.5 sigma significance quoted for the shallow valley 'b' in the full sample is computed relative to a baseline that is itself fit to the same data, namely the analytic model of Obuljen et al. (2019) with free parameters logM0, logMmin, and alpha fit to the median M_HI,grp/M_halo values. This is not an independent null hypothesis, and the significance estimate does not account for the uncertainty in the fitted baseline parameters. The authors should provide a more conservative significance estimate, for example from bootstrap resampling of the data and refitting the baseline each time, or from forward-modeled mock catalogs.","section":"§4.3, Fig. 11, valley 'b'"},{"comment":"The central AGN-HI comparison depends on the completeness and unbiasedness of the AGN inventory with respect to HI content and FSMGR at fixed halo mass. The AGN catalog is described as 'M.S. Polimera et al. 2025, in prep.' and is not yet public, and the selection function of the new SF-AGN class is not quantified here. The paper itself raises the inverse-causality scenario in §5.3.1: reduced star formation may make AGN easier to detect, which would produce the same observed HI valley without AGN-driven gas removal. Because the analysis does not control for FSMGR at fixed M_halo (see the first major comment), this alternative remains viable. The authors should either provide the AGN catalog or a detailed selection-function description, and they should show that the HI valley persists after matching or controlling for FSMGR_grp.","section":"§2.1.1, §5.3.1"}],"minor_comments":[{"comment":"The paper reports that the M_halo = 10^12.6–10^13.3 Msun bin shows enhanced X-ray emission for higher-tcross groups, contrary to the trend at higher masses, but the AGN-masked SNR in that bin is only ~2.2 and consistent with XRB expectations; this should be stated more prominently as a marginal result in the main text, not just in the figure description.","section":"§4.2, Fig. 7"},{"comment":"The label and text alternate between 'FSMGR_grp' and 'FSMGRgrp' with inconsistent spacing; please standardize the notation throughout, including in the axis labels.","section":"§4.3, Fig. 8, middle panel"},{"comment":"The X-ray stacking for AGN-hosting halos includes only three mass bins, and the lowest bin shows no significant detection; the text should explicitly note the low statistical power in that bin rather than implying a monotonic trend across all three bins.","section":"§4.3, Fig. 9"},{"comment":"The PSF percentile statement '97% of the sources identified by the sigma-clipping algorithm can be enclosed by such an aperture' would benefit from a clearer statement of whether the source radii refer to the segmentation-map equivalent radii and how the masking aperture choice affects the final count rates in low-mass stacks.","section":"§3.1, Fig. 1"}],"recommendation":"major_revision","confidential_remarks":"The headline AGN-valley result cannot be fully evaluated until the Polimera et al. (2025, in prep.) AGN catalog is released or its selection function is described in sufficient detail. Given the paper's own admission in §5.3.3 that the M_halo-plus-FSMGR_grp control is missing, I recommend requiring that control before acceptance. The X-ray stacking analysis itself appears methodologically sound and could be published independently, but the AGN-related conclusions need the additional controls."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Zack, quick take on arXiv:2504.13103. The paper has one genuinely new result: a ~0.25 dex valley in MHI,grp/Mhalo for AGN-hosting halos below Mhalo ~ 10^12.1, which shows up as a shallower ~0.1 dex valley in the full sample near 10^11.8. That sits below where SAMs (Baugh et al., Chauhan et al.) put their predicted AGN feedback trough, so it is a direct challenge to those models. The X-ray stacking is careful: random sky controls, 13 XRB estimators, bootstrapping, and explicit robustness checks on the crossing time split. The main X-ray detections of hot gas above 10^12.6 look solid, and the inverse HI/X-ray trend with halo mass is cleanly presented.\n\nThe soft spots are real but mostly acknowledged by the authors. The all-halo valley is only 2.5 sigma against a baseline fit to the same data, so the headline significance is modest. More importantly, the key AGN versus non-AGN split rests on an AGN catalog that is still in prep, and the authors explicitly note in Sec 5.3.1 that reduced star formation could make AGN easier to detect, producing the same HI valley without any AGN-driven gas removal. They also admit in Sec 5.3.3 they have not controlled for FSMGR at fixed Mhalo. Given how tightly FSMGR and HI content correlate, that missing control is the load-bearing gap. The >5 sigma KS difference between AGN and non-AGN halos is consistent with either physical interpretation; it does not by itself validate the valley as intrinsic. None of this is hidden — the discussion is honest about the ambiguity.\n\nI think the paper deserves a serious referee. The measurement, if it holds up, is a step forward in pinning where AGN feedback actually matters in the HI-halo mass relation, and the comparison to theory is useful even if the physical origin remains open. Conditional acceptance is the right call. I would ask the authors to release the group catalogs, stacked maps, and the AGN catalog when available, and to attempt the fixed-Mhalo, fixed-FSMGR control if possible. Deeper X-ray data or mock-catalog tests would also help firm up the valley. The self-citation is fine; the G3 catalog and Polimera AGN work are the natural references.","headline":"A careful stacking analysis that plausibly pins an AGN-associated HI valley to low halo masses, though the central causal interpretation rests on an unpublished AGN catalog and an untested FSMGR control.","tokens_in":33473,"tokens_out":1907,"would_cite":true,"duration_ms":16914,"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":"The long-sought AGN-driven dip in galaxy-group H I sits at lower halo mass than models predict.","keywords":["galaxy groups","neutral hydrogen","active galactic nuclei","X-ray emission","halo mass","AGN feedback","galaxy evolution","virialization"],"falsifier":"Measure the $M_{\\rm HI,grp}/M_{\\rm halo}$ gap using only AGN selected by X-ray luminosity or broad-line emission, whose detectability does not depend on H I content; if the 0.25 dex valley shrinks to noise at fixed $M_{\\rm halo}$ below $10^{12.1}\\,M_\\odot$, the valley is a selection effect rather than AGN feedback.","tokens_in":32301,"feed_emoji":"🔭","tokens_out":10038,"duration_ms":85391,"temperature":0.7,"pith_summary":"This paper tries to establish that the long-sought dip in the relation between galaxy-group atomic hydrogen and halo mass is real, but sits at lower halo masses than theory predicts, and is driven by halos hosting active galactic nuclei (AGN). Using a nearly complete, volume-limited census of local galaxy groups, the authors find that below $M_{\\rm halo}\\sim10^{12.1}\\,M_\\odot$, AGN-hosting halos have roughly 0.25 dex less atomic hydrogen per unit halo mass than non-AGN halos at the same mass, together with lower fractional stellar mass growth. Diluted by the majority of non-AGN halos, this produces a shallow valley near $M_{\\rm halo}\\sim10^{11.8}\\,M_\\odot$, below the $10^{12.1}\\,M_\\odot$ bimodality scale where current semi-analytic models place an AGN-feedback dip. The paper also links H I content inversely to stacked X-ray hot gas emission and to virialization state, with the H I-to-halo-mass ratio spreading markedly for crossing times below about 2 Gyr. If correct, the results imply that AGN activity removes cold gas from dwarf-group halos, a population that theoretical models have not yet incorporated.","feed_headline":"AGN mark a hidden gas valley in low-mass galaxy groups","feed_subtitle":"Haloes with active black holes host 0.25 dex less hydrogen near 10^11.8 solar masses, lower than models predict.","key_machinery":"The central object is the group-integrated H I-to-halo mass ratio $M_{\\rm HI,grp}/M_{\\rm halo}$ measured as a function of group halo mass, split by AGN presence, crossing time, and fractional stellar mass growth. The argument is carried by a nearly complete AGN inventory that reaches dwarf galaxies through optical emission-line diagnostics and mid-IR color diagnostics, by the ${\\rm FSMGR_{grp}}$ metric (the ratio of stellar mass formed in the past Gyr to preexisting stellar mass, summed over group members), and by $t_{\\rm cross}$ (the mean projected transverse distance divided by the mean line-of-sight velocity, approximating virialization state). On the X-ray side, the key mechanism is stacking of archival all-sky X-ray images with source masking, random-position control stacks, and a range of X-ray binary background estimates used to isolate hot gas emission.","core_discovery":"The central claim is that AGN-hosting halos below $M_{\\rm halo}\\sim10^{12.1}\\,M_\\odot$ carry a broad, roughly 0.25 dex deep valley in $M_{\\rm HI,grp}/M_{\\rm halo}$ relative to non-AGN-hosting halos at fixed halo mass, with a matching reduction in group-integrated fractional stellar mass growth. Because the fraction of halos containing AGN crosses 50% near $M_{\\rm halo}\\sim10^{11.7}\\,M_\\odot$, the dilution of this deficit over all halos yields a narrow, shallower valley centered near $M_{\\rm halo}\\sim10^{11.8}\\,M_\\odot$. The paper interprets this as the physical origin of the dip in the overall $M_{\\rm HI,grp}/M_{\\rm halo}$ versus $M_{\\rm halo}$ relation, and emphasizes that both this valley and a possible second one near $M_{\\rm halo}\\sim10^{13}\\,M_\\odot$ are shallower and located at lower masses than the dips predicted by current semi-analytic models. It further claims that stacked X-ray data detect hot gas confidently at $M_{\\rm halo}=10^{12.6-14}\\,M_\\odot$ and unambiguously above $10^{14}\\,M_\\odot$, while the H I-to-halo mass ratio drops by about 1.1-1.2 dex over the same range.","pith_inferences":["If the valley reflects genuine AGN feedback, the same survey should show an excess of warmer or molecular gas phases in AGN-hosting dwarf halos if gas is being heated rather than expelled; this is testable with molecular gas observations.","The possible second valley near $10^{13}\\,M_\\odot$ could mark a transition to a different AGN feedback mode; that hypothesis needs AGN subtype classifications (radio, optical, X-ray) that the present inventory does not provide.","Because crossing time is projection-sensitive, the $t_{\\rm cross}\\sim2$ Gyr transition should be checked against virialization metrics insensitive to merging, such as the magnitude gap; disagreement between metrics would point to projection effects rather than physical virialization."],"forward_implications":["The shallow valley near $M_{\\rm halo}\\sim10^{11.8}\\,M_\\odot$ in the all-halo relation is the diluted signature of a deeper deficit among AGN-hosting halos, giving the long-sought dip a physical origin rather than a binning artifact.","Current semi-analytic predictions of an AGN-feedback dip at or above $10^{12.1}\\,M_\\odot$ will need revision to include low-metallicity, actively star-forming dwarf AGN populations and their feedback.","The inverse relation between H I fraction and stacked X-ray emission outlines a cold-to-hot gas handover: hot gas becomes detectable around $10^{12.6}\\,M_\\odot$ and dominates by $10^{14}\\,M_\\odot$.","The wider H I scatter below $t_{\\rm cross}\\sim2$ Gyr indicates that virialization or merger state, not halo mass alone, regulates cold gas in small groups.","AGN-hosting low-mass halos show reduced fractional stellar mass growth as well as reduced H I, tying black-hole activity to suppressed recent star formation in the same halos."],"supporting_citations":[{"why":"Supplies the G3 group catalog, halo masses, and H I mass estimates for every group, so the measurement of $M_{\\rm HI,grp}/M_{\\rm halo}$ rests on it.","marker":"H23"},{"why":"Defines the dwarf AGN census including the SF-AGN class used to split halos into AGN-hosting and non-AGN-hosting samples.","marker":"Polimera et al. 2022"},{"why":"Provides the FSMGR stellar mass growth metric used to show that AGN-hosting low-mass halos also have reduced recent star formation.","marker":"Kannappan et al. 2013"},{"why":"One of the semi-analytic models predicting an AGN feedback dip in the H I-halo mass relation at higher masses; the observed valley is compared against it.","marker":"Baugh et al. 2019"},{"why":"The other main theoretical prediction of the AGN-driven dip; the paper argues it does not match the observed lower-mass valley.","marker":"Chauhan et al. 2020"},{"why":"Supplies the X-ray stacking method and hot-gas detection framework used to measure X-ray emission in halo mass bins.","marker":"Anderson et al. 2015"},{"why":"Provides the preferred X-ray binary scaling relation used to subtract non-hot-gas backgrounds and identify hot gas at high halo mass.","marker":"Lehmer et al. 2019"}],"fun_headline_variants":["AGN create hidden gas valley in small galaxy groups","Black holes carve 0.25 dex gas hole in low-mass halos","AGN-hosting halos show HI deficit, models miss it","Gas-rich groups dip where AGN dominate","Active galactic nuclei dent HI content in small groups"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The AGN inventory must be as complete in gas-rich, star-forming low-mass halos as in gas-poor ones; if low star formation simply makes AGN easier to detect, the H I valley would appear even without AGN-driven gas removal.","fun_headline_variants_meta":{"raw":{"variants":["AGN create hidden gas valley in small galaxy groups","Black holes carve 0.25 dex gas hole in low-mass halos","AGN-hosting halos show HI deficit, models miss it","Gas-rich groups dip where AGN dominate","Active galactic nuclei dent HI content in small groups"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00026,"raw_usage":{"total_tokens":1814,"prompt_tokens":1391,"completion_tokens":423,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":1007,"completion_tokens_details":{"reasoning_tokens":342}},"tokens_in":1007,"tokens_out":423,"duration_ms":4669,"temperature":1.0,"reasoning_tokens":342,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T12:14:27.527431+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the $M_{\\rm HI,grp}/M_{\\rm halo}$ gap using only AGN selected by X-ray luminosity or broad-line emission, whose detectability does not depend on H I content; if the 0.25 dex valley shrinks to noise at fixed $M_{\\rm halo}$ below $10^{12.1}\\,M_\\odot$, the valley is a selection effect rather than AGN feedback.","supporting_citations":[{"cited_title":"J., Stark, D","cited_arxiv_id":null,"evidence_quote":"Provides the FSMGR stellar mass growth metric used to show that AGN-hosting low-mass halos also have reduced recent star formation."}],"review_version":1}