{"id":"66621b2e-df33-4ab2-81d2-f6652e40cc0d","arxiv_id":"2412.07835","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"Ultra-strong MgII absorbers at z=0.4-1.7 are mostly hosted by low-mass, star-forming galaxies, with the absorbing gas preferentially located along the disk and wind axes, and many absorbers lacking a detected host.","lead":"This paper searches deep Subaru images and SDSS spectra for the galaxies that produce ultra-strong magnesium absorption in quasar light, finding 136 likely host galaxies out of 418 absorbing systems. It reports that these hosts are ordinary star-forming galaxies, often aligned with the galaxy disk or wind, and that many absorbers have no detected stellar counterpart.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Host identification is the load-bearing step: 86/136 hosts lack individual [O ii] and rely on permissive photo-z/SED matches; even a ~30% contamination fraction would bias the SFR/mass and wind/disk geometry conclusions.","rationale":"The reader's weakest assumption correctly identifies the photo-z-only host association as the most fragile link in the chain. The central claim has two components: that USMgII hosts are faint, star-forming, main-sequence galaxies at small impact parameters, and that the absorbing gas aligns with disk and wind axes. Both rest on the physical characterization of the 130-galaxy SED-fitted sample, of which 86 objects have no individual [O ii] detection. The paper offers a stacked [O ii] detection for this subset, but at 3.1σ it is not sufficient to certify individual associations. The geometry claim is additionally strained because it uses only N≈40 systems and reports a KS probability of 0.084, so the wording 'strong preference' overreaches, but this is downstream of the host-identification problem. I am not arguing that the direct [O ii] detections are invalid or that the paper is internally inconsistent; the 50 individually detected hosts and the full-sample stack are real supporting evidence. The issue is that the demographic and geometric conclusions are not shown to survive on the spectroscopically secure subset alone. A control-quasar false-match test plus a posterior-based association criterion would settle the matter with existing data, and until then CONDITIONAL remains the appropriate verdict.","tokens_in":20863,"tokens_out":8985,"duration_ms":92831,"concrete_test":"Run the exact host-finding pipeline (PSF subtraction, Source Extractor on coadds, BAGPIPES fitting, Δz/(1+z)≤0.15 cut) on ~100 control quasar sightlines from HSC that are not known MgII absorbers, and count how many unrelated r<26 galaxies inside 1.5″ are classified as 'hosts.' If the expected false-match rate is ≥15–20% of 86, redo the SFR/M* and azimuthal-angle analyses using only the 50 galaxies with direct [O ii] detections; if the main-sequence and disk/wind bimodality conclusions disappear, the central claim is not supported. A complementary check is to require P(|z−z_abs|/(1+z_abs)≤0.15)>0.5 in the BAGPIPES posterior instead of the binary cut and see how many of the 86 survive.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim depends on the '136 bona fide USMgII galaxies' sample, but 86 of those 136 (63%) are included without an individual [O ii] detection. They are accepted either because a BAGPIPES photo-z agrees with z_abs within Δz/(1+z)≤0.15 or because a z-fixed SED fit is flagged as good. With only 4–5 HSC bands and shallow VISTA/unWISE data, this tolerance is broad: at z≈1 it corresponds to Δz≈0.3, of order the full SED redshift degeneracy. The search radius is only 1.5″, where the prior probability of an unrelated r≲26 galaxy is non-negligible, so the host catalogue will contain some chance alignments. The paper's collective test—a 3.1σ stacked [O ii] line in the 86-object subset—does not exclude sizable contamination: if only ~40–60% of the objects are real, a stack of the real members can still reach ~3σ. All later steps (SFR, stellar mass, main-sequence offset, 21% starburst fraction, and the N≈40 major/minor-axis geometry) use this SED-fitted sample, so they inherit any contamination. The 50 direct [O ii] detections are secure, but the paper does not demonstrate that the photo-z-only subset gives the same physical answers.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper searches for host galaxies of ultra-strong Mg II absorbers (W2796 >= 3 Å) at small impact parameters (5–23 kpc) using deep Hyper Suprime-Cam imaging and SDSS spectroscopy. From 418 absorbers, the authors identify 50 galaxies through individual [O II] emission at >= 2σ and add 86 further candidates using BAGPIPES SED fits with photometric redshifts matched to the absorber redshift, or with redshift fixed to z_abs, yielding a total of 136 'bona fide' hosts. They report a ~38% detection rate, typical star-forming main-sequence host properties (21% starbursts), a preference for absorption along galaxy major and minor axes, a W2796–impact parameter relation with a wind/disk dichotomy, and an overdensity of galaxies around USMgII sightlines. The paper argues that USMgII absorbers trace disk gas and large-scale outflows, with some systems possibly hosted by unseen faint galaxies.","tokens_in":21186,"tokens_out":2917,"duration_ms":31306,"significance":"If the host identifications are reliable, this is one of the largest samples of ultra-strong Mg II absorber hosts in the small-impact-parameter regime, and the stacking analysis plus the public catalog would be valuable for future studies of CGM gas flows. The direct [O II] detections (50 systems) and the high-significance stacked [O II] signal for the full host sample are solid observational results. The paper also uses a carefully constructed control sample for the environmental overdensity measurement. However, the central claim that the SED-fitted sample of 130 galaxies represents genuine USMgII hosts rests on a photo-z matching procedure that has not been validated against a control field, and several of the physical conclusions (main-sequence offset, starburst fraction, wind/disk geometry) depend on this same SED-fitted sample. The statistical evidence for the geometric bimodality is also moderate (N≈40, KS P=0.084). With appropriate robustness tests and a clearer separation of secure and candidate hosts, the paper would make a solid contribution.","major_comments":[{"comment":"The 'bona fide' host sample includes 86 systems with individual [O II] below 2σ that are accepted based on a photo-z match within Δz/(1+z) ≤ 0.15 or a z-fixed SED fit. With only 4–5 HSC bands and sparse, shallow NIR data, this tolerance is broad (Δz ≈ 0.3 at z = 1), and the search radius of 1.5″ admits a non-negligible prior of unrelated faint galaxies. The stacked [O II] detection at 3.1σ for this subset does not exclude large contamination: even if only ~40–60% of the candidates are real, a stack of the real members could still reach a similar significance. The paper should compare the photo-z-only subset with the 50 direct detections in terms of redshift distribution, stellar mass, SFR, and azimuthal-angle distribution, and should estimate the chance-alignment rate using random quasar positions. Without this, the physical results in Sections 4.2–4.4, which use all 130 SED-fitted galaxies, may be biased by contamination.","section":"Sec. 3.3 / Table 1"},{"comment":"The claim that USMgII absorbers preferentially align along the major and minor axes rests on a small subsample (40 galaxies with ellipticity e>0.2, 38 with reliable PA) and a KS test between the 'wind' and 'disk' distributions that gives P=0.084, i.e., not significant at the usual threshold. The binning into 30-degree intervals with 30%, 52%, and 18% is suggestive but no significance level is quoted for the departure from a uniform distribution. The thresholds α≤30° and α≥50° are chosen post hoc and not varied; the paper should test the robustness of the bimodality to these thresholds and to the inclusion of the 10 additionally incorporated MEGAFLOW systems.","section":"Sec. 4.4 / Fig. 7"},{"comment":"The W2796–ρ analysis uses a literature compilation with heterogeneous selection functions and redshift ranges, and the wind/disk subsets are separated by arbitrary azimuthal-angle cuts. The quoted difference in mean W2796 (3.59±0.08 Å for wind versus 3.28±0.06 Å for disk) is small relative to the scatter, and no significance test for the difference between the two subsets is given. Moreover, because most of the USMgII galaxies in this figure come from the photo-z-selected sample, any systematic error in the host identifications propagates directly into the claimed wind/disk dichotomy. The authors should report a KS or rank-sum test between the wind and disk W2796 distributions and should repeat the fit excluding the photo-z-only candidates.","section":"Sec. 4.3 / Fig. 6"}],"minor_comments":[{"comment":"The text reports both 'about 40%' and 'a detection rate of 38%' for the host detection rate; the denominator (all HSC-covered sightlines with 4+ passbands, or those with a detected potential host) should be stated unambiguously.","section":"Sec. 4.1 / Fig. 3"},{"comment":"The phrase 'average logM⋆ of 9.90 M⊙' should read 'average log(M⋆/M⊙) = 9.90', and the quoted average SFR of 30.72 M⊙ yr−1 should be accompanied by a dispersion or uncertainty.","section":"Sec. 4.2"},{"comment":"The caption mentions 'shaded and hatched regions' but does not identify which region corresponds to the 1σ confidence interval of the main-sequence fit or to the starburst definition; please clarify.","section":"Fig. 5"},{"comment":"The summary counts are hard to reconcile: 120 systems have [O II] < 2σ among the 170 within 1.5″, yet categories IB, II, and IIIB sum to 120 only if one reads the table carefully; a brief note explaining the column logic would improve readability.","section":"Table 1"},{"comment":"The 2.8σ [O II] detection along 'clear' sightlines is marginal; the interpretation of this signal as evidence of faint or dark galaxies should be presented as tentative, especially given the possibility of extended [O II] from a galaxy group or a distant source.","section":"Sec. 3.2"}],"recommendation":"major_revision","confidential_remarks":"The paper is timely and uses a valuable dataset, but the central host-identification step is not yet demonstrated to be robust against chance alignments. I would encourage the editor to request a revision that adds a random-field control sample and a subset analysis limited to the 50 direct [O II] detections; if the physical conclusions survive those tests, the paper would be a strong candidate for acceptance. The current version, however, overstates the certainty of the 'bona fide' host sample and the geometric bimodality."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read this if you work on MgII absorbers and CGM gas flows. The paper delivers the largest census yet of ultra-strong MgII (W2796 >= 3 A) host galaxies: 136 hosts at z ~ 0.4-1.7, median impact parameter 11.4 kpc, from HSC imaging plus SDSS/BOSS spectra. The 50 hosts with direct [O II] emission are secure and useful. The PSF subtraction and SED fitting are careful, and the stacked [O II] detections give decent collective evidence that many of the photo-z candidates are real. The catalog itself is a contribution.\n\nThe soft spot is exactly where the stress-test note points. 86 of the 136 hosts (63%) are included without individual [O II]; they are accepted on a BAGPIPES photo-z match within Delta z/(1+z) <= 0.15 or a z-fixed SED fit. With 4-5 HSC bands and shallow NIR, that tolerance is broad, and the 1.5\" search radius has a non-negligible chance of an unrelated faint galaxy. The stacked [O II] at 3.1 sigma in the 86-object subset does not rule out 30-50% contamination, since a stack of the real members can still reach that. And the paper's host-property results - SFR, stellar mass, 21% starburst fraction, and the wind/disk azimuthal geometry (N ~ 40) - all use the SED-fitted sample that includes these photo-z hosts. The authors do not show that the [O II]-confirmed subset alone gives the same answers. That is a gap, not a fatal flaw.\n\nThe 'dark galaxy' reading of non-zero [O II] along clear sightlines is speculative, as they admit; the detection is 2.8 sigma. The environment overdensity result is a nice addition but also rests on photo-z matching.\n\nBottom line: this is an incremental but genuinely useful observational sample. It deserves peer review, but the referee should require a clean split between spectroscopic and photo-z hosts, and a demonstration that the physical conclusions survive on the spectroscopically confirmed subset. If you are in the field, cite the catalog; just do not lean on the geometry or starburst fraction until the contamination question is settled.","headline":"Largest USMgII host sample to date, but 63% of hosts rest on photo-z matches that the paper never shows are clean; still worth refereeing.","tokens_in":21821,"tokens_out":2180,"would_cite":true,"duration_ms":20486,"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 identifies 136 faint host galaxies for ultra-strong Mg II absorbers at small impact parameters and argues that the gas traces disks and winds in normal star-forming galaxies.","keywords":["ultra-strong Mg II absorbers","quasar absorption lines","host galaxy identification","circumgalactic medium","galactic winds","main-sequence galaxies","photometric redshifts","impact parameter"],"falsifier":"A decisive test is to obtain deep rest-frame optical spectra (targeting [O II] or H$\\alpha$) for the 86 galaxies classified as hosts only on photometric redshift and count how many show emission at the absorber redshift; if the confirmed fraction drops well below the ~80% implied by the stacked detection, chance alignment contaminates the sample and the claimed geometry results lose their footing.","tokens_in":20636,"feed_emoji":"🔭","tokens_out":12978,"duration_ms":104732,"temperature":0.7,"pith_summary":"Ultra-strong Mg II absorbers---quasar absorption systems with rest equivalent width $W_{2796} \\ge 3$ Å---are rare and thought to trace violent gas outflows, but the galaxies that make them have been hard to identify because they sit in the quasar's glare. This paper searches deep multi-band optical images around 418 such absorbers along 412 sightlines and identifies 136 bona fide host galaxies at small impact parameters (median 11.4 kpc, range 5.2--23 kpc), a detection rate of about 38%. The hosts are mostly ordinary star-forming main-sequence galaxies, with about 21% starbursts, and the absorbing gas is preferentially located along the galaxies' major and minor axes, which the authors interpret as disk gas and large-scale winds. A factor-three galaxy overdensity within 50 kpc points to group and interaction environments as an additional route to the strongest Mg II absorption. If correct, these results would place ultra-strong Mg II absorption in normal, actively star-forming galaxies at kiloparsec scales rather than exclusively in extreme starbursts.","feed_headline":"136 faint galaxies pinned as hosts of ultra-strong Mg II absorbers","feed_subtitle":"Normal star-forming galaxies at ~11 kpc, not just starbursts, make the strongest Mg II absorbers.","key_machinery":"The machinery that carries the argument is the host-identification pipeline: model and subtract the quasar point spread function in deep HSC images, stack the bands, extract sources at 3$\\sigma$ within the SDSS fiber radius, fit each candidate with the BAGPIPES SED code at or near the absorber redshift, and search the coadded SDSS spectrum for [O II] $\\lambda\\lambda3727,3729$ emission at $z_{\\rm abs}$. Individual [O II] detections at $\\ge 2\\sigma$ anchor 50 hosts; the remaining 86 are assigned by photometric redshift match within $\\Delta z/(1+z)\\le0.15$ together with a stacked [O II] detection at $3.1\\sigma$. The geometric part of the argument is carried by the azimuthal angle $\\phi$ between the quasar sightline and the galaxy major axis, measured from Sersic fits to galaxies with ellipticity $e>0.2$; splitting the sample into $\\phi<30^\\circ$ ('disk') and $\\phi>50^\\circ$ ('wind') subsets is what produces the bimodal distribution and the contrasting $W_{2796}$--$\\rho$ behavior.","core_discovery":"The paper's central claim is that a statistically significant fraction of ultra-strong Mg II absorbers are hosted by faint galaxies at small impact parameters: 136 bona fide hosts along 412 quasar sightlines, with a median impact parameter of 11.4 kpc and a 38% detection rate when at least four optical passbands are available. The hosts span stellar masses $8.65 \\le \\log M_\\star/M_\\odot \\le 11.67$ with an average star formation rate of $30.7\\,M_\\odot\\,\\mathrm{yr}^{-1}$, and they lie on the main sequence of star-forming galaxies (21% are starbursts, and all meet the SFR threshold for launching strong outflows). Geometrically, the absorbers prefer the major and minor axes of their hosts: among 40 systems with robust position angles, 52% lie within 30° of the minor axis and 30% within 30° of the major axis, and the strongest absorbers ($W_{2796}>4$ Å) all belong to the minor-axis 'wind' subset. The wind subset shows near-constant $W_{2796}$ out to $\\sim$23 kpc while the disk subset shows smaller scatter, which the paper argues explains the large scatter in the classical $W_{2796}$--impact-parameter anti-correlation. Non-zero [O II] emission along sightlines with no detected stellar counterpart further suggests that some absorbers arise in galaxies fainter than the image depth.","pith_inferences":["A statistical test the paper does not perform: place random HSC sources at the same impact parameters and ask what fraction would match the absorber redshift within $\\Delta z/(1+z)\\le0.15$; that would quantify the chance-alignment contamination in the 86 photometric-only hosts.","If the wind-subset interpretation is right, the SDSS spectra of those systems should show broader and more asymmetric Mg II profiles than the disk subset; a line-profile analysis of the existing spectra would be a cheap, immediate check.","The 'dark galaxy' candidates along clear sightlines connect naturally to the low-mass satellites that simulations predict account for a large fraction of absorbers; targeted narrow-band [O II] or integral-field searches around those quasars could find them directly.","The photo-z matching criterion itself is likely redshift- and mass-dependent, so the claimed host fractions should be re-derived in narrow redshift bins once spectroscopic confirmation is available."],"forward_implications":["If the 38% detection rate holds, the median impact parameter of 11.4 kpc makes ultra-strong Mg II absorption a kiloparsec-scale phenomenon rather than a diffuse halo phenomenon.","The main-sequence nature of the hosts means $W_{2796}\\ge3$ Å can be produced by moderate star formation, and the 21% starburst fraction bounds how much extreme star formation contributes to the class.","The bimodal azimuthal distribution implies that integral-field spectrograph surveys should see biconical outflows along the minor axes and co-rotating gas along the major axes of USMgII hosts.","The wind subset's flat $W_{2796}$ with impact parameter and the factor-three overdensity within 50 kpc require both outflows and group environments in models of the strongest Mg II absorbers.","The non-zero [O II] along 'clear' sightlines suggests a population of host galaxies fainter than $r\\sim26$ that deeper imaging or spectroscopy should reveal."],"supporting_citations":[{"why":"Supplies the parent catalog of 107,263 quasar sightlines and 159,524 Mg II absorbers from which the USMgII sample is drawn.","marker":"Anand et al. (2022)"},{"why":"Provides the $W_{2796}$--impact-parameter anti-correlation and the log-linear likelihood model used to fit the relation.","marker":"Chen et al. (2010)"},{"why":"Gives the MAGIICAT baseline and the prediction that ultra-strong absorbers should be found at $\\rho\\lesssim5$ kpc, which this sample tests and revises.","marker":"Nielsen et al. (2013)"},{"why":"Recent USMgII host-galaxy sample (28 hosts, average 19 kpc) that this paper's detection rate and impact parameters are compared against.","marker":"Guha et al. (2024)"},{"why":"Defines the main-sequence and starburst SFR--stellar-mass relations used to classify the hosts.","marker":"Bisigello et al. (2018)"},{"why":"Supplies the MEGAFLOW azimuthal-angle measurements and the minor-axis wind interpretation that the disk/wind partition extends.","marker":"Zabl et al. (2019)"},{"why":"Simulation showing that galactic winds raise cold-gas covering fraction to large radii, used to interpret the flat $W_{2796}$ of the wind subset.","marker":"Suresh et al. (2019)"},{"why":"The BAGPIPES SED-fitting machinery used to derive photometric redshifts, stellar masses, and star formation rates.","marker":"Carnall et al. (2018)"}],"fun_headline_variants":["Normal star-forming galaxies, not just starbursts, make ultra-strong Mg II","Ultra-strong Mg II absorbers trace galaxy disks and winds","136 faint galaxies host ultra-strong Mg II absorbers","Some ultra-strong Mg II absorbers hint at unseen faint galaxies"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that a galaxy seen near the quasar is genuinely the absorber host when its photometric redshift agrees with the absorber redshift within $\\Delta z/(1+z)\\le0.15$, even though 86 of the 136 hosts have no individual spectroscopic confirmation; if a large fraction of these are chance alignments, the detection rate, the host mass/SFR distributions, and the disk/wind geometry results would be biased.","fun_headline_variants_meta":{"raw":{"variants":["Normal star-forming galaxies, not just starbursts, make ultra-strong Mg II","Ultra-strong Mg II absorbers trace galaxy disks and winds","136 faint galaxies host ultra-strong Mg II absorbers","Some ultra-strong Mg II absorbers hint at unseen faint galaxies"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001788,"raw_usage":{"total_tokens":7212,"prompt_tokens":1277,"completion_tokens":5935,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":893,"completion_tokens_details":{"reasoning_tokens":5869}},"tokens_in":893,"tokens_out":5935,"duration_ms":38294,"temperature":1.0,"reasoning_tokens":5869,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T18:29:55.009296+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive test is to obtain deep rest-frame optical spectra (targeting [O II] or H$\\alpha$) for the 86 galaxies classified as hosts only on photometric redshift and count how many show emission at the absorber redshift; if the confirmed fraction drops well below the ~80% implied by the stacked detection, chance alignment contaminates the sample and the claimed geometry results lose their footing.","supporting_citations":[{"cited_title":"2022, MNRAS, 513, 3210","cited_arxiv_id":null,"evidence_quote":"Supplies the parent catalog of 107,263 quasar sightlines and 159,524 Mg II absorbers from which the USMgII sample is drawn."},{"cited_title":"K., Srianand, R., & Petitjean, P","cited_arxiv_id":null,"evidence_quote":"Recent USMgII host-galaxy sample (28 hosts, average 19 kpc) that this paper's detection rate and impact parameters are compared against."},{"cited_title":"I., Grogin, N., & Koekemoer, A","cited_arxiv_id":null,"evidence_quote":"Defines the main-sequence and starburst SFR--stellar-mass relations used to classify the hosts."}],"review_version":1}