{"id":"3b20455c-7ec9-418c-a9ca-99dbfa4dc3a3","arxiv_id":"1908.02331","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"PCA-based stellar masses for MaNGA galaxies run about 0.1 dex higher than photometric masses, and luminosity-weighting resolved mass-to-light ratios biases integrated stellar masses low, predominantly due to patchy dust attenuation.","lead":"This paper calibrates PCA-based stellar mass estimates from the MaNGA survey against kinematic and photometric data, and quantifies how unresolved galaxy spectra bias total stellar mass. It will release an aperture-corrected stellar mass catalog and shows that uneven dust extinction, not just young bright stars, drives the bias from spatially unresolved measurements.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Section 4.3's mass deficit is defined with a luminosity-weighted proxy for coadded-spectrum M/L, not with actual spectral coadding; the central claim about differential dust attenuation depends on this unvalidated proxy.","rationale":"The reader's weakest assumption identifies exactly the same load-bearing issue: the paper equates the mass inferred from a galaxy-coadded spectrum with the luminosity-weighted average of resolved PCA M/L values without actually performing the spectral coadding and refitting. This is not a minor methodological footnote; it defines the dependent variable in the regression that supports the paper's principal new physical conclusion. I agree with the reader that the analysis would be substantially strengthened by directly fitting coadded MaNGA spectra. The paper is otherwise careful: the aperture-correction comparison is clearly framed, the dynamical comparison is explicitly exploratory and limited to three galaxies, and the authors warn against overinterpreting the OLS regression because predictor uncertainties are omitted. Those omissions are secondary to the proxy issue. The reader's CONDITIONAL verdict already captures this concern, so no verdict adjustment is needed; the recommendation is to require the direct coadding test (or a convincing linearity demonstration from the Paper I fitting procedure) before promoting the differential-attenuation conclusion to a firm result. The proposed concrete test is feasible with existing MaNGA data and the authors' own PCA pipeline, and it would settle whether the proxy is accurate across the relevant range of inclination and dust dispersion.","tokens_in":19281,"tokens_out":2930,"duration_ms":37146,"concrete_test":"Select a stratified subsample including all 20 galaxies in Table 3 plus random controls across the axis-ratio range. For each galaxy, construct a galaxy-coadded spectrum by summing or inverse-variance-weighting the cleaned spaxels, rerun the Paper I PCA fit on this coadded spectrum to obtain log Υ*_coadd, and compute M_coadd = L_total × Υ*_coadd. Compare log(M*/M_coadd) with the proxy log(M*/M_LW) used in Figure 7 and Table 2. If the two agree to within roughly 0.02 dex across the sample, the proxy is safe; if the discrepancy correlates with στV or b/a, rerun the OLS regression of Eq. 4 using the direct coadd masses as the response and check whether β1 still clearly dominates and excludes zero.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that differential dust attenuation (στV) dominates the luminosity-weighting mass deficit rests entirely on the regression in Section 4.3 (Eq. 4, Table 2), whose response variable is log(M*/M_LW). M_LW is constructed by luminosity-weighting the resolved PCA mass-to-light ratios and multiplying by total IFU luminosity; the paper never coadds the MaNGA spaxel spectra and reruns the PCA fit. This substitution is exact only if the PCA M/L estimator is a strictly linear functional of the input spectrum. In practice, the estimator involves PCA projection with covariate uncertainties and a nonlinear mapping from PC scores to M/L, so the mass inferred from a true coadded spectrum need not equal the luminosity-weighted average of spaxel M/L values. If the nonlinearity is signal-to-noise dependent or correlates with dust and inclination, the mass deficit could be an artifact of the proxy rather than a physical luminosity-weighting effect. The only external support invoked, Ibarra-Medel et al. (2019), uses simulations with a different spatial-binning scheme and does not validate this particular proxy. Because the headline conclusion that differential attenuation is dominant is derived from regressing this proxy quantity on predictors from the same PCA fits, an unvalidated proxy is load-bearing for the paper's main new result.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper (Paper II of a series) presents applications of PCA-based stellar mass-to-light ratio estimates to resolved MaNGA spectroscopy. It compares PCA-derived stellar mass surface densities with DiskMass dynamical surface densities for three galaxies, finding that moderate overestimates of disk scale-heights (by a factor of roughly 1.5) would reconcile the two for two of the three galaxies. It constructs a total stellar mass catalog using two aperture-correction methods (the 'ring' method and a color-based CMLR method), recommends the CMLR method, and compares the resulting masses with NSA and JHU-MPA photometric masses, finding typical discrepancies of about 0.1 dex. The paper's most novel result is an analysis of luminosity-weighting biases: comparing IFU-summed masses with masses obtained by luminosity-weighting the resolved mass-to-light ratios, the authors find mass deficits that are largest for edge-on systems and systems with dust lanes, and a multivariate regression attributes the deficit primarily to the spatial dispersion of V-band optical depth (differential attenuation) rather than to average attenuation or central star formation. The paper also describes a planned SDSS Value-Added Catalog.","tokens_in":19497,"tokens_out":5747,"duration_ms":66726,"significance":"If validated, this is a useful contribution for the MaNGA community and for resolved-spectroscopy mass estimation more broadly. The external comparisons to kinematical measurements and to independently derived photometric masses provide helpful anchoring for the PCA method, and the paper is transparent about the main caveats, including the DiskMass scale-height systematics and the absence of uncertainties in the regression predictors. The luminosity-weighting analysis is the most novel element, and the paper correctly identifies a potential source of systematic error in unresolved or spatially binned mass estimates. However, the central claim about differential dust attenuation rests on a proxy whose equivalence to a true coadded-spectrum measurement is not directly demonstrated; this is a load-bearing assumption that must be validated or substantially de-emphasized before the main result can be fully accepted.","major_comments":[{"comment":"The response variable in the central regression is log(M*/M_LW), but M_LW is not obtained by coadding the MaNGA spaxel spectra and rerunning the PCA fitting, as the 'galaxy-coadded spectrum' framing implies. It is instead the total IFU luminosity multiplied by the luminosity-weighted average of the resolved PCA mass-to-light ratios. This replacement is exact only if the PCA-based M/L estimator is a linear functional of the input spectrum; PCA projection with covariate uncertainties followed by a nonlinear mapping from PCA scores to M/L is not such a functional. If the nonlinearity has a signal-to-noise dependence or correlates with dust geometry, the measured mass deficit, and hence the conclusion that sigma_tauV dominates, could be an artifact of the proxy. The external support cited (Ibarra-Medel et al. 2019) uses simulations with a different spatial-binning scheme and does not validate this specific proxy. I request a direct validation: coadd the spectra (or realistic synthetic spectra) and rerun the full PCA fit, or alternatively demonstrate with synthetic MaNGA-like observations that the luminosity-weighted proxy reproduces the coadded-spectrum result to within the claimed 0.05–0.1 dex.","section":"Section 4.3, Eq. (4), Table 2"},{"comment":"The ordinary least-squares fit in Eq. (4) uses predictors (tauV, sigma_tauV, eta) that are derived from the same PCA fits that produce the response variable, and the uncertainties on these predictors are not propagated into the reported confidence intervals or p-values. The authors note this caveat, but because the claim that differential attenuation is the dominant factor is one of the paper's main new results, the significance should be checked for robustness to correlated errors, for example by Monte Carlo sampling the PCA posteriors or by performing a split-sample test by inclination and star-formation activity. As presented, the p-values in Table 2 likely overstate the confidence in the beta1 coefficient.","section":"Section 4.3, Table 2"}],"minor_comments":[{"comment":"The abstract contains a typo ('galaxys' should be 'galaxy's'); similar minor typographical errors appear elsewhere in the text.","section":"Abstract"},{"comment":"The two Gallazzi et al. entries for 2005a and 2005b are identical in the reference list; one of these citations is likely intended to refer to a different article.","section":"References"},{"comment":"The abstract's statement that the DiskMass tension 'could be resolved if the disk scale-heights were overestimated by a factor of ~1.5' is only supported by two of the three galaxies (f_hz = 1.83 and 1.68), while UGC3997 gives f_hz = 0.90. The abstract should make this sample size and heterogeneity explicit, or the claim should be softened.","section":"Abstract and Table 1"},{"comment":"Several of the twenty galaxies with the largest mass deficits are flagged as containing foreground stars, overlapping companions, or poor fits, so the visual impression from Figure 11 may overstate the dust-lane/edge-on connection. The claim is better supported by the regression than by the table, and the text should clarify that the table is illustrative rather than definitive evidence.","section":"Section 4.3, Table 3"}],"recommendation":"major_revision","confidential_remarks":"The paper is within scope for ApJ and represents a useful catalog/application contribution. The main technical risk is the unvalidated luminosity-weighting proxy in Section 4.3; if the authors supply a direct validation or reposition that section as an exploratory analysis, the paper would be acceptable. The external comparisons are not circular, and the DiskMass section is honestly caveated. The duplicate Gallazzi reference should be corrected."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nYou should know this paper is a solid application of the PCA-based M/L estimator from Paper I to MaNGA. It delivers three things: a comparison of resolved stellar mass surface densities to DiskMass dynamics for three galaxies, an aperture-correction scheme for total masses, and an analysis of how luminosity-weighting biases total mass estimates. The planned value-added catalog will be genuinely useful.\n\nWhat is new: the luminosity-weighting result. The paper shows the deficit between spaxel-summed mass and luminosity-weighted mass correlates most strongly with the dispersion in inferred V-band optical depth (στV), not with star formation rate or mean attenuation. That points to differential dust attenuation as the dominant cause of the bias, rather than simple outshining. The figures and the worst-offender gallery give the claim visual support. The DiskMass comparison is honestly framed as exploratory, and the paper declines to claim a specific scale-height correction.\n\nThe soft spot is real but not fatal: Section 4.3 never coadds actual MaNGA spectra and reruns the PCA fit. M_LW is constructed by luminosity-weighting the resolved M/L ratios directly, which is exactly the right quantity for isolating the effect of luminosity weighting on M/L, but it is a proxy for what a true coadded spectrum would yield. If the PCA estimator is nonlinear in the spectrum, the two can differ, and the regression driving the dust conclusion sits on that proxy. The paper flags missing predictor uncertainties and cautions against overreading the regression, but it does not separately validate the proxy. The Ibarra-Medel et al. citation supports the general direction, not this specific construction. The claim would be materially strengthened by direct spectral coadding.\n\nThe DiskMass comparison is limited to three galaxies and a fitted scale-height factor, so treat it as illustrative. The comparison to NSA and JHU-MPA masses is routine but sensible, with the 0.1-0.15 dex offset handled carefully.\n\nWho should read this: anyone using MaNGA total or resolved stellar masses, and anyone working on unresolved-spectrum mass systematics. It deserves a serious referee. My recommendation is conditional acceptance: the luminosity-weighting claim should be checked with actual spectral coadding, and the code/data should be released with the catalog.\n\nBest.","headline":"A workmanlike MaNGA application paper with a useful catalog; the main new claim about dust-driven luminosity-weighting bias is built on a clearly flagged proxy that deserves a direct check.","tokens_in":20085,"tokens_out":3028,"would_cite":true,"duration_ms":34935,"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":"This paper shows that stellar masses from galaxy-coadded (unresolved) spectra are systematically low, with the largest deficits in dusty, edge-on galaxies, and identifies differential dust attenuation as the dominant cause.","keywords":["stellar mass","mass-to-light ratio","principal component analysis","MaNGA survey","integral field spectroscopy","dust attenuation","aperture correction","galaxy stellar mass catalog"],"falsifier":"Coadd the MaNGA spaxel spectra for a sample of galaxies into one spectrum per galaxy, run the same PCA fit on the coadded spectrum, and compare the resulting stellar mass with the sum of the resolved PCA masses; if the coadded-fit masses do not show a deficit that correlates with $\\sigma_{\\tau_V}$ and inclination, the luminosity-weighting proxy—not unresolved spectroscopy itself—is creating the reported bias.","tokens_in":19053,"feed_emoji":"🌌","tokens_out":10927,"duration_ms":104893,"temperature":0.7,"pith_summary":"Galaxy stellar masses are often derived from light that is summed or binned before fitting, and this paper asks whether such unresolved measurements lose mass. Using spatially resolved, PCA-based mass-to-light maps for 6,356 MaNGA galaxies (built in Paper I), the authors emulate an unresolved galaxy by luminosity-weighting the resolved mass-to-light ratios and compare it with the sum of resolved masses. They find that this proxy for coadded spectroscopy systematically undercounts stellar mass by roughly 0.05–0.1 dex in log mass-to-light ratio, with the largest deficits in dust-lane and edge-on galaxies. A regression against average dust attenuation, the spatial dispersion of attenuation, and central star-formation rate shows that differential dust attenuation across the galaxy is the dominant cause, with star formation playing only a secondary role. The paper also presents an aperture-corrected total-mass catalog, finds ~0.1 dex offsets relative to photometric mass catalogs, and reports tension with DiskMass dynamical masses that could be eased if disk scale-heights were overestimated by about a factor of 1.5.","feed_headline":"Coadding a galaxy's spectra loses stellar mass—dust lanes worst","feed_subtitle":"MaNGA resolved maps show the deficit is driven by differential dust, not star formation; a caution for fiber surveys.","key_machinery":"The central working object is the PCA-based i-band stellar mass-to-light ratio, $\\log \\Upsilon^*_i$: a basis set of six eigenspectra, trained on a library of about 40,000 composite stellar population spectra, onto which each observed spectrum is projected to recover a mass-to-light ratio with covariate uncertainties. Because it is evaluated spaxel-by-spaxel across MaNGA datacubes, it yields resolved stellar mass maps that power every comparison in the paper: luminosity-weighting those maps reproduces the effect of an unresolved galaxy spectrum; summing inside and outside the IFU sets up the aperture corrections; and deprojected radial profiles can be placed against kinematic mass surface densities. The luminosity-weighted combination rule is therefore the load-bearing operation for the paper's main negative finding about spatial binning.","core_discovery":"The central claim is that stellar mass is not invariant under spatial binning: when per-spaxel PCA mass-to-light estimates are combined by luminosity weighting to mimic a single galaxy spectrum, the resulting total stellar mass is lower than the sum over the resolved spaxels. The deficit is small on average but strongly structured—edge-on disks and galaxies with prominent dust lanes show the largest shortfalls, while face-on, low-dust galaxies are nearly unbiased. An ordinary least-squares fit of the deficit to average V-band optical depth, its spaxel-to-spaxel dispersion, and central specific star-formation rate finds that the dispersion term, $\\sigma_{\\tau_V}$, is the dominant driver, with confidence intervals excluding zero; central star formation matters only at roughly a tenth of the dust term's strength. The paper interprets this as evidence that differential dust attenuation, rather than simple outshining by young stars, is the main mechanism behind unresolved-spectrum mass deficits. The same mass estimates also appear heavier than photometric catalog masses by about 0.1–0.15 dex at red colors, and the comparison with DiskMass dynamical surface densities can be brought into qualitative agreement by reducing the adopted disk scale-heights by a factor of about 1.5.","pith_inferences":["A testable extension the paper leaves implicit: the fitted regression could serve as an empirical correction for unresolved surveys, if the spatial scatter in optical depth $\\sigma_{\\tau_V}$ can be estimated from multiband imaging or attenuation maps.","Because the test uses luminosity-weighted resolved mass-to-light ratios rather than a real coadded refit, the true bias of actual coadded spectra could be larger or smaller; a direct coadd-and-refit experiment on the same datacubes would settle the size of the effect.","If the disk scale-height tension survives in a larger sample, disks would be closer to maximal than the DiskMass fiducial assumptions imply, shifting dark-matter halo fits for disk galaxies.","The ~0.12 dex scatter between measured attenuation and axis ratio implies that inclination-based dust corrections in unresolved surveys introduce galaxy-to-galaxy scatter at about that level."],"forward_implications":["Single-fiber and low-spatial-resolution surveys will tend to underestimate total stellar masses, with the deficit growing for edge-on and dusty galaxies.","The correction for unresolved-spectrum masses must include differential dust attenuation; average attenuation alone and star-formation rate are not enough.","The color-based mass-to-light ('CMLR') aperture-correction method is preferred over the 'ring' method, and the two agree to better than 0.1 dex in total mass for most galaxies.","If the PCA/SPS masses are close to correct, the DiskMass scale-heights would need to shrink by about a factor of 1.5 to reconcile the two mass scales.","The soon-to-be-released SDSS value-added catalog will provide both resolved mass maps and aperture-corrected total stellar masses for MaNGA galaxies."],"supporting_citations":[{"why":"Supplies the PCA fitting method, the six-vector basis, and the resolved mass-to-light maps that every analysis here uses.","marker":"Pace et al. 2019"},{"why":"Introduced the PCA spectral-fitting approach that Paper I adapts to MaNGA.","marker":"Chen et al. 2012"},{"why":"Showed integrated colors yield mass-to-light ratios that are systematically low, the effect Section 4.3 revisits spectroscopically.","marker":"Zibetti et al. 2009"},{"why":"Quantified the photometric luminosity-weighting deficit and proposed the outshining explanation tested against dust here.","marker":"Sorba & Sawicki 2015"},{"why":"Provides external simulation evidence that coadded MaNGA-like spectra lose up to 0.15 dex in mass, worsening with inclination.","marker":"Ibarra-Medel et al. 2019"},{"why":"Defines the DiskMass dynamical mass surface density method and data that the PCA surface densities are compared to.","marker":"Bershady et al. 2010a"},{"why":"Provides the kinematic decomposition and radial bins for the three DiskMass–MaNGA galaxies.","marker":"Martinsson et al. 2013a"},{"why":"NSA photometry and masses anchor the aperture-correction flux deficits and the photometric comparison.","marker":"Blanton et al. 2011"},{"why":"JHU-MPA catalog masses are the second published mass scale compared against the PCA masses.","marker":"Kauffmann et al. 2003a"}],"fun_headline_variants":["Unresolved spectra undercount galaxy mass, dustiest most","Coadding spectra hides stellar mass via differential dust","Dust differential drives mass deficit in unresolved spectra","Fiber surveys' mass estimates skewed by dust lanes","Dust differential, not star formation, biases coadded masses"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The analysis assumes that the stellar mass-to-light ratio a real coadded galaxy spectrum would produce equals the luminosity-weighted average of the per-spaxel PCA mass-to-light ratios, but the paper never coadds the MaNGA spectra and reruns the fit; if spectral fitting responds nonlinearly to signal-to-noise or mixed stellar populations, the claimed unresolved-spectrum bias could be misstated.","fun_headline_variants_meta":{"raw":{"variants":["Unresolved spectra undercount galaxy mass, dustiest most","Coadding spectra hides stellar mass via differential dust","Dust differential drives mass deficit in unresolved spectra","Fiber surveys' mass estimates skewed by dust lanes","Dust differential, not star formation, biases coadded masses"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000877,"raw_usage":{"total_tokens":3891,"prompt_tokens":1138,"completion_tokens":2753,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":754,"completion_tokens_details":{"reasoning_tokens":2674}},"tokens_in":754,"tokens_out":2753,"duration_ms":20570,"temperature":1.0,"reasoning_tokens":2674,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:47:25.487599+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Coadd the MaNGA spaxel spectra for a sample of galaxies into one spectrum per galaxy, run the same PCA fit on the coadded spectrum, and compare the resulting stellar mass with the sum of the resolved PCA masses; if the coadded-fit masses do not show a deficit that correlates with $\\sigma_{\\tau_V}$ and inclination, the luminosity-weighting proxy—not unresolved spectroscopy itself—is creating the reported bias.","supporting_citations":[{"cited_title":"J., Tremonti, C., Chen, Y., et al","cited_arxiv_id":null,"evidence_quote":"Supplies the PCA fitting method, the six-vector basis, and the resolved mass-to-light maps that every analysis here uses."},{"cited_title":"A., et al","cited_arxiv_id":null,"evidence_quote":"Introduced the PCA spectral-fitting approach that Paper I adapts to MaNGA."},{"cited_title":"2015, MNRAS, 452, 235","cited_arxiv_id":null,"evidence_quote":"Quantified the photometric luminosity-weighting deficit and proposed the outshining explanation tested against dust here."},{"cited_title":"J., Avila-Reese, V., S´ anchez, S","cited_arxiv_id":null,"evidence_quote":"Provides external simulation evidence that coadded MaNGA-like spectra lose up to 0.15 dex in mass, worsening with inclination."}],"review_version":1}