{"id":"cda41890-d0cb-4c2f-a68d-71ddb646decf","arxiv_id":"2505.08858","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"In 683 massive quiescent galaxies at 0.6<z<1.0, spectroscopically measured metallicity rises perpendicular to the quiescent sequence in the UVJ diagram, opposite to fsps and BC03 model predictions.","lead":"A team measured ages and metal content for about 700 distant quiet galaxies using their spectra, not just their colors, and found that metals increase in a different direction in the classic color-color diagram than standard models predict. The mismatch implies that galaxy properties derived from photometric fitting alone, including stellar masses, may be systematically off.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The perpendicular [Fe/H] trend rests on alf metallicities near the blue UVJ edge, where the paper's own composite test still permits roughly half the trend to be a star-formation artifact; independent line-strength verification is needed.","rationale":"The reader's verdict is CONDITIONAL, with the weakest assumption exactly that alf-derived [Fe/H] values are unbiased across the UVJ quiescent box, especially near the blue edge. My independent reading agrees: the perpendicular metallicity gradient is the paper's central new empirical result, and the low-metallicity anchor at the blue edge is the part of the diagram where alf is least tested. The paper does several things well: it uses continuum-normalized spectra so ages and metallicities are independent of continuum shape, it provides stacked spectra in Fig. 5 that visually support the trend, and it explicitly tests a box-edge shift and a young-population contamination scenario in Section 4.4. Those tests mitigate but do not eliminate the concern, because the composite test itself uses alf to recover the injected [Fe/H] and can only calibrate one specific form of systematic bias, not the full range of template mismatch, dust, or abundance-pattern effects at the edge. A direct line-strength measurement on the existing stacks is the most direct settlement: it bypasses alf's response functions entirely and checks whether the spectra themselves carry the iron-abundance signature implied by the reported [Fe/H] gradient. If the line strengths confirm the gradient, the central claim is substantially strengthened. If they do not, the headline conclusion about perpendicular metallicity and the inferred bias in photometric SPS fitting would need to be softened. Since the reader already conditioned acceptance on validating edge-of-box metallicities, my recommendation is UNCHANGED: the conditional verdict remains the right call, with this test as the concrete path toward resolution.","tokens_in":28973,"tokens_out":4805,"duration_ms":53795,"concrete_test":"Measure iron- and magnesium-sensitive Lick indices (e.g., Fe5270, Fe5335, Mgb) directly from flux-calibrated stacked spectra of the same galaxies in the three perpendicular bins 2a, 2b, 2c of Fig. 5/Table 1. If the alf [Fe/H] gradient is real (median [Fe/H] approximately -0.05, -0.12, and -0.22 from 2a to 2c), the Fe indices should decline from 2a to 2c by the amount expected from about 0.17 dex in iron abundance, while Mgb should show a weaker or similar trend depending on [Mg/Fe]. If the Fe line strengths are flat or inconsistent with the alf metallicities, the perpendicular [Fe/H] trend is an artifact of the alf response functions rather than a property of the galaxy spectra. This requires no new observations and can be done on the existing stacks.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that stellar [Fe/H] increases approximately perpendicular to the quiescent sequence in the rest-frame UVJ diagram, opposite to the predictions of fsps and BC03. This claim depends on the alf-derived [Fe/H] values at the blue edge of the UVJ box (low U-V, intermediate V-J), where the data anchor the low-metallicity end of the trend. If alf systematically underestimates [Fe/H] there—because of residual star formation, dust, young sub-populations, or template mismatch in the continuum-normalized absorption-line fit—the observed perpendicular gradient could be partly or wholly artificial. The paper's Section 4.4 composite-spectrum test speaks directly to this: adding a 6 per cent young component to 33 quiescent spectra shifts U-V by about 0.24 mag and lowers fitted [Fe/H] by about 0.05 dex, whereas the observed decrease in [Fe/H] across a similar colour span in Fig. 2 is about 0.11 dex. Thus low-level star formation can generate only about half of the observed gradient, leaving a substantial unexplained residual. The same section also reports that shifting the UVJ box by 0.1 mag redward preserves both trends, which is reassuring, but this does not validate the absolute alf [Fe/H] calibration at the edge. No independent metallicity measurement (e.g., Lick indices or a different full-spectrum code) is provided for the edge-of-box galaxies. The claim that continuum-only SPS fitting is systematically biased in age, metallicity, and stellar mass therefore inherits the uncertainty in alf's edge-of-box [Fe/H] values. This is the single most load-bearing assumption in the paper.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper measures SSP-equivalent ages and [Fe/H] abundances for 683 massive quiescent galaxies at 0.6 ≲ z ≲ 1.0 from LEGA-C, fitting continuum-normalized absorption-line spectra with alf, and compares these measurements to rest-frame U−V and V−J colours. The main findings are that age increases approximately along the quiescent sequence in the UVJ diagram, while [Fe/H] increases approximately perpendicular to that sequence. Forward predictions from fsps and BC03, computed by injecting the measured ages and metallicities into the models, reproduce the age trend but predict metallicity to increase roughly along the quiescent sequence rather than perpendicular to it. The paper further compares dynamical mass-to-light ratios with model M/L trends and argues that photometric SED fitting may therefore yield systematically biased ages, metallicities, and stellar masses. Robustness tests address dust, star-formation history, UVJ box shifts, and composite young populations.","tokens_in":29260,"tokens_out":4809,"duration_ms":53031,"significance":"If the central claim holds, the paper is an important empirical result: it uses a large sample, extracts ages and metallicities independently of continuum shape, and makes a non-circular forward-model comparison rather than fitting the models to the observed colours. The study connects to the long-standing problem of age–metallicity–dust degeneracies and has direct implications for photometric SED fitting, stellar mass estimates, and SPS model development. The manuscript also contains a useful set of sanity tests: the UVJ box shift, the dust and SFH variants, and the composite-spectrum experiment in Section 4.4. The forward-model comparison in Figs 3, 4, and 7 is not circular because the measured ages and metallicities are inputs, not fitting outputs, and the model tracks are evaluated over the observed parameter range.","major_comments":[{"comment":"The composite-spectrum test shows that adding a 6 per cent young component shifts U−V by about 0.24 mag and lowers [Fe/H] by about 0.05 dex, whereas the observed decrease in [Fe/H] over a similar colour span is about 0.11 dex. Thus low-level recent star formation can account for roughly half of the observed metallicity gradient, and the paper's own test does not establish that the remaining half is robust to other edge-of-box systematics. Since the blue edge of the UVJ box anchors the low-metallicity end of the perpendicular trend, the central claim needs either a quantitative correction for this bias or an independent metallicity diagnostic (e.g., Lick indices or a different full-spectrum fitting code) applied to the edge-of-box galaxies. Without this, the claim that metallicity increases perpendicular to the quiescent sequence is only partially supported.","section":"Section 4.4, composite-spectrum test"},{"comment":"The perpendicular metallicity trend and the data–model differences in M/L are presented as qualitative visual trends, and the text explicitly states in Section 3.1 that the strength of the trends is not quantified. Given that the paper's headline result is a specific direction in UVJ space, the authors should provide a quantitative measure, such as a bootstrapped gradient vector in (U−V, V−J) space, a permutation test comparing the observed orientation with the model-predicted orientation, or a significance estimate for the residual trend after the young-population bias discussed in Section 4.4. This is a load-bearing addition rather than a cosmetic improvement.","section":"Section 3.1 and Figs 2, 4, 7"}],"minor_comments":[{"comment":"Please state the LOESS span, robust iteration settings, and any edge-weighting choices so that the smoothed maps in Figs 2, 4, and 7 are reproducible.","section":"Fig 2 and LOESS smoothing"},{"comment":"The sSFR column is labelled with units of M_sun/yr, but the quantity is a specific star-formation rate averaged over 0.01 Gyr; the units should be yr^-1, and the definition should be clarified in the table caption.","section":"Table 1"},{"comment":"The reference entry for Beverage et al. (2021) appears to be malformed, with a missing journal name in the bibliography.","section":"Reference list"},{"comment":"The statement '[Fe/H] ≈ log(Z/Zsun)' is used to map alf measurements to fsps and BC03 metallicities, but the paper should explicitly note that alf fits [Fe/H] with variable abundance patterns while fsps and BC03 assume solar-scaled abundances, so the mapping is an approximation that may affect the quantitative comparison.","section":"Section 3.2"},{"comment":"The code name 'fast' (Kriek et al. 2009, 2018) should be set in a consistent style and defined at first use in the sentence describing the dust test.","section":"Section 4.4"}],"recommendation":"major_revision","confidential_remarks":"The paper is a solid empirical study and the authors are well positioned to address the major comments with existing data: quantifying the gradient orientation and checking the edge-of-box metallicities against an independent diagnostic would substantially strengthen the central claim. I do not see a fundamental flaw in the measurement strategy, but the headline result needs more than the current qualitative support."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague — here is my read. The genuinely new thing is the metallicity trend: using continuum-normalized absorption-line spectra from LEGA-C, they find that [Fe/H] increases roughly perpendicular to the quiescent sequence in the UVJ diagram, while age increases along it. That is an empirical result independent of continuum fitting, on 683 galaxies, and it directly challenges how standard SPS models (FSPS, BC03) map metallicity onto UVJ colors. The age trend was known; the metallicity direction is new, and the direct confrontation with model tracks is useful.\n\nWhat the paper does well: the measurement design is sound. Continuum normalization decouples absorption-line properties from continuum shape, the sample is large, and the model \"predictions\" are forward computations from the measured ages and metallicities, not fits to the observed UVJ colors. So the circularity concern some people raise is not really present. The robustness checks are extensive: dust, SFH variations, UVJ box shifts, composite young populations. They also show that M/L ratio trends from dynamical masses disagree with the models, which is a genuinely independent cross-check.\n\nThe soft spots are real but not fatal. The headline [Fe/H] gradient is presented qualitatively—there is no formal significance estimate, just LOESS-smoothed maps and arrows. That should be fixed with a bootstrap or regression analysis. More importantly, the low-metallicity end of the trend is anchored near the blue edge of the UVJ box, where alf metallicities are most vulnerable to contamination from residual star formation, dust, or young sub-populations. The paper's own composite test shows that a 6% young component shifts [Fe/H] down by about 0.05 dex, while the observed gradient over a similar color span is about 0.11 dex. So low-level star formation can explain roughly half the trend. That does not kill the result, but it means the claim that continuum-only SED fitting is systematically biased inherits a model-dependent alf calibration at the edge. An independent metallicity measurement—Lick indices or a different full-spectrum code—would substantially firm it up. The M/L comparison is also a little softer than the abstract suggests, because the dynamical masses include dark matter and the treatment is not clean.\n\nOverall, this is a solid paper with one load-bearing assumption that deserves direct testing. I would send it to a serious referee, and I would ask for a quantitative significance statement for the [Fe/H] gradient and an explicit edge-of-box validation. A cautious reader should treat the perpendicular metallicity trend as interesting but not yet fully closed.","headline":"The new result here is that stellar metallicity in quiescent galaxies increases perpendicular to the UVJ quiescent sequence while age increases along it, measured directly from absorption-line spectra; the paper is solid but the edge-of-box metallicity calibration needs independent verification.","tokens_in":29920,"tokens_out":1829,"would_cite":true,"duration_ms":18597,"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":"Absorption-line measurements show stellar metallicity increases across the UVJ diagram, perpendicular to the age trend, while standard stellar population models predict the opposite direction.","keywords":["quiescent galaxies","stellar metallicity","UVJ diagram","absorption-line spectroscopy","stellar population synthesis","age-metallicity degeneracy","mass-to-light ratio"],"falsifier":"A direct check is to obtain higher signal-to-noise absorption-line spectra (or Lick-index measurements interpreted through a different set of models) for the galaxies near the blue edge of the UVJ box. If those galaxies' [Fe/H] values come out near solar rather than subsolar, the perpendicular metallicity gradient disappears; if they remain subsolar, the trend is robust.","tokens_in":28754,"feed_emoji":"🔭","tokens_out":5394,"duration_ms":47195,"temperature":0.7,"pith_summary":"This paper confronts two independent ways of measuring the stellar content of quiescent galaxies: broad-band rest-frame $U-V$ and $V-J$ colours, which trace the shape of the stellar continuum, and ages and iron abundances extracted from continuum-normalized absorption-line spectra, which do not depend on that shape. Using about 700 massive quiescent galaxies at $0.6\\lesssim z\\lesssim1.0$, it finds that age increases along the quiescent sequence in the $UVJ$ diagram, while [Fe/H] increases approximately perpendicular to that sequence, with higher metallicities at redder $U-V$ and bluer $V-J$. Standard stellar population synthesis models (fsps and BC03) predict the opposite direction for metallicity. If the result holds, photometric fits that use continuum shapes alone can systematically misassign metallicity to age, dust, or star-formation history, biasing derived stellar masses and the interpretation of quiescent galaxy evolution.","feed_headline":"Metallicity runs across UVJ, not along it","feed_subtitle":"Absorption-line ages and metallicities of ~700 quiescent galaxies show age and metallicity track different axes, unlike model predictions.","key_machinery":"The central machinery is the combination of full-spectrum absorption-line fitting with the code alf, which continuum-normalizes each spectrum before fitting so that age and [Fe/H] are independent of the overall SED shape, together with LOESS smoothing of the $UVJ$ diagram to reveal the directional gradients in age and metallicity. The comparison uses stellar evolution tracks generated with fsps and BC03 models, fed with the sample's measured ages and metallicities to predict $UVJ$ colours, and dynamical mass-to-light ratios from virial masses as an independent check. The load-bearing step is the perpendicularity: the model tracks predict [Fe/H] increasing along the quiescent sequence, while the data show it increasing across it.","core_discovery":"The paper claims that for massive quiescent galaxies at intermediate redshift, ages derived from continuum-normalized absorption-line spectra increase approximately along the quiescent sequence in the rest-frame $UVJ$ diagram, from roughly 1 Gyr in the bluest regions to 3–5 Gyr in the reddest regions. In contrast, iron abundances [Fe/H] increase roughly perpendicular to that sequence, from subsolar values near the blue edge of the $UVJ$ box to supersolar values at redder $U-V$ and bluer $V-J$. This is presented as the first detection of a metallicity trend from absorption lines that is independent of continuum shape. Two popular stellar population synthesis models, fsps and BC03, reproduce the age trend but predict that metallicity increases approximately along the quiescent sequence, nearly perpendicular to what is observed. Dynamical mass-to-light ratio trends derived from virial masses also diverge from model predictions at the reddest tip of the sequence. The paper concludes that relying on model fits to continuum shapes alone may lead to systematic biases in ages, metallicities, and stellar masses of quiescent galaxies.","pith_inferences":["If the perpendicular metallicity gradient is real, the $UVJ$ diagram carries a second, approximately orthogonal axis of information: one axis (along the sequence) encodes age and the other (across it) encodes metallicity, suggesting that joint colour and line measurements could separate age and metallicity in photometric samples without full spectroscopy.","A direct test of the explanation would be to generate $UVJ$ model tracks with non-solar or $\\alpha$-enhanced abundance patterns; if those tracks rotate the metallicity direction to match the data, the discrepancy is explained by the models' solar-scaled assumption rather than by new physics in the galaxies.","Because the models mispredict $M/L$ trends along the reddest tip, published stellar masses of the most massive red quiescent galaxies may carry systematic offsets that propagate into inferred formation redshifts and quenching timescales."],"forward_implications":["Reddening along the quiescent sequence in the $UVJ$ diagram is primarily an age effect, as previously suggested, but now confirmed with ages that do not depend on the continuum shape.","The $UVJ$ diagram can be used to rank quiescent galaxies as relatively more or less metal-rich, although the scatter is large and precise metallicities cannot be read off the colours.","Standard solar-scaled stellar population synthesis models systematically mispredict the colours associated with metallicity, so photometric SED fitting alone may bias derived metallicities, ages, dust, and star-formation histories.","Dynamical mass-to-light ratio trends disagree with model predictions, implying that stellar masses inferred from SPS-based $M/L$ ratios may be systematically offset for the reddest, most massive quiescent galaxies.","Resolving the data-model discrepancy matters for higher-redshift quiescent galaxies observed with JWST, where spectroscopy is sparse and photometric continuum fitting remains the primary tool."],"supporting_citations":[{"why":"Provides the sample selection, spectral extraction, and initial alf fitting of LEGA-C galaxies on which this study builds.","marker":"C24"},{"why":"Releases the LEGA-C spectroscopic survey data that supply the spectra for the full sample.","marker":"van der Wel et al. 2021"},{"why":"Describes the alf full-spectrum fitting code used to derive absorption-line ages and [Fe/H] values.","marker":"Conroy et al. 2018"},{"why":"Provides one of the two stellar population synthesis model sets whose predicted UVJ colour trends are compared with the data.","marker":"Bruzual & Charlot 2003"},{"why":"Provides the fsps stellar population synthesis models used as the other comparison set.","marker":"Conroy et al. 2009; Conroy & Gunn 2010"},{"why":"Supplies the previous age-colour relation for quiescent galaxies that the paper tests with its independent ages.","marker":"Belli et al. 2019"},{"why":"Provides the dynamical virial masses used to compute independent mass-to-light ratios.","marker":"van der Wel et al. 2022"}],"fun_headline_variants":["Age and metallicity split across UVJ diagram","Quiescent galaxy metallicity runs across UVJ, not along","UVJ colors mislead quiescent galaxy age and metallicity","Models fail to predict quiescent galaxy metallicity pattern"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the [Fe/H] values measured by alf from continuum-normalized absorption lines are unbiased across the whole UVJ quiescent box, especially near the blue edge where the lowest-metallicity galaxies anchor the perpendicular trend; if alf systematically underestimates [Fe/H] for galaxies with residual star formation, a young population, or dust near that edge, the observed trend would be partly or wholly artificial.","fun_headline_variants_meta":{"raw":{"variants":["Age and metallicity split across UVJ diagram","Quiescent galaxy metallicity runs across UVJ, not along","UVJ colors mislead quiescent galaxy age and metallicity","Models fail to predict quiescent galaxy metallicity pattern"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000224,"raw_usage":{"total_tokens":1547,"prompt_tokens":1118,"completion_tokens":429,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":734,"completion_tokens_details":{"reasoning_tokens":360}},"tokens_in":734,"tokens_out":429,"duration_ms":4504,"temperature":1.0,"reasoning_tokens":360,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T21:46:12.588234+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct check is to obtain higher signal-to-noise absorption-line spectra (or Lick-index measurements interpreted through a different set of models) for the galaxies near the blue edge of the UVJ box. If those galaxies' [Fe/H] values come out near solar rather than subsolar, the perpendicular metallicity gradient disappears; if they remain subsolar, the trend is robust.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the dynamical virial masses used to compute independent mass-to-light ratios."}],"review_version":1}