{"id":"2b9c7486-c5eb-4fe3-b1fe-1359d4519e26","arxiv_id":"2501.12342","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"In FIRE-2 simulations, alpha-element bimodality in Milky Way-mass galaxies is not universal; its low-alpha sequence appears only in galaxies that accrete significant metal-poor gas at late times.","lead":"This paper analyses 11 Milky Way-mass galaxies from the FIRE-2 simulations and finds that a split between high- and low-alpha stellar populations appears in only about half of them. The presence of the low-alpha sequence tracks recent infall of metal-poor gas from the circumgalactic medium, not mergers alone.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The bimodality classification is threshold-sensitive and internally inconsistent (prominence 0.3 vs 0.03; distance 0.025 vs 0.015), and the inflow/disk-size contrast is not significance-tested; small threshold changes could reshuffle categories and erase the central contrast.","rationale":"The reader's weakest assumption focuses on the unstated selection of 11 galaxies and the four deferred strong-bimodal cases. That is a legitimate representativeness concern, but I consider the more load-bearing issue to be internal to the 11: the classification that defines the categories used in the causal comparison is threshold-dependent, and the paper states contradictory peak-finding parameters (Section 3.1: prominence 0.3, distance 0.025; Appendix A: prominence 0.03, distance 0.025; weakly-bimodal added at distance 0.015). The central Figs. 5 and 6 compare median accretion histories of categories built on this classification, and no statistical test establishes that the bimodal/non-bimodal difference is significant. The claim in the abstract that low-alpha presence is 'related to recent infall of metal-poor gas' is therefore only as strong as the stability of these labels. I credit the paper for using public FIRE-2 data and a transparent algorithmic classification, which makes the requested threshold-sensitivity and permutation checks straightforward. The paper's own deferral of the star-formation-history connection to Barry et al. (in prep) further compounds the uncertainty, but the classification robustness is the more direct threat to the central claim. The correct verdict remains conditional pending these checks, so I do not change the reader's conditional verdict.","tokens_in":20884,"tokens_out":6016,"duration_ms":60451,"concrete_test":"Using the public FIRE-2 data, reproduce the [Fe/H]-[Mg/Fe] distributions for all 11 galaxies and rerun the find_peaks classification with both prominence values (0.3 and 0.03) and both distance thresholds (0.025 and 0.015), then recompute the median inflow-mass and gas-disk-size tracks for the resulting category assignments as in Figs. 5 and 6. If any of the five labeled bimodal galaxies changes category, or if the bimodal versus non-bimodal difference over the last 5–6 Gyr falls within the 1-sigma scatter under either threshold choice, the central claim does not survive as stated. Additionally, run a permutation test on the category labels to quantify the probability of the observed inflow contrast under the null hypothesis that category assignment is unrelated to accretion history.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim requires a robust partition of the 11 galaxies into bimodal, weakly-bimodal, and non-bimodal classes, because the entire accretion-history contrast (Figs. 5–7) is computed between these classes. That partition is not as secure as the argument needs. Section 3.1 specifies find_peaks with prominence at least 0.3 and peak distance at least 0.025 dex; Appendix A specifies prominence 0.03 with the same distance, and the weakly-bimodal class is admitted only after lowering the distance to 0.015. A prominence threshold ten times higher will suppress many candidate double peaks, so the number of metallicity bins in which bimodality is 'detected' can change substantially, potentially reassigning galaxies across categories. The classification also uses a 20% low-alpha fraction threshold to define the onset of the low-alpha sequence, and for converging bimodals (Louise and Remus) the separation line is extrapolated into the merging region, adding further subjectivity. Moreover, the headline contrast rests on median tracks over only 5 bimodal, 2 weakly-bimodal, and 3 non-bimodal galaxies, with shaded 1-sigma spreads; no significance test (e.g., permutation or bootstrap) is applied to the inflow-mass or gas-disk-size differences in Figs. 5 and 6. Since the abstract's statement that the low-alpha sequence is 'related to recent infall' is inferred from this category contrast, threshold sensitivity and the absence of a significance test are load-bearing. The paper defers the detailed connection between star formation history and bimodal patterns to Barry et al. (in prep), which weakens its ability to rule out star-formation-history-driven alternatives to the gas-accretion interpretation.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript analyzes 11 Milky Way-mass galaxies from the FIRE-2 cosmological zoom-in simulations and classifies them by the presence of double sequences on the [Fe/H]-[Mg/Fe] plane. Five galaxies are classified as bimodal, two as weakly-bimodal, three as non-bimodal, and one (m12b) as strongly-bimodal and deferred to a companion paper. Using a peak-finding algorithm on [Mg/Fe] distributions in narrow [Fe/H] bins, the paper then compares the gas accretion histories of these categories over the last 8 Gyr. It reports that bimodal galaxies experience increasing inflow of metal-poor gas over the past 5-6 Gyr, grow larger gas disks, and form a low-alpha sequence from stars born in the diluted outer gas disk. The paper contrasts smooth CGM accretion with gas-rich mergers as two delivery mechanisms and concludes that both can produce bimodal patterns, with slightly different morphologies. The analysis is built on public FIRE-2 simulations, and the paper is explicit that its agreement with Milky Way trends is qualitative; it also states several limitations, including the deferral of the strongest bimodal cases to Barry et al. (in prep).","tokens_in":21223,"tokens_out":8634,"duration_ms":79075,"significance":"If the central correlation is robust, the paper makes a valuable contribution by showing that alpha-bimodality in Milky Way-mass galaxies is not a universal feature and that its presence in FIRE-2 is tied to late-time accretion of metal-poor gas. The accretion metrics (inflow mass, gas disk size, infall metallicity) are defined independently of the bimodality classification, so the observed correlation is not forced by construction. The paper also benefits from using the publicly available FIRE-2 simulations, from stating data availability, and from candidly acknowledging that the simulated bimodal sequences are less separated than in the Milky Way. However, the small number of galaxies, an internal inconsistency in the peak-detection thresholds, and the absence of significance tests currently limit the strength of the central claim; these issues are identifiable and likely fixable within the manuscript's scope.","major_comments":[{"comment":"The classification threshold is not uniquely specified: Section 3.1 states that find_peaks is applied with 'a prominence of at least 0.3' and a peak separation of 0.025 dex, while Appendix A states that the same procedure uses a prominence of 0.03 and the same 0.025 dex separation. These thresholds differ by an order of magnitude, and the number of [Fe/H] bins in which a double peak is detected depends directly on this parameter. The weakly-bimodal class is then admitted only after separately lowering the peak distance to 0.015 dex, so the three categories are defined under different detection criteria. Because the entire accretion-history contrast in Figs. 5-7 is computed over these categories, the central claim of the paper is sensitive to this choice. Please specify a single, justified detection criterion, verify that the categorical assignments in Table 1 are stable to reasonable threshold variations (e.g., a grid in prominence and distance), and report the resulting category assignments. The extrapolation of the separation line into the converging region for Louise and Remus, described in Section 3.1, should also be quantified as a source of uncertainty.","section":"Section 3.1 and Appendix A"},{"comment":"The claim that bimodal galaxies experience 'rapidly increasing' inflow and 'significant growth of the gas disk' is based on median tracks for 5 bimodal, 2 weakly-bimodal, and 3 non-bimodal galaxies, with shaded 1-sigma spreads. No significance test is applied to the differences between categories, and with n=5, 2, 3 the visual separation could be driven by one or two galaxies (for example, Louise has the largest R_edge,gas and R_infall in Table 1). The word 'significant' in the abstract and Section 5 is therefore not supported by a statistical test. Please add a permutation or bootstrap test that compares the bimodal and non-bimodal distributions of inflow mass and gas-disk radius at fixed lookback times (e.g., at 2, 4, and 6 Gyr ago), and report effect sizes or p-values. This is load-bearing because the abstract's causal statement about recent infall is inferred from the category contrast in these figures.","section":"Section 3.3, Figs. 5-6"},{"comment":"The sample of 11 galaxies is not demonstrated to be representative of Milky Way-mass FIRE-2 systems, because Section 1 states that 'four additional galaxies with stronger bimodality' are examined in the companion paper Barry et al. (in prep), and Section 3.1 defers the strongly-bimodal case m12b to that same paper. If those systems were included, the bimodal fraction (currently 5 of 11) and the contrast in accretion histories could change materially, and the paper's conclusion that bimodality is specifically tied to late metal-poor gas accretion would not be falsifiable within this manuscript's scope. Please state the selection criteria that produced the 11 galaxies, or include the deferred systems in at least a summary comparison (e.g., their inflow mass and disk-size tracks) so the reader can assess whether the central correlation holds across the full FIRE-2 Milky Way-mass sample.","section":"Section 1 and Section 3.1"}],"minor_comments":[{"comment":"There is a typo in 'NIHAO-UHD simluations'; it should be 'simulations'.","section":"Section 1"},{"comment":"The caption relies on frame colors ('red frame', 'blue frame', 'navy frame') to identify galaxy categories; please add explicit panel letters or symbols so the figure is interpretable without color.","section":"Figure 1 caption"},{"comment":"The column 'R_infall' is described as 'the radius of the 10-kpc shell enclosing gas infall particles'; this wording is ambiguous about whether the shell is a spherical shell of thickness 10 kpc or a shell starting at a radius 10 kpc. Please clarify the geometric definition in the text or caption.","section":"Table 1 and Section 3.3"},{"comment":"The text refers first to 'Fig. 9' for the metallicity distributions and then to 'Fig. 8' for the radial distribution; since these are sequential sections, it would be easier for the reader if the figures were referenced in the order they are discussed, or if each reference carried a brief content descriptor.","section":"Section 3.4"}],"recommendation":"major_revision","confidential_remarks":"The paper's central dependence on the unpublished companion paper (Barry et al., in prep) for the strong-bimodal subset is a matter for the editor: if the companion paper is not forthcoming, the present claim about the frequency and drivers of bimodality in FIRE-2 cannot be fully evaluated. The internal inconsistency between the prominence threshold in Section 3.1 (0.3) and Appendix A (0.03) also suggests a possible error that the authors should be asked to resolve explicitly before acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Worth a read, and worth sending out, but the central contrast needs a bit more spine before I'd take the accretion story as firmly established.\n\nWhat's new: this is the first systematic census of [Mg/Fe] bimodality across 11 FIRE-2 MW-mass galaxies, and the split between smooth CGM accretion and gas-rich mergers producing slightly different bimodal patterns is a useful addition. The paper does a good job of showing that the low-alpha sequence appears with late-time metal-poor infall and outer disk growth, and that the high-alpha population is older and kinematically hotter, in qualitative accord with the Milky Way. It is also careful to note that the accreted gas is not primordial and that outflow-enriched CGM matters. The data are public (FIRE-2), which is real evidence.\n\nSoft spots in proportion: the bimodality classification uses different peak-detection thresholds in the main text (prominence 0.3, distance 0.025) and Appendix A (prominence 0.03, same distance), and the weakly-bimodal class only appears after lowering the distance to 0.015. That is a genuine inconsistency, and since the whole accretion-history contrast is computed between the resulting classes, small threshold changes could reshuffle a galaxy or two. It is not fatal—the qualitative figures show bimodal galaxies with rising inflow and larger gas disks—but the absence of any significance test on those median tracks leaves the headline correlation weaker than the abstract suggests. The sample of 11 galaxies omits four stronger-bimodality cases deferred to Barry et al. (in prep); selection here is not described, so I can't tell if the 5/11 bimodal fraction is representative. The paper also defers the SFH-bimodality connection, so the interpretation leans on the accretion correlation alone.\n\nNone of this is disqualifying. The authors are honest about the qualitative nature and the limitations. The result is a useful mechanistic correlation, not a derivation, and it should be judged as such. I'd send it to a referee, mainly to push on threshold robustness and to ask for a permutation test on the inflow/gas-disk contrast. For anyone in galactic chemical evolution, this is a solid citation for the FIRE-2 census and the smooth-vs-merger distinction. I'd bring it to reading group; I'd cite it, though I'd wait for the companion paper before treating the SFH-attribution as settled.","headline":"Useful FIRE-2 census tying low-alpha sequences to late metal-poor gas accretion, but the classification thresholds and sample selection need scrutiny before the correlation is taken as established.","tokens_in":21824,"tokens_out":1904,"would_cite":true,"duration_ms":18112,"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":"Late infall of metal-poor gas from the circumgalactic medium creates the low-alpha stellar sequence in Milky Way-mass galaxies.","keywords":["alpha-element bimodality","low-alpha sequence","gas accretion","circumgalactic medium","FIRE-2 simulations","[Fe/H]-[Mg/Fe] plane","disk galaxy evolution","Milky Way analogues"],"falsifier":"Track the four strong-bimodal galaxies analyzed in the companion paper: if any shows a clear low-alpha sequence without a concurrent rise in metal-poor gas inflow and gas-disk growth over the last 5–6 Gyr, the claimed mechanism fails. Alternatively, find a simulated or observed Milky Way-mass galaxy whose low-alpha sequence formed while its accretion was declining and metal-rich.","tokens_in":20685,"feed_emoji":"🌌","tokens_out":8628,"duration_ms":81758,"temperature":0.7,"pith_summary":"This paper asks why some disk galaxies show two distinct stellar populations on the iron-magnesium abundance plane—a high-alpha sequence of old, hot stars and a low-alpha sequence of younger, colder stars—whereas others show only one. Using 11 simulated Milky Way-mass galaxies, it finds that the low-alpha sequence appears in galaxies that have recently accreted large amounts of metal-poor gas from the circumgalactic medium; this gas lands in the outer disk, dilutes the interstellar medium, and grows the gas disk. Galaxies without such late accretion remain unimodal or only weakly bimodal. If correct, the bimodality seen in the Milky Way and some other disks is a direct signature of late-time gas accretion, not an inevitable outcome of disk formation.","feed_headline":"Metal-poor gas inflow creates the low-alpha stellar sequence","feed_subtitle":"In simulated Milky Way-mass galaxies, a second stellar population appears only when metal-poor gas rains onto the outer disk.","key_machinery":"The load-bearing object is the $[\\mathrm{Fe/H}]$–$[\\mathrm{Mg/Fe}]$ abundance plane, with Mg used as the $\\alpha$ element most purely produced by core-collapse supernovae in the simulation. The paper separates high- and low-$\\alpha$ sequences by locating double peaks in $[\\mathrm{Mg/Fe}]$ within narrow $[\\mathrm{Fe/H}]$ bins. The causal analysis then rests on accretion diagnostics: gas particles with negative radial velocity inside a 10 kpc shell are tracked over the final ~8 Gyr, their mass-weighted metallicity distributions are compared across time, and the radius where gas surface density falls below 5 solar masses per square kiloparsec tracks disk growth. The mechanism that carries the argument is that metal-poor, relatively high-angular-momentum gas from the circumgalactic medium joins the galaxy in the outskirts, dilutes the star-forming interstellar medium, and fuels the young, extended low-$\\alpha$ population.","core_discovery":"The paper argues that the presence of a low-$\\alpha$ sequence on the $[\\mathrm{Fe/H}]$–$[\\mathrm{Mg/Fe}]$ plane is not a universal property of disk galaxies but a symptom of late-time metal-poor gas accretion. In five of eleven simulated Milky Way-mass galaxies, two distinct stellar sequences emerge: an old, kinematically hot, centrally concentrated high-$\\alpha$ population and a younger, cooler low-$\\alpha$ population that extends to the outer disk. Comparing gas inflow over the last 8 Gyr, the bimodal galaxies are the ones whose circumgalactic gas becomes increasingly metal-poor and abundant, settles at large radii, and grows the gas disk; unimodal galaxies show declining, more metal-rich inflow. The paper further shows that both smooth CGM accretion and gas-rich mergers can deliver this gas, producing somewhat different bimodal patterns: smooth accretion builds a metal-poor tail at large radii, while mergers dilute $[\\mathrm{Fe/H}]$ across all radii and can rapidly shift stars to lower $[\\mathrm{Mg/Fe}]$.","pith_inferences":["If the causal link holds, a Milky Way-mass galaxy with a prominent low-alpha sequence should also show a current or recent extended, low-metallicity gas reservoir at large radii; this is testable with atomic-hydrogen and molecular-gas mapping of nearby disks.","The four strong-bimodal galaxies deferred to the companion paper are a natural falsification check: if they acquired their low-alpha sequences without a late metal-poor inflow, the proposed mechanism is not the only route to bimodality.","The FIRE-2 picture resembles the two-infall chemical-evolution scenario, but with the second infall realized as prolonged cosmological accretion rather than a single delayed episode, which may change how the Milky Way's age–$[\\mathrm{Fe/H}]$ relation is interpreted.","One might predict that galaxies with long steady star-formation phases and extended gas disks are more likely to show bimodality, since those are the conditions under which late metal-poor gas can form a distinct young low-alpha population."],"forward_implications":["In the FIRE-2 sample, alpha-element bimodality is not universal: five galaxies are bimodal, three are unimodal, two are weakly bimodal, and one is strongly bimodal, so the Milky Way's two sequences are a common but not guaranteed outcome.","The low-alpha population in bimodal galaxies is younger, kinematically cooler, and more radially extended than the high-alpha population, matching the qualitative age and kinematic split seen in the Milky Way.","Bimodal galaxies show rising metal-poor gas inflow and gas-disk growth over the last 5–6 Gyr, whereas non-bimodal galaxies show declining inflow and little disk growth; the onset of low-alpha star formation lags the accretion by 1–2 Gyr.","Both smooth accretion from the circumgalactic medium and gas-rich mergers can supply the metal-poor gas, but they leave distinguishable patterns: smooth accretion creates an extended metal-poor tail in the outer disk, while mergers cause rapid dilution in $[\\mathrm{Fe/H}]$ at all radii.","A galaxy can experience a metal-poor merger without becoming bimodal if the merger happens during the bursty star-formation phase, so the timing of accretion relative to star-formation mode matters."],"supporting_citations":[{"why":"Supplies the FIRE-2 simulation physics model, including feedback, metal diffusion, and element tracking, that all abundance measurements in the paper rely on.","marker":"Hopkins et al. 2018"},{"why":"Introduces the Latte suite of isolated Milky Way-mass galaxies used as part of the sample.","marker":"Wetzel et al. 2016"},{"why":"Provides the Local Group-like galaxy pairs from the ELVIS on FIRE runs that make up the rest of the sample.","marker":"Garrison-Kimmel et al. 2019a"},{"why":"Defines the bursty-to-steady star formation transition used to normalize gas inflow and to set the epoch when the thick disk is already in place.","marker":"Yu et al. 2021"},{"why":"Previous FIRE analysis of gas accretion modes that the paper relies on to describe the late-time hot-mode rotating cooling flow.","marker":"Anglés-Alcázar et al. 2017"},{"why":"Quantifies the angular momentum of gas accreting in the hot mode, used to argue that the high-angular-momentum inflow in bimodal galaxies settles far out and grows the disk.","marker":"Hafen et al. 2022"},{"why":"Provides the Auriga simulation scenarios for bimodality that the paper compares against and distinguishes from the FIRE-2 behavior.","marker":"Grand et al. 2018"},{"why":"Earlier attribution of low-alpha sequence formation to gas-rich mergers, against which the smooth-accretion result is contrasted.","marker":"Buck 2020"}],"fun_headline_variants":["Gas infall shapes alpha bimodality in simulated disk galaxies","Low-alpha stars in simulated disks trace metal-poor gas inflow","Inflowing metal-poor gas builds the low-alpha sequence","Bimodal alpha sequences linked to late gas accretion","Simulated galaxies reveal gas-driven alpha bimodality"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The analysis assumes the 11 simulated galaxies are a representative sample of Milky Way-mass galaxies; four galaxies with stronger bimodality are deferred to a companion paper, and if those galaxies lack the late metal-poor inflow, the proposed causal link would not cover all bimodal cases.","fun_headline_variants_meta":{"raw":{"variants":["Gas infall shapes alpha bimodality in simulated disk galaxies","Low-alpha stars in simulated disks trace metal-poor gas inflow","Inflowing metal-poor gas builds the low-alpha sequence","Bimodal alpha sequences linked to late gas accretion","Simulated galaxies reveal gas-driven alpha bimodality"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000186,"raw_usage":{"total_tokens":1373,"prompt_tokens":1039,"completion_tokens":334,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":655,"completion_tokens_details":{"reasoning_tokens":252}},"tokens_in":655,"tokens_out":334,"duration_ms":3621,"temperature":1.0,"reasoning_tokens":252,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T17:15:18.410635+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Track the four strong-bimodal galaxies analyzed in the companion paper: if any shows a clear low-alpha sequence without a concurrent rise in metal-poor gas inflow and gas-disk growth over the last 5–6 Gyr, the claimed mechanism fails. Alternatively, find a simulated or observed Milky Way-mass galaxy whose low-alpha sequence formed while its accretion was declining and metal-rich.","supporting_citations":[],"review_version":1}