{"id":"6dfb7613-9a97-459c-bf88-0c2550a7487b","arxiv_id":"2501.04084","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Low-mass galaxies with high perturbation indices are more extended and dark-matter poor at fixed stellar mass, with both environment and halo mass contributing to the scatter.","lead":"Using the FIREbox cosmological simulation, this study finds that low-mass galaxies in strongly tidal environments are systematically larger and poorer in dark matter than similar-mass galaxies in calmer regions. The result supports the view that a galaxy's local surroundings, not only its halo mass, shape its structure, with implications for upcoming dwarf galaxy surveys.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The direct/indirect environmental decomposition is compromised because the Perturbation Index (Eq. 5) contains the target galaxy's own Mhalo and rhalo, so high PI correlates with low Mhalo mechanically; the paper's evidence does not isolate physical environment from this construction.","rationale":"The reader's weakest assumption identifies the same load-bearing issue: PI is defined with the target halo's own mass and radius, so the environmental variable is mechanically entangled with Mhalo. I agree with the conditional verdict. The paper has genuine strengths: a well-resolved cosmological volume simulation, a clearly described sample, and repeated random train-test splits that demonstrate stability of the Random Forest importances. However, the headline claim of direct and indirect environmental effects rests on PI as an independent measure of environment, and that premise is not secure. The proposed check replaces PI with an environment metric that does not contain the target halo mass; this directly tests whether the observed correlations are physical or built into Eq. 5. If the trends survive, the paper's interpretation would be substantially stronger. If they do not survive, the abstract overstates the support for the direct/indirect decomposition. I would therefore keep the current conditional verdict, contingent on this test or an equivalent mediation analysis.","tokens_in":21627,"tokens_out":3442,"duration_ms":38630,"concrete_test":"","verdict_should_be":"UNCHANGED","load_bearing_attack":"The abstract's central claim requires an environmental measure independent of halo mass. Equation 5 defines PI = sum_c (M_c/M_p)(r_p/D)^3, where M_p and r_p are the target galaxy's own halo mass and radius. For satellites, Mhalo is the tidally truncated mass, so stripping lowers both M_p and r_p, inflating PI even if the neighbor distribution is unchanged. At fixed stellar mass, lower Mhalo is independently associated with larger r50 and lower M50_DM (Fig. 7). Thus the positive Delta-log-r50 versus PI trend in Fig. 5 and the PI feature importance in Fig. 6 are partly guaranteed by the variable's construction. The Random Forest cannot resolve this: feeding both log PI and log Mhalo as features lets the model exploit the algebraic link rather than a physical environmental effect. Figure 8 attempts to control for Mhalo by binning, but the PI axis still uses Mhalo in its definition, so residual trends can reflect within-bin Mhalo variation. Therefore the claimed direct and indirect environmental effects are not established by the current evidence.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript uses the FIREbox cosmological volume simulation to study roughly 1,200 low-mass galaxies (M_star between 10^6 and 10^9 M_sun), analyzing the size–mass relation, the inner dark-matter–stellar-mass relation, and the halo-mass–stellar-mass relation at fixed stellar mass. The environment is quantified by the Perturbation Index (PI) defined in Eq. (5), and the analysis compares PI > 1 and PI < 1 subpopulations, then trains a Random Forest regressor on log M_halo, log M_star/M_halo, and log PI to predict residuals from the median central-galaxy scaling relations. The paper reports that galaxies with higher PI are larger and less massive in their halos at fixed stellar mass, that all three features contribute to predicting size residuals while M_halo dominates inner dark-matter residuals, and concludes that environment influences galaxy size and inner dark-matter content both directly and indirectly through its effect on halo mass.","tokens_in":21852,"tokens_out":4405,"duration_ms":48893,"significance":"If the claimed direct/indirect decomposition were established, the paper would provide a notable result: environment, not only halo mass, would be a fundamental driver of low-mass galaxy structure, with implications for dwarf galaxy diversity and future survey analyses. The manuscript has real strengths: it draws on a high-resolution cosmological volume simulation, uses a sizable low-mass sample, and the Random Forest importance scores are reported as stable across 500 train/test splits. However, the central decomposition is not established, because the PI as defined in Eq. (5) includes the target galaxy's own halo mass and radius, so the paper's environmental variable is partly a re-expression of the very quantity it claims to act alongside. This is a load-bearing issue for the abstract's main claim, and it requires a substantive revision rather than a local fix.","major_comments":[{"comment":"The Perturbation Index is not an independent environmental metric: PI = sum_c (M_c/M_p)(r_p/D)^3 contains the target galaxy's own halo mass M_p and halo radius r_p in the denominator. For satellites, M_p and r_p are the tidally truncated values defined in Sec. 2.2, so tidal stripping lowers both and inflates PI even if the surrounding neighbor distribution is unchanged. At fixed stellar mass, the paper itself shows (Fig. 7) that lower M_halo correlates with larger Delta log r50 and smaller Delta log M50_DM. Therefore the positive Delta log r50 versus PI trend in Fig. 5 and the lower M_halo sequence for PI > 1 in Fig. 4 are partly guaranteed by the construction of PI, not by a separately measured environment. The abstract's concluding statement that environment acts indirectly through its impact on halo mass is not supported, because the causal chain is entangled with a definitional link.","section":"Eq. (5), Sec. 2.4"},{"comment":"The control attempted in Fig. 8 does not solve the circularity. Even after binning by M_halo, the PI axis is still a function of M_halo through the M_p and r_p terms in Eq. (5). Within a halo-mass bin, galaxies at the lower-mass edge will have mechanically larger PI values, so the reported weak positive Delta log r50 versus PI correlation in the intermediate bins may simply reflect residual within-bin M_halo variation. To claim a direct environmental effect at fixed mass, the authors need an environmental measure that does not use the target's own M_halo or r_halo, or they need to demonstrate explicitly that within-bin M_halo variation cannot drive the trend.","section":"Sec. 3.4, Fig. 8"},{"comment":"The Random Forest analysis feeds both log PI and log M_halo as features even though the two are algebraically related through Eq. (5) at fixed stellar mass. With correlated features, tree-based importance scores can be split arbitrarily between the two, so Fig. 6 cannot identify a 'direct' environmental contribution separate from the halo-mass contribution. The moderate test-set R2 values (0.32 and 0.41) do not mitigate this issue, because the model can exploit the algebraic link rather than a physical environmental effect. A cleaner test would replace PI with an external tidal-field or neighbor-density metric that excludes the target-object properties, or use permutation importance with a deliberately constructed control feature.","section":"Secs. 2.5 and 3.3, Fig. 6"}],"minor_comments":[{"comment":"The text says halos are 'spherical systems with viral radii'; 'viral' should be 'virial.'","section":"Sec. 2.2"},{"comment":"The hyperparameter table is hard to read: the columns labeled 'max features' and 'min samples leaf' contain values 'sqrt' and '3,' but it is not stated which target each row's optimized values refer to; please state explicitly that 'sqrt' is the max_features setting and 3 is min_samples_leaf for both targets.","section":"Table 1"},{"comment":"The phrase 'lower masses' in the abstract is ambiguous; the manuscript actually claims lower halo masses, while the inner dark-matter relation is reported as similar for PI > 1 and PI < 1 except at low stellar mass. Please specify which mass is meant.","section":"Abstract and Sec. 3.2"},{"comment":"Four galaxies with M_star below 4 x 10^6 M_sun are excluded from Fig. 2 but included in the rest of the analysis; please state whether any reported median trends change if these four galaxies are excluded everywhere.","section":"Sec. 2.3"},{"comment":"The claim of a 'weak positive correlation' between Delta log r50 and PI in the two intermediate mass bins of Fig. 8 is not quantified; reporting a Spearman correlation coefficient and the associated uncertainty for each panel would make the visual claim testable.","section":"Sec. 3.4"},{"comment":"The text says GridSearchCV optimizes max_features and min_samples_leaf, then notes that setting max_features=None changes absolute importances but not rankings; please provide these alternative values or a figure version so the reader can assess the sensitivity quantitatively.","section":"Sec. 2.5.2"},{"comment":"There is a typo in the summary section: 'through it's effect' should be 'through its effect.'","section":"Sec. 4"}],"recommendation":"major_revision","confidential_remarks":"The paper's qualitative trends are visible in the simulation, but the central 'direct plus indirect' claim is compromised by the definition of PI in Eq. (5). I recommend major revision rather than rejection because the issue can, in principle, be addressed by redoing the analysis with an environmental metric that does not include the target galaxy's own M_halo and r_halo, and by reframing the claims to distinguish what is measured from what is inferred."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe paper is a solid simulation analysis with a clear qualitative result, but the headline claim outruns the evidence. The new piece is the specific application: roughly 1,200 low-mass galaxies from FIREbox, split by Perturbation Index, with a Random Forest ranking Mhalo, Mstar/Mhalo, and PI as predictors of size and inner-DM residuals. The qualitative trends are visible and credible: at fixed stellar mass, high-PI galaxies are more extended and live in lower-mass halos. The RF importances are stable across 500 train/test splits, and the quadrant analysis showing satellites as DM-poor and extended is a nice touch. The writing is clear and the sample selection is documented.\n\nThe soft spot is the causal decomposition. PI by construction includes the target's own Mhalo and rhalo (Eq. 5), so using it as an independent environmental variable and then concluding that environment affects Mhalo is partly circular. For centrals the dependence largely cancels because r_vir scales as M_vir^{1/3}, so the concern is weaker there than the stress-test note implies. But for satellites, r_halo is the truncated radius, and the coupling does not cancel cleanly. More importantly, the abstract's 'direct and indirect' language is a causal claim, and the paper never actually tests mediation. It shows correlations, not pathways. The RF feature importance is descriptive; without permutation tests or a mediation analysis, the 'indirect through halo mass' step is asserted, not established. Figure 8's binning by Mhalo is a reasonable attempt, but residual within-bin Mhalo variation remains, and the trend is weak.\n\nNone of this kills the paper. The empirical trends are useful, and the simulation is not misrepresented. But the title and abstract overstate the case. I'd suggest the authors either soften the causal language or add a direct test, e.g., tracking individual stripping histories or using an environment metric that does not contain the target's own mass.\n\nThis is for people working on dwarf scaling relations and environmental effects. It deserves a serious referee; I'd send it to review, but with the expectation that the indirect-effect claim be toned down or re-analyzed.","headline":"Useful simulation trends, but the direct/indirect environmental claim is not established because PI is entangled with the target's own halo mass.","tokens_in":22427,"tokens_out":4575,"would_cite":false,"duration_ms":45124,"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":"In a cosmological simulation, low-mass galaxies in stronger tidal environments end up systematically larger and with less dark matter, making environment a driver on par with halo mass.","keywords":["dwarf galaxies","scaling relations","galaxy environment","Perturbation Index","tidal interactions","dark matter content","cosmological simulation","Random Forest"],"falsifier":"Replace the Perturbation Index with an environment measure that does not include the target galaxy's own halo mass or radius, such as the number of neighbours within a fixed 1 Mpc aperture or the distance to the nearest halo more massive than $10^{11.5}\\,M_\\odot$, and rerun the residual and Random Forest analyses; if the claimed direct environmental correlations disappear or shrink drastically, the paper's direct/indirect decomposition is an artifact of the PI normalization.","tokens_in":21436,"feed_emoji":"🌌","tokens_out":11324,"duration_ms":91439,"temperature":0.7,"pith_summary":"This paper argues that a galaxy's local surroundings are a central ingredient in how small galaxies end up, not a secondary correction to halo mass. At fixed stellar mass in a cosmological volume simulation, galaxies experiencing stronger tidal perturbations are systematically more extended, live in lower-mass halos, and tend to be poorer in inner dark matter. A Random Forest regression shows that environment, quantified by the Perturbation Index, and halo mass are both significant predictors of how far a galaxy sits above or below the median size and dark-matter relations. Because halo mass itself is depressed by the environment at fixed stellar mass, the authors conclude that environment acts on galaxy structure both directly and indirectly. If correct, this places environment on par with halo mass as a fundamental driver of dwarf galaxy formation and evolution.","feed_headline":"Strong tides stretch dwarf galaxies and drain their dark matter","feed_subtitle":"If true, environment ranks with halo mass in setting dwarf size and dark matter content.","key_machinery":"The central object is the Perturbation Index, $\\mathrm{PI} = \\sum_c (M_c^{\\rm halo}/M_p^{\\rm halo})\\,(r_p^{\\rm halo}/D_{cp})^3$, the ratio of the tidal force a galaxy feels from its neighbours to its own binding force. It splits the sample into strongly perturbed ($\\mathrm{PI}>1$) and weakly perturbed ($\\mathrm{PI}<1$) galaxies and is used as one of three Random Forest features, together with halo mass $M_{\\rm halo}$ and the stellar-to-halo mass ratio $M_\\star/M_{\\rm halo}$. The targets are residuals, $\\Delta\\log r_{50}$, $\\Delta\\log M_{\\rm DM}^{50}$, and $\\Delta\\log M_{\\rm halo}$, defined as logarithmic offsets from the median relations of central galaxies at fixed stellar mass. The machine-learning model provides feature importance scores that the paper uses to separate the direct environmental channel from the mass channel.","core_discovery":"Using roughly 1,200 galaxies (886 centrals and 332 satellites) with stellar masses $10^6$–$10^9\\,M_\\odot$ from FIREbox, the paper shows that the residuals around the size–mass, inner dark matter mass–stellar mass, and halo mass–stellar mass relations of central galaxies correlate with the Perturbation Index. Galaxies with $\\mathrm{PI} > 1$ sit, on average, above the median $r_{50}$–$M_\\star$ relation (more extended) and below the median $M_{\\rm halo}$–$M_\\star$ relation (lower halo mass) compared with galaxies with $\\mathrm{PI} < 1$; the inner dark matter mass relation is similar for the two populations except at the lowest stellar masses, where environment and feedback become more important. Random Forest models trained on $\\log M_{\\rm halo}$, $\\log(M_\\star/M_{\\rm halo})$, and $\\log(\\mathrm{PI})$ achieve mean test $R^2 \\approx 0.32$ for relative size and $R^2 \\approx 0.41$ for relative inner dark matter content. In the model, all three features matter comparably for relative size, while halo mass is the dominant predictor of inner dark matter content. Since $M_{\\rm halo}$ at fixed stellar mass is itself lowered by the environment, the paper's central claim is that environmental conditions shape galactic sizes and inner dark matter content directly and indirectly through halo mass.","pith_inferences":["A testable consequence the paper does not run: if the Perturbation Index is replaced by an environment metric that does not divide by the target galaxy's own halo mass and radius, such as fixed-aperture neighbour counts or distance to the nearest massive halo, a physical direct environmental channel should survive, while a pure normalisation artifact should weaken or vanish.","The paper's indirect channel implies that some of the scatter in low-mass galaxy rotation curves now attributed to baryonic feedback or halo concentration could instead record tidal histories, a prediction that could be checked with zoom-in simulations tracking halo mass loss separately from star formation.","Upcoming wide surveys of dwarf galaxies could look for field dwarfs near massive neighbours whose sizes and inner dark matter deficits match high-PI simulated systems, turning the simulation result into an observational test.","The same logic suggests that the most extended ultra-diffuse galaxies may preferentially be found in moderately perturbed, low-halo-mass systems rather than only in dense clusters, a sharper prediction than the paper states."],"forward_implications":["At fixed stellar mass, low-mass galaxies with $\\mathrm{PI}>1$ are systematically more extended and have lower halo masses than their $\\mathrm{PI}<1$ counterparts.","Relative size scatter is set by a combination of halo mass, stellar-to-halo mass ratio, and environment, while relative inner dark matter content is dominated by halo mass.","Because environment suppresses halo mass at fixed stellar mass, environmental effects reach galaxy structure both directly and through the halo-mass channel.","More than half of satellite galaxies in the sample are dark-matter-poor and extended, consistent with tidal stripping during infall into a host.","At the lowest stellar masses ($M_\\star < 10^7\\,M_\\odot$), shallow potential wells make inner dark matter content especially sensitive to environment and feedback."],"supporting_citations":[{"why":"Supplies the FIREbox cosmological volume simulation and the FB1024 run that produces the galaxy sample.","marker":"Feldmann et al. 2022"},{"why":"Provides the parent catalog from which the 1,218 galaxies are drawn and identifies the seven dark-matter-deficient outliers later excluded.","marker":"Moreno et al. 2022"},{"why":"Defines the FIRE-2 feedback physics used in FIREbox for star formation and stellar feedback.","marker":"Hopkins et al. 2018"},{"why":"Establishes the Perturbation Index formula adopted to quantify the tidal environment.","marker":"Verley et al. 2007"},{"why":"Gives the low-mass size–mass relation used as reference for the size residuals.","marker":"Yoon et al. 2023"},{"why":"Defines the inner dark matter mass–stellar mass relation that anchors the $M_{\\rm DM}^{50}$ residuals.","marker":"Kravtsov 2024"},{"why":"Provides an observational size–environment correlation at intermediate stellar masses that motivates studying the low-mass regime.","marker":"Ghosh et al. 2024"},{"why":"Introduces the Random Forest regression method used for feature importance.","marker":"Breiman 2001"}],"fun_headline_variants":["Tidal forces stretch dwarf galaxies and strip their dark matter","Environment ranks with halo mass in setting dwarf size and dark matter","Tidal environment inflates low-mass galaxies and cuts dark matter","Galaxy environment rivals halo mass in shaping dwarf size and dark matter","FIREbox: strong tides stretch dwarf galaxies and drain dark matter"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the Perturbation Index measures the surrounding environment independently of the target galaxy's own properties, but the index is defined with the target galaxy's halo mass and radius in the denominator, so at fixed stellar mass a galaxy with a smaller halo automatically has a larger PI even if its surrounding tidal field is unchanged.","fun_headline_variants_meta":{"raw":{"variants":["Tidal forces stretch dwarf galaxies and strip their dark matter","Environment ranks with halo mass in setting dwarf size and dark matter","Tidal environment inflates low-mass galaxies and cuts dark matter","Galaxy environment rivals halo mass in shaping dwarf size and dark matter","FIREbox: strong tides stretch dwarf galaxies and drain dark matter"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000716,"raw_usage":{"total_tokens":3306,"prompt_tokens":1120,"completion_tokens":2186,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":736,"completion_tokens_details":{"reasoning_tokens":2099}},"tokens_in":736,"tokens_out":2186,"duration_ms":12824,"temperature":1.0,"reasoning_tokens":2099,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T21:41:20.190884+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Replace the Perturbation Index with an environment measure that does not include the target galaxy's own halo mass or radius, such as the number of neighbours within a fixed 1 Mpc aperture or the distance to the nearest halo more massive than $10^{11.5}\\,M_\\odot$, and rerun the residual and Random Forest analyses; if the claimed direct environmental correlations disappear or shrink drastically, the paper's direct/indirect decomposition is an artifact of the PI normalization.","supporting_citations":[{"cited_title":"2023, , 957, 59, 10.3847/1538-4357/acfed5","cited_arxiv_id":null,"evidence_quote":"Gives the low-mass size–mass relation used as reference for the size residuals."}],"review_version":1}