{"id":"b66fe3a7-6c4a-4a93-a69f-a77c9102c606","arxiv_id":"2505.16231","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Coupling atmospheric retrievals to a Gibbs free energy model shows modern Earth-like chemical disequilibrium is hard to constrain: reflected light is limited by methane, and M dwarf transit constraints need 1 to 2 ppm noise.","lead":"This paper tests whether future telescopes could measure the chemical disequilibrium of an Earth-like exoplanet's atmosphere, a proposed sign of life. It finds that reflected-light observations cannot pin down methane on a modern Earth twin, and only extremely low-noise transit observations could do so.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The AGFE posteriors are built by independently sampling marginal retrieval posteriors rather than the joint posterior, so unrecognized parameter covariances could change the quoted detectability thresholds.","rationale":"Reader's weakest_assumption is precisely the marginal-sampling step in Section 2.3. I agree that this is the single most load-bearing concern. The central qualitative message — reflected-light CH4 is hard and transit AGFE needs very low noise — is supported by the retrieval marginal posteriors themselves (CH4 upper limits at SNR 10-40; O2 and CH4 broadening at 5 ppm), so it is likely robust. The quantitative thresholds (AGFE peaks at order-of-magnitude low; tight at 1-2 ppm; unconstrained at 5 ppm) depend on the AGFE propagation, which is where independent marginal sampling can fail. I would not reject or accept; the paper already presents a conditional verdict, and the requested joint-posterior check is the natural condition to satisfy. The SNR-192 methane detection threshold in Section 4 is asserted without derivation and would also benefit from a written calculation, but it is secondary because the retrieval results already show CH4 unconstrained below SNR 40.","tokens_in":14727,"tokens_out":3468,"duration_ms":28837,"concrete_test":"Recompute the AGFE posteriors using draws from the joint posterior instead of independent marginals: take post-burn-in MCMC samples from the 10 chains (or a Kernel-Density-Estimate of the joint distribution) and feed each full parameter vector through the Gibbs model, repeating for the SNR 20 and 40 reflected-light cases and the 1, 2, and 5 ppm transit cases. If the resulting AGFE credible intervals and median biases stay within, say, 20% of Figures 7 and 10, the marginal-sampling concern does not land; if they shift by more than that or the 5 ppm case becomes constrained (or the 1 ppm case becomes unconstrained), the headline detectability thresholds need revision.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 2.3 and Figure 2 describe coupling retrievals to the thermodynamics model by 'randomly sampl[ing] the marginal posterior distributions' of O2, CH4, H2O, CO2, O3, P0, and T0. This is equivalent to drawing from the product of the marginals, which equals the joint posterior only if all seven parameters are independent. The paper does not establish independence, and the appendix corner plots (Figures 12-17) are the natural place to check; text in Section 3.2 already notes profile-induced biases in P0 and O3, and CH4 and T0 are both retrieved from overlapping near-IR features, so correlation is plausible. If parameters are correlated, the propagated AGFE distribution is not the true posterior: it can be too narrow (if correlated parameters partially cancel in Phi) or too broad or wrongly centered (if they add), directly affecting the claims that reflected-light AGFE is biased low and that 1-2 ppm transit noise yields tight constraints while 5 ppm loses the signal. Because every quantitative AGFE result in Figures 7 and 10 passes through this step, this is the most load-bearing assumption in the paper.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper couples the rfast atmospheric retrieval code to a Gibbs free energy thermodynamics model in order to infer the available Gibbs free energy (AGFE) of modern Earth-like exoplanet analogs. For reflected-light observations of an Earth-Sun twin at V-band SNRs of 10, 20, and 40, the authors find that CH4 is only upper-limit constrained and that the resulting AGFE posterior is biased low by about an order of magnitude. For a TRAPPIST-1e analog observed in transit with JWST MIRI at 1, 2, and 5 ppm noise, they report tight AGFE constraints only at 1-2 ppm, with the signal becoming unconstrained at 5 ppm. The AGFE posterior is constructed by independently sampling the marginal posteriors of seven retrieved parameters and passing those draws to the thermodynamics model (Section 2.3, Figure 2).","tokens_in":14930,"tokens_out":11656,"duration_ms":97093,"significance":"If the quantitative thresholds survive a proper joint-posterior propagation, this is a useful benchmark for future life-detection observing strategies. The study is an injection-recovery test in which the same forward model generates the synthetic observation and is used in the retrieval, so it measures pipeline self-consistency rather than an independent measurement; this is appropriate for a detectability study. Strengths include explicit burn-in and convergence checks (Section 2.1), a constant-profile control retrieval that removes known isoprofile/isothermal biases (Appendix Figure 11), and public data and code links. The central quantitative claims, however, rest on an unvalidated independence assumption in the coupling step, so the reported noise thresholds should be treated as conditional until that assumption is tested.","major_comments":[{"comment":"The AGFE posterior is built by \"randomly sampling the marginal posterior distributions\" of P0, T0, O2, H2O, CO2, O3, and CH4. This is equivalent to drawing from the product of seven one-dimensional marginals, which equals the joint posterior only if the parameters are independent. The paper does not establish independence: the appendix corner plots (Figures 12-17) are the natural diagnostic, but no covariance summary or independence test is reported, and correlations are plausible (e.g., CH4 and T0 both affect overlapping water/methane features; Section 3.2 already reports profile-induced biases in P0 and O3). Because every AGFE posterior in Figures 7 and 10 is produced through this step, the central quantitative claims—CH4-limited AGFE detection for reflected light at SNRs 10-40, and tight AGFE only at 1-2 ppm transit noise with loss at 5 ppm—are not robust until the propagation is repeated with draws from the joint MCMC posterior or the independence assumption is explicitly validated. The independent-marginal scheme also risks generating unphysical parameter combinations, such as mixing ratios summing above unity; the manuscript does not describe any rejection or constraint applied during the N2 back-fill.","section":"Section 2.3, Figure 2"},{"comment":"The reflected-light retrieval adopts isothermal, constant-mixing-ratio profiles while the synthetic data are generated from altitude-dependent Atmos profiles. The authors show that this assumption biases P0 substantially high (about an order of magnitude in Section 3.2; a factor of 5 in Section 4) and biases O3 relative to the column average. They assert that these parameters have a negligible effect on the AGFE, but no sensitivity test is provided. Since Eq. (4) contains an explicit pressure term, RU T ln(NiP/NP°), a quantitative check—for example, propagating the constant-profile control retrieval (Figure 11) through the thermodynamics model, or recomputing the AGFE with the biased P0—is needed to separate the claimed low-bias AGFE result from retrieval-systematics effects.","section":"Section 3.2, Eq. (4)"}],"minor_comments":[{"comment":"Equations (1)-(3) mix scalar and vector notation in a way that is hard to follow; a short notation table would improve reproducibility.","section":"Section 2.1"},{"comment":"The text says observations have \"constant noise specified at V-band,\" while Section 2.1 says the noise model simulates \"constant signal-to-noise along the full wavelength range\"; these statements should be reconciled.","section":"Section 3.2"},{"comment":"The calculation of the SNR 192 requirement for a CH4 detection is not described; the readership cannot reproduce this number without the spectral-differencing formula and the assumed noise scaling.","section":"Section 4"},{"comment":"The AGFE posterior distributions should be summarized numerically (median and credible interval) in the text or captions, since the visual histograms alone do not support the stated \"tight\" versus \"unconstrained\" thresholds.","section":"Figures 7 and 10"},{"comment":"Several typographical and wording issues remain, including \"Therfastretrieval model\" in Section 2.1, \"affect\" for \"effect\" in Section 3.2, and \"repoted\" in Section 4.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The paper is a carefully executed injection-recovery study and the writing is generally clear. The main technical risk is the independent-marginal propagation of the AGFE posterior; I would ask the authors to rerun the AGFE calculation using post-burn-in joint-chain samples, or to provide an explicit independence/covariance test from the corner plots. The second concern about P0 bias can likely be addressed with a short sensitivity test. If the joint-chain propagation changes the AGFE widths or centers, the conclusions may still hold, but the current manuscript does not demonstrate that."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague, the headline is that the paper gives the community a clean, sobering number: detecting modern Earth-like O2/CH4 disequilibrium is really hard. In reflected light, CH4 basically cannot be constrained below SNR 40, and the needed SNR for a detection is ~192. In transit around an M dwarf, the AGFE is only tightly constrained at 1-2 ppm noise with MIRI, which is likely implausible. These thresholds are new and will be used in mission planning, so it's worth getting the details right.\n\nThe paper is careful in most ways. The retrieval setup is standard, noise injection is explicit, burn-in is checked, and they include a constant-profile control to show the known isoprofile biases. The method itself (coupling retrievals to the Gibbs free energy code) was published in the authors' earlier Nature Astronomy paper; this is an extension to a modern Earth twin and a TRAPPIST-1e-like transit case. Code and data are available, which is good.\n\nThe soft spot is the coupling step. They build the AGFE posterior by independently sampling the marginal posteriors of O2, CH4, H2O, CO2, O3, P0, and T0 (Section 2.3, Figure 2). That is the product of marginals, which is only the joint posterior if the parameters are independent. Retrieval parameters are rarely independent—CH4 and T0 overlap in the near-IR, and the corner plots in the appendix likely show covariances. If there are correlations, the propagated AGFE distribution can be too narrow or shifted, and that directly affects the claimed tightness at 1-2 ppm and the low bias in reflected light. The paper does not validate marginal sampling against joint-chain draws, so this is a real concern, though it is fixable: just draw samples from the actual MCMC chain and run them through the thermodynamic model. I don't think the main qualitative conclusion changes, but the quantitative thresholds should be re-derived.\n\nA smaller issue: the SNR 192 methane requirement is asserted without a derivation. It's a side comment in Section 4, but if it's going to be quoted, it needs a method.\n\nWho is this for? The direct imaging and JWST/MIRI mission planning community, and anyone building life detection strategies. It is a legitimate, if incremental, contribution.\n\nI would send it to peer review. The right referee report would ask for the joint sampling validation and a derivation of the SNR 192 number. Fix those and the quantitative outputs are trustworthy.","headline":"Modern Earth chemical disequilibrium is a tough biosignature to constrain—this paper gives useful numbers, but the AGFE coupling step should be fixed before quoting them.","tokens_in":15497,"tokens_out":2691,"would_cite":true,"duration_ms":20942,"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":"Chemical disequilibrium on a modern Earth twin is only inferable when methane is pinned down; reflected-light spectra up to SNR 40 cannot do it, and JWST MIRI transit spectra need 1–2 ppm noise.","keywords":["chemical disequilibrium","available Gibbs free energy","biosignature gases","exoplanet atmospheric retrieval","reflected light spectroscopy","transit spectroscopy","JWST MIRI","TRAPPIST-1e"],"falsifier":"Re-run the same simulated retrievals but compute the Gibbs free energy by drawing all seven inputs from the full MCMC joint chain instead of from independent marginals; if the resulting available Gibbs free energy posterior shifts significantly from the paper's distributions, the coupling scheme is the limiting assumption. Alternatively, a real JWST MIRI transit observation of TRAPPIST-1e reaching 5 ppm noise that still recovers both O$_2$ and CH$_4$ would rule out the claimed 5 ppm loss of the disequilibrium signal.","tokens_in":14521,"feed_emoji":"🔭","tokens_out":12517,"duration_ms":91940,"temperature":0.7,"pith_summary":"Modern Earth's atmosphere is out of chemical equilibrium because biology pumps oxygen and methane into it at the same time, and the paper asks whether that imbalance can be seen from another star. It couples atmospheric retrievals of simulated spectra to a thermodynamics model that computes the available Gibbs free energy $\\Phi$, the gap between the observed composition and its theoretical equilibrium state. The answer depends almost entirely on how well oxygen, methane, and temperature are known. In reflected light at signal-to-noise ratios of 10, 20, and 40, methane at modern Earth abundance is effectively unconstrained, so the inferred $\\Phi$ is biased low by about an order of magnitude and only upper limits are possible. In transit around TRAPPIST-1e with JWST MIRI, tight constraints on $\\Phi$ require noise of 1–2 ppm and disappear at 5 ppm.","feed_headline":"Detect methane first to see an Earth-like life signal","feed_subtitle":"Even SNR 40 spectra give only upper limits; JWST transit data need 1–2 ppm noise to catch it.","key_machinery":"The available Gibbs free energy, $\\Phi \\equiv \\Delta_f G(T,P,\\{N_i\\}_{\\rm obs}) - \\Delta_f G(T,P,\\{N_i\\}_{\\rm eqm})$, is the metric that carries the argument. The pipeline that produces it works by running an MCMC retrieval on a synthetic spectrum, then randomly drawing the marginal posterior distributions of seven retrieved parameters — surface pressure, atmospheric temperature, and the O$_2$, H$_2$O, CO$_2$, O$_3$, and CH$_4$ mixing ratios — and feeding those draws into a Gibbs free energy minimization model that computes the distance between the observed composition and the equilibrium composition. The paper's sensitivity claims follow from how this pipeline responds to different noise levels: the CH$_4$ marginal posterior, and the temperature that enters the thermodynamic calculation, are what control whether the resulting $\\Phi$ distribution peaks near the truth or collapses into an upper limit.","core_discovery":"The central discovery is that the detectability of modern Earth-like chemical disequilibrium biosignatures is limited by the observable gases that carry the disequilibrium, not by the thermodynamics itself. For a modern Earth analog observed in reflected light, O$_2$ is constrained to within an order of magnitude at all tested SNRs, but CH$_4$ at its 2 ppm surface mixing ratio is only an upper limit, and the resulting available Gibbs free energy posterior is biased low by roughly an order of magnitude. For a modern Earth analog transiting a late M dwarf and observed with JWST MIRI from 5 to 12 µm, the higher CH$_4$ abundance of the M-dwarf case makes the available Gibbs free energy ($\\sim 320$ J mol$^{-1}$ versus $\\sim 1$ J mol$^{-1}$ for the Sun case) potentially constrainable, but only at instrument noise of 1–2 ppm; at 5 ppm both O$_2$ and CH$_4$ go unconstrained and the disequilibrium signal is lost. The paper gives the minimum SNR for a reflected-light CH$_4$ detection at modern Earth abundance as 192, computed by summing the wavelength-dependent SNR across the 1.64–1.7 µm feature.","pith_inferences":["The paper does not validate the independent-marginal sampling scheme against joint-chain draws; if CH$_4$ correlates with temperature or clouds, the quoted Gibbs free energy posteriors are not the true joint posteriors.","Because the transit simulations are cloud-free, they likely represent an optimistic bound; including high-altitude clouds would probably require even lower noise than 1–2 ppm.","Using O$_3$ as an O$_2$ proxy in the coupled pipeline, which the paper mentions as a possibility, could tighten the oxygen side of the disequilibrium constraint in reflected light.","The same retrieval-to-thermodynamics coupling could be applied to any exoplanet spectrum to screen for thermodynamic imbalance, making it a general agnostic-biosignature tool rather than an Earth-specific one."],"forward_implications":["For direct imaging of a modern Earth twin, available Gibbs free energy will usually be an upper limit unless methane can be detected, which requires roughly SNR 192 in the 1.64–1.7 µm band.","A reflected-light observation at SNR 40 that fails to constrain methane will place the inferred Gibbs free energy about an order of magnitude below the true value, making a biotically active planet look closer to thermodynamic equilibrium than it is.","For a transiting Earth-like planet around a late M dwarf, JWST MIRI can deliver tight Gibbs free energy constraints only if the noise floor is near 1–2 ppm; at 5 ppm the O$_2$–CH$_4$ disequilibrium signal is lost.","In the M-dwarf case the available Gibbs free energy is comparable in magnitude to Mars's abiotically generated value, so a chemical disequilibrium detection alone cannot distinguish biology from photochemistry; identifying the species driving the signal is required.","Chemical disequilibrium is best treated as one line of evidence in a hierarchy of biosignatures rather than a standalone life-detection metric."],"supporting_citations":[{"why":"Defines the available Gibbs free energy metric and supplies the modern Earth gas-phase value ($\\sim 1$ J mol$^{-1}$) against which the paper's inferred posteriors are compared.","marker":"Krissansen-Totton et al. (2016)"},{"why":"Supplies the rfast retrieval and forward model used to generate and invert both the reflected-light and transit spectra.","marker":"Robinson & Salvador (2023)"},{"why":"Motivates the JWST MIRI 5–12 µm window for the TRAPPIST-1e transit simulations and the detectability of the O$_2$ collision-induced absorption feature.","marker":"Fauchez et al. (2020)"},{"why":"Provides the cold-terminator mixing ratios for the modern Earth-like TRAPPIST-1e atmospheric scenario.","marker":"Pidhorodetska et al. (2020)"},{"why":"Is the prior work this paper extends, showing that a Proterozoic Earth analog's available Gibbs free energy can be constrained to within an order of magnitude at SNR 50.","marker":"Young et al. (2024)"},{"why":"Establishes the retrieval chain setup of 10 chains, 200 walkers, and 100,000 steps with randomized spectral noise.","marker":"Feng et al. (2018)"},{"why":"Further develops the Gibbs free energy disequilibrium framework used in the thermodynamic model.","marker":"Krissansen-Totton et al. (2018b)"}],"fun_headline_variants":["Methane is the bottleneck for Earth-like biosignature detection","JWST needs 1-2 ppm noise to see Earth-like disequilibrium","Chemical disequilibrium detection hinges on methane constraints","Earth-like life signal requires ultra-precise transit spectra","Reflected-light spectra can't pin down Earth-like disequilibrium"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that randomly drawing each retrieved parameter from its own marginal posterior is as good as drawing from the full joint posterior; if methane is correlated with temperature or clouds, the available Gibbs free energy distribution produced this way is not the true joint distribution.","fun_headline_variants_meta":{"raw":{"variants":["Methane is the bottleneck for Earth-like biosignature detection","JWST needs 1-2 ppm noise to see Earth-like disequilibrium","Chemical disequilibrium detection hinges on methane constraints","Earth-like life signal requires ultra-precise transit spectra","Reflected-light spectra can't pin down Earth-like disequilibrium"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000727,"raw_usage":{"total_tokens":3327,"prompt_tokens":1086,"completion_tokens":2241,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":702,"completion_tokens_details":{"reasoning_tokens":2154}},"tokens_in":702,"tokens_out":2241,"duration_ms":13479,"temperature":1.0,"reasoning_tokens":2154,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T15:04:40.573184+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-run the same simulated retrievals but compute the Gibbs free energy by drawing all seven inputs from the full MCMC joint chain instead of from independent marginals; if the resulting available Gibbs free energy posterior shifts significantly from the paper's distributions, the coupling scheme is the limiting assumption. Alternatively, a real JWST MIRI transit observation of TRAPPIST-1e reaching 5 ppm noise that still recovers both O$_2$ and CH$_4$ would rule out the claimed 5 ppm loss of the disequilibrium signal.","supporting_citations":[{"cited_title":"S., & Catling, D","cited_arxiv_id":null,"evidence_quote":"Defines the available Gibbs free energy metric and supplies the modern Earth gas-phase value ($\\sim 1$ J mol$^{-1}$) against which the paper's inferred posteriors are compared."},{"cited_title":"J., Villanueva, G","cited_arxiv_id":null,"evidence_quote":"Motivates the JWST MIRI 5–12 µm window for the TRAPPIST-1e transit simulations and the detectability of the O$_2$ collision-induced absorption feature."}],"review_version":1}