{"id":"b167239c-ab0f-488f-9a51-ed7f6f96eb28","arxiv_id":"2505.13407","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":5,"one_line_summary":"A joint re-analysis of NIRISS, NIRSpec, and MIRI spectra of K2-18 b finds no statistically significant evidence for DMS or DMDS, with ethane offering an equally good fit.","lead":"A team re-analyzed all JWST observations of the sub-Neptune K2-18 b and found no robust evidence for the proposed biomarker gases DMS and DMDS. Their joint 0.6 to 12 micron spectrum shows that simpler hydrocarbons like ethane fit the data just as well.","discovery_kind":"replication","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The non-detection relies on 1-bar DMS/DMDS opacities whose effect on the model evidence is asserted as conservative but not demonstrated; a shape change could shift the marginal preferences across the adopted threshold.","rationale":"The reader's verdict ACCEPT is well argued: the paper uses multiple independent reductions, two retrieval codes, explicit Bayesian evidence comparisons, and an ethane alternative, and the joint conclusion is consistent with external reanalyses. The weakest assumption identified by the reader is indeed the 1-bar DMS/DMDS cross-section validity, and this is the most load-bearing element of the central claim. My concern sharpens the reader's point: the authors' statement that overestimated absorption makes the analysis conservative is about abundance upper limits, not about model evidence. A change in the line-shape model could plausibly shift the marginal Δln Z values (2.3-2.8 in the favorable SCARLET reductions) across the paper's own strong-evidence threshold of 5, or could eliminate them entirely. Because the effect is unquantified and the central claim is a definitive 'no statistical significance', the appropriate verdict is conditional on a targeted sensitivity test. The secondary concern about generalization to 'other molecules' is real but less load-bearing; even if ethane were not representative, the non-detection of DMS/DMDS would stand unless another model produced strong evidence. The paper's internal explanation that marginal preferences reflect 'limiting the number of molecules considered' is not fully demonstrated by the included models, since adding C2H6 to the SCARLET baseline does not remove the DMS/DMDS preference, but this does not invalidate the main negative claim.","tokens_in":16346,"tokens_out":14589,"duration_ms":154144,"concrete_test":"Re-run the PLATON and SCARLET retrievals (Table C.1 configurations) on the exoTEDRF and exoTEDRF+SPARTA spectra with DMS/DMDS opacity tables de-broadened from the 1-bar Sharpe et al. (2004) measurements to ~1 mbar (e.g., deconvolving the Lorentzian component and recomputing R=100-200 band-averaged cross sections, or using new low-pressure laboratory data), and compare the Δln Z values for the with/without DMS-DMDS model pairs. If all Δln Z remain below 5, the non-detection stands; if any crosses 5, the central claim needs revision. A cheaper preliminary check is to flatten the band wings of the DMS/DMDS cross sections by a smooth wavelength-dependent scaling and repeat the SCARLET exoTEDRF retrieval to see whether the 2.3-2.8 preference is stable.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central negative claim is computed with DMS/DMDS cross sections measured at 278 K and 1 bar by Sharpe et al. (2004), as described in Appendix B.1 and B.2, while the transmission photosphere is near 1 mbar. The paper states that this 'likely overestimates DMS/DMDS absorption and thus underestimates their abundances'. That statement addresses the abundance scale, not the model evidence: if the 1-bar spectra contain broad Lorentzian wings that fill in inter-band opacity, the predicted DMS/DMDS band shapes at the data resolution (R ~ 100-200) differ from the true ~1 mbar band shapes. An overall opacity scaling can be absorbed by shifting the retrieved log-abundance without changing the maximum likelihood, but a shape change can alter Δln Z in either direction. The SCARLET reductions already give Δln Z = 2.3-2.8 for DMS/DMDS; a modest shape correction could push these above the Δln Z = 5 threshold used for detection claims, or eliminate them. The paper overstates its robustness when calling the 1-bar assumption conservative in Section 5: it is conservative for abundance upper limits, but its effect on the Bayes factors supporting 'insufficient evidence' is untested. The incomplete C2H6 line list is in the opposite direction and does not threaten the ethane degeneracy argument, since a more complete list would only make ethane fit better.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper reanalyzes the three JWST transit observations of K2-18 b (NIRISS/SOSS 0.6-2.5 um, NIRSpec/G395H 2.9-5.1 um, MIRI/LRS 5-12 um) and asks whether the claimed dimethyl sulfide (DMS) and dimethyl disulfide (DMDS) detections of Madhusudhan et al. (2023, 2025) survive a joint, panchromatic, model-comparison analysis. The authors combine the published JExoRES reductions with three independent reductions (exoTEDRF for all three instruments; Eureka! and SPARTA for NIRSpec and MIRI) and two atmospheric retrieval codes (PLATON at R=100,000 on the JExoRES spectra; SCARLET at R=20,000 on the independent reductions), all with Bayesian evidence comparison using the Trotta (2008) thresholds. Their central quantitative result is that adding DMS/DMDS to a baseline model decreases ln Z by 0.3 on the JExoRES spectra and increases it by only 2.3-2.8 on the independent reductions, never reaching the paper's Delta-ln-Z >= 5 strong-evidence threshold. They further show that ethane, a photochemically expected methyl-bearing hydrocarbon, provides a comparably good fit; that the hot photospheric temperature needed to explain the MIRI feature amplitudes (M25) is inconsistent with the cool temperature favored by the full spectrum and with energy balance; and that roughly 25 additional MIRI transits (sigma = 13) would be needed to reject a flat line at 3 sigma under the best-fitting DMS/DMDS model.","tokens_in":16575,"tokens_out":17302,"duration_ms":160700,"significance":"If the central claim holds, this is an important and timely contribution to the K2-18 b biosignature debate and to JWST exoplanet methodology more broadly. Its strengths are concrete: it is the first joint analysis of all three instruments from Program 2722; it uses three independent reduction pipelines and two retrieval codes; it reproduces the M23 CH4 and CO2 abundances on the same data; it shows explicitly that the DMS/DMDS evidence is below a Delta-ln-Z = 5 threshold under every tested reduction and of either sign depending on the reduction; and it demonstrates a physical inconsistency between the hot MIRI-only photosphere and the full-spectrum constraints. The ethane test generalizes the degeneracy to an entire class of methyl-bearing molecules, and the paper ships reproducible products (Zenodo data products, control/parameter files, public exoTEDRF support for MIRI). The ~25-transit cost estimate, with its large scatter, is a concrete planning figure for the community.","major_comments":[{"comment":"The central quantitative claim is the set of model-evidence differences reported in Sec. 4 and Table C.1, all computed with DMS/DMDS absorption cross sections measured at 278 K and 1 bar (Sharpe et al. 2004) while the transmission photosphere is near ~1 mbar. The Sec. 5 statement that the absence of low-pressure cross sections 'lead[s] to overestimated DMS/DMDS absorption features' is an abundance-scale statement: broadening at 1 bar changes band shapes and fills in inter-band opacity, so the sign of the effect on Delta ln Z is not established a priori. This is not a purely academic concern because Table C.1 shows the independent-reduction SCARLET runs already return Delta ln Z = 2.3-2.8, within a factor of two of the paper's own 'strong evidence' threshold, and an opacity-shape correction could move these values in either direction. I request a quantitative sensitivity test (for example, repeating the DMS/DMDS model comparison with differently broadened or band-averaged cross sections) or, failing that, a quantitative argument that at the data resolution (R ~ 100-200) the band-averaged opacities are insensitive to the 1 bar versus ~1 mbar assumption; the wording of Sec. 5 should then be amended to state what is and is not covered by the claimed conservatism.","section":"Appendix B.1, B.2, and Sec. 5"}],"minor_comments":[{"comment":"The sentence 'To explore the effect of forcing higher temperatures, we perform a retrieval with DMS/DMDS where we impose a minimum temperature of 100 K instead of 200 K' appears to have the comparison reversed: allowing 100 K does not force higher temperatures, while the reported lowering of the DMS/DMDS upper limits is the expected consequence of raising the minimum to 200 K; please restate the experiment and its purpose.","section":"Sec. B.1"},{"comment":"The statement that 'any marginal preferences are the result of limiting the number of molecules considered in the model' is stronger than the ethane test supports: in Table C.1, including C2H6 in the baseline leaves the DMS/DMDS preference essentially unchanged (exoTEDRF: Delta ln Z ~ 2.6 with and without C2H6); recommend harmonizing the abstract with the more hedged wording of Sec. 4 ('likely inflated'), since the ethane test demonstrates the degeneracy but not the causal claim.","section":"Abstract and Sec. 4.1"},{"comment":"The Eureka! MIRI reduction is shown in Fig. 1 but no retrieval is performed on it; add one sentence explaining the choice of the two SCARLET combinations so readers do not infer that this reduction was omitted inadvertently.","section":"Fig. 1 and Sec. 3"},{"comment":"Gordon et al. 2022a and 2022b refer to the same HITRAN2020 paper and should be merged into a single reference.","section":"References"},{"comment":"The abstract quotes '~25 more MIRI transits' while Sec. 4.2 reports an average of 26 transits from 1,000 simulations with a standard deviation of 13; please make the numbers consistent and state the scatter alongside the point estimate.","section":"Abstract and Sec. 4.2"},{"comment":"The caption explains that 'N/A' for CO means an unconstrained posterior, but the same symbol is used for C2H6 rows where the molecule was simply not included; please distinguish the two meanings.","section":"Table C.1 caption"}],"recommendation":"major_revision","confidential_remarks":"The reader's report recommended acceptance, and I agree with its assessment of the paper's internal consistency and value. My independent concern is narrow but real: the paper's headline Bayes factors are the load-bearing numbers, and their robustness to the 1-bar DMS/DMDS cross-section assumption is asserted rather than demonstrated. If the authors supply the requested sensitivity analysis or a quantitative resolution argument, I would expect this to be the last round. The manuscript is well within A&A Letters scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know this is the paper that finally puts the K2-18 b DMS/DMDS story under one continuous 0.6–12 µm lens. Previous work disagreed over NIRISS/NIRSpec or MIRI in isolation; this one models all three instruments together, reducing the data with three independent pipelines (JExoRES, exoTEDRF, SPARTA/Eureka!) and running two retrieval codes. The results are public, and the effort is real. The central negative claim holds up better than I expected: on the JExoRES reduction, adding DMS/DMDS actually lowers the Bayesian evidence (Δln Z = −0.3), while the most favorable independent reductions give only Δln Z = 2.3–2.8, below their own strong-evidence threshold of 5. The ethane test is the strongest new piece—structurally similar to DMS, photochemically plausible, and it fits the data just as well without invoking sulfur. That is a genuinely useful degeneracy demonstration.\n\nThe soft spot is the one the stress-test flags, but I think it is half a point. The DMS/DMDS cross sections come from 278 K, 1 bar nitrogen measurements (Sharpe et al. 2004), while the transmission photosphere is near 1 mbar. The paper calls this conservative, and that is true for abundance upper limits—broader 1-bar bands overestimate absorption, so retrieved abundances are lower limits. But it is not automatically conservative for model evidence. Pressure broadening changes band shapes, not just overall opacity, and a shape change could shift Δln Z in either direction. The paper never tests this. I do not think it overturns the conclusion, because the positive evidence is already marginal and reduction-sensitive, and the JExoRES reduction actually disfavors DMS/DMDS. You would need a large, systematic shape correction to push all reductions past Δln Z = 5. Still, the authors should not have used the word \"conservative\" as broadly as they did.\n\nMinor points: only ethane is actually fitted as an alternative hydrocarbon—propyne and diethyl sulfide are only cited, as the paper admits. The \"25 more MIRI transits\" estimate is noise-dependent and they flag the ±13 scatter. The C2H6 line-list incompleteness cuts the other way (it makes ethane look worse, not better), so that does not threaten their argument.\n\nThis is a paper worth sending to a serious referee. The central argument is sound, the multi-reduction design is the right way to test claim robustness, and the limitations are mostly acknowledged in the text. My only requirement for acceptance would be a sentence or two clarifying that the effect of low-pressure DMS/DMDS cross sections on the model evidence is untested, not merely conservative for abundances.","headline":"A credible joint reanalysis that spells real trouble for the K2-18 b DMS/DMDS claim, with one honest caveat about pressure-broadened cross-sections.","tokens_in":17293,"tokens_out":2844,"would_cite":true,"duration_ms":30099,"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":"A joint analysis of the full 0.6–12 µm JWST transmission spectrum of K2-18 b finds insufficient evidence for DMS or DMDS, with ordinary hydrocarbons like ethane fitting the data equally well.","keywords":["K2-18 b","transmission spectroscopy","dimethyl sulfide","dimethyl disulfide","exoplanet atmospheres","JWST","Bayesian model comparison","biosignatures"],"falsifier":"A decisive test would be to measure DMS and DMDS cross sections near 1 mbar and 175–250 K and recompute the Bayesian evidence for the same three reductions; if the marginal $\\Delta\\ln Z \\approx 2.3$–$2.8$ preferences in the independent reductions drop below 1, the non-detection is confirmed, whereas if they sharpen into a consistent $\\Delta\\ln Z \\geq 5$ preference, the central claim would be overturned.","tokens_in":16020,"feed_emoji":"🪐","tokens_out":12584,"duration_ms":119198,"temperature":0.7,"pith_summary":"This paper asks whether the signs of dimethyl sulfide (DMS) and dimethyl disulfide (DMDS) claimed in parts of the K2-18 b spectrum hold up when all JWST observations are modeled together. The answer is no: across the full 0.6–12 µm wavelength range there is insufficient evidence for either molecule, and ordinary hydrocarbons such as ethane reproduce the same features just as well. The authors re-reduced the raw data with several independent pipelines and ran two retrieval codes, finding that any preference for DMS/DMDS depends on which reduction is used and on how many molecules are allowed in the model. This matters because DMS has been floated as a potential biosignature; the finding removes the strongest spectral support for that idea without ruling out an ocean world.","feed_headline":"Joint JWST analysis finds no DMS/DMDS on K2-18 b","feed_subtitle":"Combining NIRISS, NIRSpec, and MIRI data leaves ordinary hydrocarbons as the better explanation.","key_machinery":"The machinery is a panchromatic 0.6–12 µm transmission spectrum assembled from the three JWST instruments, each reduced by several independent pipelines and analyzed by two retrieval codes, with Bayesian model comparison (comparing log-evidence differences $\\Delta\\ln Z$) used to judge whether any molecule is detected. The decisive physical object is the methyl functional group, a carbon atom bound to three hydrogen atoms: DMS, DMDS, and ethane all carry two methyl groups, so their C–H stretching band near 3.4 µm and H–C–H bending band near 6.9 µm nearly coincide, making the sulfur molecules hard to distinguish from ordinary hydrocarbons. The comparison is carried out with DMS/DMDS cross sections measured at 278 K in 1 bar nitrogen, even though the planet's photosphere sits near ~1 mbar; the authors state this likely overestimates DMS/DMDS absorption and underestimates their abundances, while incomplete ethane line lists likely do the opposite.","core_discovery":"The paper claims that when the NIRISS, NIRSpec, and MIRI transmission spectra of K2-18 b are modeled jointly over 0.6–12 µm with a chemically reasonable set of molecules, the DMS and DMDS signals reported from single-instrument analyses do not reach statistical significance. In the analysis run on the reductions from the original publications, adding DMS/DMDS to the baseline model changes the Bayesian evidence by $\\Delta\\ln Z = -0.3$; in two independent reductions the preference rises to $\\Delta\\ln Z \\approx 2.3$–$2.8$, still below the $\\Delta\\ln Z \\geq 5$ threshold the authors take as a detection claim. The mid-infrared features can be fit equally well by ethane ($\\mathrm{C_2H_6}$), a photochemically plausible molecule whose methyl-group absorption bands overlap those of DMS/DMDS, so the features are not unique to the sulfur-bearing molecules. The paper also argues that the hot photospheric temperature required to explain the MIRI features alone conflicts with the near-infrared data and stellar energy balance, and estimates that roughly 26 additional MIRI transits would be needed for a $3\\sigma$ rejection of a flat line even under the best DMS/DMDS model.","pith_inferences":["Reader inference: the same methyl-group degeneracy likely applies to any temperate sub-Neptune with a hydrocarbon haze, so a DMS claim from a single spectral window should first be checked against ethane, propyne, and diethyl sulfide.","Reader inference: completing laboratory cross-section measurements at low pressure and temperature for DMS/DMDS, and completing the ethane line list, would directly test whether the remaining $\\Delta\\ln Z \\sim 2$ preferences are real or an artifact; the authors' caveats imply the evidence for DMS/DMDS could move in either direction.","Testable extension: fit a single shared methyl-group absorber opacity parameter instead of individual DMS/DMDS/ethane abundances; if the data cannot separate the molecules, the evidence of that aggregated model should match the best individual models, quantitatively confirming the degeneracy."],"forward_implications":["The earlier DMS/DMDS detection claims for K2-18 b are not supported once all three JWST instruments are modeled together; the observed features are consistent with ordinary hydrocarbons such as ethane.","Any biosignature interpretation of this planet must confront the methyl-group degeneracy: the diagnostic bands at 3.4 and 6.9 µm are shared by many simple organic molecules.","The methane detection in K2-18 b's atmosphere remains robust across reductions, while the sulfur-bearing molecules do not survive the joint analysis.","A secure DMS/DMDS detection with MIRI would require roughly 25 additional transits, or several years of JWST time, before a 3σ deviation from a flat line could be claimed.","Future searches should analyze the complete wavelength range rather than individual instruments to avoid reduction-dependent artifacts."],"supporting_citations":[{"why":"Original NIRISS/NIRSpec spectrum and tentative DMS detection whose robustness the joint analysis tests.","marker":"M23"},{"why":"MIRI spectrum and claimed DMS/DMDS detection that the full-spectrum analysis does not confirm.","marker":"M25"},{"why":"Prior reanalysis of the near-infrared data that found no significant CO2 or DMS and supplied independent reductions reused here.","marker":"S25"},{"why":"Proposal that diethyl sulfide can explain the NIRISS/NIRSpec features, supporting the claim that DMS/DMDS are not unique.","marker":"P-C25"},{"why":"Proposal that propyne and other hydrocarbons explain the MIRI features and provide photochemical context for ethane.","marker":"Welbanks, Nixon et al. (2025)"},{"why":"Source of the 278 K, 1 bar DMS/DMDS cross sections used in both retrieval codes; the pressure mismatch is the key caveat.","marker":"Sharpe et al. (2004)"},{"why":"Abiotic photochemistry models predicting ethane as a common hydrocarbon, making it the paper's chosen alternative.","marker":"Huang et al. (2024)"},{"why":"Earlier critique of the MIRI DMS/DMDS detection that motivates quantifying the claim on the full spectrum.","marker":"Taylor (2025)"}],"fun_headline_variants":["No DMS/DMDS in K2-18 b when you look at all JWST data","K2-18 b: joint JWST analysis undercuts DMS/DMDS claims","Ethane fits K2-18 b as well as DMS/DMDS, JWST shows","Sulfur biosignatures on K2-18 b? JWST joint spectrum says no","K2-18 b's DMS/DMDS signals fade under joint JWST analysis"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the laboratory measurements of DMS and DMDS absorption at 278 K in 1 bar of nitrogen apply to the thin upper atmosphere the telescope actually sees; the authors themselves note this likely overstates how strongly these molecules absorb, and if the real low-pressure behavior differs, the statistical verdict could change.","fun_headline_variants_meta":{"raw":{"variants":["No DMS/DMDS in K2-18 b when you look at all JWST data","K2-18 b: joint JWST analysis undercuts DMS/DMDS claims","Ethane fits K2-18 b as well as DMS/DMDS, JWST shows","Sulfur biosignatures on K2-18 b? JWST joint spectrum says no","K2-18 b's DMS/DMDS signals fade under joint JWST analysis"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001469,"raw_usage":{"total_tokens":6040,"prompt_tokens":1212,"completion_tokens":4828,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":828,"completion_tokens_details":{"reasoning_tokens":4708}},"tokens_in":828,"tokens_out":4828,"duration_ms":32380,"temperature":1.0,"reasoning_tokens":4708,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T20:14:46.912776+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive test would be to measure DMS and DMDS cross sections near 1 mbar and 175–250 K and recompute the Bayesian evidence for the same three reductions; if the marginal $\\Delta\\ln Z \\approx 2.3$–$2.8$ preferences in the independent reductions drop below 1, the non-detection is confirmed, whereas if they sharpen into a consistent $\\Delta\\ln Z \\geq 5$ preference, the central claim would be overturned.","supporting_citations":[{"cited_title":"W., Johnson , T","cited_arxiv_id":null,"evidence_quote":"Source of the 278 K, 1 bar DMS/DMDS cross sections used in both retrieval codes; the pressure mismatch is the key caveat."},{"cited_title":"2024, , 975, 146","cited_arxiv_id":null,"evidence_quote":"Abiotic photochemistry models predicting ethane as a common hydrocarbon, making it the paper's chosen alternative."}],"review_version":1}