- The paper models M4 and M8 stellar spectra from 650 Myr to 5 Gyr with coupled climate and photochemical simulations, showing that host-star aging strongly reshapes Earth-like atmospheres.
- Anoxic Archean atmospheres around aging M4 stars develop 36-fold more methane, while ozone can fall by over 5 orders of magnitude and abiotic oxygen reach 50–100 ppm.
- The study finds that UV Hartley-band signals can mimic weak biological oxygen features, so biosignature searches must measure both total UV flux and the FUV/NUV ratio while considering CO and stellar contamination.
Overview
Davis et al. investigate how age-dependent quiescent ultraviolet emission from M dwarfs alters the photochemical composition and observable spectra of Earth-like planetary atmospheres, and how those changes affect the detection and interpretation of biosignatures. The study couples PHOENIX stellar atmosphere models of M4 and M8 stars at ages from 650 Myr to 5 Gyr with the Atmos photochemical model and VPL Climate, iterating both to convergence for two archetypal atmospheres: a Pre-Industrial Earth (21% O₂) and an anoxic Archean Earth (10% CO₂), each held at 66% of Earth's instellation to remain temperate (~284 K). Planetary spectra are generated with SMART at resolutions matched to JWST/NIRSpec Prism and a LUVOIR-B/Habitable Worlds Observatory-class imager. The work deliberately excludes flares and considers only quiescent UV emission.
The authors construct quiescent spectra for 0.35 M☉ M4 stars at 650 Myr, 1 Gyr, 3 Gyr, and 5 Gyr by selecting PHOENIX upper-atmosphere models from Peacock et al. that reproduce median FUV/NUV fluxes of Hyades members and field M0–M4 stars, calibrated against GALEX-derived power-law decay relations (FFUV∝t−0.96, FNUV∝t−0.85). For the M8 star, spectra at 650 Myr and 5 Gyr suffice because saturation lifetimes for such late-type stars extend to ~4 Gyr, making intermediate ages nearly degenerate with 650 Myr. Because GALEX measurements exhibit a 1–2 dex spread in FUV/NUV flux at each epoch, the authors define the "considered range of UV emission" (CRUVE): the span between upper-quartile young-star and lower-quartile old-star states, representing population-level rather than single-star variability. This is a deliberate framing choice — the results characterize the likely range across stellar populations, not the evolution of any individual system.
Atmospheric response around M4 hosts
The Archean Earth responds strongly to the M4 CRUVE. Methane surface abundance rises from 173 ppm to 6234 ppm — a factor-of-36 increase — as the host ages, deepening the 1.7 μm CH₄ transit feature by 2 ppm (68%) and the corresponding reflectance feature by 0.07 in reflectivity. Conversely, abiotic ozone column depth falls by more than 5 dex (from ηO3≈1.1×1017 to 4.5×1011 cm⁻²) with increasing stellar age. Around the 650 Myr median and upper-quartile and 1 Gyr upper-quartile stars, CO₂ photolysis drives surface O₂ to 50–100 ppm and produces a Hartley band (0.26 μm) reflectance feature roughly half as reflective as planets around old M4s. Notably, this feature is spectrally similar to the O₃ signal produced by biogenic O₂ during seasonal photosynthesis on Proterozoic Earth models, establishing the central interpretive hazard of the paper.
The Pre-Industrial Earth is far less responsive: methane rises only one dex, O₃ falls one dex, and the Hartley band remains saturated at all ages, with variation confined to the band wings (~50% change in reflectivity near 0.3 μm). The authors attribute this difference to photochemical shielding: abundant O₂ absorbs shortwave UV high in the atmosphere, suppressing lower-atmospheric photolysis rates, whereas the anoxic Archean atmosphere relies on weaker shielding by CO₂.
Atmospheric response around M8 hosts
Trends around the M8 star are directionally consistent but smaller in magnitude, consistent with the shallower decline in UV flux over its longer saturation lifetime. Across the M8 CRUVE, Archean Earth CH₄ increases 0.4 dex and O₃ falls 3.1 dex; Pre-Industrial CH₄ rises 0.5 dex while O₃ drops only 0.28 dex. The most observationally relevant result here is that M8-hosted atmospheres show transmission feature variability of up to ~8 ppm (Pre-Industrial) and ~5 ppm (Archean) in the 1.7 μm CH₄ band — comparable to or above JWST's approximate 5 ppm noise floor, though still entangled with much stronger stellar contamination signals.
Net UV flux versus FUV/NUV ratio
A key diagnostic contribution is the separation of the two governing stellar parameters. Ozone column depth depends on both net UV flux (118–300 nm integrated) and the FUV/NUV ratio (FNR): it plateaus above ~300 mW m⁻² net flux but collapses by 3 dex within a ~30% flux decrease near 200 mW m⁻², and when flux is fixed at ~240 mW m⁻², high FNR (>0.55) yields ozone-rich states while low FNR (~0.2) suppresses O₃ by ~3 dex. This follows mechanistically from FUV-driven CO₂ photolysis supplying atomic oxygen versus NUV photolysis destroying O₃. Methane, by contrast, correlates primarily with net UV flux and shows little sensitivity to FNR (only 0.2 dex change over an FNR drop from 0.6 to 0.2), because its destruction is governed by direct FUV photolysis aloft and OH chemistry below — the latter fed mainly by NUV-pumped water photolysis. An instructive comparison: the 650 Myr M8 delivers higher HZ net UV than its M4 counterpart yet produces 0.5 dex less O₃ owing to its ~20% lower FNR, demonstrating that spectral shape matters independently of total flux. Practically, this means accurate biosignature interpretation requires measuring both the integrated UV flux and the FNR of the host star, not merely one.
Robustness of the abiotic O₂/O₃ buildup
Because spurious O₂/O₃ accumulation is a known pathology of photochemical models, the authors explicitly test three failure modes. First, all atmospheres conserve redox balance to within ≤40 ppm (86% within 10 ppm), and maximum oxidizing fluxes into the ocean are only 8% of estimated early-Earth floor values — ruling out redox imbalance as the driver. Second, the adopted Lincowski et al. CO₂ cross-section lies at or below recommended values relative to Broussard et al.'s recent analysis, keeping surface O₂ within ~0.5 dex of their most conservative cases. Third, the atmospheric grid extends to 98–104 km (38–59 nanobar), sufficient to resolve the CO₂ photolysis peak per Ranjan et al.'s criterion. The buildup mechanism itself is physical: slow spin-forbidden CO + O recombination leaves free oxygen available for O₂/O₃ formation, while CO₂'s own FUV absorption shields H₂O and CH₄ from dissociation into reducing radicals that would destroy the products. A limitation acknowledged by the authors is that the assumed 10% CO₂ — needed for habitability at reduced instellation — exceeds true Archean levels (~640 ppm), although it remains near the low end expected for outer-HZ carbonate-silicate buffered planets, meaning the effect could plausibly apply broadly across the HZ.
Observational and biosignature implications
Most transmission changes fall below JWST's noise floor, with the possible exception of M8-hosted CH₄ features near 5–8 ppm contingent on mitigating stellar contamination. The more consequential finding concerns direct imaging: an LUVOIR-B-class observatory could detect ~9 warm terrestrials including ~3 "exo-Earths" around M stars, and for anoxic, CO₂-rich planets orbiting ≤1 Gyr hosts, the abiotic Hartley band feature would be present and strong. Without stellar UV context, this feature could be misread as evidence of low-level biogenic O₂ — the paper's headline cautionary claim. The 4.6 μm CO feature (varying by 1.0–9.5 ppm across the CRUVEs) offers a potential discriminant, since abundant CO alongside O₂/O₃ implicates CO₂ photolysis; seasonal variability of the O₃ signal provides another possible discriminator, though eccentric orbits or tidally maintained obliquity could confound it. The trend may also extend to young early-M and K stars, which can deliver ~200 mW m⁻² of habitable-zone UV despite lower typical FNRs, though dedicated modeling is required to confirm this.
Limitations and open questions
Several caveats bear directly on the results. All simulations are steady-state responses to discrete stellar spectra, not continuous co-evolution of atmosphere and star, so transient behavior during stellar spin-down is unexplored. Flares are excluded entirely despite radiation-plus-proton events being capable of ≥90% O₃ depletion in prior work. Hydrocarbon hazes were not simulated; the authors justify this via low CH₄/CO₂ ratios (≤0.07), but Teal et al. showed haze formation is sensitive to M-dwarf UV continua at observable levels, leaving an open question about hazy variants of these atmospheres. The choice of Peacock et al.'s TRAPPIST-1 spectrum over Wilson et al.'s semi-empirical MegaMUSCLES spectrum introduces systematic uncertainty: the former overpredicts some FUV emission lines (which would inflate O₂/O₃) but better captures NUV continuum (which would deflate them via lower implied FNR); Cooke et al. found O₃ differing by 1.4 dex between these inputs for TRAPPIST-1 e. The authors note these opposing effects have not been disentangled, and that the M8 models represent population averages rather than TRAPPIST-1 specifically. Finally, sensitivity of the O₂/O₃ buildup to alternative redox source/sink assumptions remains unresolved.
Conclusion
This work establishes that stellar age-dependent UV emission is a first-order control on the abundances and spectral expressions of O₃ and CH₄ in Earth-like M-dwarf planet atmospheres: young (≲1 Gyr) hosts drive >5 dex ozone enhancement and up to 90 ppm abiotic O₂ via CO₂ photolysis in anoxic, CO₂-rich atmospheres, producing a UV Hartley-band feature that mimics a weak biogenic O₂ proxy, while old hosts permit methane accumulation factors of tens relative to young-star equivalents. The dual dependence of O₃ — but not CH₄ — on both net UV flux and FNR implies that panchromatic UV characterization of target stars must precede any biosignature interpretation for Habitable Worlds Observatory targets, particularly those resembling the early Earth.