Oxygen bounty for Earth-like exoplanets: spectra of Earth through the Phanerozoic

ABSTRACT

In the search for life in the Universe, Earth provides a template of evolution for the one habitable planet we know. Earth’s atmospheric composition has changed significantly throughout its history. The last 500 Myr – the Phanerozoic Eon, which includes the origins of animals, dinosaurs, and land plants – saw oxygen rise from ≤10 per cent to 35 per cent. But the resulting transmission spectra are a crucial missing piece in our search for signs of life in exoplanet atmospheres. Here, we simulate the atmosphere and transmission spectra of the Phanerozoic, using estimates from established climate models, and present the first high-resolution transmission spectra for Phanerozoic Earth. We demonstrate that the spectral biosignature pairs O₂ + CH₄ and O₃ + CH₄ in the atmosphere of a transiting Earth-like planet would indicate a biosphere, with O₂ and O₃ features potentially stronger than for modern Earth. The full model and high-resolution transmission spectra, covering 0.4–20 µm, are available online and provides a tool to plan and optimize observations, train retrieval methods, and interpret upcoming observations with ground- and space-based telescopes.

astrobiology, Earth, exoplanets

1 INTRODUCTION

Earth’s geological evolution (e.g. Catling & Zahnle 2020) provides a crucial template for finding life on rocky exoplanets. Several teams have modelled the spectra of Earth for selected epochs of geologic time (Meadows 2005; Kaltenegger et al. 2007, 2020; Rugheimer & Kaltenegger 2018) from pre-biotic to modern Earth, and have shown that the strongest absorption bands for CH₄, O₂, and O₃ are detectable for and change with major stages in atmospheric evolution throughout Earth’s history. These models showed that a remote observer could identify a biosphere for about 2 billion yr in Earth’s history (Kaltenegger et al. 2007, 2020). Together, these recent studies provide high-resolution spectra templates as a guide for observations of habitable exoplanets.

Through Earth’s geological evolution, as O₂ levels rose, spectral features of O₂ and O₃ became more prominent. However, no model spectra exist for the diverse Phanerozoic Eon, the span of Earth’s history that encompasses the earliest fossil evidence of animals up to the present day. During this time, O₂ increased from a few per cent to a major atmospheric constituent (Kasting 1987; Holland 2006; Lyons Reinhard & Planavsky 2014). The rise of oxygen to atmospheric prominence on Earth over the last ∼540 Ma is linked to the biosphere. Evidence of the connection between the histories of O₂ and complex life is recorded in biogeochemical proxy data, such as the ‘fire window’ bounds on O₂ (i.e. Jones & Chaloner 1991; Glasspool & Scott 2010; Glasspool & Gastaldo 2022). Atmospheric O₂ must have been between 16 per cent and 35 per cent for most of the last ∼400 Ma in order for charcoal deposits observed in the rock record to have formed. Below ∼16 per cent O₂, fires cannot ignite regardless of dryness/fuel availability; above ∼35 per cent, plants burn irrespective of drying (meaning fire, once ignited, would theoretically burn uncontrollably until it exhausted all available fuel). So ambient O₂ must have been within the ‘fire window’ limits for most of the Phanerozoic once the presence of terrestrial forests established a supply of burnable material on the continents.

The rise of O₂ in the Phanerozoic was due, in large part, to two major milestones. The advent and spread of terrestrial forests in the Carboniferous and Permian periods (∼359–251 Ma) facilitated atmospheric O₂ becoming a major constituent gas (Bergman et al. 2004; Berner 2009), peaking between ∼350–300 Ma due to the formation of coal deposits made from increasingly abundant vascular (i.e. woody and/or stemmed) plants on the continents that resulted in extensive organic carbon burial (e.g. Royer 2013). Another possible rise in atmospheric O₂ around ∼100 Ma – likely due to changes in continental geography after the Pangea supercontinent breakup (Berner 2004) – saw O₂ concentrations exceed that of the modern atmosphere (Lenton Daines & Mills 2018) before settling at modern levels of 21 per cent in the last 50 Ma.

While similar processes may or may not be ongoing on exoplanets, the Phanerozoic Eon nevertheless provides an underexplored template for a habitable planet with varying levels of atmospheric oxygen. Here, we model five stages (approximating 500, 400, 300, 200, and 100 Ma) and their transmission spectra. Each modelled stage broadly represents a noteworthy phase of Phanerozoic time: i) 500 Ma: a global environment after the origin of shelled organisms but devoid of land plants; ii) 400 Ma: the evolutionary midpoint after vascular plants had first evolved, forming the earliest forests; iii) 300 Ma: the dominance of highly productive global swamp forests, when models agree that O₂ likely reached its highest concentrations; iv) 200 Ma: the calibration of O₂ sources and sinks, with the output from the photosynthetic biosphere combatting drawdown from outgassing; and v) 100 Ma: the later potential peak in O₂ during the Cretaceous.

We first model the atmospheric composition of five stages in Earth’s Phanerozoic Eon based on geological tracers and then calculate the high-resolution transmission spectra of these environments to evaluate expressed spectral features in these stages. Models and high-resolution transmission spectra covering a wavelength range from 0.4–20 µm are available online. These provide a tool to train retrieval algorithms, plan, optimize observations, and interpret upcoming observations with ground- and space-based telescopes.

2 METHODS

To assess what spectral features are expressed in the spectrum of a transiting Phanerozoic Earth and any exoplanet Earth-analouges in a similar evolutionary phase, we used exo-prime2 (see Kaltenegger et al. 2007; Madden & Kaltenegger 2020a, b), an established 1D radiative–convective atmosphere code, to model the planet and resulting transmission spectra for select scenarios (see Fig. 1 and Table 1). We model CO₂ concentrations of 3000 ppm (500 Ma), 2000 ppm (400 Ma), and 1000 ppm (300 Ma, 200 Ma, and 100 Ma). The corresponding O₂ concentrations from the GEOCARB model (Berner & Canfield 1989; Berner 2009) broadly suggest atmospheric O₂ was ≤21 per cent for most of the Phanerozoic, while the COPSE model (Bergman et al. 2004, Lenton et al. 2018) suggests a more drastic rise in O₂ from far lower early concentrations to levels equal to or in excess of modern 21 per cent O₂ from the mid-Phanerozoic onward. Modelling the progression of O₂ posited by both models allows us to explore the strength of spectral features in transmission, including the atmospheric biosignature pairs O₂ + CH₄ and O₃ + CH₄ (Lovelock Shoemake and Zlatkis 1964; Lederberg 1965; Kasting 2014).

Figure 1.

Estimates of (top) CO₂ (ppm) and (bottom) O₂ (per cent) throughout the Phanerozoic: GEOCARB (Berner 2009; orange), COPSE (Lenton et al. 2018; light blue), corresponding shaded region show uncertainty range. O₂ reconstructions from charcoal deposits (Glasspool & Scott 2010; dashed grey) are shown as existing geochemical proxy data for O_2.

Table 1.

Phanerozoic model CO₂ and O₂ mixing ratios estimated from GEOCARB (Berner 2009) and COPSE (Lenton et al. 2018) and the resulting global mean surface temperature (T_Surf).

CO2 (ppm) estimate	3000	2000	1000
Phanerozoic Time (Ma)	∼500	∼400	∼300	∼200	∼100
O2 GEOCARB	15 per cent	15 per cent	30 per cent	15 per cent	20 per cent
TSurf (K)	292.7	291.6	288.6	287.9	288.3
O2 COPSE	5 per cent	10 per cent	30 per cent	20 per cent	30 per cent
TSurf (K)	292.8	291.1	288.6	288.1	288.6

Using the incident stellar spectrum and planetary outgassing rates, exo-prime2 couples 1D climate and photochemistry models to compute the global mean temperature, as well as vertical profiles of each chemical species and temperature in 1-km layers from the surface to 100 km using wavelength-dependent albedo for surfaces and clouds (details in Madden & Kaltenegger 2020b). exo-prime2 has been validated from the visible to infrared by comparing to Earth seen as an exoplanet by Mars Global Surveyor, earthshine observations, EPOXI, and shuttle data missions (Kaltenegger & Traub 2009; Rugheimer et al. 2013).

We model the high-resolution transmission spectra from 0.4 to 20 µm at a resolution of 0.01 cm⁻¹ including the spectroscopically relevant molecules in our calculations (from the HITRAN2016 line list: C₂H₆, CH₄, CO, CO₂, H₂CO, H₂O, H₂O₂, H₂S, HNO₃, HO₂, N₂O, N₂O₅, NO₂, O₂, O₃. OCS, OH, and SO₂; Gordon et al. 2017) as well as Rayleigh scattering. For each atmospheric layer, line shapes and widths are calculated individually with Doppler- and pressure-broadening with several points per line width. Deeper atmospheric regions block or deflect light from a distant observer (Kaltenegger and Traub 2009; Sidis and Sari 2010; Garcia-Munoz et al. 2013; Bétrémieux and Kaltenegger 2014; Robinson Fortney and Hubbard 2017). Earth can be probed in primary transit to about 13 km in the modelled wavelength. Clouds at the terminator that could obscure underlying spectral features generally lie below that height. We chose not to scale total solar or ultraviolet (UV) radiation for each scenario, in order to isolate the effect of changing O₂ (and CO₂). The difference in radiation and UV is small for the Phanerozoic – the solar constant at 500 Ma is 0.96 times the modern (Ribas et al. 2005), with between 6–10 per cent less UV flux (Claire et al. 2012). A sensitivity study we conducted for our cases at 500 Ma showed only a negligible effect to surface temperature and transmission spectra.

3 RESULTS

3.1 Climate-photochemical model results

The general decline in surface temperature (Fig. 2) from 500 Ma to 100 Ma is largely due to the decreasing CO₂ concentrations in our models: −4.43 K for the GEOCARB, −4.17 K for COPSE. This is broadly in agreement with estimated climate sensitivity to changes in atmospheric CO₂ from individual (i.e. Holland & Bitz 2003) and aggregated (i.e. Krissansen-Totton & Catling 2017) 3D climate models.

Figure 2.

Temperature (left) and mixing ratios (right) of major atmospheric constituents for Phanerozoic models from 500 Ma to 100 Ma using approximate (top) GEOCARB and (bottom) COPSE estimates for O₂. Modern Earth is shown as dashed grey.

The differences in surface temperature between GEOCARB and COPSE values can be attributed to changes in atmospheric photochemistry: our GEOCARB-based Phanerozoic models see an overall O₂ and O₃ increase over time of only 5 per cent (from 15 per cent at 500 Ma to 20 per cent at 100 Ma) and resulting decrease in CH₄ due to a rise in available reactive oxygen from the photolysis of stratospheric O₃ (Fig. 2). The Phanerozoic models based on COPSE values see a much more pronounced increase in atmospheric O₂, from 5 per cent to 10 per cent to 30 per cent, as CO₂ falls from 3000 ppm to 2000 ppm to 1000 ppm, respectively. Upper atmospheric O₃ in particular sees a marked increase as O₂ increases, which in turn most influences temperature profiles in the mid-atmosphere above ∼35 km, especially across the distinct shift in atmospheric composition before and after 300 Ma (when O₂ reached its peak). A greater abundance of O₃ would have helped compensate for weakened greenhouse warming as CO₂ decreased through the Phanerozoic in Earth, though at the cost of a loss in warming by CH₄.

3.2 Transmission spectra results

The transmission spectra (Fig. 3) show models for a Phanerozoic Earth from 0.4–20 µm and indicate the major spectral features and commonly used filters. The modern Earth transmission spectrum (see also Kaltenegger & Traub 2009) is shown in grey for comparison. The change in CO₂ and O₂ levels for the different cases – and the resulting changes in atmospheric chemistry, like in the O₃ and CH₄ levels – shape the strength of the spectral features in the transmission spectra throughout the Phanerozoic. The decline in CO₂ over time results in a decrease of the prominent CO₂ feature. As O₂ increases, the O₂ and O₃ spectral features increase in strength (see also Fig. 4 for a detailed view of the high-resolution spectral features). Spectral features of CH₄ decrease with increasing O₃ and O₂ abundance.

Figure 3.

Five modelled transmission spectra during the Phanerozoic epochs from 500 Ma to 100 Ma using (top) GEOCARB and (bottom) COPSE estimates for O₂ estimates. The vertical axis indicates both effective height and transit depth (the ratio of planetary to stellar flux, R²_P/R²*) for transmission spectroscopy. Major spectral features and commonly used filters are indicated. Modern Earth is shown in grey for comparison. All spectra are shown at a resolution of λ/Δλ > 300 for ease of comparison and are available online at a resolution of 100 000.

Figure 4.

High-resolution (resolution = 100 000) spectra for biosignature pairs O2 + CH4 and O3 + CH4 for Phanerozoic epochs from 500 Ma to 100 Ma, showing an increase in spectral feature strength for O2 and O3, and a decrease of CH4 feature strength.

4 DISCUSSION

Deciphering the evolution of atmospheric O₂ and CH₄ is vital to understanding the development of the biosphere on a planet (Kasting 2014; Kaltenegger 2017; Meadows et al. 2018). But despite its significance, many unknowns remain even for Earth like the abundance of O₂ during key stages of the Phanerozoic. Geochemical records – namely from charcoal, as in Glasspool & Scott (2010) – have established a compelling likely minimum concentration for Phanerozoic O₂. Charcoal evidence indicates that O₂ remained at least as high as in the modern atmosphere after its peak at ∼400 Ma when the continental landscape was dominated by the coal forests of the Carboniferous and Permian periods. However, while geochemical data posit a lower limit on O₂, few can constrain the upper limit more narrowly than the 16–35 per cent range given by the charcoal ‘fire window’ (Royer et al. 2014).

As a result, attempts to place constraints on O₂ over the last ∼540 Ma have primarily been done with climate models. These are also limited since O₂ has a negligible effect on global climate even with an established ozone layer (Payne et al. 2016; Wade et al. 2019). The most well-known models for Phanerozoic O₂ are arguably the models we based our simulated stages on: GEOCARB – namely Berner (2009), building on the mathematical model of Berner & Canfield (1989) – and the COPSE model used in Bergman et al. (2004) and Lenton et al. (2018). Both GEOCARB and COPSE broadly agree that a large-scale decline in CO₂ in the early Phanerozoic was accompanied by an increase in O₂ due to biological expansion on the continents. However, there is considerable disagreement after the peak at ∼300 Ma, and estimates from GEOCARB are generally much lower than those from COPSE (Fig. 1). The COPSE estimates are in overall better agreement with the lower O₂ bound established by the charcoal proxy, suggesting O₂ remained above about 20 per cent after ∼350 Ma. GEOCARB’s O₂ estimates for the latter half of the Phanerozoic are near or less than the lower limit posited by the fire window. This discrepancy may be partially due to differences in handling geochemical data. Both GEOCARB and COPSE use C isotopes to constrain rates of carbon burial and sequestration. COPSE then uses a process-driven model to integrate the effect on O₂ based on rates of biological activity, while GEOCARB instead applies the isotope data to the C-O cycle to find the O₂ abundance. But C isotope constraints for surface processes can be ill conditioned when applied to O₂ cycling (Kasting 2013; Krissansen-Totton Buick & Catling 2015; Saltzman & Edwards 2017) and thus lead to much lower modelled O₂. Variation in measured δ¹³C of just 1 per mil can underestimate atmospheric O₂ by several per cent (for details, see Payne et al. 2016), which may contribute to GEOCARB’s lower estimates. Recent estimates from COPSE (Lenton et al. 2018) have the closest agreement with the finite geochemical data available for determining Phanerozoic O₂.

Support for higher Phanerozoic O₂ is also important from an astrobiology perspective. For example, relatively high (≥15 per cent) O₂ on Earth was likely a consequence of a global photosynthetic biosphere. Even higher O₂ levels are suggested as a requirement for later evolutionary advancements in animals on Earth (i.e. Falkowski et al. 2005; Reinhard et al. 2016; Bozdag et al. 2021).

The lack of well-constrained upper bounds on O₂ from geochemical data or models, and disagreement between models, means questions remain pertaining to the level of Phanerozoic O₂. However, the presence of O₂ + CH₄ and O₃ + CH₄ as biosignature pairs are even more compelling indicators of life (i.e. Lovelock & Whitfield 1982; Kasting 2014; Harman et al. 2015; Kaltenegger 2017; O'Malley-James & Kaltenegger 2019) that are expressed for all model stages.

5 CONCLUSIONS

We chose two established geological models to assess whether spectral features of O₂, O₃, and CH₄ could be discernable as biosignature pairs to identify life on a Phanerozoic Earth as well as exoplanets at similar stages in their evolution. While the individual models indicate different levels of these gases (see Figs 1 and 2), their spectral features are expressed in the transmission spectra of planets in all analogue phases of Earth’s Phanerozoic (see Figs 3 and 4).

The full model and high-resolution transmission spectra data base (resolution = 100 000), covering 0.4–20 µm, for exoplanets through the evolution of Phanerozoic Earth is available online. It is a tool to plan and optimize our observation strategy, train retrieval methods, and interpret upcoming observations with ground- and space-based telescopes in order to identify life on Earth-like exoplanets.

ACKNOWLEDGEMENTS

The authors acknowledge funding from the Brinson Foundation and the Carl Sagan Institute.

DATA AVAILABILITY

The data underlying this article is on zenodo, at doi.org/10.5281/zenodo.8302086.