Abiotic O2 Levels on Planets around F, G, K, and M Stars: Effects of Lightning-produced Catalysts in Eliminating Oxygen False Positives

Over the last few years, a number of authors have suggested that, under certain circumstances, molecular oxygen (O2) or ozone (O3) generated by abiotic processes may accumulate to detectable concentrations in a habitable terrestrial planet’s atmosphere, producing so-called “false positives” for life. But the models have occasionally disagreed with each other, with some predicting false positives, and some not, for the same apparent set of circumstances. We show here that photochemical false positives derive either from inconsistencies in the treatment of atmospheric and global redox balance or from the treatment (or lack thereof) of lightning. For habitable terrestrial planets with even trace amounts of atmospheric N2, NO produced by lightning catalyzes the recombination of CO and O derived from CO2 photolysis and should be sufficient to eliminate all reported false positives. Molecular oxygen thus remains a useful biosignature gas for Earth-like extrasolar planets, provided that the planet resides within the conventional liquid water habitable zone and has not experienced distinctly non-Earth-like, irrecoverable water loss.


Introduction
In our search for life on planets orbiting other stars, detection strategies most often focus on how life here on Earth has modified its environment. Over geological timescales, photosynthesis by cyanobacteria (and later, by algae and plants) has increased the atmospheric molecular oxygen (O 2 ) concentration from negligible values to 21% by volume today. The increase in O 2 is thought to have occurred mostly in two or three major steps. The first was the Great Oxidation Event (GOE) some 2.5 billion years ago (Gya); the second is sometimes called the Neoproterozoic Oxidation Event (NOE), which took place ∼0.8 Gya (Kump 2008;Lyons et al. 2014). A third major increase may have occurred at around 400 Ma during the late Devonian period (Wallace et al. 2017). Atmospheric O 2 may have climbed even higher (up to ∼35% by volume) during the Carboniferous (e.g., Berner & Canfield 1989; see also Lenton 2001), but this increase was not sustained. Oxygen would have been difficult to observe remotely at low spectral resolution prior to the GOE and should have been easily detectable following the NOE (Segura et al. 2003;Harman et al. 2015). Whether O 2 and its photochemical byproduct, ozone (O 3 ), were remotely observable during the intervening Proterozoic Eon depends on how much of it was present at that time. Estimates for Proterozoic O 2 range from 0.5 PAL (times the present atmospheric level) (Kump 2008) to <0.1% PAL ). Ozone may have been detectable throughout the Proterozoic, even at lower O 2 concentrations, due to its strong UV absorption features (Segura et al. 2003).
Whether oxygenic photosynthesis would evolve on another Earth-like planet is unknown. Some authors have argued that it is a natural consequence of the ready availability of liquid water (H 2 O) and carbon dioxide (CO 2 ) (e.g., Kiang et al. 2007;Léger et al. 2011). If so, then photosynthetic O 2 could be common on terrestrial planets within the habitable zone (Kasting et al. 1993b;Kopparapu et al. 2013) of their parent star (Kiang et al. 2007;Léger et al. 2011;Meadows 2017). Molecular oxygen or ozone could thus be one of the first biosignature gases detected remotely using a next-generation space telescope like the James Webb Space Telescope (JWST) (Schwieterman et al. 2016). O 2 and O 3 satisfy both the survivability and detectability criteria for biosignature gases (Seager et al. 2013;Meadows 2017;Meadows et al. 2017). Whether they also satisfy the third necessary criterion, reliability, has been a topic of recent debate. Understanding their reliability is critical and urgent, as JWST will begin observations within the next two years, and the detection of O 2 and O 3 are also central to the design of instruments for other future ground-and space-based telescopes Harman et al. 2015;Meadows 2017;Meadows et al. 2017). Here, we argue that O 2 and O 3 are reliable bioindicators under certain conditions. We outline what those conditions are, and discuss the secondary measurements and constraints necessary to demonstrate the presence of such conditions on another world.

False Positives in the Literature
For the purposes of this paper, a "false positive" for life is defined as any abiotically derived O 2 concentration that exceeds that estimated to follow the GOE (Harman et al. 2015). If the low estimates by Planavsky et al. (2014) are correct, then that level is at or below 0.1% PAL. Possible mechanisms by which that abiotic O 2 level might be achieved have been recently reviewed by Meadows (2017) and Meadows et al. (2017). Broadly speaking, these scenarios can be divided into two categories: (1) false positives driven by changes in the redox budget of a planet's combined atmosphere/ocean system (as defined in Catling & Kasting 2017, Ch. 8), and (2) false positives driven by photochemistry within the atmosphere.
Examples of the redox-driven false positives include planets that experience substantial water loss because they are located inside the inner edge of the habitable zone (Kasting 1988;Kasting et al. 1993a), or their host (M) stars are relatively bright during their pre-main sequence lifetimes (Ramirez & Kaltenegger 2014;Luger & Barnes 2015;Tian 2015), or they lack non-condensable gases like N 2 that help keep water vapor contained in the lower atmosphere (Wordsworth & Pierrehumbert 2014). In each of these cases, hydrogen is irreversibly lost to space, oxidizing the atmosphere and oceans and, ultimately, the planet itself. The ultimate fate of this oxygen is uncertain but it could potentially be drawn down to undetectable concentrations by absorption into the mantle (Wordsworth et al. 2018). Other types of redox-driven false positives essentially correspond to hypothetical planetary scenarios that lack a volcanic source of reduced gases (e.g., Hu et al. 2012;Domagal-Goldman et al. 2014;Gao et al. 2015). If volcanic outgassing is assumed to be nearly zero, or the recombination of CO and O 2 is inefficient, or there are no surface sinks for O 2 , then O 2 can build up in the atmosphere. If reduced volcanic gas emissions are entirely neglected, O 2 can and must accumulate as H 2 O is photolyzed and the resulting hydrogen escapes to space. But we argue that a false positive driven by such low outgassing rates is unlikely if the planet is habitable, so that liquid water can facilitate reactions of O 2 with its crust, and if the planet remains volcanically active for long time periods, as the Earth has done. (We return to this last point in Section 6.) The false positive scenario described by Hu et al. (2012) is more complicated in that it has balanced hydrogen escape with both surface deposition and volcanic emission. For a 90% CO 2 , 10% N 2 atmosphere on an abiotic planet orbiting a G star, they predict an O 2 mixing ratio of 1.3×10 −3 , or 0.006 PAL (see their Figure 6 and Table 7). This scenario has no H 2 outgassing but a finite source of volcanic H 2 S. The H 2 S source exceeds the rate of hydrogen escape to space (after stoichiometric weighting to put these fluxes in the same redox units). Hence, the scenario has the excess hydrogen balanced by deposition of reduced species to the surface. As such, the redox budget of the ocean is unbalanced. This is equivalent to assuming production and burial of organic carbon on a planet that has no biological means of producing it. We have simulated additional scenarios (not shown) using the same model as Hu et al. (2012). For one scenario, we remove the previously assumed volcanic outgassing of H 2 S, and for the other we remove both the H 2 S source and hydrogen escape. These two additional scenarios produce similar O 2 and O 3 mixing ratio profiles to that in Hu et al. (2012), and they both balance global redox. Importantly, all of the scenarios discussed in this paragraph allow O 3 to be deposited at the surface, but not O 2 . This contributes to the buildup of atmospheric O 2 , but it is not a physically realistic assumption, as reduced species (e.g., dissolved ferrous iron) should react with both. The assumption of zero volcanic outgassing of reduced gases is also not geologically plausible for an Earth-like planet, as rocky planetary interiors are expected to be generally reduced and as volcanism is difficult to suppress entirely on a planet as large as Earth. To put this another way, the presence or absence of a false positive for Earth-like planets around Sun-like stars is sensitive to the boundary conditions but, ultimately, the choice of boundary conditions must reflect ageologically plausiblecase.
Most of the redox-driven false positives should be identifiable remotely, given enough observations. Runaway greenhouse planets like early Venus would be suspect because of their position near or inside the inner edge of the habitable zone (Kasting 1988;Kopparapu et al. 2013;Kane et al. 2014;Way et al. 2016). Planets around late M stars should all be initially suspect, for the reasons just mentioned. One would need to verify that they have retained some of their water, potentially through spectral observations of the strong water vapor absorption feature at ∼0.95 μm, before assigning any significance to observed O 2 or O 3 . Planets with very low N 2 (<0.5 bar) could, in principle, be identified by the absence of absorption by the N 2 dimer, by decreased Rayleigh scattering, or by sufficiently detailed spectra to show the effect of pressure on other absorption features . The total pressure for the Archean atmosphere may have been <0.5 bar (Som et al. 2012(Som et al. , 2016, suggesting that N 2 would have been largely undetectable during this interval. Methods for identifying these false positive-generating mechanisms have been extensively discussed in the recent literature (Harman et al. 2015;Meadows 2017;Meadows et al. 2017).
The second category of false positives, those caused by photochemical processes, are more difficult to rule out, as these can ostensibly occur on worlds nearly indistinguishable from modern or ancient Earth. The case that we focus on here concerns planets around M dwarf host stars Tian et al. 2014;Harman et al. 2015). This case is of great interest because JWST may be able to obtain transit spectra of such planets in the near future.
This particular photochemical false positive arises for the following reason. On Earth-like planets around any type of star, atmospheric CO 2 should be photolyzed by UV radiation shortward of 200 nm: The reaction between CO and O to reform CO 2 is spinforbidden and slow, so the O atoms can recombine with each other to form O 2 : On planets like the early Earth, the rate of O 2 formation is relatively slow because CO and O can also recombine with each other by catalytic cycles involving the byproducts of water vapor photolysis: ) accumulated to detectable levels. All three of these models were redox-balanced and should, in principle, have given similar results. (Indeed, all three models are derivatives of the same Kasting-group photochemical model.) The present study was motivated by the desire to figure out why they disagreed.
Having completed the analysis, we find that most of the differences between the Tian/Harman results and the Domagal-Goldman results were caused by the neglect of lightning in the first two models and its inclusion in the third, as well as methodological differences in addressing global redox. Lightning creates nitrogen oxides that can help catalyze the recombination of CO with O. Below, we briefly review how lightning affects planetary atmospheres in general, and then discuss its importance for the question of false positives on exoplanets.

Lightning Occurrence
The modern Earth experiences a global average of two to three lightning flashes km −2 yr −1 (∼2×10 −4 flashes km −2 hr −1 ), while the gas giants Jupiter and Saturn have lower lightning flash rates (∼1-50×10 −7 flashes km −2 hr −1 ), and Venus has even fewer lightning flashes (∼4×10 −11 flashes km −2 hr −1 ) (Hodosán et al. 2016 and references therein). Lightning on Earth is approximately an order of magnitude less frequent over oceans than over land (Hodosán et al. 2016), due to several potential contributing factors, including greater efficiency in converting the convective available potential energy (CAPE) into strong updrafts over land (e.g., Williams & Stanfill 2002). Lightning occurrence is a strong function of CAPE, so warmer, wetter atmospheres tend to produce more lightning (Wong et al. 2017). Additionally, some authors have suggested a connection between galactic cosmic rays (GCRs) and lightning (Stozhkov 2003 and references therein), which may enhance lightning for M dwarf planets experiencing higher GCR fluxes (Airapetian et al. 2017). We will return to GCRs briefly below.

Lightning Production of NO
On the modern Earth, lightning affects the chemical composition of the atmosphere by converting (or "fixing") N 2 and O 2 into more reactive NO x species, which can influence the concentrations of gases such as O 3 and OH (e.g., Chameides 1978). Given the high temperatures and pressures within the lightning flash, the equilibrium reactions can be written as Thus, the oxygen used to form NO can be derived from CO 2 , O 2 , or water vapor. The ancient Earth would also have featured lightning-driven nitrogen fixation (Chameides & Walker 1981;Navarro-Gonzalez et al. 2001), with lower NO production compared to modern, as the early Earth's atmosphere is thought to have lacked significant amounts of O 2 (Kasting 1979). Other planets in our solar system, including Venus and Mars, could also have lightning-influenced chemistry (e.g., Nna Mvondo et al. 2001). On Earth today, each lightning flash produces 30-670 moles of NO, with a typical lightning flash producing ∼250 moles of NO (Schumann & Huntrieser 2007). When integrated over the surface of the Earth, this yields a global NO production rate of ∼5 Tg yr −1 , or ∼6×10 8 NO molecules cm −2 s −1 (Schumann & Huntrieser 2007). We use this as our estimate for the global lightning production of NO on the modern Earth. Figure 1 shows the production rate of NO from lightning in 1 bar, CO 2 -O 2 -N 2 -dominated atmospheres, using a lightning parameterization updated from Kasting (1979) that assumes a freeze-out temperature of 3500 K for all constituents (Kasting 1990), which broadly agrees with other lightning simulations (e.g., . We have assumed the same surface temperature (288 K) for each composition, broadly  (2014) as being susceptible to abiotic O 2 accumulation from water loss. NO production in excess of modern (>∼6×10 8 cm −2 s −1 ) is highlighted by the white boundary, with arrows pointing toward higher NO production. For low concentrations of CO 2 and O 2 (<100 ppm), NO production is supported by oxygen derived from water vapor, remaining greater than ∼3×10 7 cm −2 s −1 (∼5% of modern); O 2 concentrations >0.9 bar (gray region) were not tested. The modern Earth's atmospheric composition is shown as the blue circle. consistent with the presence of a carbonate-silicate feedback cycle that helps to stabilize a planet's climate (e.g., Kasting et al. 1993a;Kopparapu et al. 2013). This results in a water vapor mixing ratio of ∼1% at the surface.
We have chosen the ternary diagram in Figure 1 to highlight the dependence of NO production on composition, and we acknowledge that some of the mixtures represented may not correspond to physically self-consistent gas assemblages. That said, we note several effects. (1) NO production is highest for approximately present-day N 2 concentrations with higher-thanpresent-day O 2 concentrations. We have highlighted this region with a white boundary, with arrows pointing toward higher NO production.
(2) NO production remains within an order of magnitude of modern NO production throughout the diagram, except at very low CO 2 and O 2 concentrations, where oxygen contributions to NO formation are sustained by water vapor in the atmosphere and where the "floor" is approximately 5% of modern NO fluxes (Kasting 1979).
(3) At low N 2 concentrations (down to ∼1% N 2 ), NO production remains within a factor of 5 of the modern value. However, these low N 2 concentrations would likely result in abiotic accumulation of O 2 from water loss (Wordsworth & Pierrehumbert 2014)-a redox-based false positive. This region of parameter space is shown by the white-hatched area in the lower-right corner of the diagram. Taken together, this suggests that a temperate terrestrial planet with even modest amounts of N 2 will exhibit rates of lightning production of NO within an order of magnitude of the modern terrestrial rate.
Below, we describe the photochemical models we used to investigate the effect of lightning on the photochemistry of hypothetical abiotic planetary atmospheres, along with the results from those models.

Model Description
For this study, we used two one-dimensional, horizontallyaveraged photochemical models, which share a common heritage. The model of Harman et al. (2015) is similar to those of Segura et al. (2007) and Tian et al. (2014), and by extension that of Domagal-Goldman et al. (2014), which used the Segura et al. (2007) model with minor modifications. A detailed description of this model can be found in Appendix B of Catling & Kasting (2017). The second model, Atmos, is described in detail in Arney et al. (2016), and while it shares its origins with the Harman et al. model, Zahnle (1986) began significant alterations. These models have differing chemical species, reaction lists, and physics parameterizations, which explains much of the inter-model variability. However, these differences cannot explain the qualitatively different behavior recently reported in the literature, which arises instead from their differing treatments of lightning.

Results
To compare with the previous calculations of Domagal-Goldman et al. scenario-set to maximize potential abiotic O 2 productionassumes that surface sinks for O 2 are nonexistent, and that CO has an (abiotic) deposition velocity of 10 −8 cm s −1 (Harman et al. 2015). (The abiotic deposition velocity is derived by assuming that the only sink for CO in solution is hydration to formate.) Both conditions tend to magnify any abiotic O 2 accumulation derived from CO 2 photolysis (Harman et al. 2015). Both the Harman et al. and Atmos models produced atmospheres dominated by CO and O 2 for the M dwarf cases (Figure 2, panel (A)), in line with previous estimates of abiotic O 2 accumulation (Tian et al. 2014;Harman et al. 2015). The Atmos model did not accumulate O 2 in the absence of lightning-produced NO for the K dwarf case (panel (C), dashed curves), which is due to differences in how species like CO 2 and O 3 are treated numerically, as well as differences in model physics.
We then repeated this same calculation with lightning included. The results are shown in Figure 2, panels (B) and (D). In these cases, surface O 2 concentrations are vanishingly small, consistent with estimates for O 2 on the prebiotic Earth and on lifeless planets orbiting F and G stars (Segura et al. 2003;Harman et al. 2015). Similar low-O 2 results are obtained even if the atmospheric CO 2 concentration is increased to as much as 90% by volume.
We conclude that the O 2 false positives reported by Tian et al.  Harman et al. 2015.) Because lightning is to be expected on any planet with a wet surface that experiences moist convection (Wong et al. 2017), it seems unlikely that this represents a real false positive. M-star planets may build up high abiotic O 2 levels in other ways (e.g., pre-main sequence water loss), as discussed earlier. But photochemical false positives like those described in Tian et al. (2014) and Harman et al. (2015) are unlikely to actually occur.

Discussion
To attempt to quantify how the production of nitrogen oxides by lightning helps to reduce abiotic O 2 concentrations, we have made a list of possible catalytic cycles for recombining CO with O (see Table 1). Our methodology follows that of Stock et al. (2012), who looked at several possible catalytic cycles for the Martian atmosphere. Mars is a reasonable analog for the high-CO 2 , low-outgassing planets we have simulated here, except that the Martian atmosphere is much colder and contains much less H 2 O.
Several of the catalytic cycles listed share individual chemical reactions between them, but the slow step in each cycle should determine its overall contribution to CO + O recombination (shown as the "net" value below each cycle). In Table 1, we provide side-by-side comparisons of cases for the young Sun, the modern Sun, ò Eridani  (Claire et al. 2012). For each star, we have shown a case where lightning is assumed to be producing NO (the "L" columns) and where it is not (the "L-off" columns). In cases where no false positive for O 2 exists, the approximate rate of catalytic destruction of CO and O is equal to or greater than the rate of photolysis of CO 2 . For model planets around G-type stars, catalytic destruction of CO and O always outpaces CO 2 photolysis, even with lightning turned off in the model. For K-and M-dwarf Importantly, there is an overlap between where N 2 -derived NO may be sufficient to drive the recombination of CO and O 2 , and where N 2 is undetectable. Below approximately 0.5 bar N 2 , the N 2 dimer feature would be largely undetectable , but the NO production rate would still be large enough to catalytically recombine O 2 with CO, as long as pN 2 is greater than ∼0.01 bar. However, the lower limit for pN 2 ∼0.08 bar would be set by temperate water loss and O 2 accumulation (Wordsworth & Pierrehumbert 2014). For 0.5 bar<pN 2 <∼0.08 bar, then, Note.Summary of the major catalytic cycles suggested by Stock et al. (2012) occurring in the Martian atmosphere. Each value is the column-integrated rate (cm −2 s −1 ) for the reaction listed, except the mixing ratios. This is an analog for the 5% CO 2 , low instellation (1.3 au equivalent), low volcanic outgassing atmospheres shown here and in Harman et al. (2015). First column, cycles 1-5; note that we have modified cycle 3. separating these two false positives using the N 2 dimer alone is not possible, and additional constraints on the atmospheric composition or total pressure would be needed. Secondary features such as the absence of the O 2 dimer would rule out O 2 >0.5 bar (Misra et al. 2014), and high signal-to-noise measurements of several spectral features may place a useful lower limit on total atmospheric pressure (Des Marais et al. 2002;Misra et al. 2014), although these types of observations would be difficult with the JWST (Batalha et al. 2018). Additional information from Rayleigh scattering (e.g., Selsis 2004;Benneke & Seager 2012) or thermal phase curves (e.g., Koll & Abbot 2015) could also be used to constrain atmospheric pressure.

Additional Sinks for Abiotic Oxygen
In addition to the catalytic recombination of CO and O 2 by NO, other geological sinks for abiotic O 2 exist on Earth-like planets. Plate tectonics exposes new material that would chemically react with free oxygen, but whether and under what circumstances plate tectonics started on Earth (e.g., Condie & Pease 2008;Korenaga 2013), let alone on other planets (e.g., Valencia et al. 2007;Korenaga 2010;van Summeren et al. 2011;van Heck & Tackley 2011;Foley et al. 2012;Noack & Breuer 2014), remains uncertain (and potentially unanswerable; see Lenardic & Crowley 2012). Oxidative weathering may be relatively insensitive to the exposed land fraction, if it behaves similarly to silicate weathering (Abbot et al. 2012). In the absence of plate tectonics, however, the addition of reducing gases from the interior of the planet would act as a sink for O 2 . Volcanic outgassing is tangentially related to the tectonic activity of a planet, both being driven principally by internal heating (from planetary formation and radioactive decay). However, the radionuclide budget of a planet is expected to decrease with time, but also as a function of its formation age . The composition and flux of volcanic gases is also incredibly complex, depending on redox state of the mantle (Kasting et al. 1993a), the source material (Schaefer & Fegley 2007), and the volcanic setting (e.g., Burgisser & Scaillet 2007;Gaillard et al. 2011;Gaillard & Scaillet 2014), but may remain consistent over the lifetime of the planet (e.g., Trail et al. 2011). All three terrestrial planets in our solar system demonstrate varying amounts of volcanic activity (e.g., Smrekar et al. 2010;Hauber et al. 2011), despite their divergent evolutionary paths.

Other Considerations
Smaller amounts of lightning-produced NO, consistent with suggestions for ocean worlds (Hodosán et al. 2016), can still be effective in limiting abiotic O 2 buildup. Even if the rate of NO production is decreased by an order of magnitude, O 2 concentrations in our "worst case" GJ 876 scenario remain at or below ∼1 ppb at the surface. However, NO production rates more than 30 times less than modern are insufficient to prevent significant accumulations of O 2 in our 5% CO 2 GJ 876 case. That said, the models used here have parameterized lightning production of NO based on the modern Earth, which would underestimate lightning occurrence (and thus NO production) for warmer, wetter atmospheres (e.g., Wong et al. 2017).
In our model, lightning is assumed to be the only source of NO. However, other sources of NO exist, and NO is not the only catalyst capable of recombining CO and O 2 . Volcanoes are a known source of NO (e.g., von Glasow et al. 2009), and could have introduced fluxes of NO into the early Earth's atmosphere comparable to that produced by lightning fixation of NO on the modern Earth (Martin et al. 2007). GCRs and coronal mass ejections (CME) can also produce nitric oxides in the atmosphere (e.g., Nicolet 1975;Scalo et al. 2007;Airapetian et al. 2016). Planets orbiting M dwarfs experience larger and more frequent flares (e.g., Airapetian et al. 2017) and enhanced GCR fluxes due to the proximity of a planet in the habitable zone to its host star (e.g., Grenfell et al. 2007), suggesting that the NO production estimates outlined below are conservative. Chlorine and bromine radicals could potentially operate in the same way as NO (even interacting with NO x , which further complicates the story). These radicals can be derived from sea salt spray (Finlayson-Pitts 2003), and are relevant to the O 2 and O 3 chemistry on Venus (e.g., Mills et al. 2006). Further work could quantify the effects of these species on abiotic O 2 concentrations in more Earth-like atmospheres.
As always, constraints on the atmospheric composition of an exoplanet will be invaluable in determining whether an O 2 absorption signal is really an indicator of life. Substantial concentrations of CH 4 , or of H 2 O, N 2 , CO 2 , and CO, combined with the absence of a substantial O 2 dimer spectral feature (indicative of a post-runaway greenhouse atmosphere), would effectively exclude all known false positive mechanisms (Schwieterman et al. 2016;Wang et al. 2016). Several methods along these lines meant to minimize false positives have been proposed (e.g., Desch et al. 2018;Harman & Domagal-Goldman 2018). This work suggests that, even in the absence of such constraints, the simultaneous detection of O 2 + H 2 O + N 2 would be strongly suggestive of the presence of life on a terrestrial exoplanet orbiting in the habitable zone.

Conclusion
In the cases outlined here, lightning eliminates the reported photochemical O 2 false positives in the atmospheres of terrestrial planets around small stars. We conclude that: (1) a self-consistent habitable (but lifeless) terrestrial planet is likely to have several mechanisms at work that reduce photolysisdriven disequilibrium, including lightning and outgassing of reduced gases, and (2) physical processes beyond the scope of gas-phase chemistry control the chemical composition of a planet's atmosphere. On Earth, biology is the dominant control. For lifeless worlds, these controls include boundary conditions reflecting the assumed chemistry of the ocean, and phenomena like lightning, as we have shown here. Modelers must weigh the potential impact of these secondary processes and reservoirs and work to connect assumptions to geologically plausible scenarios. The production of NO is not limited to lightning, however, and M-dwarf host stars, with their myriad other complications, have several substantial alternative NO production routes. Consequently, the presence of O 2 , alongside reasonable concentrations of H 2 O, CO 2 , and N 2 , should be regarded as a robust biosignature of terrestrial exoplanets orbiting within habitable zones. Future plans to explore the influence of lightning on atmospheric composition will couple photochemical and climate simulations to better capture the interaction of lightning occurrence and O 2 accumulations. This work was funded by NASA's Habitable Worlds proposal #NNX15AQ11G, and by the NASA Astrobiology Program through the Nexus for Exoplanet System Science (NExSS).