Co-evolution of primitive methane-cycling ecosystems and early Earth’s atmosphere and climate

The history of the Earth has been marked by major ecological transitions, driven by metabolic innovation, that radically reshaped the composition of the oceans and atmosphere. The nature and magnitude of the earliest transitions, hundreds of million years before photosynthesis evolved, remain poorly understood. Using a novel ecosystem-planetary model, we find that pre-photosynthetic methane-cycling microbial ecosystems are much less productive than previously thought. In spite of their low productivity, the evolution of methanogenic metabolisms strongly modifies the atmospheric composition, leading to a warmer but less resilient climate. As the abiotic carbon cycle responds, further metabolic evolution (anaerobic methanotrophy) may feed back to the atmosphere and destabilize the climate, triggering a transient global glaciation. Although early metabolic evolution may cause strong climatic instability, a low CO:CH4 atmospheric ratio emerges as a robust signature of simple methane-cycling ecosystems on a globally reduced planet such as the late Hadean/early Archean Earth.


Infuence of biological parameters on ecosystem viability
By comparing the distribution of parameter values from the subset of simulations with persistent biological activity to the distribution of all parameter values, we can delineate the region in parameter space that corresponds to ecosystem viability (Supplementary Figure  1C). We find that viability is significantly conditioned by high values of , low values of basal mortality , low values of , and high values of maximum division rate , with a predominant effect of (Supplementary Figure 1D and E). Next, we use the subset of ecologically viable simulations to examine how model parameters influence the equilibrium biomass and biogenic methane flux, ( 4 ). We find that the level of biogenic methane emission is positively related to the maximum metabolic rate through parameter and negatively related to the maintenance cost through , while biomass production negatively correlates with both and (statistical analysis not shown).
Finally, we compare the distribution of outputs in the subset of ecologically viable simulations to the default parameterization outcome (Fig. 1). The distribution is relatively narrow (95% interval envelope is about one order of magnitude wide), highlighting the fact that the model is more strongly constrained by its structure than parameterization. In most scenarios, default parameter values yield results that are close to the median of the subset of ecologically viable simulations (Fig. 2). With the MG ecosystem, the results of the default parametrization are within the 95% confidence intervals, but close to the boundaries (lower limit for methane emission, upper limit for biomass; data not shown). This greater sensitivity to parameters is due to the global redox equilibrium being strongly impacted by the metabolic rate of methanogens; this is in contrast to the other ecological scenarios where the global redox equilibrium is determined chiefly by photochemical processes. The default value of is near the lower end of the viability range for that parameter, so most of the ecologically viable simulations correspond to higher , hence larger CH4 emission at equilibrium and lower equilibrium biomass. This is because the redox state of the system is closer to its thermodynamic equilibrium, and therefore metabolism is less efficient. Interestingly, exploring higher values of is equivalent to releasing kinetic constraints on biology. This explains why our predictions of CH4 emission then get closer to ref12.
Global redox balance of the planet By tracking the global hydrogen budget of the atmosphere, computed as f(H2) + 4 f(CH4) + f(CO) following ref6, we check the evolution of the atmospheric global redox budget in the simulations presented in Fig 4. We find that in most cases the atmospheric redox budget is very similar whether the planet is populated by a primitive biosphere or not (Supplementary Figure 11). The only two exceptions are the ecological scenarios in which AG is present in the biosphere in the absence of AT. When this is the case, some of the redox budget of the atmosphere is transferred to the ocean in the form of acetate.
The conservation of a steady atmospheric redox budget from a lifeless to a living Earth highlights that the biomass production of primitive biospheres was so low that it did not constitute a significant sink of H2 relative to atmospheric escape (the main sink of reduced species during the Archean).
Ecological feedback of methanogenesis on climate warming and resilience Supplementary Figure 3 shows the atmospheric and climatic impact of the biosphere as a function of H2 volcanic outgassing,  Fig. 3 where temperature is set by pCO2. Quantitatively, with the MG ecosystem the effect of temperature variation on biological activity is even stronger when TGeo and pCO2 are independent. This is because the negative effect of higher temperature on thermodynamics is partially offset if pCO2 is concomitantly higher. As a consequence, hydrogenotrophic methanogenesis is expected to have an even greater effect on climate when temperature varies independently of pCO2. In contrast, climate warming by AG+AT metabolisms is weak, irrespective of H2 outgassing and abiotic temperature when the latter varies independently of pCO2. Warming will occur, however, in an AG+AT ecosystem in which MG evolves, the effect being as strong as in the MG-only ecosystem (Supplementary Figure 3A).
Climate resilience to variation of pCO2 is shown in Supplementary Figure 4. With the MG ecosystem, the ecological feedback to the atmosphere has a buffering effect on temperature variation above ca. 5 °C (Supplementary Figure 4A) and an amplifying effect below 5 °C (Supplementary Figure 4B). With the AG+AT ecosystems, the amplification effect prevails irrespective of the temperature range (Supplementary Figure 4C and D). Once MG, AG and AT have all evolved, we can however conclude from Fig. 3 that the ecosystem has almost no effect on the resilience of the climate.
Atmospheric and climatic impact of methanotrophy CH4 emissions by MG and AG+AT metabolisms create conditions favorable for the evolution of methanotrophy (MT). The evolutionary rate has a critical influence on the MT environmental feedback. Supplementary Figure 6 shows the environmental impact of fastevolving MT that arises on a 103 years timescale after the establishment of MG and/or AG+AT metabolisms. In this case, MT evolution takes place under atmospheric and climatic conditions set by the atmosphere-ecosystem equilibrium of MG and/or AG+AT (Figs. 2 and 3, Supplementary Figure 3). Irrespective of H2 volcanic outgassing rate and abiotic temperature (Supplementary Figure 6A), the environmental effect of MT is to consume most of the atmospheric CH4 produced by methanogens (Supplementary Figure 6C), driving the surface temperature close to its abiotic value, (Supplementary Figure 6B). The timescale over which this happens is very short, of the order of 103 years (Supplementary Figure 6B and C). The outcome is a new atmosphere-ecosystem equilibrium at which all metabolisms coexist, under a methane-poor atmosphere resulting in a cool climate.
The previous scenario will hold provided the evolutionary timescale is much shorter than the timescale of the carbon cycle. If the timescale of MT evolution is of the order of the C cycle characteristic time (107 yrs), or longer, then the environmental impact of MT evolution will depend on the long-term effect that the carbon cycle has on the environment inhabited by MG and/or AG+AT ecosystems (Supplementary Figure 12). As explained in the main text, the carbon cycle response to the evolution of methanogenesis leaves the biogenic outflux of CH4 relatively unaltered. However, the equilibrium temperature, , is much lower than at the short-term equilibrium shown in Fig. 3, in the absence of carbon cycle feedback. Under such conditions, the evolution of methanotrophy drives both pCH4 and pCO2 down (Fig. 5A), causing dramatic climate cooling and putting the planet at high risk of global glaciation (Fig. 5B and C). Supplementary Figure 8C shows that relatively low abiotic temperature combined with a high rate of H2 volcanic outgassing favors the global glaciation outcome.
Supplementary Figure S4. Climate resilience in response to pCO2 variation. The central color panels are from Fig. 3.
Side panels A-D show the climatic response of the planet, either inhabited ( , plain curves) or lifeless ( , dashed curves), to periodic variation in pCO2, depending on ecosystem composition (MG or AG+AT). The corresponding abiotic temperature variation amplitude is ∆TGeo = 20 °C (indicated by the white dots and arrows in the central color panels). On a warm planet inhabited by MG (TGeo ranging from 30 to 50 °C), the climate response is buffered by about 20 % (A). However, on a cool planet (TGeo ranging from -20 to 0 °C, panel) the MG ecosystem amplifies the climate response by up to 33% (B). On a planet inhabited by AG+AT, the climate response to pCO2 variation is always amplified, but much less on a warm planet (C) (5% for TGeo ranging from 30 to 50 °C) than on a cool planet (33% for TGeo ranging from -10 to 10 °C, bottom-right panel) (D).  Fig. 4 (1,000 simulations in each scenario), obtained by drawing uniformly the model abiotic parameter values in log-uniform priors based on the litterature (see Table 1). The white dots represent the median of the distributions, the thick gray lines the interquartile range, and the thin gray lines the rest of the distribution.