Bayesian analysis of Enceladus’ plume data to assess methanogenesis

The NASA Cassini mission revealed the existence of an ocean and hydrothermal activity underneath the Saturnian moon’s icy surface. Earth's hydrothermal systems team with microbial life. How likely are Enceladus’ hydrothermal vents habitable to Earth-like microorganisms?

Like Comment
Read the paper

Hydrothermal circulation is likely responsible for what is perhaps the most notable feature of Enceladus: a water-rich plume erupting from cracks in the ice at the South pole. The Cassini spacecraft traversed the plume multiple times and was able to perform mass-spectroscopy measurements of its composition. The plume was found to contain a relatively high concentration of dihydrogen (H2), methane (CH4), and carbon dioxide (CO2). In Earth’s hydrothermal vents, methane can be produced abiotically in H2-rich fluids, but at a slow rate. Most of the production is due to microorganisms that harness the chemical disequilibrium of hydrothermally produced H2 as a source of energy, and produce methane (CH4) from carbon dioxide (CO2) in a process called methanogenesis. Searching for methanogens at Enceladus’ seafloor would require extremely challenging deep-dive missions that are not in sight for several decades. We took a different, easier route: we constructed mathematical models to quantify the likelihood that different candidate processes, including biological methanogenesis, might explain the Cassini data. 

Artist's rendition of Enceladus' interior (credit: NASA/JPL)

We looked at Enceladus’ plume composition as the end result of several chemical and physical processes taking place in the moon’s interior. We first aimed at assessing what hydrothermal H2 production would best fit Cassini’s observations, and whether this production could provide enough ‘food’ to sustain a population of Earth-like hydrogenotrophic methanogens. To do so, we developed a model for the population dynamics of a hypothetical hydrogenotrophic methanogen, whose thermal and energetic niche was parameterized after known strains. We were then able to assess whether a given set of chemical conditions (e.g., H2 concentration in the hydrothermal fluid) and temperature would provide a suitable environment for these microbes to grow, and also what effect such a population would have on its environment (e.g., on the H2 and CH4 escape rates in the plume). So not only could we evaluate whether Cassini’s observations are compatible with a habitable environment, but we could also make quantitative predictions about observations to be expected should methanogenesis actually occur at Enceladus’ seafloor.

We found that the amount of dihydrogen escaping in the plume is compatible with deep-ocean conditions allowing the growth of the modeled hydrogenotrophic methanogen. Interestingly, if such microorganisms were actually consuming this dihydrogen, the change in the plume’s hydrogen content would likely be negligible. Thus, the detection of dihydrogen in the plume may not be interpreted against biological methanogenesis in Enceladus’ putative hydrothermal environment (or in analogous terrestrial hydrothermal vents). Furthermore, even the highest possible estimate of abiotic methane production (without biological aid) from known hydrothermal chemistry is far from sufficient to explain the methane concentration measured in the plume. In contrast, biological methanogenesis could produce enough methane to match Cassini’s observations.

Should we conclude that Earth-like microbes might inhabit the depths of Enceladus’ ocean? Absolutely not. The story that our study tells is actually one of a negative (albeit intriguing) result. Science is built on hypotheses and predictions that can be tried and possibly rejected. When we are looking at proving the effect of a medical treatment, we are actually hoping that patients administered with the treatment show a positive response significantly different from the test group. This significance means that, under the assumption that the treatment has no effect, the observed outcome would be unlikely. If we know a little bit more about how the treatment works, we might even be able to propose a quantitative prediction on the treatment's effects and see if it matches observations. Here, our reasoning is essentially the same, although we are not working on a controlled experiment like it would be the case when testing the efficiency of a drug or a vaccine. We only have partial knowledge of the system and can only study some of the processes at work, and that introduces some nuance to result interpretation. We developed our model to test, or even reject the hypothesis of biological methanogenesis using the Cassini data. What we found is that our current understanding of Enceladus’ interior, of hydrothermal chemistry, and of methanogenesis does not allow to reject this hypothesis ; in contrast, the hypothesis of only abiotic Earth-like hydrothermal activity to explain the Cassini data is rejected.

Schematic view of bayesian hypothesis testing, or model selection. In this schematic example, the hypothesis A appears to be a very inaccurate prediction of x, and is thus unlikely (and rejected). Hypothesis B, however does predict the observation with a somewhat high probability, definitely higher than hypothesis A. Hypothesis B is thus preferred to hypothesis A, but there might be yet other preferable hypotheses.

So, what does this work actually teach us? First, it gives us some guidance on research warranted for a better understanding of the observations made by Cassini. We need to elucidate the abiotic processes that could produce enough methane to explain the data. For example, methane could come from the pyrolysis of primordial organic matter putatively present in Enceladus’ core, which could be partially turned into dihydrogen, methane and carbon dioxide through the hydrothermal process. We are not able yet to quantify methane production by this process, but improving our understanding of Enceladus’ history would be helpful. In particular, if Enceladus formed through the accretion of organic-rich material (e.g.  cometary), the hypothesis of the plume methane coming from pyrolysis may be very plausible. An alternate mechanism is the outgassing of primordial methane buried in Enceladus since its formation. Having such mechanisms in mind, we asked: how would our results change if arbitrarily large quantities of methane could be produced by an abiotic mechanism in Enceladus’ interior? We found that the combination of an elevated abiotic methane production together with biological methanogenesis could yield the observed abundance of methane in the plume. In this thought experiment, rejecting either hypothesis (abiotic methane production vs. biological methanogenesis) partly boils down to how probable we believe these hypotheses are a priori. For example, if we assign a low probability for life emergence in Enceladus, the hypothesis of pyrolysis turns out to be strongly supported.

What we find most promising here is the methodology: looking at habitability and inhabitation as hypotheses that can be cast precisely and rigorously within a quantitative inference framework. Our process-based Bayesian approach is general (not limited to specific systems such as interior oceans of icy moons) and paves the way to deal with exoplanet atmospheric data as they become available in the coming decades.

You can find the full article here.

Antonin Affholder

PhD student, École Normale Supérieure & Observatoire de Paris