Strong Evidence that Abiogenesis Is a Rapid Process on Earth Analogs

1. The Inference Hurdle Posed by the Weak Anthropic Principle

The earliest fossil evidence for life on Earth comes from 3.7 Gyr old metamorphosed sedimentary rocks in southwest Greenland, which contain 1–4 cm high stromatolites (Nutman et al., 2016). Earth plausibly became suitable for the emergence of life once the oceans formed (4.404 ± 0.008) Gya (Wilde et al., 2001), and thus life clearly began quickly (≤700 Myr). It is tempting to assume that this timescale would be typical on similar Earth-like exoplanets and therefore that “Life is not a fussy, reluctant and unlikely thing,” to quote a commentary article associated with the stromatolite discovery (Allwood, 2016).

Consider, however, the possibility that the timescale for evolution to run its course and produce self-aware organisms capable of statistics, geology, paleontology, and so on ¹ is consistently long, say 3.7 Gyr, as occurred here. It has been estimated that Earth’s biosphere will collapse as a result of the Sun’s growing luminosity in approximately 0.9 Gyr (Caldeira and Kasting, 1992), which would mean that the latest epoch for life to start and still have time to lead to something like us would be 2.8 Gya. In this picture, life must start (3.6 ± 0.8) Gya—else we would not be here to talk about it. Hence, the observed value of 3.7 Gya is hardly surprising. This is an example of a selection effect influencing our bias, specifically a case of what has become known as the “weak anthropic principle” (Carter, 1974).

Bayesian statisticians have previously attempted to assess the case for a rapid abiogenesis, conditioned upon this chronology (Spiegel and Turner, 2012). However, that initial work did not allow both the abiogenesis and evolutionary timescales to be jointly inferred, which is critical given the covariance imparted by the weak anthropic principle. This author remedied this issue (Kipping, 2020) and reframed the problem in terms of hypothesis comparison (rather than posterior inference), which dissolves the influence of seemingly subjectively chosen priors on the emergence rates. The result is an expression for the minimum odds ratio between a fast and slow abiogenesis scenario, given byZ fast − life Z slow − life ≥ { T 2 t ′ L i f T ≥ 2 t ′ L + t ′ I T t ′ L 4 ( T − t ′ I ) t ′ L − 2 t ′ L 2 − ( T − t ′ I ) 2 if T < 2 t ′ L + t ′ I(1) where T is the lifespan of Earth’s biosphere, t′ _L is the time it took for the first evidence of life to appear in the historical record (which necessarily succeeds the true emergence date), and t′ _I is the time from then until intelligent life emerges (presumed to be the current epoch). Using previously published estimates of T = 5.304 Gyr (Caldeira and Kasting, 1992; Wilde et al., 2001), t′ _L = 0.7 Gyr from microfossils (Nutman et al., 2016) and t′ _I + t′ _L = 4.404 Gyr (i.e., intelligence emerged recently), one obtains a 3.8:1 odds ratio. A 400 Myr earlier start to life (4.10 ± 0.01 Gya) was indicated via the analysis of carbon-13/12 isotope ratios (Bell et al., 2015), and this adjusts the odds to 8.7:1. However, both fall short of the 10:1 threshold (Kass and Raftery, 1995) usually used to define “strong evidence.” Accordingly, the evidence up to now has been tantalizing but not substantive.

A recent study (Moody et al., 2024) revises these numbers by dating the Last Universal Common Ancestor (LUCA) using divergence time analysis of pre-LUCA gene duplicates and calibration using microbial fossils and isotope records. Updating Equation (1) to account for LUCA living 4.20 − 0.11 + 0.13 Gya (Moody et al., 2024) (95% interval) finally pushes us over the threshold to 13.0 - 3.1 + 5.4 : 1 (68.3% interval). However, the result is even more profound than this.

One might contest two numbers used in Equation (1): T and t′ _I . Concerning T, it was recently argued (Graham et al., 2024) that the lifespan may be much greater than previously calculated (0.9 Gyr), out to 1.8 Gyr, and this adjusts the odds to 15.2 - 3.6 + 6.3 : 1 . However, one might question whether a civilization could truly emerge during the tail-end of Earth’s waning biosphere (O’Malley-James et al., 2013) and thus argue for a more conservative T. Figure 1 shows the impact of varying T in Equation (1). We certainly know T ≥ 4.404 Gyr (since that is our arrival date), and thus one finds that in all cases, the odds exceed 11.3 with a minimum at T = 4.6 Gyr (200 Myr from now).

OPEN IN VIEWER

FIG. 1.

Odds ratio of the hypotheses that abiogenesis is a fast versus slow process. Each curve represents the odds ratio, computed using Equation 1, as a function of the biosphere’s ultimate lifespan, T—although we have subtracted off the modern epoch to yield remaining time on the x-axis. Previous estimates (Bell et al., 2015; Nutman et al., 2016) for the earliest evidence of life yield the dotted and dashed lines, but the new LUCA date of 4.2 Gya (Moody et al., 2024) greatly improves the odds factor. Indeed, the three solid lines show how the odds are always above 10:1 irrespective of how one chooses T or indeed fairly extreme choices (Carboniferious and Cambrian period) for the earliest civilization on Earth, t′ _I —evident by the circles which denote the minima. LUCA, Last Universal Common Ancestor.

The result is even more robust than this, though. For one might also contest t′ _I . The Silurian hypothesis suggests that a civilization could have emerged as early as the Carboniferous period (350 Mya), and we would have no record of its existence today (Schmidt and Frank, 2019). Figure 1 reveals that even this scenario leads to a minimum odds factor ² of 10.4:1 for all possible T. Finally, perhaps the most extreme possibility is a civilization emerging during the Cambrian explosion (535 Mya), the very start of complex life on Earth. Yet even this scenario still has a minimum odds factor of 10.0:1. The LUCA date is so far back that its effect overwhelms the weak anthropic principle, as well as any plausible variations on the other parameters involved.

Our work assumes that life began on Earth rather than via panspermia. Consider instead that life began on Mars and then transferred to Earth. The cooling timescale is proportional to planetary radius, and thus if Earth cooled from the Moon-forming impact (4.48 Gya [Canup et al., 2023]) to oceans in ∼76 Myr (Wilde et al., 2001), then Mars could have cooled in 40 Myr and had oceans no earlier than 4.50 Gya. The latest (most pessimistic) abiogenesis date would be the LUCA value, giving Z fast - life / Z slow - life > 9.0 assuming 0.9 Gyr for Earth’s habitability (Caldeira and Kasting, 1992). This becomes 10.5 using the more modern 1.8 Gyr value (Graham et al., 2024). Accordingly, despite invoking contrivances to engineer this number as small as possible, it still floats around the “strong evidence” threshold. For nondirected interstellar panspermia, it is shown in the Appendix that this would require essentially all interstellar objects to be life-bearing in order to work, which is simply implausible.

There are two caveats to the main result presented. First, the LUCA date is a recent result that may not stand up to scrutiny; our result is conditional in this sense. Second, our result does not establish that life is common, since Earth’s conditions could be incredibly rare (Ward and Brownlee, 2000). Our next task is clearly to look out and address this question: How common are conditions analogous to those of Earth?