The Ediacaran Period (ca. 635-540 Ma) marks a turning point in Earth history when groups of morphologically complex multicellular eukaryotes, including algae and animals, attained ecological dominance, irrevocably changing Earth System dynamics (Lenton et al., 2014). These biological radiations took place after two prolonged episodes of global glaciation during the Cryogenian Period (ca.715-660 and ca. 650-635 Ma), and were accompanied by further short-lived (ca.580 Ma and ca.550 Ma) regional ice ages, interspersed with warmer intervals (Condon et al., 2005; Pu et al., 2016; Le Heron et al., 2019). The world’s oceans also became episodically more oxygenated during the Ediacaran with the extent of oxygenated seafloor reaching near-modern levels at times by the early Cambrian (Chen et al., 2015). Both climate and oxygenation are regulated by Earth’s long-term carbon cycle, and so perhaps unsurprisingly this interval is characterised by extreme carbon isotope instability (Kaufman et al., 1997). Since its discovery over 30 years ago (Burns & Matter, 1993; Calver, 2000), the uniquely high amplitude of the Shuram δ¹³C anomaly has led some to question conventional interpretations of carbonate carbon isotopes (Melezhik et al., 2005; Schrag et al., 2011). However, its pivotal position beneath fossil evidence for the first biomineralising, bioturbating and bilaterian animals encourages us to venture a more satisfying, parsimonious explanation for the Shuram anomaly, one that ties together carbon cycle perturbations with life’s evolution.

Conventional carbon isotope mass balance is based on the principle that the isotopic composition of carbon input via outgassing and weathering, and that of sedimentary carbon outputs are equal on time scales of >10⁵ years (Garrels & Lerman, 1984). This δ¹³C value is considered to be unchanging at -5‰: the average composition of crustal carbon. Because organic matter is depleted in ¹³C, and carbonate rocks precipitate in isotopic equilibrium with ambient dissolved inorganic carbon (DIC), the mean δ¹³C value of carbonate rocks and fossils can be used to determine the proportion that sedimentary organic matter makes up of the total sedimentary carbon sink. This proportion is generally referred to as f_org, which has varied over Earth history between 0.1 and 0.3 (Krissansen-Totton et al., 2015). Using a conventional mass balance approach, extreme negative anomalies can only be explained by negative rates of organic burial. This is particularly true of the late Ediacaran Shuram excursion when δ¹³C remained below -8‰ for at least ~10 Myr (Condon et al., 2005; Gong et al., 2017). In order to address this quandary, it was proposed that the pool of dissolved organic matter (DOM) in the Proterozoic ocean was much larger than today, and that negative excursions represent non-steady-state remineralisation of that pool (Rothman et al., 2003). However, later numerical treatments of this model pointed out that the Earth system cannot remain out of oxygen (and carbon) balance for such a long period of time (Bristow & Kennedy, 2008). In other words, there is insufficient oxidant even in the modern atmosphere and oceans to remineralize enough organic matter to drive a -8‰ δ¹³C excursion for several million years. As a result, many authors have interpreted extreme negative anomalies as diagenetic artefacts, biased sampling of authigenic cements or as purely regional phenomena (Bristow & Kennedy, 2008; Derry, 2010; Grotzinger & Fike, 2011; Schrag et al., 2013; Li et al., 2017; Shi et al., 2018). However, such arguments appeal to an inexplicable sampling bias, whereby globally correlative isotopic signatures are presumed to be unrepresentative of the global carbonate sink.

Here we take a different approach to the problem of negative δ¹³C excursions by viewing them in terms of a linked carbon-sulfur-oxygen system, whereby changes in oxidant dynamics caused an excess of organic carbon oxidation over burial, resulting in a smaller DOM reservoir. For steady state to be maintained throughout a negative δ¹³C excursion, shrinkage of the DOM pool would need to match surplus oxidant production for the duration of the anomaly. If we consider plausible δ¹³C values of -10‰ and -35‰ for deposited carbonates and kerogen (globally averaged carbon sinks), respectively, and -30‰ and -5‰ for the DOM reservoir and crustal carbon (globally averaged carbon sources), respectively (Lu et al., 2013; Krissansen-Totton et al., 2015; Lee et al., 2015), then organic carbon oxidation would need to increase over the background rate by approximately a factor of three. In other words, and following conventional C-isotope mass balance [δ¹³C_carb = (δ¹³C_carb-org) . f_org + δ¹³C_input at steady state], when f_org = 0.1 , δ¹³C_input = -12.5‰ and f_DOM = 0.3, whereby f_DOM is the proportion that marine organic matter oxidation contributes to the overall carbon flux. The oxidant imbalance (f_DOM = 0.3) would have needed three times as much oxygen as could have been supplied by organic burial (f_org = 0.1) alone, and so requires a contribution from other sources, most likely pyrite burial.

Although bacterial sulfate reduction coupled with pyrite burial releases on a mole-for-mole basis almost twice as much oxygen as organic burial , it is generally assumed that the oxygen released by pyrite burial is approximately matched by the oxygen consumed during pyrite weathering. However, most riverine sulfate derives from the weathering of evaporites (Burke et al., 2018), rates of which due to the sporadic and regional nature of evaporite deposits, will vary considerably over time (Wortmann & Paytan, 2012). During parts of the Proterozoic, when oceans were both iron-rich and anoxic, and so prone to sulfate reducing conditions (euxinia) at productive margins, one might realistically suppose that the amount of oxidising power transferred from rock sulfate to the surface environment would also have varied considerably, particularly during times when no basin-scale evaporite deposits were forming. We refer to this as ‘sulfur cycle imbalance’, meaning that the oxidant required to oxidise pyrite S was not balanced by the oxidant released by pyrite S burial and/or that the SO₄ flux to the ocean from evaporite weathering was not balanced by evaporite deposition (on timescales over which the long-term carbon and oxygen cycles must be in balance).

By writing a simple steady state mass-balance for the surface carbon cycle we can map the broad relationship between the evaporite sulfate input rate, net DOM oxidation and δ¹³C composition of the ocean-atmosphere system. These calculations show that an evaporite weathering flux of around 1.1 x 10³ mol S yr^-1 (around 10 times the modern flux, as proposed for the dissolution event during the early Cenozoic could sustain a carbon isotope excursion of between -10‰ and -15‰, depending on the proportion of the riverine sulfate flux that is eventually buried as pyrite. Thus, the amount of oxidant required to achieve a deep negative carbon isotope excursion through net organic carbon oxidation may reasonably result from basin-scale evaporite dissolution.

The second half of the Tonian Period (ca. 830-770 Ma) witnessed the largest evaporite depositional event of pre-Ediacaran time during the break-up of the supercontinent Rodinia (Evans, 2006; Turner & Bekker, 2016; Schmid, 2017). Estimates for the current preserved volume of Tonian evaporites range from 375,000 km³ to 912,400 km³, which places Tonian deposits among the largest basin-scale evaporite deposits ever (Evans, 2006; Prince et al., 2019). They include: the Centralian Superbasin, Australia (683,000 km³); Amundsen and Mackenzie-Ogilvie basins, Canada (193,000 km²); Adelaide fold belt, South Australia (100,000 km²); Duruchaus basin, Namibia (30,000 km²); and the Central African copperbelt (50,000 km²). Despite diverse geodynamic settings, these Tonian evaporite basins underwent inversion during middle to late Ediacaran time, mostly after ca. 570 Ma, and broadly coincident with the ‘Pan-African orogeny’ or ‘Transgondwanan supermountain’ events (Campbell & Squire, 2010). Specifically, Centralian evaporites were exhumed by deformation associated with the Kuunga Orogen; Amundsen-Mackenzie-Ogilvie evaporites were exhumed on a passive margin; South Australian evaporites were exhumed during development of a raised volcanic rift margin; Duruchaus was exhumed by early Damaride accretion of terranes onto an active margin; and the Copperbelt was exhumed by intrusion of an anorogenic large igneous province. Subsequent erosion means that initial volumes must have been substantially higher. The high sulfate mineral content of these evaporite deposits (Fakhraee et al., 2019) and relatively unchanging seawater δ³⁴S values (Strauss, 1993) are consistent with the build-up of a large marine sulfate reservoir during the early Neoproterozoic . Sulfur cycle imbalance (excess sulfate weathering over deposition, coupled with excess pyrite deposition over weathering) is likely to have continued until massive evaporite deposition began in the ‘Palaeotethyan’ realm after c.550 Ma (Schroder et al., 2004), at which point the imbalance seems likely to have been reversed.

This evaporite dissolution / DOM oxidation scenario appears to be the most parsimonious solution to the Shuram C-isotope conundrum in that it predicts the extent of oxic seafloor to increase towards the end of the excursion (Chen et al., 2015; Kendall et al., 2015), while maintaining high sulfate concentrations, which is in line with geochemical studies (Shi et al., 2018). Net oxidation of organic carbon (via pyrite burial) should cause a substantial rise in atmospheric CO₂ concentration, leading to global warming and enhanced continental weathering, with the potential to drive further evaporite dissolution and therefore sustain oxidant delivery.

Although our model fits best the late Ediacaran Shuram anomaly, sulfur cycle imbalance may have also played a role in other carbon isotope excursions. The interval of major carbon cycle disruption from ca.770 Ma until ca.550 Ma was a time of little or no basin-scale evaporite deposition (Guilbaud et al., 2015), suggesting that, as in the Cenozoic, the sulfate weathering-deposition cycle was not in steady-state. However, unlike the Cenozoic, the low atmospheric oxygen and anoxic deep ocean of the Neoproterozoic allowed evaporite-derived oxidizing power to be effectively transmitted into a negative δ¹³C signal. Such imbalance, potentially driven by weathering events and related positive feedbacks, may be particularly relevant to the ending and immediate aftermath of ‘Snowball Earth’ glaciations, which are marked by euxinia and anomalously high levels of pyrite deposition (Sahoo et al., 2012; Lang et al., 2018).

The existence of a series of negative carbon isotope excursions during much of the Neoproterozoic Era indicates that the DOM pool underwent dynamic size changes throughout this time and served as a buffer against oxygenation and climate change, but only when the pool was sufficiently large. The absence of a large DOM buffer in the ocean, climate and oxygen levels would have been rendered more vulnerable to change. There are tantalising hints that this may have been the case after c.550 Ma (Tostevin et al., 2019), and that the ocean redox balance remained sensitively balanced throughout the Ediacaran-Cambrian transition interval (He et al., 2019). If exhaustion of the DOM pool occurred during the Shuram anomaly, it is plausible to suppose that the subsequent expansion of aerobic Ediacaran fauna was an opportunistic radiation in response to a transient oxidant surplus.