General context

It is well established that the Earth's surface reservoirs have undergone major changes of their chemical composition over time. The development of new geochemical redox proxies in the last 10 years (e.g. iron and trace metals speciation and isotopes, sulfur isotopes, multiple sulfur isotopes) has allowed a tremendous refinement in our knowledge of the timing of the redox evolution of the atmosphere and oceans, and of their respective chemical composition. These studies roughly converge towards the current consensus scenario:

·      Until 2.4 Ga, the Earth oceans and atmosphere were anoxic. The widespread deposition of banded iron formations (BIF) indicates that the oceans were ferruginous. A first redox increase seems to have occurred at 2.7 Ga (e.g. Thomazo et al., 2009a; 2009b; 2011) but it remains unclear if very low concentration of free oxygen were accumulating or not in the environment before the Great Oxidation Event (GOE) at 2.4 Ga (e.g. Farquhar et al., 2011).

·      After 2.4 Ga, the delivery of oxygen by oxygenic photosynthesis would have oxygenated the atmosphere and surface oceans. The atmosphere would then have contained more than 1% of the present atmospheric oxygen level (PAL; e.g. Holland, 2009). Oxidative weathering on the emerged continents could have started, oxidizing pyrite and delivering sulfate to the surface oceans. The reduction of sulfate to sulfide in the anoxic waters underlying the oxic surface waters would have progressively titrated out the dissolved iron by pyrite precipitation (Canfield, 1998). Patches of sulfidic waters thus likely started to develop, initially on continental platforms, possibly extending to the deep oceans with time (see Planavsky et al., 2011). As a result, BIF deposition stopped after 1.8 Ga, with an episodic recurrence in the Neoproterozoic at around 0.7 Ga. The deep oceanic waters would have remained essentially anoxic until the second major oxidation event, the Neoproterozoic oxidation event at 0.7-0.5 Ga (NOE, Och and Shields-Zhou, 2012).

·      After 0.5 Ga, the oxygen concentration in the atmosphere would have reached at least 10% of the present value, and the oceans would have become dominantly oxic with short episodes of widespread anoxia.

The biosphere, via its metabolic activities, is believed to have been one of the main drivers of these redox changes, photosynthesis coupled to organic matter and/or pyrite burial supplying oxidants. In turn, redox conditions could impart a strong control on metabolic activities. Direct fossil evidences for these linkages remain elusive, but a wealth of indirect evidences have been found. The most compelling ones are the major changes identified in the carbon, sulfur, iron and nitrogen isotope secular variations in sedimentary deposits, which seem to be correlated to the major redox increases and are interpreted as reflecting coupled changes in the biogeochemical cycle of oxygen, carbon, nitrogen, sulfur and iron (e.g. Thomazo et al., 2009a, 2009b; 2011; Sansjofre et al., 2011; Och and Shields-Zhou, 2012).

The most documented isotope secular evolution through geological times is the carbon one. It shows a relative stability around 0‰ from at least 3.5 Ga but with some so-called excursions towards more positive or more negative values (e.g. Schidlowski, 2001; Bartley and Kah, 2004). In particular, two periods of positive d13Ccarb excursions stand out because of their durations (100 to 200 Ma) and extreme amplitudes, locally reaching values of up to +14‰ (e.g. Schidlowski, 2001). The first positive excursion (commonly referred to as the Lomagundi excursion) occurs at 2.2-2.0 Ga, right after the GOE. The second one occurs during the Neoproterozoic (0.8-0.55 Ga) (e.g. Och and Shields-Zhou, 2012) at the same time period as the NOE. Both are thus temporally associated with periods of indisputable redox increase of the Earth’s surface.

These positive d13Ccarb excursions have long been (and still are by a large majority) interpreted as a global increase in the proportions of carbon burial as organic matter relative to carbonates. Because an increase in organic matter burial results in an increase in oxidant supply to the oceans and atmosphere, periods of high d13Ccarb have been proposed as periods of increased oxygen supply to the ocean and atmosphere (Schidlowski, 2001). However this interpretation entails that this isotope signal is global, so that all sedimentary carbonates deposited at the same period should show the same positive excursion. Recent increases in the spatial and temporal resolution of these d13Ccarb excursions has started to reveal instead strong regional and temporal variability of d13Ccarb (Ader et al., 2009; Frauenstein et al., 2009; Frimmel, 2010), suggesting that the mechanisms responsible for these positive d13Ccarb excursions may rather be regionally controlled and may not record global variations in the proportion of organic matter burial. Two mechanisms are most likely to generate a local to regional d13CDIC increase: methanogenesis (hydrogenotrophy: CO2 + 4 H2 → CH4 + 2H2O or acetoclastic methanogenesis: CH3COOH → CH4 + CO2), which leads to 13C-enriched DIC (Hornibrook et al., 2000) and/or a high primary productivity, photosynthesis preferentially taking up 12C. Both have been proposed as mechanisms for the positive d13Ccarb excursions in restricted basins or in sediment pore-water (Ader et al., 2009; Hayes and Waldbauer, 2006; Frimmel, 2010) but lack support from direct observation of modern systems in which these mechanisms can be shown to lead to the precipitation of carbonates with high d13Ccarb values. Moreover, even if these mechanisms were demonstrated as able to induce positive d13Ccarb in modern analogs, their unambiguous identification in the rock record would still require confirmation by other proxies yet to be identified, since as exposed above, a positive d13Ccarb signature alone can also be produced by an elevated organic matter burial rate, the most often invoked hypothesis to explain the Lomagundi excursion. It is therefore essential to identify in these modern analogs other distinctive biogeochemical features and the chemical, mineralogical or isotopic signatures they leave in rock record.

Up to now, few modern systems have been identified as analogs for Precambrian oceans and are essentially limited to marine sulfidic basins (i.e. Black Sea; Cariaco Basin, Framvaren Fjord). These systems exhibit negative d13CDIC and are thus inadequate analogs for the Precambrian time periods recording positive d13Ccarb excursions. Their negative d13CDIC results from the fact that DIC is issued mainly from organic matter mineralization by sulfate reduction because of their relatively high sulfate contents. Moreover for most of the Precambrian times the oceans are thought to have been sulfate-poor. Permanently stratified freshwater (and thus sulfate-poor) lakes with ferruginous deep waters (e.g. Lake Matano and Lake Pavin) are currently investigated. In these lakes, because of the low sulfate concentrations, organic matter is mostly remineralized by methanogenesis so that the bottom waters have accumulated methanogenesis-derived DIC, which is characterized by a positive d13CDIC (Assayag et al., 2009; Crowe et al., 2010). However, these freshwater lakes do not contain enough Ca2+ and Mg2+ to allow carbonates precipitation, except for minor amounts of siderite, which may precipitate from dissolved Fe2+. The analogy with Precambrian environments precipitating carbonate rocks with a positive d13Ccarb excursion therefore remains unsatisfactory.

Our consortium has recently identified the saline crater lake Dziani Dzaha in Mayotte (Indian Ocean) as an extremely promising modern analog for Precambrian environments in which carbonates precipitate with a positive d13Ccarb. It is radically different from all previously studied modern analogs of Precambrian oceans, and possibly unique on Earth. According to our unpublished preliminary results (see section 2.2) this saline lake remains permanently anoxic below 1.5 m and precipitates carbonates both as microbialites and carbonate mud. Our preliminary isotope measurements have shown positive d13C of +13‰ both in the DIC and in the carbonates, similar to those of the Precambrian positive excursions, while the nearby ocean d13CDIC is close to 0‰. Moreover, the lake ecosystem is massively dominated by prokaryotes, which is a relatively scarce feature in modern ecosystems while it is believed to have been the rule at the time of the first positive excursion at 2.2-2.0 Ga (Close et al., 2011). This lake may thus allow us to get an insight into the conditions liable to have led to regional d13Ccarb enrichments at 2.2-2.0 Ga.

photo: S. Turay / Sur une île ; (c) DCO 2014