The grandest of them all: the Lomagundi–Jatuli Event and Earth's oxygenation

Evidence of the oxygenation of Earth's surface environments during the Paleoproterozoic is based on a now familiar list of temporally linked phenomena that includes geological features (decline of banded-iron formation, cessation of uraninite and pyrite as resedimented detrital particles, advent of non-marine red beds; e.g. Roscoe 1969), loss of sulphur mass-independent isotope fractionation (Farquhar et al. 2000), Lomagundi–Jatuli and Shunga C-cycle perturbations (Schidlowski et al. 1976; Karhu and Holland 1996; Melezhik et al. 1999a) and the hypothesised window of Cyanobacteria evolution (e.g. Fischer et al. 2016; Sánchez-Baracaldo and Cardona 2020). Of those, the Lomagundi–Jatuli Event (LJE) has become essentially synonymous with tracking the trend of atmospheric oxygenation (e.g. Karhu and Holland 1996; Bekker and Holland 2012; Lyons et al. 2014). The LJE is the largest positive carbonate–carbon isotope (⁠$δ$¹³C_carb) excursion in Earth history, marked by $δ$¹³C_carb values between +5 and 10‰ (and higher) in marine carbonate units tens to several hundreds of metres thick that were deposited between c. 2.3 and 2.1 Ga (Melezhik et al. 2007; Martin et al. 2013; Gumsley et al. 2017). Two contrasting hypotheses have been championed for the genesis of the LJE. The first is the consensus view, which interprets the LJE as a synchronous and global-scale disturbance of the carbon cycle linked to changes in the oxygenation state of the ocean–atmosphere system such that, following the Great Oxidation Event (Holland 2006; although see Ohmoto 2020 for an alternative perspective), a biogeochemical feedback was established with atmospheric O₂ rising and falling in concert with co-varying fluctuations between oxidative weathering of continental landmasses, nutrient fluxes and primary productivity (Karhu and Holland 1996; Bekker and Holland 2012; Planavsky et al. 2012; Canfield et al. 2013; Scott et al. 2014; Bellefroid et al. 2018; Miyazaki et al. 2018; Ossa Ossa et al. 2018; Hodgskiss et al. 2019). The second view is advocated by a minority of workers such as Melezhik et al. (1999b) and Frauenstein et al. (2009) who interpret high C-isotope values of the LJE as due to processes bespoke to palaeoenvironmental settings within individual sedimentary basins (e.g. organic burial rate, sediment flux, evaporation, methanogenesis) that enhance a plausibly global signal. Central specifically to the consensus view is the conjecture that the highly positive $δ$¹³C_carb values of the LJE are representative of the dissolved inorganic carbon (DIC) reservoir of the global ocean which, in turn, underpins the paradigm that the carbon cycle of the LJE and atmospheric oxygenation are directly interrelated (Schidlowski 1988; Karhu and Holland 1996; Planavsky et al. 2012; Lyons et al. 2014; Daines et al. 2017; Bellefroid et al. 2018). Here we assess this paradigm by analysing the relationship between carbon isotope data and facies (i.e. palaeoenvironment) for LJE-bearing marine carbonate rocks worldwide (Fig. 1). Because many LJE-bearing successions are lean in organic carbon, our focus is on carbonate–carbon isotope data.

The connection between key fluxes of carbon as carbonate precipitated out of the marine DIC reservoir and the amount buried as organic matter, including the respective isotopic compositions of each flux, has been commonly evaluated within a framework of mass balance and used in evaluating the pattern and tempo of the global carbon cycle (e.g. Schidlowski 1988; Kump 1991; Hayes and Waldbauer 2006; Mason et al. 2017). With few exceptions, the $δ$¹³C value of marine DIC has remained largely within a window of 0 ± 4‰ for most of geological time (e.g. Schidlowski 1988; Prokoph et al. 2008). The LJE is one of the most remarkable of those exceptions, and hypotheses relying on mass balance have struggled to explain the vast amounts of organic matter that would need to be buried to generate such positive $δ$¹³C_carb values (e.g. Karhu and Holland 1996; Aharon 2005; Holland 2006) or have opted to offer speculations invoking an unprecedented spike in primary productivity (e.g. Bekker and Holland 2012). The postulated net outcome of either is an extraordinary pulse in atmospheric oxygen. Other models invoke tectonics driving exhumation and erosion of older sedimentary rocks that, in turn, drive fluctuations in the isotopic composition of the carbon flux entering the marine realm to generate C-isotope excursions in lieu of changes to primary productivity (e.g. Kump 1991; Miyazaki et al. 2018), or attribute C-isotope excursions to changing intensities and proportions of chemical versus physical weathering, in part linked to tectonics, such as supercontinent cycles (e.g. Shields and Mills 2017). Our purpose is not to evaluate the veracity of these various models. Rather, our work assesses the key premise that underpins the LJE paradigm, namely, that the magnitude and duration of its $δ$¹³C_carb values are an accurate archive of the C-isotopic composition of the global ocean between 2.3 and 2.1 Ga. All concepts and models that use the LJE as evidence of the redox evolution of Earth's surface environments and initial rise and proliferation of free di-oxygen rely on that premise to be true.

We compiled a database consisting of 14 943 published $δ$¹³C_carb data: 11 557 data are for the time interval 3.0–1.0 Ga, which includes 2038 $δ$¹³C_carb data from LJE-bearing rocks worldwide. Of the latter, 188 are new $δ$¹³C_carb and $δ$¹⁸O_carb isotope data obtained on carbonate rock samples from outcrops of the LJE-bearing Tulomozero Formation in the Onega Basin, NW Russia, at Raiguba (62°22.107′N, 033° 47.129′E) and Shunga (62°36.697′N, 034° 49.457′E) and the portion of the Onega Parametric Hole drill core containing the Tulomozero Formation and lower part of the overlying Zaonega Formation. The remaining 3386 data are from Phanerozoic carbonate rocks deposited in environments ranging from sabkha to open ocean and used as a comparator for the LJE. All data and their sources are reported in the Supplementary material. For the new data, measurements were made on a Thermo Scientific Delta V Advantage continuous flow isotope ratio mass-spectrometer in the Department of Geology, University of Tartu, Estonia, following standard protocols. C and O data are reported in per mil (‰) deviation relative to the Vienna PeeDee Belemnite (V-PDB) and standard reproducibility was better than ±0.2‰ at 1$σ$⁠. In constructing the database of published $δ$¹³C_carb values for the 3.0–1.0 Ga rocks, we included only those data that record primary depositional (or nearly so) conditions as per original authors’ reporting of petrographic features, C–O isotope cross-plots and key element ratios. We excluded data from rocks that experienced significant post-depositional isotopic resetting and from features of clearly later diagenetic origin (e.g. concretions, cements), as well as carbonate rocks associated with banded-iron formation. Data are evaluated in 100 Myr bins (geons of Hofmann 1990), using the inferred depositional ages reported in the literature.

Sample means were determined by the bootstrap method using an approach described in Keller and Schoene (2012) and Cox et al. (2018) which enables estimation of the sampling bias on statistics (mean, median, etc.) whereby a synthetic dataset (bootstrap sample) is created by random independent sampling with replacement from an existing sample (population). The statistics of interest are estimated for each bootstrap sample and each step is repeated numerous times obtaining estimates that can be treated for additional statistical inferences (mean, standard deviation, confidence interval, etc.). The bootstrapped means and confidence intervals reported in Table 1 and Supplementary material were calculated from 10 000 repetitions. The sample bias of the binned data was assessed further by jackknife estimates (Tukey 1958) that were in all cases in agreement with bootstrapped means (jackknife bias = 0). Jackknife procedure is similar to the bootstrap method but sampling is done without replacement and was developed for estimating the variance and bias of large datasets. During the jackknife procedure, a statistic estimate is calculated leaving out one observation at a time from the sample set. Similar to the bootstrap method, the statistic of an estimate for the bias can be calculated from the population of the repeated calculations.

We focus here on two archetypal LJE-bearing successions: the Franceville Basin, Gabon, and the Karelia (Onega Basin)–Kola (Pechenga–Imandra Varzuga Belts) regions, Russia. These have been studied extensively by many workers including ourselves, the former by Préat et al. (2011), Canfield et al. (2013), Ossa Ossa et al. (2018) and Bakakas Mayika et al. (2020) and the latter by Galimov et al. (1968), Yudovich et al. (1990). Akhmedov et al. (1993), Karhu (1993) and Tikhomirova and Makarikhin (1993) and via the International Continental Scientific Drilling Program's Fennoscandia–Arctic Russia Deep Early Earth Project (FAR-DEEP; Melezhik et al. 2013a). Because exposure is poor in both regions, building sedimentological and stratigraphic frameworks relies on drill core to supplement outcrop-based observations. As examples, we highlight the Onega Parametric Hole and FAR-DEEP cores in the Onega Basin and LST12 core from the Lastoursville area of the Franceville Basin. These cores are representative of the sedimentology and stratigraphy common to LJE-bearing successions in those regions and, incidentally, to many other LJE-bearing localities elsewhere.

In both the Franceville and Onega Basins the LJE-bearing successions begin with interbedded pale grey to pink dolostone and red–brown–grey mudstone and dolomarl containing variable proportions of shallow-water sedimentary structures including stromatolites, evaporite fabrics and (palaeo)karst, tidal couplets and cross-bedded grainstones, and ripple and herring-bone cross-lamination replete with reactivation surfaces and mud drapes (Fig. 2). These facies occur in units many tens to several hundreds of metres thick and pass upward into very fine dolostone, dolomarl, mudstone and shale in units many tens of metres thick marked by sets of planar- to wavy-parallel laminae with small ripples, in places hummocky cross-stratification, and millimetre- to centimetre-thick graded beds and rhythmite (Fig. 3); colours vary from grey to black excepting one 10–30 m thick unit in the Onega Basin that is cream–pink–green in colour (more on this unit below). The vertical facies successions in both basins offer clear sedimentological evidence for a transition from nearshore-marine–intertidal and in places coastal sabkha settings for the pink–grey dolostone interval into open and deeper marine settings (i.e. depths near or below storm-wave base) for the overlying fine dolostone–dolomarl–mudstone interval. This interpretation is in general agreement with all previous workers (e.g. Brasier et al. 2011; Préat et al. 2011; Melezhik et al. 2013b; Ossa Ossa et al. 2018; Bakakas Mayika et al. 2020) and permits us to recognize three main palaeoenvironmental clusters: intertidal and in places sabkha settings, nearshore-marine–inner shelf settings, and open and deeper marine settings.

The LJE-bearing successions in the Franceville Basin and Karelia–Kola regions share another trait; they display an overall declining stratigraphic trend in $δ$¹³C_carb values from c. +6–10‰ in the coastal–nearshore-marine intervals to 0–5‰ in the overlying open- and deeper-marine intervals. In the consensus model, this trend is interpreted as recording the termination of the LJE (e.g. Melezhik et al. 2007; Canfield et al. 2013; Ossa Ossa et al. 2018). A salient observation, though, is that this trend coincides with facies changes such that C-isotope values correlate directly with palaeoenvironment (Table 1; Fig. 4a): the mean value for coastal intertidal–sabkha facies is +9.3 ± 1.7‰, for nearshore marine–inner shelf facies it is +6.8 ± 1.5‰ and for open- and deeper-marine facies it is +2.0 ± 2.1‰. Thus, the high $δ$¹³C_carb values that define the LJE in the Karelia–Kola and Francevillian successions are an expression of facies; they are present in nearshore-marine–coastal–sabkha settings but minor in open- and deeper-marine settings. The question that arises is: is it merely a coincidence that facies changes and the trend of $δ$¹³C_carb values are synchronized in those regions?

Golovkinsky–Walther's Law is a fundamental guiding principle in stratigraphy. First articulated in the mid-1800s by Nikolai Golovkinsky working in the Perm district of Russia (Nurgalieva et al. 2007) and later independently by Johannes Walther working in the Austrian Alps (Walther 1894), it states that in a vertical succession of strata unbroken by significant hiatal surfaces, rocks layered one on top of the other were deposited in coeval, laterally adjacent environments. Given that C-isotope values and facies changes co-vary in the Francevillian and Karelia–Kola LJE-bearing successions, it is important to assess if the trend in $δ$¹³C_carb in those successions can be attributed to original environmental isotopic gradients now superposed stratigraphically, as per Golovkinsky–Walther's Law, rather than recording global ocean DIC as per the consensus view.

To determine this, we expanded our analysis to encompass published data for LJE-bearing rocks worldwide and focused on studies in which authors provided sufficient sedimentological details to enable assigning their facies interpretations and $δ$¹³C_carb data into our palaeoenvironmental categories of intertidal–coastal–sabkha, nearshore marine–inner shelf, and open- and deeper-marine. Strikingly, even though $δ$¹³C_carb values in the worldwide dataset range between c. −10‰ and c. +26‰, for the entire 200–300 million year duration of the LJE there is a strong co-dependence between facies changes and $δ$¹³C_carb values (Table 1, Fig. 4b): the mean value for the open-marine realm was +1.5 ± 2.4‰, for the nearshore-marine–inner shelf +6.2 ± 2.0‰ and for intertidal–sabkha settings +8.1 ± 3.8‰. To place the LJE in context, we extended our analyses further to include marine carbonate rocks from 3.0 to 1.0 Ga, and also from times of major Phanerozoic evaporite deposits (Permo-Triassic, Messinian, modern sabkhas), given that LJE-bearing successions can contain variable amounts of evaporite pseudomorphs. This analysis shows that $δ$¹³C_carb values of open- and deeper-marine carbonate rocks for all assessed time periods, including LJE-bearing successions, were consistently within 0 ± 4‰ and comparable to that for the modern open ocean DIC reservoir (Fig. 4c). The data confirm that the uniquely highly positive $δ$¹³C_carb values of the LJE are themselves facies dependent, found dominantly in nearshore-marine and coastal settings and mostly absent in open- and deeper-marine settings. Hence, these highly positive $δ$¹³C_carb values are not representative of either the global DIC pool or the operation of the global carbon cycle.

In a few LJE-bearing successions (e.g. Tulomozero–Zaonega Formation; Silverton Formation, South Africa; Frauenstein et al. 2009) high $δ$¹³C_carb values are known for rocks interpreted as recording more open- and deeper-marine settings. We have first-hand knowledge of one of those and offer it as both an exemplar and explanation for such occurrences (see Fig. 5). In the Onega Basin, the upper part of the LJE-bearing Tulomozero Formation consists of sabkha and shallow-marine evaporite–fabric-bearing dolostone and mudstone (Brasier et al. 2011; Melezhik et al. 2013b) with $δ$¹³C_carb between +6 and 12‰. Above these rocks is a several tens-of-metres thick unit consisting of centimetre- to decimetre-thick beds of grey to dark grey dolostone–dolomarl–mudstone with $δ$¹³C_carb from +3 to 5‰, overlain by cream–pink–green fine dolostone–dolomarl marked by $δ$¹³C_carb between +6 and 9‰ (these dolostones are termed krivozerite by Russian workers) and then a return to grey–dark grey dolostone–dolomarl–mudstone in which $δ$¹³C_carb declines to 0‰. This interval passes upward into organic-rich and pyritiferous deep-marine mudstones of the Zaonega Formation (e.g. Črne et al. 2013, 2014; Kreistmann et al. 2019; Paiste et al. 2020).

Assessed within the construct of conventional thinking about the LJE, such a fluctuating C-isotope pattern is taken as evidence of oscillating productivity–oxygenation episodes (e.g. Bekker and Holland 2012). However, there is a more likely, albeit less grand, explanation. Detailed petrography, mineralogy and sedimentology by Črne et al. (2014) showed that many carbonate beds in the lower part of the Zaonega Formation were resedimented (e.g. Fig. 3e), emplaced by turbidity currents inferred to have originated from laterally adjacent shallower settings of the Tulomozero Formation. Such an interpretation is consistent with evidence for erosion and transport of carbonate detritus as per the abundance of cross-bedded doloarenite and dolorudite in the Tulomozero rocks (e.g. Fig. 2i). Applying Occam's razor, the episodic return of high $δ$¹³C_carb values in the deeper-water lower Zaonega Formation represents sediment derived from the ¹³C-rich shallow-water Tulomozero Formation intermixed with carbonate of the open- and deeper-marine DIC pool with its ‘normal’ C-isotopic composition. We propose that other postulated open- and deeper-marine units with very positive $δ$¹³C_carb values (e.g. Silverton Formation) may also reflect redeposition of ¹³C-rich carbonate transported from adjacent shallower environments.

Our analyses reveal that the highest-amplitude, longest-lived deviation from what is considered to be ‘normal’ C-cycle functioning over the past four billion years is tied directly to facies changes. Everywhere during the 200–300 million year duration of the LJE, the C-isotopic composition of facies recording open- and deeper-marine conditions remained ‘normal’ whereas those deposited in nearshore–coastal realms sustained ¹³C-enriched isotopic compositions. Thus, the canonical image of the LJE as a high-amplitude C-isotope bulge punctuating an otherwise undisturbed c. two billion year continuum of relatively normal marine C-isotope values is misleading (Fig. 6). Instead, our findings show that the C-isotope pattern of the LJE is a record of contemporaneous lateral gradients in the isotopic composition of DIC pools, ranging from normal open-marine values of c. 0‰ $δ$¹³C_carb to extremely high values of +16‰ and even higher, in nearshore-marine–coastal settings. Stated simply, trends of C-isotope profiles in LJE-bearing successions are a consequence of Golovkinsky–Walther's Law, i.e. strata superposed during sedimentation occurring concurrently during lateral migration of coeval depositional environments characterized by distinct local DIC pools. We thus conclude that the $δ$¹³C_carb trends do not record a global-scale perturbation of the carbon cycle, nor can they be used to infer a worldwide C-isotopic composition of marine DIC. In these aspects, the LJE is not a singularity in that the link between facies changes and positive $δ$¹³C_carb excursions marked by sustained values of +5–10‰ and locally higher for stratigraphically significant intervals is known from other times in Earth history. Studies combining careful sedimentology and petrography with isotope geochemistry show that co-variation between facies and $δ$¹³C_carb values is common and can modify global signals; examples include the Cryogenian Tayshir Member of the Tsagaan Oloom Formation in Mongolia (Macdonald et al. 2009; Bold et al. 2016) and Etina Formation in Australia (Rose et al. 2012), Ediacaran Hüttenberg Formation in Namibia (Cui et al. 2018), end-Ordovician Hirnantian strata (Jones et al. 2020) and Silurian Ireviken and Lau Events in the Baltic Basin (Wigforss-Lange 1999; Rose et al. 2019). Studies of C-isotope process-response in recent and modern depositional environments provide additional credence to our scenario that the LJE is a record of lateral isotopic gradients of contemporaneous DIC pools. Those studies show shallow-marine settings may have 2–5‰ or more variability in $δ$¹³C_carb relative to that of the open ocean (e.g. Swart 2008; Geyman and Maloof 2019). These findings are apropos to the LJE, consistent with and substantiating interconnectedness between facies changes and $δ$¹³C_carb values. It is worth noting that, although exceptionally high $δ$¹³C_carb values (>15‰) are known for some LJE-bearing strata, the number of such values total 24 out of the 2038 values we amassed (i.e., c. 1% of the data) making them more a curiosity than a norm, much like modern environments where carbonates form with $δ$¹³C_carb in excess of +15‰ (e.g. Birgel et al. 2015).

Neither the modern nor ancient examples, though, explain why nearshore-marine and coastal settings during the LJE reached such high enrichments in ¹³C. For this, we do not yet have an answer and at this time prefer not to offer speculative scenarios. We highlight, however, that many LJE-bearing successions are shallow-marine carbonate rocks associated with evaporitic and sabkha-like features. Evaporation pond brines are known to have $δ$¹³C values of +8–16‰ and brine-evaporation experiments have generated values of +16–35‰ (Stiller et al. 1985), that overlap those of the LJE. Thus, modern and ancient examples hint of the possibility of a yet undetermined process-response link between C-isotope composition, evaporation and microbial ecosystems to create conditions for widespread ¹³C-enrichment in shallow-marine environments during mid-Paleoproterozoic time.

Our analysis integrating 14 943 $δ$¹³C_carb measurements with sedimentology for marine carbonate rocks 3.0–1.0 Ga in age and, including appropriate Phanerozoic deposits, reveals that ¹³C enrichment during the 200–300 Myr span of the LJE was restricted to specific environments: $δ$¹³C_carb values from nearshore-marine and coastal–sabkha settings yield means of 6.2 ± 2.0‰ and 8.1 ± 3.8‰, respectively. In contrast, carbonate sediments deposited in open- and deeper-marine settings are marked by $δ$¹³C_carb values averaging 1.5 ± 2.4‰, consistent with the mean C-isotope composition of oceans throughout most of geological time. These results indicate that high $δ$¹³C_carb values, the hallmark feature of the LJE, archive locally heterogeneous C-isotopic composition of contemporaneous DIC pools of coeval depositional settings. Our findings necessitate rethinking of models that rely on the LJE as a proxy for the global carbon cycle and atmospheric oxygen levels and highlight an unresolved question: why, during the middle part of the Paleoproterozoic Era, did coastal–nearshore-marine and evaporite settings acquire such high ¹³C-enriched values?

We thank the Institute of Geology, Karelian Research Centre of the Russian Academy of Sciences, Petrozavodsk, Russia, and the Department of Geology, Université des Sciences et Techniques de Masuku, Franceville, Gabon, for their hospitality and help in accessing drill cores and outcrops. We thank E. Stüeken and J. Raven for insightful comments that improved earlier versions of this manuscript, and D. Fike and A. Brasier for reviews. We also thank G. Shields for providing raw data compiled originally for Shields and Vezier (2002).

ARP: conceptualization (equal), investigation (equal), writing – original draft (equal), writing – review & editing (equal); KK: conceptualization (equal), investigation (equal), writing – original draft (equal), writing – review & editing (equal); AL: conceptualization (equal), investigation (equal), writing – original draft (equal), writing – review & editing (equal); AEF: conceptualization (equal), writing – original draft (equal), writing – review & editing (equal); TK: conceptualization (equal), investigation (equal), writing – original draft (equal), writing – review & editing (equal); YED: conceptualization (equal), investigation (equal); AER: conceptualization (equal), investigation (equal); DVR: conceptualization (equal), investigation (equal); PVM: conceptualization (equal), investigation (equal), writing – original draft (equal); MM: conceptualization (equal), investigation (equal); KB: conceptualization (equal), investigation (equal); MSWH: writing – original draft (equal), writing – review & editing (equal)

K.K., A.L. and T.K. received funding from Estonian Science Agency Project PRG447 and Yu.D., A.R., D.R. and P.M. were supported by the state assignment of the Institute of Geology, Karelian Research Centre of the Russian Academy of Sciences.

The datasets generated during and/or analysed during the current study are available in the Supplementary material included with this manuscript.

Scientific editing by Heda Agic