The record of sea water chemistry evolution during the Ediacaran–Cambrian from early marine cements

The Ediacaran–Cambrian Radiation marks the widespread appearance of metazoans and calcareous biomineralised hard parts. These innovations occurred during an interval of dynamic changes in marine redox and sea water chemistry. Here, changing carbonate mineralogy, Mg/Ca ratios and rare earth element concentrations including the relative abundance of cerium (Ce anomaly: Ceanom) are documented to track sea water oxygen levels, in well‐preserved early marine cements from shallow marine reefs from Cambrian Stages 2–4 (ca 525–512 Ma). First, integrating the mineralogical data with published records, several shifts in dominant carbonate mineralogy are inferred: ‘dolomite‐aragonite seas’ in the late Ediacaran; ‘aragonite/high‐Mg calcite seas’ in Cambrian Stage 2; a temporary shift to a ‘calcite sea’ during early Cambrian Stage 3; an ‘aragonite sea’ between late Cambrian Stage 3 and late Cambrian Stage 4, then a gradual shift from mixed ‘aragonite–calcite seas’ during the middle and upper Cambrian towards a ‘calcite sea’ by the early Ordovician. Second, based on measured mMg/Ca in early marine cements, calculated sea water mMg/Ca at 15 and 35°C ranges from 1.2 to 0.8 in Cambrian Stage 2, 0.7–0.4 in Stage 3 and 1.4–0.9 in Stage 4 respectively. Finally, analysed Ceanom data combined with existing Ceanom data suggest potentially three phases of global oxic expansion. First, a long‐lived phase of progressive oxygenation during the late Ediacaran to Fortunian (ca 550–540 Ma; average Ceanom from 0.99 to 0.41), and possibly two shorter phases during early Cambrian Stage 3 (ca 519 Ma; average Ceanom from 0.91 to 0.40) and Stage 4 (ca 512 Ma; average Ceanom from 1.02 to 0.49), bounded by intervals of more dominant anoxia. Summarising, these data demonstrate that early marine cements offer an underused and high‐resolution archive of shallow marine redox and sea water chemistry through this critical transition in Earth's evolution.


| INTRODUCTION
The Cambrian Radiation (starting ca 541 Ma) is a pivotal event for the evolution of life, marking the transition from a microbial to an animal (metazoan) biosphere, with the rise of calcareous biomineralisation, predation, bioturbation and complex ecological networks (Erwin et al., 2011). These innovations occurred in parallel with globally, highly dynamic biogeochemical cycles, with dramatic changes in the carbon cycle, sea water chemistry and ocean redox conditions (Figure 1). Oxygenation is suggested to have proceeded via a series of oceanic oxygenation events (OOEs; Sahoo et al., 2016), although records are incomplete and correlation between regions throughout the Ediacaran-Cambrian transition are unresolved. Where known, carbonate uranium isotope (δ 238 U) values show an antithetical pattern with δ 13 C values, where decreasing values of δ 238 U are proposed to represent global expansions of anoxic, or more specifically euxinic, conditions (Dahl et al., 2019). Values of δ 238 U reach a nadir during the δ 13 C peak in the middle of Cambrian Stage 2 indicating increasing anoxia, and recover to a more positive mean value during the subsequent terminal Cambrian Stage 2 to lower Stage 3 interval, which may reflect a gradual transition to a less reducing (or less euxinic) global deep ocean characterised by continued redox stratification (Dahl et al., 2019). A causal relationship between biotic and physicochemical records has been proposed (He et al., 2019), but there remains, however, considerable debate as to whether oxygenation was the main driver of early metazoan evolution after an initial physiological requirement was met (Wood et al., 2019. Throughout the entire Ediacaran-Cambrian radiation, oxygen levels may have been relatively low, but dynamic, fluctuating redox conditions were prevalent on local as well as global, temporal and spatial scales. Open marine conditions were often typified by highly dynamic oxygen minimum zones (OMZs) overlying potentially oxic basinal waters (see Bowyer et al., 2017 for review). However, different geochemical proxies yield information with varying levels of spatial and temporal resolution, and local redox proxy data are unavailable for many important successions. The targeting of fine-grained, clastic facies has also led to considerable bias in existing data (Sperling et al., 2015), and many global proxies (such as δ 238 U) only enable estimates of expanded sea floor anoxia and do not differentiate between deep and shallow marine settings. Consequently, it is not clear if such expansions restricted the habitable area of the shallow shelf, where most biodiversity resides.
F I G U R E 1 Schematic evolution of redox conditions based on compiled iron speciation data and proposed widespread anoxic intervals and oceanic oxygenation events. Diagonal stripes indicate known regional differences in redox state. Paucity of truly basinal shale deposits prevents determination of the redox state of the global oceanic deep basin (indicated by white question marks). 'Global Marinoan' and 'Regional Gaskiers' indicate 'snowball Earths'. OOE, oceanic oxygenation event. Ediacaran and Cambrian C-isotope (δ 13 C carb ) compilation and isotopic excursions from Bowyer et al. (2022). Timing of the first appearances of the earliest inferred soft-bodied metazoans, skeletal metazoans and trilobites. FORT Bulk sampling can lead to an averaging or contamination of the signal, whereas early marine cements often precipitate directly from open water so potentially provide a more accurate record of original sea water chemistry (Nothdurft et al., 2004;Webb & Kamber, 2000;Whittaker et al., 1994). Mimetic preservation by dolomite (i.e., retention of original crystallographic orientation) of originally aragonite and/or high-Mg calcite grains (Corsetti et al., 2006;Tucker, 1982) and dolomite cements (Hood & Wallace, 2015) provides evidence that early marine dolomite precipitation dominated Cryogenian to early Ediacaran oceans (ca 740 to 630 Ma). The controls for this are not clear, but may be due to widespread anoxic, ferruginous oceans and high-Mg/Ca sea water (possibly up to molar 10 mMg/Ca; Hood et al., 2011). The presence of high iron (ferroan) concentrations in inferred early dolomite cements (Hood & Wallace, 2015) and ferroan dolomite concretions in shales supports the inference that these oceans were anoxic and ferruginous (Spence et al., 2016). These so-called 'aragonite-dolomite seas' (Hood et al., 2011) were largely replaced by 'aragonite seas' with aragonitic cements and inferred lower Mg/Ca ratios by the early (Corsetti et al., 2006;Hardie, 2003) or late Ediacaran . These in turn were succeeded by 'calcite seas' with dominant low-Mg calcite early marine cements indicating even lower Mg/Ca ratios by Cambrian Stage 3 (early Atdabanian). Fluid inclusions from Oman and Siberia from the late Ediacaran to early Cambrian show a rapid decrease in Mg/Ca ratios from 4 to 1 due to a marked rise in the concentration of calcium, supporting a transition from an 'aragonite' to 'calcite sea' (Brennan et al., 2004). This reduction in sea water Mg/Ca ratios is also supported by models (Hardie, 2003), and by elemental analysis of echinoderm stereom cements (Dickson, 2002). Such oscillations are significant, as early metazoan skeletal clades commonly co-opted carbonate minerals in concert with ambient ocean chemistry. All skeletal taxa from the late Ediacaran to Cambrian Stage 2 bear aragonite or high-Mg calcite hard parts, and low-Mg calcite skeletal hard parts not known until the appearance of 'calcite seas' in Cambrian Stage 3 (Porter, 2007;Zhuravlev & Wood, 2008).
The chemical behaviour of rare earth elements and yttrium (REEY) are proxy indicators for the average history of local and basinal/regional redox conditions respectively German & Elderfield, 1990). Unlike the other REEY, which normally exist in trivalent oxidative states, cerium (Ce) occurs in a tetravalent oxidative state with lower solubility (Ling et al., 2013). In a well-oxygenated marine setting, the soluble Ce 3+ is always oxidised to insoluble Ce 4+ with the catalysis of Mn and Fe oxides, removing the Ce from sea water (Bau & Koschinsky, 2009;German & Elderfield, 1990). Normalisation of Ce abundance relative to the neighbouring rare earths can therefore be used as a measure of oxidative Ce removal, known as the Ce anomaly or Ce anom .
Shallow water cement signatures record a strong negative Ce anom in fully oxic marine basins, and a partial Ce anom signature (i.e. a weaker negative Ce anom ) if there are anoxic bodies at intermediate or deep waters (Wallace et al., 2017). The Ce anom value for marine cements is well preserved during diagenesis  but aragonite to calcite conversion in meteoric settings can very slightly increase Ce anom values (Banner et al., 1988;Webb et al., 2009). Normalised values less than 1.0 (the 'negative' Ce anom ) are an indication of an oxic state, and more positive values always indicate anoxic or hypoxic states (McLennan, 1989;Wallace et al., 2017).
Compiled iron speciation data from shales do not show any statistical increase in the sea water oxygen level across the Ediacaran-Cambrian transition (Sperling et al., 2015), but Ce anom values indicate notable changes (Wallace et al., 2017). The Ce anom values show a moderate increase in oceanic oxygenation during the Ediacaran from the Cryogenian (average Cryogenian Ce anom = 1.1, average Ediacaran Ce anom = 0.62), followed by a decrease in oxygen levels during the early Cambrian (average Cambrian Ce anom = 0.90), with significant ocean anoxia persisting through the early and mid-Palaeozoic (average Early Cambrian-Early Devonian Ce anom = 0.84). It was not until the Late Devonian that deep marine ocean oxygenation reached levels comparable to the modern (average of all post-middle Devonian Ce anom = 0.55), which may be related to the rise of the land plants which probably increased weathering regimes and the extent of organic matter burial (Algeo & Scheckler, 1998;Lenton et al., 2016;Wallace et al., 2017).
Here, early Cambrian marine cements were analysed within primary cavities from shallow marine reefs precipitated from unaltered marine fluids, where such cements have been widely documented (Whittaker et al., 1994). First, the dominant carbonate mineralogy was reconstructed, and then cathodoluminescence (CL) imaging was used to screen for well-preserved earliest marine cements, which were targeted via microanalysis to infer sea water chemistry (Mg/Ca ratios), and oceanic redox state via REEY profiles.

| SAMPLES AND METHODS
Analysed samples were collected from four localities from the Lena River, Siberia, Mongolia and Southern Labrador, Canada, of Cambrian Stage++ 2 to 4 age (Table 1). Each region occupied a different microcontinent during the early Cambrian. Samples are from shallow marine reefs rich in archaeocyath sponges, calcimicrobes, that have undergone differing diagenetic histories (see Astashkin et al., 1990;James & Klappa, 1983;James & Kobluk, 1978;Kruse et al., 1996;Zhuravlev et al., 2015). Synsedimentary or very early marine cements are fibrous and formed isopachous crusts that either lined framework cavities associated with geopetal sediment infill, or directly encrusted upon, or within the internal pores of, archaeocyaths and calcimicrobes.
Highly polished thin sections were examined under transmitted light by Nikon SMZ800 Stereo and Leica DMLP microscopes, and under CL using a cold cathode, CITL 8200 Mk 3A mounted on a Nikon Optiphot petrological microscope to identify well-preserved early marine cements and undertake detailed petrographic descriptions. The well-preserved early marine cements always show pure dull or non-luminescence, and areas in patchy luminescence are avoided in analysis.
Only the earliest marine cement generations showing dull to non-luminescence were further analysed geochemically, with avoidance of patchy luminescence zones showing signs of recrystallisation.
The Mg, Ca, Fe, Mn and Sr elemental compositions were measured on targeted early marine cement zones on highly polished thin sections using a Cameca SX100 Electron Microprobe with an 80 s count time, a beam diameter of 3 μm, an accelerating voltage of 15 kV, and a beam current of 60 nA. A calcite-Silicarb All early marine cement zones were also analysed for REEY using a laser ablation (LA)-inductively coupled plasma mass spectrometer (ICP-MS) directly on highly polished thin sections using an Analyte Excite 193 nm ArF eximer based laser ablation system connected with an Attom HR-ICP-MS. Ablation frequency was 15 Hz, with the ablation time <1 min for every 100 μm size laser spot. In the absence of certified calcite standards, the well-characterised NIST 612 glass standard was used to calibrate the LA-ICP-MS analyses, with the REEY measurements normalised using 88 Sr concentrations established both by LA-ICP_MS and Electron Microprobe in the samples.
All imaging and analyses were undertaken at the School of Geosciences, the University of Edinburgh, UK.

| Mineralogy of early marine cements
Cambrian Stage 2 samples show thick, layered, fibrous cements (up to 2.5 mm) that are widely distributed within internal pores of archaeocyaths, attached to the calcimicrobe Renalcis, or associated with micritic, geopetal sediments lining framework cavities (Figure 2A,E). Fibrous cements are brownish and cloudy with abundant inclusions and composed of bladed to acicular crystals arranged in bundles (Figure 2A,D,E). Each bundle ranges up to 0.85 mm in length and 0.25 mm in width, with pointed terminations, and shows sweeping extinction with divergent optics, indicative of fascicular-optic cements ( Figure 2C). Under CL, most fibrous cements show very dull to nonluminescence, with a few rhombic crystals of microdolomite showing brighter luminescence ( Figure 2F). Remaining pore space is filled with thin (<50 μm) late equant cements of prismatic calcite with abundant inclusions, and clear, non-luminescent, blocky calcite cements. Based on the pointed crystal terminations and microdolomite inclusions, the original mineralogy of the fascicularoptic cements is inferred to be high-Mg calcite (Lohmann & Meyers, 1977;Marshall & Davies, 1981).
Cambrian Stage 3 (early Atbabanian) samples from Oy Muran show columnar cements that form thick (up to T A B L E 1 Geological distribution and ages of lower Cambrian reef samples analysed. Straits of Belle Isle, Labrador, Canada 1.5 mm) palisade structures of bladed, columnar crystals lining cavities ( Figure 3). Crystals reach 0.7 mm in height and 0.13 mm in width ( Figure 3B), and show incomplete divergent, undulatory extinction ( Figure 3E), indicating fascicular-optic cement. These cements are dull or non-luminescent under the CL, but with some patches of brighter luminescence ( Figure 3C). Clear, blocky, euhedral cements occlude remaining pore space. Some microdolomites ( Figure 3F) illustrate that the original mineralogy is inferred to be high-Mg calcite (Lohmann & Meyers, 1977). Slightly younger Cambrian Stage 3 (late Atdabanian) samples from Zuun-Arts show acicular crystal cements which infill the pores of radiocyaths, and are found within geopetal micrite ( Figure 4A). In the latter case, it is not clear if they formed replacively within the micrite, or are overlain by, micrite. These form botryoids, which show individual crystals reaching up to 0.5 mm in length with flat terminations ( Figure 4B) and show sweeping extinction, but most are calcitised to microspar aggregates. The acicular crystals and flat terminations of these early cements are inferred to have been originally aragonitic precipitates (Folk & Assereto, 1976). Some archaeocyath pores also show fascicular fibrous cements with sweeping extinction, inferred to have been high-Mg calcite ( Figure 4C). Remaining pore space is filled with blocky calcite.

Formations Siberian stratigraphy International stages and ages Locality
Cambrian Stage 4 samples from Labrador, Canada, show widespread and thick early cements up to 390 μm thick, within internal cavities of archaeocyaths and surrounding Renalcis ( Figure 5). The first generation of marine cements are fibrous and cloudy or brownish with many inclusions, and precipitate with, or before, geopetal sediment infill and within archaeocyath pores ( Figure 5A,C). These fibrous cements always occur as bundles up to 45 μm in length and 30 μm in width, with pointed crystal terminations ( Figure 5A). Some crystals up to 0.1 mm long are attached to Renalcis, pre-dating  Figure 5C). Some crystals are dolomitised. All show imperfect, sweeping extinction and divergent axes, indicating a fascicular optic crystal form, with dull luminescence ( Figure 5B-D). A second generation of clear, fibrous cements occur within reef framework cavities, with brighter luminescence ( Figure 5D,E). These early marine cements are interpreted as being originally high-Mg calcite (Aissaoui, 1988).
In some voids of archaeocyaths, acicular crystals form a thin layer, where crystals are about 20 μm long with  triangular cross-section, blunt terminations and with no luminescence ( Figure 5F). These are interpreted as originally aragonitic cements (Folk & Assereto, 1976).

| Elemental compositions of early cements
Maximum, minimum and mean values (ppm) for Fe, Mn, Sr and calculated mMg/Ca ratios were compiled for analysed earliest marine cements, with the addition of data from the late Ediacaran of the Nama Group, Namibia (ca 547 Ma) for comparison, from Wood et al. (2018) ( Table 2). Elemental concentrations and detection limits (DL) data for each analysed point are in Table S1.
Iron concentrations are very low below DL in all samples except for an elevated and variable range in Cambrian Stage 4 (mean = 1035 ppm; Figure 6A). Manganese is similarly low below DL except for an elevation in Stage 4 (mean = 135 ppm; Figure 6B). Strontium is low in all samples except for the late Ediacaran where levels reach a mean of 1890 ppm, and a moderate elevation of mean values to 430 ppm in Stage 2 ( Figure 6C Figure 7B). Calculated sea water mMg/Ca values vary depending on different formation temperatures. Based on measured mMg/Ca values, the following equations were used (Coggon et al., 2010;Rimstidt et al., 1998) to calculate the sea water mMg/Ca values, where T is temperature in Kelvin, and temperatures are constrained to 15-35°C for shallow marine cement formation.
At 15°C, mean values are 1.2 in Cambrian Stage 2, 0.7 in Stage 3 and 1.4 in Stage 4, and at 35°C mean values are 0.8 in Stage 2, 0.4 in Stage 3 and 0.9 in Stage 4 ( Figure 6D).

| Ce anomaly profiles from early marine cements
Normalisation of Ce abundance relative to the neighbouring rare earths (Ce anom ) is used as a measure of oxidative Ce removal.
The Ce anom can be calculated in two ways. First via 2Ce n /(La n + Nd n ) , which uses the neighbouring elements La and Nd as these appear with overabundant concentrations in the sea water, where n denotes normalisation of concentrations relative to the post-Archean Australian shale (PAAS; McLennan, 1989). In this calculation, Pr/Pr* = 2Pr n /(Ce n + Nd n ) can be used to constrain the true negative Ce anom . When the Pr/Pr* > 1.05 and the Ce/Ce* < 0.95, this represents the true negative Ce anom . Second, Ce n /(Pr n 2 /Nd n ) can be used after Lawrence et al. (2006). This calculation avoids normalisation relative to La, which typically has an over-abundance relative to neighbouring REEs (a La anomaly) in sea water (De Baar et al., 1985). In this calculation, depletion of Ce on normalised REEY profiles (referred to as a negative Ce anom and indicating greater sea water oxygenation) produces values less than 1.0, where 1.0 is equivalent to no Ce anom .
After normalisation, modern sea water REE profiles show an enrichment in heavy REE (HREE), a negative Ce anom , and a positive yttrium (Y) anomaly (Elderfield, 1988;Shields & Stille, 2001;Tostevin et al., 2016). This is due to the preferential absorption of light rare earth elements (LREE) (Bolhar et al., 2004;Byrne & Kim, 1990) and more rapid removal of Holmium (Ho) than Y (Frimmel, 2009). The enrichment of HREE in sea water is calculated via Nd PAAS /Yb PAAS , which can also be used for identifying sample contamination

A B
all data. The PAAS normalised REEY profiles in Cambrian Stage 2 show variable trends with negative and slightly positive Ce anomalies, a positive Europium (Eu) anomaly, and negative and positive Y anomalies ( Figure 8A). Cambrian Stage 3 REEY profiles also have a negative Ce anomaly, but a few samples also exhibit a negative Y anomaly ( Figure 8B). Cambrian Stage 4 REEY profiles show a negative Ce anomaly and a positive Y anomaly ( Figure 8C). When data are plotted as Pr versus Ce, Ce anom , different localities show distinctive transitions (Figure 9). Using Ce n /(Pr n 2 /Nd n ), the Ce anom in Cambrian Stage 2 data are the most variable, ranging from 0.57 to 1.38 (mean value 0.91, n = 18), with a strongly negative Ce anom and slightly positive Ce anom , and Y/Ho values range from 2.7 to 41.3 (mean value 22.5, n = 18). Samples from Stage 3 show Ce anom that range from 0.49 to 0.69 (mean value 0.56, n = 10), with a strong negative Ce anom . This indicates the removal of Ce and an increase in the sea water oxygen content. The Y/Ho values range from 10.9 to 61.7 (mean value 32.1, n = 10). The Ce anom values from Stage 4 are slightly higher, ranging from 0.61 to 0.84 (mean value 0.72, n = 23). The Y/Ho values range  .

A B
from 36.5 to 56.5 (mean value 43.4, n = 23). The Ce anom therefore decreases slightly from Stages 2 to 3, then increases in Stage 4. These results are consistent with the outcome of the first calculation.

| Evolution of carbonate mineralogy
During the terminal Ediacaran, the botryoidal form, blunt crystal terminations and high strontium in the early marine cements from the Nama Group, Namibia has been suggested to indicate original aragonite precipitates (Wood et al., 2018). In the samples used for this study, all early marine cements from Cambrian Stages 2 to 4 are inferred to have been either high-Mg calcite and/or aragonite.
A compilation of published inferred mineralogy of early marine cements and ooids from the Ediacaran to early Ordovician (Table S3, n = 98) reveals some trends ( Figure 10). Most marine cements and ooids now preserved as low-Mg calcite were precipitated mainly as high-Mg calcite and aragonite during the Ediacaran to early Cambrian Stage 2 (Tommotian, n = 27). Dolomite is present in the Dengying Formation of South China and the early Ust'Yudoma and Aim formations in Siberia during the terminal Ediacaran (n = 2, Hu et al., 2020;Wood et al., 2017), and inferred original calcitic precipitates are documented from South China and Oman (n = 2, Osburn et al., 2014;Zhao et al., 2020). The occurrences of low-Mg calcite precipitates increased during early Cambrian Stage 3 (n = 9, early Atdabanian), documented from South Australia, Siberia, South China and Morocco (Álvaro & Debrenne, 2010;Li et al., 2021;Tucker, 1989; F I G U R E 1 0 Distributions of inferred carbonate mineralogy from early marine cements and ooids during the terminal Ediacaran to early Ordovician, 580-480 Ma. A, Aragonite Sea; C, Calcite Sea; Dru., Drumian stage; Guz., Guzhangian stage; Pai., Paibian stage (see Table S3). Wood et al., 2017). This suggests the appearance of a 'calcite sea'. But other contemporary localities in Siberia show inferred high-Mg or aragonite cements, such as documented here. High-Mg calcite and aragonite dominance returns in late Cambrian Stage 3 (late Atdabanian), and continues until the late Cambrian Stage 4. During the middle Cambrian to the late Furongian (Stage 10), calcite, high-Mg calcite and aragonite co-precipitate together in the same succession, and existing data show no preference for any specific mineralogy (Powell, 2009;Riaz et al., 2021). Low-Mg calcite precipitates only are known in the early Ordovician, indicating the potential start of a 'calcite sea'. This transition is poorly constrained, however, due to limited documented data. In sum, a shift in dominant carbonate mineralogy is inferred from 'dolomite-aragonite' to 'aragonite/ high-Mg calcite seas' in the late Ediacaran to Cambrian Stage 2, a temporary shift to a 'calcite sea' during Cambrian early Stage 3, returning to 'aragonite sea' between late Cambrian Stage 3 and late Cambrian Stage 4, mixed 'aragonite-calcite seas' during the middle and upper Cambrian, and a 'calcite sea' by the early Ordovician. Balthasar and Cusack (2015) suggest that aragonite and calcite can co-precipitate in 'calcite seas' above 20°C. The climate may have changed from icehouse conditions in the Neoproterozoic to a dominant greenhouse condition in the Cambrian, suggesting that factors such as regional sea water temperature may explain the presence of both low-Mg and high-Mg calcite under intermediate sea water mMg/Ca values during the same interval. Indeed, other possible controls such as pCO 2 , or local differences in salinity, can also influence the precipitation of CaCO 3 polymorphs (Zhuravlev & Wood, 2008).

| Evolution of sea water mMg/Ca values
Numerous proxies have been developed to reconstruct changes of sea water mMg/Ca values through geological history, including geochemical models, fluid inclusions, echinoderm data and early marine carbonate precipitates. Changes of carbonate mineralogy in early marine cements and ooids may indicate variations of sea water mMg/Ca values but, as noted above, other local factors may override this.
In samples from Cambrian Stage 2, the fibrous inferred high-Mg cements are preserved with relatively high concentrations of Mg, ranging from 1.2 to 3.2 mol.% MgCO 3 . Stage 3 samples show a decrease in Mg concentrations to a mean of 1.1 mol.% MgCO 3 , and Stage 4 samples show an increased mean of 2.25 mol.% MgCO 3 . These values are much lower than modern high-Mg marine cements, potentially indicating varying degrees of Mg-loss during diagenesis. The calculated mMg/Ca values at 15 and 35°C gained here are 0.8-1.4 in Cambrian Stages 2 and 4, and 0.4-0.7 in Stage 3 ( Figure 6D). The original sea water mMg/Ca may, however, have had higher values, as high-Mg calcite cements lose Mg during transformation to low-Mg calcite during diagenesis. This may indicate more extensive Mg removal for Stage 3 samples from the Siberian Platform.
The calculated sea water mMg/Ca values during Cambrian Stages 2 and 4 based on marine cements in this study are generally also slightly higher than values from geochemical models (consistently <1; Hardie, 1996), slightly lower than fluid inclusion data (mMg/Ca <2; Lowenstein et al., 2001), but far lower than data based on echinoderm stereom (mean of 3 mMg/Ca; Dickson, 2002) ( Figure 11). Hardie's model predicts low sea water mMg/ Ca values so supporting a dominantly 'calcite sea' during the whole Cambrian period, but this is contradicted by the widespread precipitation of aragonite and high-Mg calcite early marine cements (Dickson, 2002;Figure 11). A similar contradicting scenario has been noted during the late Ordovician, where Laurentian Sanbian rocks principally host aragonitic ooids, during the supposed 'calcite sea' predicted by the geochemical model (James et al., 2020).
In addition to the diagenetic loss of Mg, and the operation of local factors such as temperature and salinity, other possible explanations for these common discrepancies between formulated models and carbonate mineralogical evidence could be the inaccuracy of modelled mMg/Ca values. The geochemical model is based on continental weathering and basalt-brine interaction driven by different ocean crust production rates, with inferred rapid increases in the rate of ocean crust production leading to transitions from an 'aragonite-dolomite' to 'calcite sea'. Ocean crust production during the Cambrian, however, is poorly constrained (Petach, 2015).

| Evolution of Ce anom to track redox
The petrographic features of early cements documented here, as well as precipitation before or with geopetal sediments, support their early marine origins (Davies, 1977;Hood et al., 2011). Therefore, it is proposed that the chemical compositions of these well-preserved marine cement generations might provide robust sea water REEY chemistry profiles.
The PAAS normalised REEY profiles during Cambrian Stage 2 have a variable range of Ce anom , values between 0.57 and 1.38. The co-occurrences of negative and positive Ce anom values capture a highly dynamic sea water redox state on the Siberia platform during Cambrian Stage 2.
While most Ce anom values are below 1.0, reflecting sea water oxygenation, some early marine cements also develop positive Ce anom values indicating the interaction of oxic and anoxic sea waters. The Y/Ho values may decrease in anoxic waters when experiencing scavenging by iron oxides (Wallace et al., 2017), so the Y/Ho values in Cambrian Stage 2 compared to those shown during Cambrian Stages 3 and 4, also suggest episodic anoxia.
A positive Eu anomaly also appears in some analysed marine cements from Siberia during Cambrian Stage 2. Europium occurs with divalent or trivalent states, and Eu 2+ preferentially partitions in the dissolved phase due to its large ionic radius and different charge (Bau, 1991;Bau et al., 2010;Sverjensky, 1984). Positive Eu anomalies normally occur in marine hydrothermal systems, where under such reducing environments Eu always occurs as divalent values (Bau, 1991;Sverjensky, 1984). In oxic oceans, high-temperature fluid causes feldspar to release Eu 2+ , which is then easily oxidised to Eu 3+ and preferentially scavenged by Mn-Fe oxides, resulting in no Eu anomaly (Bau, 1991). Conversely, a positive Eu anomaly in Archean microbialite is interpreted to reflect anoxic deep ocean waters (Kamber & Webb, 2001). Hence, some notable Eu anomalies in Cambrian Stage 2 might be explained by local anoxia in the Siberia platform, as is consistent with Ce anomaly profiles. The PAAS normalised REEY profiles during Cambrian Stages 3 and 4 show similar patterns to modern oxic sea water, and both have negative Ce anom values. The data, however, also show an increasing trend from an average 0.58 in Stage 3 to an average 0.72 in Stage 4, suggesting a decreasing oxygen concentration in sea water during this interval.
The Ce anom data have been combined here with published data from different components from Ediacaran to early Cambrian successions (Cui et al., 2016;Frank et  F I G U R E 1 1 Inferred variation in sea water Mg/Ca molar ratio from Ediacaran (550 Ma) to Ordovician (480 Ma). Dru., Drumian stage; Guz., Guzhangian stage; Pai., Paibian stage (Erhardt et al., 2020;Ries, 2010). Ward et al., 2019;Wei et al., 2018;Yang et al., 2022) in order to compare how oxygenation in distinct basins developed and to determine whether oxygenation events correlate globally (Figure 12).
Data show broadly consistent values and follow the same trends globally. In the Yangtze block, South China (Dengying, Doushantuo, Zhujiaqiang, Yanjiahe and Baiyanshao formations), the Ce anom data show a decreasing trend from a mean of 0.84 at ca 557 Ma to the strongly negative mean of 0.40 by the start of the Cambrian, ca 540 Ma. Data from Australia (Wonoka and Julie formations) ca 560-550 Ma and Siberia (Khatyspyt Formation) ca 547 Ma are slightly elevated but are broadly consistent with this trend, suggesting increasing oxygenation over this period. From 538 to ca 520 Ma, however, Ce anom values rise from 0.36 to 1.38 in samples from both South China (Yanjiahe, Liuchapo, Niutitang and Jiumenchong formations) and in the Siberian data presented here, suggesting the recurrence of anoxic sea water conditions. During late Cambrian Stage 2, through Stage 3 to early Stage 4, Ce anom values from Australia (Todd River Formation and Harkless Formation) and South China (Wulongqin Formation), and the Siberian data presented here, show a decrease again to 0.36 by ca 519 Ma, an increase to 1.08 by ca 518 Ma, then a decrease to 0.42 by ca 513 Ma. This may suggest a short oxygenation event in early Cambrian Stage 3, a return to anoxia, then back to oxic conditions during early Cambrian Stage 4. After this time, Ce anom values from Australia (Shannon, Giles and Wilkawillina formations), and the Labrador data presented here, increase to nearly 1.0 by the Drumian, suggesting a further interval of expanding anoxia.
Although more data are needed to confirm these trends, this compilation of Ce anom data is suggestive of three global phases of oxic expansion. First, a longlived phase of progressive oxygenation during the late Ediacaran to Fortunian (ca 550 to 540 Ma; average Ce anom values from 0.99 to 0.41), and two shorter phases during F I G U R E 1 2 Complied Ce anomaly data from carbonates (Cui et al., 2016;Frank et al., 2019;Li et al., 2015;Ling et al., 2013;Wallace et al., 2017;Ward et al., 2019;Wei et al., 2018;Yang et al., 2022) (Wei et al., 2018), and an inferred anoxic event, the Sinsk Event, ca 513 Ma (Zhuravlev & Wood, 1996).
There remains considerable debate as to the drivers and dynamics of these oxygenation events. Data from the early Cambrian Siberian record are characterised by multiple, transient (0.5-2 Myr) coupled carbonate δ 13 C and carbonate-associated sulphate δ 34 S cycles, each inferred to be an OOE (He et al., 2019). Biogeochemical modelling of these cycles suggests that each positive carbon isotope excursion represents a pulse of oxygenation within highly productive anoxic oceans with relatively low sulphate concentrations. Additional positive feedbacks between ocean redox and phosphorus retention in sediments may have driven rapid bottom-water deoxygenation, leading to the reestablishment of anoxia, potentially creating the repetitive oxygenation-deoxygenation cycles. The Sinsk event coincides with a decoupling of the C-S isotope records, suggesting a further shrinking of the marine sulphate reservoir, as well as expanded shallow marine anoxia (He et al., 2019).
Pulses of increased oxygenation (and potentially related changes in productivity) during the Cambrian Radiation are hypothesised to have temporarily increased the extent of shallow-ocean oxygenation and therefore habitable shelves and hence to have promoted diversification of metazoans (He et al., 2019). Indeed, dynamic and synchronous increases in biodiversity, rates of origination and body size in archaeocyath sponges, hyoliths and helcionelloid molluscs through the early Cambrian on the Siberian Platform have been quantified, with these trends proposed to follow OOEs (Zhuravlev & Wood, 2020). By contrast, the inferred anoxia event ca 513 Ma, the Sinsk Event, caused a mass extinction which removed over 50% of species and some major groups of the Cambrian fauna entirely (Zhuravlev & Wood, 1996).

| CONCLUSIONS
Integrating the samples analysed in this study with the compiled literature, a shift in dominant carbonate mineralogy is inferred from 'dolomite-aragonite' to 'aragonite seas' in the late Ediacaran. This was followed by a temporary shift to a 'calcite sea' during Cambrian early Stage 3, returning to an 'aragonite sea' during the late Stage 3 and late Stage 4. Subsequently there was a gradual transition from mixed 'aragonite-calcite seas' during the middle to upper Cambrian with a 'calcite sea' reached by the early Ordovician.
The calculated sea water mMg/Ca values from analysed cements during Cambrian Stages 2 to 4 are mostly ca 1.0. Some geochemical models (Hardie, 1996) predict low sea water mMg/Ca values supporting a dominantly 'calcite sea' during the Cambrian, but this is contradicted by the widespread precipitation of aragonite and high-Mg calcite early marine cements. Higher sea water mMg/Ca values may have been present during the Cambrian but local factors such as temperature and salinity are also important in controlling carbonate mineralogy.
The Ce anom data gained in this study combined with published Ce anom values suggest potentially three phases of global oxic expansion. First, a long-lived phase of progressive oxygenation during the late Ediacaran to Fortunian (ca 550 to 540 Ma; average Ce anom values from 0.99 to 0.41), and two shorter phases during early Cambrian Stage 3 (ca 519 Ma; average Ce anom values from 0.91 to 0.40) and Early Cambrian Stage 4 (ca 512 Ma; average Ce anom values from 1.02 to 0.49), bounded by intervals of more dominant anoxia. These phases of dynamic oxia and anoxia are consistent with phases of increased rates of origination and extinction, respectively, of Cambrian benthic metazoans.
In conclusion, early marine cements offer a valuable and potentially high-resolution temporal and spatial archive of shallow marine redox and sea water chemistry through this critical transition in Earth's evolution.