Authigenic clay mineral evidence for restricted, evaporitic conditions during the emergence of the Ediacaran Doushantuo Biota

The Ediacaran Period witnessed major environmental change and an expansion of eukaryotic life following the Marinoan glaciation. The Doushantuo Formation of south China has furnished much of our understanding of Earth System and biosphere evolution during this period, but its depositional setting–marine or non-marine–is still debated. Here we conduct mineralogical, petrographic and geochemical analyses of samples from the lower Doushantuo Formation in order to determine the origin of saponite clay minerals which are abundant in this interval. We find that the morphology and distribution of the saponite clays suggest they are of pre-compaction, authigenic origin. We infer that the lower Doushantuo sediments were deposited in a restricted, mildly evaporitic lagoonal basin, offering additional constraints on the environmental setting into which metazoan life emerged. Further, our findings confirm that Ediacaran seawater favored reverse weathering, with marine clay formation likely representing an important sink for various elements during the Precambrian. The Ediacaran-age Doushantuo Formation, which hosts some of the oldest animal fossils, was likely deposited in a restricted, mildly evaporitic lagoon environment, according to petrographic analyses of saponite clays that suggest an authigenic origin.

T he Doushantuo Formation of South China preserves a unique sedimentary record of major paleo-environmental change in the aftermath of the Marinoan glaciation 1 as well as the subsequent expansion of eukaryotic life. The lower Ediacaran Doushantuo Biota, which have been primarily recognised in shallow-water settings, include multicellular algae 2 and a diverse suite of acanthomorphic acritarchs 3 including embryolike fossils which are candidates for the earliest animal life on Earth 2,4 . While the metazoan affinity of the embryo-like fossils is contested 5 , it is nevertheless widely accepted that the Doushantuo Formation provides an important record of the environmental setting and nature of eukaryotic expansion during the interval in time that metazoans first diversified 6 . In this context, the disputed marine versus non-marine nature of the shallow-water, fossilbearing deposits of the basal Doushantuo Formation 7,8 represents an important knowledge gap in our understanding of the environmental controls on the emergence of early life.
Current stratigraphic and paleo-geographic models of the Yangtze Block suggest that the Doushantuo Formation was deposited on a passive continental margin spanning tidal flat--shelf lagoon settings in the North West (Yichang City) to basinal settings in the South East (northern Guangxi Province, Fig. 1) 9 . An expanded sedimentary record amenable to geochemical characterisation, together with an abundant and diverse fossil record, have made the shallow-water sequences of the Yangtze Gorges Area (YGA) in particular a key locality for geobiological and paleo-environmental studies and the epicenter of the debate over the environmental setting into which the Doushantuo Biota emerged. The Doushantuo Formation in the YGA is commonly divided into four lithostratigraphic members 9,10 : the 1 st member consists of cap carbonates that sharply overly the glacial diamictite of the Nantuo Formation; the 2 nd member comprises interbedded shale and dolostone with chert and phosphatic nodules, with abundant saponite throughout; the 3 rd member is made up of dolostone and bedded chert in the lower part and ribbon rocks in the upper part; the 4 th member consists of black shale characterized by highly illitic mixedlayered illite/smectite and discrete illite, but no saponite 7 . This sedimentary sequence is overlain by the upper Ediacaran Dengying Formation 9 .
Clay minerals are persuasive paleo-environmental indicators, provided that the origin and mode of formation of the clay can be unambiguously identified 11 . The presence of saponite, a Mg-rich, trioctahedral 2:1 clay mineral of the smectite group, in the 2 nd member of the Doushantuo Formation has been central to the debate over the sedimentary setting of the YGA 7,8 . Saponite is known to form by various mechanisms, including (1) hydrothermal alteration of (ultra)mafic igneous rocks, volcanic ash or siliceous dolostone 12-15 ; (2) transformation of clay minerals (e.g. nontronite-saponite series transformation under hydrothermal conditions 16 , or talc-saponite in submarine hydrothermal settings 17 ); (3) pedogenic weathering of (ultra)mafic silicates under humid conditions 18 ; (4) precipitation from gel precursors under alkaline conditions 19 or (5) direct precipitation from supersaturated solutions (including pore fluids) 20,21 . Crucially, although saponite can potentially form in both marine and nonmarine settings, abundant syn-depositional saponite precipitation in open marine or shelf settings has not been documented. Local accumulations of detrital pedogenic saponites are, in principle, possible, but in the Phanerozoic, a marine origin is typically restricted to the alteration of volcanic ash layers under reducing conditions or to hydrothermal systems 15 . In contrast, saponite formation in alkaline or evaporitic lakes is common and well documented 11,19,22,23 . We note, however, that reverse weathering may have been a more important feature of the Si-rich Proterozoic oceans 24 , meaning that marine clay authigenesis (including Fig. 1 Geological map of the Yangtze Gorges Area. a Location of the Yangtze Block (b) Location and generalized paleogeographic reconstruction of the Yangtze platform during Doushantuo deposition (adapted from ref. 9 ) and locations of the investigated Sixi Section and Qinglinkou Section (green rectangle).
saponite) was potentially more widespread in the Precambrian. This latter possibility remains largely untested because of the difficulty of distinguishing and quantifying clay minerals of authigenic versus detrital origin.
The spatial distribution of saponite offers clues as to its origin. The 2 nd member in the YGA contains abundant saponite, whereas equivalent sequences in open marine slope to basin settings do not 7 . Initially, this was interpreted as evidence that the lower Doushantuo Formation in the YGA was deposited in an alkaline lake setting disconnected from the ocean 7 . Subsequent work, in contrast, inferred an open marine setting for the YGA arguing that the saponite was likely of detrital terrestrial weathering origin and thus not indicative of the depositional environment 8 . However, this study did not identify or characterise the clay mineral assemblage, relying instead on indirect constraints including bulk geochemical data for the 2 nd member consistent with a mafic-ultramafic provenance 8 . The presence and origin of saponite in these sections remains unresolved despite major implications for determining the marine versus nonmarine origin of the Doushantuo Biota, informing the interpretation of isotope proxy records [25][26][27] , and testing recent models invoking widespread marine clay authigenesis in the Precambrian Ocean 24 .
A fundamental limitation of previous studies is their reliance on bulk mineralogical and indirect geochemical techniques, which are unable to differentiate saponite of detrital origin from saponite of authigenic origin. Recent technical and methodological advances in scanning electron microscope -energy dispersive spectroscopy (SEM-EDS) based petrography now allow the differentiation and quantification of clays of distinct genetic origins 28,29 . Here we couple this petrographic approach with X-ray diffraction (XRD) techniques and electron probe microanalysis (EPMA) to conclusively identify saponite mode of formation in 19 samples collected from the Doushantuo 2 nd Member in two representative sections of the YGA (Fig. 2). This provides previously unobtainable constraints on the depositional setting associated with the emergence of the Doushantuo Biota, and demonstrates a robust approach with which to identify authigenic clays and test the prevalence of reverse weathering in the Precambrian 24 .

Results and discussion
Quantitative SEM-EDS mineral mapping shows that the dolomitic shales and dolostones in the 2 nd member of the Doushantuo Formation mainly consist of dolomite (16.9-75.3 wt%), quartz (3.8-40.6 wt%), feldspar (0-33.1 wt%), and clay minerals (4.3-34.4 wt%), excluding an unusual illite-dominated sample (Qlk3) and a heavily altered sample (Qlk4) close to the cap carbonate in the Qinglinkou section (Table 1, Fig. 2). Mineral mapping identifies saponite, chlorite and muscovite as the most common phyllosilicates (Table 1). Consistent with this, XRD analysis of Ca-saturated, oriented clay preparations identifies a dominance of mixed-layered chlorite/smectite with (001) peak located between 1.40 nm and 1.48 nm (air-dried), expanding to between 1.52 and 1.66 nm after ethylene glycol solvation and then collapsing to~1.40 nm and~1.00 nm on heating to 400°C (Fig. 3a) and minor illite/mica ((001) reflection at 1.0 nm). We attribute variable (001) peak positions to varying proportions of smectite in the irregular mixed-layered chlorite/smectite (R0) of different samples. The (002) peaks (0.76-0.83 nm) in ethylene glycol solvated samples suggest a smectite proportion between 40% and 80% 30 . The smectite clay fraction is confirmed as saponite by powder XRD analysis of the <2 µm fraction showing the (060) peak located between 0.154 nm and 0.153 nm (Fig. 3b), demonstrating the trioctahedral nature of the clay 31  Saponite dominates the clay fraction (70.0 wt%, on average) (Fig. 4a), followed by chlorite (25.7 wt% on average), whereas muscovite contributes only 4.3 wt% on average. Saponite is concentrated in bedding-parallel laminae up to 120 µm in length and 5-10 µm thick (Fig. 4b) or in intergranular pore-fill domains (Fig. 4d, f). It also occurs, more rarely, as intragranular pore-fill in feldspar (Fig. 4g). Saponite is commonly partially altered to and intergrown with chlorite, but retains the morphology of surrounding unaltered saponite (Fig. 4b, d). Saponite domains contain negligible amounts of detrital material and exhibit  compaction deformation around silt-size quartz and dolomite grains (Fig. 4b, d, f). Individual saponite particles or flakes within laminae are oriented approximately parallel to bedding (Fig. 4b) whereas the saponite flakes in pore-fill domains appear more randomly oriented, consistent with variable compaction-induced orientation ( Fig. 4d, g). We distinguish two distinct chlorite types in our samples: (1) short, platy flakes <5 µm in length including some with finegrained texture similar to saponite (Fig. 5a-c) and (2) larger flakes »5 µm in length with discrete boundary (Fig. 5d, e). Short-flake chlorite is present mainly in intergranular pore-fill domains (Fig. 5a, b), commonly in association with saponite and, more rarely, associated with partially altered feldspar (Fig. 5c) or intergrown with pyrite (Fig. 5b). Some individual short-flake chlorite particles are thicker than saponite flakes but, just like saponite, show varying degrees of orientation depending on silt bridging 32 (Fig. 5a). In addition, short-flake chlorite domains are similar to saponite pore-fill domains, and saponite-chlorite intergrowths are common (Figs. 4d, f; 5a-c). Large chlorite flakes, by contrast, are discrete grains 7-60 µm in length that are oriented parallel to bedding, identical in shape and size to muscovite grains. Individual flakes commonly contain pyrite inclusions and, more rarely, show regions that are compositionally more similar to muscovite (Fig. 5d, e), suggesting chloritization of muscovite. Consistent with this, muscovite occurs usually in less altered samples (Fig. 5f). Overall, large-flake chlorite displays higher FeO and TiO 2 content compared to short-flake chlorite ( Supplementary Fig. 2).   (Table 1). In addition to clastic quartz, we also observe micro-crystalline quartz that is concentrated in laminae, often in association with apatite, saponite, or short-flake chlorite (Fig. 6a), as well as chert nodules~5 mm in size. Apatite content ranges from 0.2 to 13.8 wt% (average 3.6 wt%) and is concentrated in elongate, phosphatic nodules oriented parallel to bedding (Fig. 6b) or in laminae with micro-quartz (Fig. 6a). Both phosphatic nodules and laminae contain other minerals, including saponite, chlorite, quartz, feldspar and dolomite. In addition, phosphatic fossils (Fig. 6c) were observed in several samples (SX4, SX12 and Qlk6), where they most commonly occur  as dolomite or authigenic feldspar cemented clusters. These first occur~22 m above the lower boundary of the 2nd member, and are identified as Archaeophycus yunnanensis 33 . Framboidal pyrite is common in all samples, with minor euhedral pyrite and micropyrite inclusions in chlorite or muscovite flakes also occurring ( Fig. 6a, b). The total pyrite content ranges from 0.8 to 6.6 wt%, averaging 2.5 wt%.
Origins of clay minerals. Our petrographic evidence identifies a pre-compaction origin for saponite in the Doushantuo Formation (Supplementary Table 1). Firstly, saponite in most samples is concentrated in domains exceeding 10 μm in size (Fig. 4a). Primary (i.e. intergranular) pores of this size rarely survive compaction and burial in mudstones 34 , so that the saponite fill must predate burial-induced compaction. Consistent with this, compaction has impacted on both saponite laminae and intergranular pore-fill domains. Individual saponite particles in laminae exhibit bedding parallel orientation, consistent with compaction, whereas pore-fill domains are partially sheltered so that individual particles are randomly orientated except where pore-fill is deformed around resistant grains 32 . Clays of hydrothermal origin, by contrast, post-date compaction and are therefore expected to be randomly or radially oriented regardless of domain type (laminae vs intergranular pore) or proximity to resistant grains. We therefore conclude that the above properties demonstrate a precompaction origin for saponite, ruling out burial diagenesis or hydrothermal alteration origins.
Both detrital and authigenic clays are of pre-compaction origin, but can be distinguished petrographically (Supplementary Table 1). Detrital saponite is sourced from pedogenic weathering of (ultra)mafic silicates 18 and undergoes erosion and transportation together with other weathering products prior to deposition. Detrital saponite would therefore be expected to settle together with other detrital weathering products. However, not only is saponite concentrated in domains that are largely devoid of any detrital material, the fragile nature of the individual elongate, sheet-like saponite particles means that these are unlikely to survive transport without modification, further arguing against a detrital origin. In addition, since the sheet-like structure of clay particles typically results in preferential orientation after settling through the water column 35 , the random orientation of individual pore-fill saponite particles also argues against a detrital origin. Saponite is commonly associated with and deformed around silt-size dolomite (dolomicrite; Fig. 4a). Previous work has shown that dolomicrite in the 2 nd member of the Doushantuo Formation is of syndepositional, earliest diagenetic origin 36 . Deformation of saponite around dolomite grains could therefore be partially due to early fabric displacing dolomite growth, but is likely mainly due to burial compaction. Together, the above observations strongly argue for an authigenic origin for the Doushantuo saponite. This conclusion is further supported by the striking similarity between the saponite morphology and distribution in the Doushantuo Formation and the demonstrably authigenic saponite domains observed in representative samples of the Green River Formation (Fig. 4c, e; 22 ).
We infer that the authigenic saponite laminae formed at the sediment-water interface or, alternatively, as early diagenetic replacements of detrital clay or gel precursors 19 , whereas the intergranular pore-filling authigenic saponite precipitated from pore water shortly after sediment deposition. The intragranular saponite in feldspar is authigenic pore-fill associated with early diagenetic feldspar dissolution. A late stage hydrothermal alteration origin 12 for this feldspar-associated saponite can be ruled out because a) the samples with the greatest abundance of feldspar-associated saponite do not show any evidence for hydrothermal alteration (i.e. extensive chloritization) and b) EPMA results ( Supplementary Fig. 2, Supplementary Data 2) reveal that these are compositionally indistinguishable from laminae or intergranular-pore hosted authigenic saponite. In conclusion, the Doushantuo saponite was formed syndepositionally or during earliest diagenesis, with the required chemical constituents sourced from bottom water and from early diagenetic dissolution of detrital silicates (see extended discussion in next section).
What then is the significance of the saponite-chlorite intergrowths? Short-flake chlorite is compositionally and texturally similar to authigenic saponite (Figs. 4d, f, g; 5a-c). In addition, mineral mapping shows that saponite-chlorite intergrowths are widespread (Fig. 4b, d) and XRD analysis identifies the ubiquitous presence of mixed-layer chlorite-smectite (Fig. 3a). Previous work documents a regional hydrothermal event centered on the underlying Nantuo Formation and cap dolostones 37 , and this has been shown to have caused alteration in the 2 nd member of the Doushantuo Formation, showing increased chloritization downward toward the cap carbonate 38 . Consistent with this, we interpret short-flake chlorite and chlorite-saponite mixed layer as hydrothermal alteration products of authigenic saponite 38,39 , and infer that chlorite-enriched samples are relatively more hydrothermally altered than samples with lower proportions of chlorite. Long-flake chlorite, by contrast, is most likely a pseudomorphic alteration product of muscovite 28 , as supported by their particle morphology and the occasional presence of residual muscovite (Fig. 5d). The sample with largest proportion of muscovite (SX22, Table 1) contains no chlorite further supporting an alteration origin for long-flake chlorite. Strikingly this sample (SX22) contains only~2 wt% muscovite, underscoring the scarcity of detrital phyllosilicates in the 2 nd member of the Doushantuo Formation.
In summary, we are able to distinguish the origins of Doushantuo clays based on their distinct petrographic and geochemical characteristics. We demonstrate that clay minerals of both detrital (muscovite) and authigenic (saponite) origins were present prior to compaction, and both were affected by partial post-compaction alteration to chlorite or chlorite-saponite mixed layer ( Supplementary Fig. 3). The authigenic saponite formed syn-depositionally or during earliest diagenesis, in response to favourable basinal bottom water or pore water conditions and represents the dominant phyllosilicate constituent in the 2 nd member.

Implications for Paleo-Environmental Setting and Reverse
Weathering. Two fundamentally different paleo-environmental settings have been proposed to explain the spatially restricted abundance of saponite in the 2 nd member of the Doushantuo Formation: (1) a proximal marine setting which received an abundance of weathered (ultra)mafic detritus including pedogenic saponite 8 or (2) an alkaline (pH > 9) lake setting which favored the precipitation of authigenic saponite 7 . Our results, when considered together with published constraints, argue against both of these scenarios.
Firstly, while the detrital fraction of the 2 nd member may well have a mixed igneous to (ultra)mafic provenance, detrital saponite is essentially absent, with the detrital constituents consisting mainly of quartz and feldspar rather than phyllosilicates. Meanwhile, authigenic saponite makes up~70 wt% of the clay content throughout the 2 nd member in the YGA. Since the vast majority of Phanerozoic sedimentary saponite deposits are associated with evaporitic, alkaline lake settings 40,41 this might seem to support a similar lacustrine setting for the saponite in the YGA. However, published carbonate-hosted Sr isotope values (0.7078-0.7082, radiogenic values outside this range are restricted to a narrow interval in the uppermost 2 nd member) 42 are centered around the expected lower Ediacaran seawater value of 0.708 43 and REEs measured in chert also demonstrate a seawater-like pattern for the 2 nd member 44 . These observations argue against an alkaline lake setting. Moreover, the abundant presence of pyrite (up to 6.6 wt%, average 2.5 wt%) as well as the ubiquitous co-occurrence and common intergrowth of authigenic saponite, pyrite and apatite (Fig. 6a, b) also argue against such a lacustrine setting. Abundant pyrite formation usually implies a marine sulfate source and, since the solubility of apatite increases greatly with the incorporation of carbonate anions 45 , the predominance of the carbonate ion under highly alkaline conditions is expected to inhibit carbonate fluorapatite formation 8 . Individually, any of these observations may sporadically occur in specific lake settings (e.g. pyrite in volcanically fed lakes 46 , minor apatite deposition during early diagenesis in alkaline lakes 47 ), however the persistently marine character of each of these lines of evidence throughout the 2 nd member collectively argues against a lake setting, even a lake with intermittent marine connection 7 .
Given that an isolated lake setting is incompatible with the available geochemical and mineralogical evidence, could Ediacaran seawater composition have permitted the formation of saponite in the YGA? To guide our discussion, we consider the ingredients required for saponite formation via a reverse weathering reaction, with YGA saponite composition constrained by our EPMA data (Supplementary Data 2): 0:19K þ þ 0:07Ca 2þ þ 2:63Mg 2þ þ 0:20Fe 2þ þ 3:39H 4 48 . Reductive dissolution of ferric oxyhydroxides coupled to microbial organic carbon remineralisation is a likely source of Fe 2+ in the pore fluid 49 , consistent also with the presence of abundant pyrite. The ubiquitous early diagenetic saponite infilling of feldspar pores supports the idea that dissolution of feldspar and, potentially, detrital clays provided the Al 3+ ions required to form saponite (Fig. 4g).
Thermodynamic considerations indicate that the activity of the dissolved H 4 SiO 4 and Mg 2+ species as well as fluid pH are key for forming authigenic Mg-silicates 48 , including saponite. In the absence of a large biological sink for silica, early Ediacaran seawater was likely characterized by elevated H 4 SiO 4 compared to the present 50 and therefore would have been more favorable to clay authigenesis (i.e., reverse weathering; 24 ). Elevated seawater H 4 SiO 4 is supported by the widespread occurrence of silica cements (Fig. 6a) and chert nodules in the Doushantuo Formation 51 . Similarly, Neoproterozoic seawater was likely characterized by greater Mg 2+ concentrations compared to Cambrian seawater, with marine carbonate cement evidence for high Mg/Ca conditions and widespread, potentially primary, dolomite 52 . Consistent with this, thermodynamic calculations indicate that saponite was most likely supersaturated in Ediacaran seawater (Fig. 7). However, the scarcity of sedimentary saponite in Neoproterozoic marine sequences (we are aware of only one report outside of the YGA 53 ), and the absence of saponite in time-equivalent lower Doushantuo sediments deposited in unambiguously marine slope to basin settings 7 shows that the abundance of authigenic saponite we document in the YGA cannot simply be a product of favorable Precambrian seawater composition. We therefore conclude that the spatially restricted, regional abundance of saponite argues against a typical open marine setting for the YGA. This interpretation is further supported by geochemical evidence documenting major differences between the YGA and coeval slope to basin (open marine) settings, including the relatively low abundance of water-column sourced redox-sensitive trace elements 7 as well as the distinct stratigraphic trends of δ 15 N, δ 34 S and δ 13 C carb in the YGA 27,54,55 .
Given the above, how do we explain the abundance of saponite in the YGA? We attribute the restricted distribution of saponite to kinetic inhibition of Mg-silicate precipitation 53 , possibly because Ediacaran seawater pH was lower (less alkaline) than is favorable for saponite precipitation 56,57 . The precipitation of saponite from lower pH but highly Si and Mg enriched fluids (e.g. lowtemperature off-axis hydrothermal settings 15 ) suggests that a further increase in a[H 4 SiO 4 ] and a[Mg 2+ ], for example via evaporative concentration beyond a 'critical supersaturation', can help to overcome a kinetic barrier 58,59 (Fig. 7). That is, conditions only slightly exceeding the saponite stability curve may not be enough to overcome additional constraints (e.g. kinetic controls). The real-world precipitation of saponite likely requires a greater degree of supersaturation (larger distance from saponite stability curve) as would be the case for Ediacaran seawater because of its higher silica content than modern seawater (Fig. 7). To test this, Modern seawater (green) and Ediacaran seawater (blue). Talc stability boundary (turquoise), saponite stability boundary (orange), and sepiolite stability boundary (purple) are shown as a function of dissolved Mg, H, and Si activities. Filled circles represent starting conditions (0 evaporation), solid lines represent evaporation pathway with no mineral precipitation, and dashed lines represent evaporation pathways considering calcite precipitation (each hash along the two lines represents a 10% step in evaporation). Two different pH starting conditions are modeled for Ediacaran seawater (7.0 and 7.8) to create the envelope of potential results. Orange gradient shading is parallel to saponite stability lines and its upper and lower boundary cross the starting point at 7.0 pH and 80% evaporation point with calcite precipitation of Ediacaran seawater pathway, respectively. The area with plus sign pattern shows the differences between modern and Ediacaran seawater after 80% evaporation, and represents potential range of the critical supersaturation. Temperature = 25°C, pCO 2 = 10 −3 bar.
we used reaction path modeling, constrained by existing estimates of Ediacaran seawater composition (0.5-1.0 mM silica 50,60 and pH of 7.0-7.8 57 ), which offers support for our hypothesis. This modeling demonstrates that seawater becomes increasingly supersaturated with respect to Mg-silicates including saponite, talc and sepiolite with progressive evaporative concentration (solid blue lines in Fig. 7), even if we allow for pH buffering through calcite precipitation (dashed blue lines in Fig. 7). We attribute the absence of talc in the YGA samples to elevated pore water Al (sourced from the dissolution of Al-rich debris such as feldspar and clays), which would inhibit precipitation of nonaluminous talc and other hydrous Mg-silicates, favoring saponite formation 48,59 . Moreover, modelled pH remains below 8.5 ( Supplementary Fig. 4), a critical kinetic threshold for talc precipitation 53 , further inhibiting talc formation. In the presence of a dissolved Al pool, both the tetrahedral and octahedral sheets of a newly forming clay mineral incorporate structural Al, producing charge imbalances in these layers. These imbalances drive the formation of an interlayer sheet, meaning that smectitetype clays preferentially form 11 . Such smectites preferentially sorb K + from the pore water 58 , which expresses in the formation of K-intercalated clays. This is consistent with K-saponite in our samples ( Supplementary Fig. 2). Early diagenetic microbial sulfate reduction and organic matter decay (SO 4 2− + 2CH 2 O → H 2 S + 2CO 2 + 2OH − ) would have also contributed to saponite formation, further increasing pore water pH, whereas sulfide production favored pyrite formation before exhaustion of marine sulfate 51 allowed accumulation of dissolved Fe 2+ , 10 . The dominance of bivalent (Mg 2+ , Fe 2+ ) over trivalent (Al 3+ , Fe 3+ ) ions in the pore water resulted in the formation of a trioctahedral clay, and is consistent with the K-saponite containing some Fe 2+ which occasionally intergrow with pyrite (Fig. 5b). Dioctahedral clays arguably did not form because of Mg 2+ /Fe 2+ » Al 3+ , preventing the formation of montmorillonite 48 , and Fe 2+ » Fe 3+ , preventing the formation of nontronite 17 .
Existing paleo-geographic models suggest that an expansive, carbonate-rimmed intrashelf basin or lagoon covered a substantial portion of the Yangtze Block 9 . We hypothesize that a restricted, modestly evaporitic (degree of evaporation <90% because of absence of evaporite phases such as gypsum in the Doushantuo Formation 61 ) but marine-influenced shelf lagoon setting would have provided the highly supersaturated conditions required for the precipitation of saponite (Fig. 7). This is supported by paleo-magnetic data which place South China at ∼30°N from 635 Ma to 560 Ma 62 , a mid-latitude position (high evaporation, low rainfall) which would have contributed to the evaporative concentration of already Si-and Mg-rich Ediacaran seawater 50,63 , as well as a modest increase in pH 64 .
What are the implications of our findings in the context of recent work arguing that reverse weathering was much expanded in the Precambrian oceans? As noted above, the apparent absence of saponite in slope to basinal equivalents of the 2 nd member of the Doushantuo demonstrates that Ediacaran seawater was insufficiently enriched in key ingredients to form saponite in open marine settings. However, reaction path modelling suggests that the striking,~40 Myr dominance of authigenic saponite in the lower Doushantuo of the YGA required not only specific local environmental conditions facilitating evaporative concentration of seawater, but relied also on seawater composition favoring marine-influenced authigenic clay formation. We suggest that the enriched nature of Precambrian seawater allowed saponite to reach critically supersaturated conditions with only modest evaporation. However, the critical threshold for saponite precipitation cannot be reached by evaporation of Si-depleted Phanerozoic seawater, consistent with the absence of saponite in Phanerozoic marine evaporites. This is supported by reaction path modelling which shows that Ediacaran seawater is more saturated with respect to saponite than modern seawater for all but the lowest estimates for Ediacaran seawater pH (Fig. 7). We stress that this is a conservative estimate, as our model assumes a conservative Ediacaran seawater composition (0.5 mM H 4 SiO 4 , modern-like Mg). Our results therefore provide indirect support for the Precambrian reverse weathering hypothesis 24 , implying that authigenesis of low-Mg but Fe-and K-rich marine clays occurred more widely outside of evaporitic settings, with implications for the global carbon cycle, marine elemental cycles and planetary climate 24,29,65,66 . While systematic additional work is required to address the full extent of these implications, our study demonstrates the tools and an approach with which to robustly test for the predicted widespread abundance of authigenic clays.
Impact of restricted lagoon environment on Doushantuo Biota. How might a restricted and evaporitic lagoonal setting have influenced the Doushantuo biota? The 2 nd member of the Doushantuo Formation hosts various phosphatically or siliceously preserved multicellular acritarchs 3,67 , including embryolike fossils which are potential candidates for the earliest animal life on Earth 4 . Strikingly, fossil-bearing localities in the lower Doushantuo are largely restricted to shelf lagoon and shelf margin settings, and rare in the deeper slope to basin 9 (Fig. 1). This pattern has been attributed to challenging conditions including widespread anoxia in the Neoproterozoic Ocean 7,68 , but our results indicate an additional role for water chemistry. The lagoon setting is likely to have provided an unusually favorable environment for prokaryotic and eukaryotic organisms (including, potentially, metazoans) as it would have facilitated accumulation of nutrients such as phosphorous that were in scarce supply in the early Ediacaran Ocean 69 . Geochemical proxy-based evidence indicates that the Yangtze Lagoon received riverine inputs from weathering of a mafic to ultra-mafic hinterland 8 , lithologies associated with elevated P-and trace metal contents 70 . Limited exchange between the lagoon and the open ocean would have contributed to the retention and build-up of these terrestrialrunoff derived nutrients 71 . A local build-up of P and other nutrients derived from terrestrial runoff is also supported by paleo-geographic reconstructions showing that phosphatic intervals in the Doushantuo Formation are almost entirely restricted to shallow-water lagoonal environments 9,72 . These intervals are rare in the deeper slope and basin localities, regions which would have received abundant P if nutrient distribution was being dominantly controlled by upwelling 9 . An abundance of terrestrially derived P and evaporatively concentrated marine Si would also have contributed to the preservation of the Doushantuo fossils 7,73 . Finally, the evolution of these organisms in a restricted basin and subsequent adaptation to the lagoon's unique water chemistry may also explain the high degree of endemism of the lower Doushantuo biota 9,72 , similar to the ecological impact of an alkaline lake setting proposed by Bristow et al 7 . Their subsequent radiation and expansion into open marine environments would have been facilitated by the gradual rise in marine nutrients (including P; 69 ) as well as dissolved oxygen over the Ediacaran, which ultimately set the scene for the major environmental and biological changes that ushered in the Phanerozoic 74 .

Conclusion
In summary, the distinct petrographic and geochemical features we present here confirm the ubiquitous presence of abundant, syn-depositional authigenic saponite in the 2 nd member of Doushantuo Formation. The spatially restricted distribution of saponite and marine geochemical signatures in the YGA combined with paleo-geographic reconstructions imply a shelf lagoon setting at mid latitude. Modest evaporation within the lagoon would have concentrated key seawater constituents (Mg 2+ , H 4 SiO 4 and OH − ) in the lagoon beyond a critical supersaturation threshold, facilitating the formation of saponite. This lagoonal setting could have also allowed for the accumulation of key nutrients such as phosphorous from terrestrial runoff, providing a favourable environment for earliest metazoan life. Furthermore, the occurrence of authigenic saponite also supports the hypothesis that silica-rich Precambrian seawater was favourable for reverse weathering.

Materials and methods
Samples. Nineteen samples were selected from the shale interbeds of the 2 nd member at two different sites in the YGA (Fig. 2), namely, the Sixi Section (30°45'25.8" N, 110°55'51.0" E) and the Qinglinkou Section (30°48'0.2" N, 110°55'18.5" E). No sampling permissions were required. In addition, several samples of a classic alkaline, evaporitic lake sequence rich in authigenic saponite (Parachute Creek Member of the Eocene Green River Formation, USA) were studied for comparison (see ref. 22 for sample details). All samples were cut into blocks with a 1-2 cm 2 surface perpendicular to the bedding plane and prepared as polished mounts for subsequent scanning electron microscope imaging and mineral mapping.
Samples for powder X-ray diffraction (XRD) were crushed in a Rock Labs RM100 Disc Mill and micronized using a McCrone mill, with the addition of ethanol. Four samples with relatively high clay content based on the bulk XRD results were selected for further identification of the clay mineral suite. For the clay mineral separation, 5 g of crushed sample was reacted overnight with 1 M acetic acid-sodium acetate buffer in a 70°C water bath to remove the carbonate fraction before centrifugal separation of the clay size (<2 µm) fraction and Ca saturation 30 .
X-Ray Diffraction. Powder X-ray diffraction patterns spanning 3.5°to 95°2θ were collected using a Malvern Panalytical Aeris tabletop XRD instrument (Co radiation source with 40 kV generator voltage and 15 mA tube current, PIXcel3D-Medipix3 1×1 detector) with 0.022°step size. Oriented, air dried clay films were analysed from 2°to 40°2θ with the same equipment, followed by analysis after treatment with ethylene glycol and heating to 550°C. Rietveld-based quantification of PXRD patterns was conducted via the Panalytical Highscore Plus Software and its implemented pdf-4 database with an uncertainty of <3%.
Scanning electron microscope and mineral mapping. Backscatter electron (BSE) images and mineral maps were obtained with a FEI Teneo LoVac field emission scanning electron microscope (SEM) equipped with dual Bruker XFlash Series 6 energy dispersive X-ray spectroscopy (EDS) detectors. BSE images of the entire sample (100 nm pixel resolution) and EDS spectra (2 µm step size, 8 ms acquisition time) for mineral mapping of smaller regions of interest were obtained using the FEI Maps Mineralogy software, followed by classification of the individual EDS spectra using the FEI Nanomin software (FEI Nanomin/Maps Mineralogy 28,29 ). Unlike other SEM-based mineral mapping techniques (e.g. QEMSCAN), the Nanomin mineral classification system can deconvolve mixed X-ray spectra and assign up to three minerals per analysed spot. This is a critical requirement for the correct interpretation of the mixed phase X-ray spectra characteristic of heterogeneous fine-grained sediments, where the X-ray generating electron interaction volume is commonly larger than the grain size. Mineral concentrations (reported as wt%) in the mapped areas are determined by converting the mineral area% to wt% using published densities for all identified sample constituents, with accuracy and precision comparable to quantitative XRD analysis 28 . Comparison of quantitative mineralogy from SEM-EDS mineral mapping to constraints from quantitative XRD shows near-identical results, demonstrating the accuracy of SEM-EDS mineral identification ( Supplementary Fig. 1, Supplementary Data 1).
Electron probe micro analysis. Electron probe micro analysis was applied on two representative samples (SX2 and Qlk7) to validate the petrographic classification of clay types, using a Cameca SX-100 electron microprobe at 15kv, 20 nA and spots sizes of 2-5 μm (See supplementary Data 2 for the standard information).
Hydrochemical modelling. To better constrain the potential depositional setting, hydrochemical modelling was applied to test the thermodynamic feasibility of forming Mg-silicates (including saponite) from seawater-derived fluids and contribution of modest evaporation (0-80%, Fig. 7), both with and without carbonate precipitation. We also compare the evaporation pathway of modern seawater with inferred Ediacaran seawater. While Ediacaran seawater remains poorly constrained, studies have suggested that seawater at this time had a pH~7.8, and possibly as low as 7.0 57 , with silica content between 0.5 and 1.0 mM 50,60 . For this model, we ran Ediacaran seawater with a starting pH value of 7.0 and 7.8, and assume a conservative silica content of 0.5 mM and a magnesium content of 52 mM 63 . Our starting values for modern seawater are pH = 8.04, silica content of 0.035 mM 53 and Mg 2+ content of 55 mM 75 .
The chemical evolution of seawater with evaporation as well as relevant stability boundaries were calculated assuming thermodynamic equilibrium. Modeling was carried out in PHREEQC v3.7.1 76 . Mineral stabilities were evaluated using the carbfix.dat database 77 . The saponite stability boundary was evaluated for the Mg, K, and Na endmembers of Mg-saponite 78 . The pitzer.dat database with its Pitzer specific-ion-interaction aqueous model was used for seawater evaporation pathways due to its suitability for high salinity fluids. The pathways assume temperature of 25°C, pCO 2 of 10 −3 bar and thermodynamic equilibrium with calcite at all times.

Data availability
All relevant data of this study are included in the article and/or https://doi.org/10.6084/ m9.figshare.19919321 79 .