The role of volcanic-derived clays in the preservation of Ediacaran biota from the Itajaí Basin (ca. 563 Ma, Brazil)

The early evolution of metazoans has been reconstructed by studies on exceptionally preserved molds in siliciclastic rocks from the Ediacaran Period. However, there remains considerable controversy regarding the formation mechanisms of this unusual ‘Ediacaran-style’ preservation. Proposed hypotheses usually include early authigenesis of minerals, but evidence for this is scarce. In a recently discovered deposit of Ediacaran biota in Brazil, we show that the classic moldic preservation is related to clay mineral authigenesis. Specifically, these clays originated from the alteration of original pyroclastic sediments, likely enhanced by microbial activity, leading to early illitization and morphological templating of the fossiliferous surfaces at a micrometric scale. Such high-fidelity preservation was made possible by rapid burial during volcanic events and the in-situ templating of tissue by clays via microbially-mediated mineralization. This newly described Lagerstätte demonstrates that a number of minerals can facilitate preservation, and that perhaps ‘Ediacaran-style’ preservation result from different processes leading to the same broad style of preservation.

The early evolution of metazoans has been reconstructed by studies on exceptionally preserved molds in siliciclastic rocks from the Ediacaran Period. However, there remains considerable controversy regarding the formation mechanisms of this unusual 'Ediacaran-style' preservation. Proposed hypotheses usually include early authigenesis of minerals, but evidence for this is scarce. In a recently discovered deposit of Ediacaran biota in Brazil, we show that the classic moldic preservation is related to clay mineral authigenesis. Specifically, these clays originated from the alteration of original pyroclastic sediments, likely enhanced by microbial activity, leading to early illitization and morphological templating of the fossiliferous surfaces at a micrometric scale. Such high-fidelity preservation was made possible by rapid burial during volcanic events and the in-situ templating of tissue by clays via microbially-mediated mineralization. This newly described Lagerstätte demonstrates that a number of minerals can facilitate preservation, and that perhaps 'Ediacaran-style' preservation result from different processes leading to the same broad style of preservation.
The typical moldic preservation of soft-bodied organisms in the Ediacaran Period (635-539 Ma) [1][2][3][4] , in particular the positive epirelief type where the fossils protrude up from bedding surfaces 4 , has few parallels in other periods of geologic time, hence the term "Ediacaran-style preservation. " One explanation points to rheological differences amongst the sediments 5 , a hypothesis that has received recent support 4 . Gehling 1 also proposed the socalled 'death-mask' model for this preservation, in which microbial mats stabilized the sediments and produced a veneer of authigenic pyrite (FeS 2 ) that replicated the morphology of the organisms. Evidence in favor of this hypothesis comes from the presence of iron oxide and pyrite concentrations along fossiliferous bedding planes in certain localities 3,6 . Other factors may include episodic event deposition 7 , early cementation by amorphous silica (SiO 2 ) 2,8 , and burial by volcanic ash-the so-called "Conception-style" preservation 9 .
It is similarly possible that multiple mechanisms could produce this type of moldic fossilization and/or that taphonomic factors varied according to the geologic setting. In this regard, the Itajaí Basin 10-12 (ca. 563 Ma, Brazil) appears to be a promising locality in which to study the possible mechanisms of preservation operating during the Ediacaran Period. This geological unit is relatively unmetamorphosed, has similar ages to the Avalon

Results
The Itajaí biota is composed of the well-known chambered taxon Palaeopascichnus, the discoidal forms Aspidella and Nimbia, and micron-sized algal or bacterial filaments 13 (Fig. 1c-f). Based on the mineralized width (33-193 µm) and bedding plane disposition of the filaments it is suggested that these could be the fossilized remnants of giant filamentous sulphur oxidizing bacteria or eukaryotic algae. Well-preserved reticulated and Arumberia-type microbial mats have also been observed (Fig. 1a,b). We detected three modes of preservation: (1) primarily impressions in positive epirelief (Palaeopascichnus, Aspidella, and micrometric filaments) ( Fig. 1d-f); (2) rare impressions as negative epirelief (Aspidella) (Fig. 1c); and (3) full-relief, three-dimensional microbial mats replicated by clays (Fig. 1a,b). To our knowledge, this is the first case in the geological record of distinctive morphologies of microbial mats (e.g., reticulated tufts and Arumberia) being three-dimensionally replicated by clay minerals. Investigations by microtomography (µ-CT) corroborated the moldic preservation of the filaments (Supplementary Fig. S1a-c) as well as the unusual 3D nature of the microbial mats ( Supplementary Fig. S1d-f, and Movie 1). Remarkably, very small filaments (c. 30 µm in width) are preserved as impressions, and some individuals seem to bear putative cellular structures (Fig. 1e). This minute detail indicates a level of morphological retention in molds that is unprecedented in the fossil record because filamentous microfossils are usually permineralized or occur as palynomorphs 14 .
All the preservation types mentioned above occur in the same lithology: pale-gray millimeter-scale rhythmites of mud and silt that were deposited in an upper slope setting subjected to rhyolitic volcaniclastic input  www.nature.com/scientificreports/ microbially induced sedimentary structures (MISS) are found in non-volcaniclastic sediments, they are preserved only as impressions in semi-relief without associated authigenic minerals (see Ref. 13 ). The upper slope fossiliferous beds, which yielded the majority of samples, are represented by ash-fall pyroclasts, including elongate, cuspate, angular, and blocky devitrified vitric clasts; euhedral phenocrysts (some broken); accretionary pellets (sensu Brown et al. 15 ); ash clusters; and coated particles ( Fig. 2d-h, Supplementary  Fig. S7a-f). Fine layers of altered ash were observed directly covering well-preserved micrometric filaments (Fig. 2d,e). Coarse-grained silt laminae interbedded with the fossiliferous horizons also contain volcaniclasts ( Supplementary Fig. S4), which likely represent a mix of primary and reworked ash-fall deposits. These clasts exhibit devitrification features with the formation of clay minerals ( Supplementary Fig. S7e,f), as well as clay rims and fluid and/or melt inclusions. Some tuffs near the fossiliferous levels also showed the presence of glass spheres ( Supplementary Fig. S7g,h).
In thin section, the surfaces with fossil impressions are characterized by localized concentrations of very fine clays (Fig. 2a, Supplementary Fig. S5). In some layers, early-formed fractures developed mainly following these bedding surfaces. These fractures were filled by fibrous radial micaceous crystallites (sericite) and later by quartz. The formation of this fibrous sericite seems to have occurred before the lithification of the deposits given their molding relationship with the surrounding sediments (Fig. 2i). The early-filled fractures are observed in some instances to crosscut the clay-mineralized fossiliferous surfaces, implying an early origin for the fossil-bearing clays.
X-ray diffraction (XRD) patterns and mid-infrared (MIR) and near-infrared (NIR) spectra further reveal that the fossil-bearing clays (clay-rich surfaces with fossil impressions and clay-replicated microbial mats; hereafter FBCs) are mineralogically similar to the clays from the volcanic sediments (volcaniclastic laminae-VL, tuffs, and tuffites), but differ from other facies throughout the basin that do not show abundant volcaniclasts or clay mineralized fossils (see Supplementary Text 2, Supplementary Figs. S8-S18, Dataset S1). Both FBCs and clays from the volcanic-derived sediments show a predominance of illite (1 M polytype) and ordered illite-smectite www.nature.com/scientificreports/ mixed-layer minerals (R3 I-S MLMs) as suggested by the shoulder on the left side of the illite peak (at 10 Å) and further confirmed by NEWMOD modelling (Supplementary Figs. S12 and S13). As a comparison, the clay assemblages from facies with no fossil-bearing clays are represented by smectite and randomly ordered R0 I/S MLMs (Fig. 3h, Supplementary Text 2, and Supplementary Figs. S8-S18), as well as kaolinite/smectite and Chl-S MLMs that originated during weathering and late diagenetic processes, respectively (Supplementary Text 2). Moreover, the prevalence of the 2 M polytype suggests that most of the illite/muscovite from these samples are detrital in origin.
At higher resolution, the XRD data show that the FBCs exhibit a higher proportion of illite than the clays from the adjacent VL of the same beds, and the tuffs and tuffites (Fig. 3h, Supplementary Figs. S11 and S12). SEM www.nature.com/scientificreports/ micrographs illustrate this difference by evidencing the crystal habit of illite for the clays inside the 3D replicated microbial mats, while the crystal habits of I-S MLMs dominated the host rock ( Supplementary Fig. S19). Scanning electron microscopy with energy dispersive X-ray spectrometry (SEM-EDS) and synchrotron radiation micro-X-ray fluorescence spectroscopy (SR-µXRF) confirmed the higher K concentration and the illite composition of the FBCs. Indeed, these analyses showed that the FBCs have high concentrations of K, Al, Fe, Ti, and Mg (Fig. 3a-d), while the host rock contains high concentrations of elements related to the nature of the clasts and cements (e.g., Si, Na, and Fe). Their illite composition was further confirmed by the oxide weight% results of EDS point analyses (Dataset S2) showing the following average composition: (Si 3.56 , Al 0.44 )O 10 (Al 1.63 , Mg 0.11 , Fe 0.20 )(OH) 2 (K 0.61 , Na 0.01 ). Finally, Raman spectroscopy of the 3D clay-replicated microbial mats additionally reveals the characteristic peaks of illite (ca. 200 cm −1 , 264 cm −1 , 396 cm −1 , 464 cm −1 , and 703 cm −1 ) (Fig. 3e,  Dataset S3). The results from the fitting of several spectra demonstrate that this illite is indeed authigenic in origin and more concentrated in the FBCs than the cement/matrix of the adjacent host rock (Supplementary Text 3, Supplementary Fig. S20).
Raman spectroscopy reveals that anatase (TiO 2 ) is the Ti-bearing phase in the 3D clay-replicated microbial mats ( Fig. 3e-g, Dataset S4). This anatase is formed of micrometer-to submicrometer-sized crystals (Fig. 3g,  Supplementary Fig. S21). Anatase is especially concentrated in the microbial mats, while the matrix contains a high proportion of another TiO 2 mineral, rutile (Supplementary Text 3, Dataset S3). The greater concentration of anatase may explain the dark appearance of the clay-rich fossiliferous surfaces and 3D replicated mats under petrographic observation.

Discussion and conclusion
Our results show that the studied fossil impressions are preserved in surfaces composed of authigenic illite and are associated with diverse volcanic sediments (Figs. 2, 3). In the case of the microbial mats, they are even threedimensionally replicated by the authigenic clays (Figs. 2b,c, 3a-g). Indeed, the fossiliferous horizons and volcanic sediments are enriched in authigenic illite when compared to other levels throughout the basin (Supplementary Text 2, Supplementary Figs. S8-S18). Despite that illite/muscovite is also a component of the clay assemblage at other (non-fossiliferous) levels (e.g., Supplementary Figs. S8-S13), it likely reflects a detrital origin given the practical absence of the 1 M polytype and the predominance of the 2 M polytype.
Illite is usually the result of the progressive alteration of smectite during diagenesis depending on temperature, time, and K + availability 16 . Therefore, the concentration of authigenic illite in the fossil horizons and volcanic materials suggest the original presence of authigenic smectite or illite-smectite clay minerals. Furthermore, the absence of prograde diagenesis (Supplementary Text 2; Supplementary Figs. S8-S10) shows that the higher illitization of the FBCs and volcanic sediments are unrelated to the diagenetic maturation of the rocks and more likely represent K + availability from the volcanic sediments.
Importantly, the mineralogical similarity and the direct association of volcanic material with the fossil-bearing clays suggest that the volcanic sediments played a major role in the formation of these authigenic clays. In fact, the alteration of unstable volcanic glass is well known to promote the formation of new silicate phases [17][18][19][20][21][22][23][24] . Their instability is due to nonbridging silicate sites, which allow easy alteration at low temperatures 18 . In this regard, both experimental and field-based studies have shown the formation of smectite and R0 I-S MLMs from the alteration of volcanic glass occurring at low temperature [19][20][21][22] .
The abundant evidence for the presence of microbial communities at the fossiliferous beds (Figs. 1a, 2b,c; Supplementary Figs. S4, and S5; see also Ref. 13 ) supports the idea that the alteration of volcanic sediments could have been enhanced by microbial activity. For instance, Kawano and Tomita 23 demonstrated that bacterially influenced alteration of pyroclastic sediments can lead to smectite formation through an allophane precursor, while Konhauser et al. 24 showed the formation of smectite directly on the cell surfaces of bacteria colonizing basaltic tephra.
Moreover, the higher amount of illite in the R3 I-S MLMs from the FBCs than in the adjacent VL and tuffs ( Fig. 3; Supplementary Figs. S10, and S11) advocates a link between the presence/activity of microbial communities and processes of early diagenetic illitization, which could have influenced fossil preservation. Two mechanisms are envisaged to explain this higher illitization: (1) further enrichment in K + resulting from microbial metabolism and/or (2) dissimilatory iron reduction (DIR) of Fe 3+ in smectite. The former hypothesis is similar to the one proposed for the microbial mats of the Paleoproterozoic Francevillian Group 25 , where the presence and/or metabolism of ancient microbial mats served as the source of K + for the illitization of these microbially related surfaces. Although K-feldspar grains from the volcanic sediments in the Itajaí Basin likely contributed to posterior diagenetic illitization, their presence does not account for the higher illitization of the FBCs than the clays in the millimeter apart volcaniclastic laminae, as well as the tuffs levels throughout the basin (Fig. 3h,  Supplementary Figs. S10, and S11).
The second mechanism (DIR) is related to the dissolution of smectite layers during the illitization process. This process could also account for the presence of disordered hematite in the FBCs (see Kim 26 ). Furthermore, Zhang et al. 27 demonstrated that illitization by DIR can be enhanced by an external supply of Al and K, which in the Itajaí Basin could have come from volcaniclastic material. The release of dissolved iron may have also contributed to the adsorption of this element by the surfaces of algal or bacterial cells-including both the cell walls and/ or extracellular polymeric substances (EPS)-as observed in modern examples 28 . This process can subsequently result in the formation of sites prone to the nucleation of clay mineral phases [28][29][30] . Several other studies have documented the role of microbes in providing sites for the nucleation and growth of clay and silicate phases [31][32][33][34] .
In addition to the higher degree of illitization, the enrichment in anatase (Fig. 3e-g, Supplementary Fig. S21), as well as disordered hematite (Fe 2 O 3 - Fig. 3e), inside the clay-replicated microbial mats further supports the influence of microbial metabolism in the microenvironmental conditions at the authigenic clay surfaces. Studies by Fe(II)-oxidizing and/or iron-scavenging bacteria (i.e., those that produce siderophores) 35 , thus creating conditions for the precipitation of new phases, such as anatase. Furthermore, experiments with photosynthetic microbial mats have shown the influence of biological activity on the precipitation of anatase when ilmenite was added to the medium 36,37 . In the Itajaí Basin, Fe-Ti oxide grains (e.g., ilmenite and titanomagnetite-now altered to leucoxene; Supplementary Fig. S22) are common in the volcaniclastic sediments and are a likely source of anatase and Fe enrichment in the ancient mats. According to our results, in the FBCs and volcanic sediments of the Itajaí Basin, smectite and/or I-S MLMs were likely the primary mineral phases derived from the alteration of volcanic sediments. This process, supported by microbial activity, resulted in microenvironments enriched in cations, such as K + , Al 3+ , and Na + , and then to the precipitation of newly formed clay minerals. Therefore, we hypothesize that a combination of abundant and easily-altered volcanic particles-in addition to microbial activity-were the essential conditions necessary for the preservation of the Itajaí biota. This complex interplay between volcanic sediments and microbial activity culminated in the formation of abundant early diagenetic clays (Figs. 4,5), resulting in the 3D substitution of microbial constructions and in templating the fossil-bearing surfaces with clay precipitates, thus preserving the external morphologies of macro-organisms and micron-sized algal or bacterial filaments. Besides, the early precipitation of clays may have also helped in the cementation and stabilization and of the sediments.
The role of volcanic material in the preservation of certain Ediacaran assemblages has been previously proposed for the conception-style preservation 9 . However, details explaining this mechanism have not been provided to support this hypothesis, and until now, no authigenic minerals have been associated with this model. Moreover, recent studies have even extended the death-mask model to these deposits, where pyritization would have played a major role in the moldic fossilization 3 . Our work shows that, at least for some basins, the fossilization of the Ediacaran macroscopic communities was the result of the interconnected processes related to volcanic and microbial activity (Figs. 4, 5).
Volcanic activity has often been considered one of the main causes of biotic events during the history of life on Earth, and here we demonstrated unequivocally that volcanism also played an important role in preserving the information of ancient ecosystems. Other examples are also known from the fossil record [38][39][40][41][42][43][44][45] and show that volcanism was not only important for exceptional preservation throughout the geologic record but also diverse in terms of processes and composition of the precursor material.  X-ray diffraction and NEWMOD modeling. Whole-rock powder and clay mineral fractions (< 2 µm) from 24 samples were analyzed with a Panalytical Xpert Pro diffractometer at the University of Poitiers using Cu (Kα = 1.541874 Å) radiation, with the following configuration: Xccelerator detector; geometry θ/θ (Bragg-Brentano); goniometer of 240 mm radius; wavelength filter of Ni (0.3 mm thick); anti-divergence slit of 1/8°2θ, anti-diffusion slit of ¼°2θ; a mask of 10 mm in diameter; soller slit of 0.04 rd spacing; and fixed sample holder. The analyses were performed with a voltage of 40 kV and a current of 40 mA in the angular range of 2-65°2θ (powder) and from 2 to 35º 2θ (oriented preparation). Twenty-two samples originated from different levels (fossiliferous and nonfossiliferous) throughout the basin, while the fossil clays and the intercalated coarse-grained laminae were carefully selected and removed from the fossiliferous beds. These fossiliferous samples were first prepared using air abrasive processes to grossly clean and remove the surrounding matrix from the 3D clay-mineralized microbial mats. Then, we extracted the clays from the selected structures and laminae (fossilized microbial mats and intercalated coarse-grained laminae) under a stereomicroscope using a scalpel. All samples were ground in a mortar and separated for powder analysis. Later, we extracted the clay size fraction (< 2 µm) for oriented preparation. These clay size fractions were then Ca saturated and mounted on glass slides for analysis in air-dried ethylene glycol and after thermal treatment (i.e., 350 °C and 550 °C).
For the characterization of the I-S MLMs, we used NEWMOD simulation to model the Reichweite ordering parameter (R), which can range from randomly interstratified (R = 0) to long-range ordered MLMs (R = 1, 2, and 3).
Scanning electron microscopy. We used a Quanta 650FEG and FEI Inspect F50 microscopes (project SEM-21836 and SEM-23684) at the Brazilian National Laboratory of Nanotechnology (LNNano/CNPEM) and a JEOL JSM IT500 scanning electron microscope, equipped with secondary electron, backscatter electron detectors and coupled with a Bruker lynxeye Energy Dispersive X-ray Spectrometer (EDX) at the IC2MP laboratory of the Université de Poitiers. Analyses were conducted in high-vacuum mode and with a current tension of