Microbial influence on dolomite and authigenic clay mineralisation in dolocrete profiles of NW Australia

Dolomite (CaMg(CO3)2) precipitation is kinetically inhibited at surface temperatures and pressures. Experimental studies have demonstrated that microbial extracellular polymeric substances (EPS) as well as certain clay minerals may catalyse dolomite precipitation. However, the combined association of EPS with clay minerals and dolomite and their occurrence in the natural environment are not well documented. We investigated the mineral and textural associations within groundwater dolocrete profiles from arid northwest Australia. Microbial EPS is a site of nucleation for both dolomite and authigenic clay minerals in this Late Miocene to Pliocene dolocrete. Dolomite crystals are commonly encased in EPS alveolar structures, which have been mineralised by various clay minerals, including montmorillonite, trioctahedral smectite and palygorskite‐sepiolite. Observations of microbial microstructures and their association with minerals resemble textures documented in various lacustrine and marine microbialites, indicating that similar mineralisation processes may have occurred to form these dolocretes. EPS may attract and bind cations that concentrate to form the initial particles for mineral nucleation. The dolomite developed as nanocrystals, likely via a disordered precursor, which coalesced to form larger micritic crystal aggregates and rhombic crystals. Spheroidal dolomite textures, commonly with hollow cores, are also present and may reflect the mineralisation of a biofilm surrounding coccoid bacterial cells. Dolomite formation within an Mg‐clay matrix is also observed, more commonly within a shallow pedogenic horizon. The ability of the negatively charged surfaces of clay and EPS to bind and dewater Mg2+, as well as the slow diffusion of ions through a viscous clay or EPS matrix, may promote the incorporation of Mg2+ into the mineral and overcome the kinetic effects to allow disordered dolomite nucleation and its later growth. The results of this study show that the precipitation of clay and carbonate minerals in alkaline environments may be closely associated and can develop from the same initial amorphous Ca–Mg–Si‐rich matrix within EPS. The abundance of EPS preserved within the profiles is evidence of past microbial activity. Local fluctuations in chemistry, such as small increases in alkalinity, associated with the degradation of EPS or microbial activity, were likely important for both clay and dolomite formation. Groundwater environments may be important and hitherto understudied settings for microbially influenced mineralisation and for low‐temperature dolomite precipitation.

and authigenic clay minerals in this Late Miocene to Pliocene dolocrete. Dolomite crystals are commonly encased in EPS alveolar structures, which have been mineralised by various clay minerals, including montmorillonite, trioctahedral smectite and palygorskite-sepiolite. Observations of microbial microstructures and their association with minerals resemble textures documented in various lacustrine and marine microbialites, indicating that similar mineralisation processes may have occurred to form these dolocretes. EPS may attract and bind cations that concentrate to form the initial particles for mineral nucleation. The dolomite developed as nanocrystals, likely via a disordered precursor, which coalesced to form larger micritic crystal aggregates and rhombic crystals. Spheroidal dolomite textures, commonly with hollow cores, are also present and may reflect the mineralisation of a biofilm surrounding coccoid bacterial cells. Dolomite formation within an Mg-clay matrix is also observed, more commonly within a shallow pedogenic horizon. The ability of the negatively charged surfaces of clay and EPS to bind and dewater Mg 2+ , as well as the slow diffusion of ions through a viscous clay or EPS matrix, may promote the incorporation of Mg 2+ into the mineral and overcome the kinetic effects to allow disordered dolomite nucleation and its later growth. The results of this study show that the precipitation of clay and carbonate minerals in alkaline environments may be closely associated and can develop from the same initial amorphous Ca-Mg-Si-rich matrix within EPS. The abundance of EPS preserved within the profiles is evidence of past microbial activity. Local fluctuations in chemistry, such as small increases in alkalinity, associated with the degradation of 1 | INTRODUC TI ON Dolomite (CaMg(CO 3 ) 2 ) is a common carbonate mineral found in both marine and continental settings and is economically significant as a component of many hydrocarbon reservoirs, sulphide ore bodies and productive groundwater aquifers (Bakalowicz, 2005;Gregg, 2004;Sun, 1995;Warren, 2000). However, uncertainties remain regarding the mechanisms of dolomite formation. Dolomite precipitation is kinetically inhibited at Earth's surface temperatures and pressures and mostly does not precipitate even when supersaturated in solution (Land, 1998;Morrow, 1982). The high enthalpy of the Mg 2+ double hydration shell is the dominant factor inhibiting dolomite precipitation and crystal growth (Lippmann, 1973;Morrow, 1982). Low CO 2− 3 activity relative to HCO − 3 typical of most natural waters may also be limiting dolomite crystal growth (Machel & Mountjoy, 1986). In recent decades, microbially influenced precipitation of low-temperature dolomite has been documented in laboratory experiments (Vasconcelos et al., 1995;Warthmann et al., 2000;Wright & Wacey, 2005) and observed directly in a number of modern environments, including lakes and hypersaline lagoons or sabkhas (e.g. Bontognali et al., 2012;van Lith et al., 2002;Wright, 1999), where microbial respiration modifies conditions to overcome the kinetic barriers of dolomite precipitation (Vasconcelos et al., 1995;Warthmann et al., 2000;Wright & Wacey, 2005). Typically, dolomite forms via precipitation of an initial very high-Mg calcite (VHMC, i.e. Ca:Mg near 50:50 but no cation ordering) or a disordered Cadolomite precursor that then matures into ordered dolomite (Gregg et al., 2015;Kaczmarek et al., 2017;Zhang et al., 2012Zhang et al., , 2015. More recently, extracellular polymeric substances (EPS) generated by microbes have been increasingly recognised as a catalyst for dolomite precipitation (Bontognali et al., 2010;Krause et al., 2012;Perri et al., 2018;Roberts et al., 2013;Zhang et al., 2015).
The co-occurrence of clay minerals and dolomite formed in the presence of EPS has been observed, most notably in microbial mats (Bontognali et al., 2010;Pace et al., 2016;Perri et al., 2018), but also in lacustrine environments (Casado et al., 2014;Leguey et al., 2010).
However, the combined association of EPS with clay and dolomite authigenesis remains poorly understood and few studies have considered the combined role of EPS and clay minerals on dolomite precipitation.
Magnesium-bearing clays including palygorskite, sepiolite and smectites are commonly associated with dolomite within semi-arid and arid environments, typically in alkaline lacustrine or groundwater environments (Akbulut & Kadir, 2003;Botha & Hughes, 1992;Bristow et al., 2012;Calvo et al., 1999;Casado et al., 2014;Cuadros et al., 2016;Mees, 2001;Yeniyol, 2014). Knowledge of the processes and conditions of authigenic clay and carbonate mineral formation in continental settings is particularly important as the presence and chemistry of the minerals are frequently utilised as palaeoenvironmental indicators (e.g. Alonso-Zarza, 2003;Arenas et al., 1997;Calvo et al., 1999;Hay & Kyser, 2001). Thus far, investigation of microbially influenced dolomite has overwhelmingly focussed on hypersaline lakes, lagoons, microbial mats and marine environments (e.g. Di Loreto et al., 2019;Dupraz et al., 2004;Meister et al., 2007;Pace et al., 2016;Spadafora et al., 2010;You et al., 2018). As groundwater systems are a major location of microbial biomass (as pointed out by Petrash et al., 2017Petrash et al., , 2021, dolomite formed in shallow aquifers may provide new insights into the possible biotic and abiotic controls on dolomite precipitation. Mather et al. (2019) identified authigenic Mg-clay minerals that appeared to be associated with both EPS and dolomite within groundwater dolocrete in the Pilbara region of inland northwest Australia; this provided a new opportunity to study this occurrence in a groundwater environment. Dolocrete aquifers in the Pilbara EPS or microbial activity, were likely important for both clay and dolomite formation.
Groundwater environments may be important and hitherto understudied settings for microbially influenced mineralisation and for low-temperature dolomite precipitation.

K E Y W O R D S
biomineralisation, clay minerals, dolocrete, EPS, groundwater, microbial dolomite, terrestrial carbonate region, renowned for high biodiversity and endemic subterranean invertebrates (the stygofauna), are known to have been present in the period preceding and overlapping with the age of dolocretisation in the Late Miocene and Pliocene, indicative of a diverse ancient groundwater ecosystem (Humphreys, 2001;Humphreys et al., 2009;Karanovic, 2007;Leys et al., 2003;Leys & Watts, 2008;Perina et al., 2019). Here we seek to determine the nature of the association between EPS, authigenic clay and dolomite in groundwater environments by further investigation of the dolocrete profiles presented in Mather et al. (2019). Using a combination of petrographic and mineralogical analyses, our aims are to determine the textural and timing relationships between EPS, dolomite and clay minerals and to establish the mechanisms of mineral precipitation including the role of microbes and/or microbial EPS.
The deposition of carbonates and other evaporites in soil and groundwater environments throughout much of inland and central Australia has been linked to a shift to arid and slightly cooler conditions in the Late Miocene (Arakel & McConchie, 1982;Bowler, 1976;Byrne et al., 2008;Mann & Horwitz, 1979;Mao & Retallack, 2019).
Similarly, in the Hamersley Basin, the major phase of dolocretisation is constrained to the Late Miocene to Pliocene, that is in the time range ~12-5 Ma (Barnett, 1981;Humphreys, 2001;Kneeshaw & Morris, 2014;Mather et al., 2019;Morris & Ramanaidou, 2007), although some Quaternary dolocretes have also been reported (Barnett, 1981;Mather et al., 2018). Prior to dolocrete formation, a period of deep weathering under more humid conditions followed by a dominantly erosional period, resulted in aggradation of valleys and drainage channels with clay and silt-rich sediments and in the development of internal drainage basins (Arakel & McConchie, 1982;Kneeshaw & Morris, 2014). The sharp reduction in vegetation following the onset of aridity (Martin, 2006), reduced transpirational fluxes and led to a rising water table, with increased potential for gaseous exchange with the atmosphere where groundwater was close to the surface (Eugster, 1980). Shallow groundwater systems with sluggish lateral flow resulted in the development of salt lakes and playas within landscape depressions and the precipitation of chemical sediments (Arakel & McConchie, 1982;Jacobson et al., 1988;Mann & Deutscher, 1978). Dolocrete formed under saline-evaporitic conditions, promoted by evaporation and CO 2 degassing from shallow and emerging Mg-rich groundwater (Kneeshaw & Morris, 2014;Mather et al., 2019). The dolocretes are typically ~5-30 m thick and are distributed in areas of groundwater emergence along structural boundaries, and large outcrops of tens of km 2 are notable in the upper reaches of creek systems (Mather et al., 2019). Weathering of Proterozoic marine dolomite (Wittenoom Formation) basement rocks that comprise the major regional aquifer has resulted in alkaline groundwater chemistry dominated by Mg 2+ , Ca 2+ and HCO − 3 ions that, given the right conditions, may rapidly become oversaturated in respect to dolomite (Dogramaci et al., 2012;Eugster, 1980). Atmospheric dust may have also contributed a considerable solute load to groundwater, delivered via precipitation, providing a further source of ions for subsequent dolocrete formation (Bowler, 1976;Chiquet et al., 1999).
The dolocrete investigated for this study is located in the upper reaches of the ephemeral Coondiner Creek system within a subbasin of the Hamersley Ranges. Approximately 12 km 2 of dolocrete is exposed at the surface (Figure 1b-d) surrounding northeast and eastward tributaries of the Coondiner Creek system where they merge to form the main north-easterly Coondiner Creek channel.
The dolocrete outcrop is relatively flat (~650-660 m a.s.l.) and is incised by small drainage lines connecting to Coondiner Creek.
The dolocrete is most strongly developed within Cenozoic alluviallacustrine sediments and extends into in situ regolith and weathered bedrock of the Hamersley Group (Mather et al., 2019). The alluviallacustrine sediments are primarily comprised of a clay-rich matrix, of both detrital and authigenic clay mineral assemblages, with varying proportions of coarser fragments of silcrete, Fe oxide pisoliths and gravels derived from erosion of adjacent weathering profiles (Killick et al., 2008;Kneeshaw & Morris, 2014;Mather et al., 2019). The main body of dolocrete extends several kilometres west and southwest of the outcrop in the sub-surface. Drill-core data also suggest dolocrete of up to 30 m thick buried within the connecting E-W trending valleys, although the distribution of dolocrete appears to be heterogeneous and patchy.
The dolocrete from the Coondiner Creek area (hereon referred to as the Coondiner dolocrete) was previously investigated by Mather et al. (2019). Mather et al. (2019) reported that the dolocrete is characteristic of valley groundwater dolocrete, comprising multi-metre thick profiles that contain phreatic cements (Mann & Horwitz, 1979;Nash & McLaren, 2003). Fissures and brecciation from displacive growth of dolocrete are well developed and the profiles are highly cemented (Mather et al., 2019). Pedogenic carbonate composed of calcite and dolomite, including characteristic nodules (glaebules), is also present within some profiles in the uppermost sediments (Mather et al., 2019). Erosional processes dominating since the Pliocene have resulted in the down-cutting of the dolocrete and dissection by modern drainage. Calcretisation and some dedolomitisation have occurred in the upper few metres due to the influence of subsequent fresher meteoric waters near the surface (Mather et al., 2019).

| Samples and approach
This study utilises three diamond drill cores (cores 010, 009 and 015) previously studied by Mather et al. (2019), which were extracted within 1 km of each other on the southeastern corner of the Coondiner Creek dolocrete outcrop ( Figure 1a). Core selection was based on access to high-quality diamond drill core that penetrated the entire profile of dolocrete and extended into the underlying bedrock. The drill cores are 6 cm in diameter and depths range from 0 to 13.4 m (core 010), 0 to 21 m (core 009) and 0 to 23 m (core 015).
The drill cores were analysed using the HyLogger-3™ hyperspectral scanner to produce continuous infrared reflectance spectral data and construct the mineralogy throughout the dolocrete profiles. Representative core sections were sampled with the aim to collect at least three samples for each characteristic section based on the properties of host sediment, dolocrete texture and colour (Mather et al., 2019). A total of 60 core sections were sampled for polished thin-section preparation and 69 for bulk mineralogical analysis by X-ray diffraction (XRD), including one sample from the intact bedrock of each core. Further sub-sampling of clay-rich sediments was undertaken for additional petrographic microanalysis of mineral chemistry and textural features (15 samples) and XRD analysis of the <2 μm fraction (10 samples).

| Mineralogical analysis from continuous spectral reflectance data
The drill cores were scanned using the HyLogger-3 system and the methods outlined by Hancock and Huntington (2010), Hancock et al. (2013) and Schodlok et al. (2016) to determine the presence and relative abundance of minerals and to establish continuous mineral logs. The HyLogger-3 scanners acquired continuous reflectance spectra along each drill core covering the visible-near-infrared (VNIR: 380-1000 nm), shortwave infrared (SWIR: 1000-2500 nm) and thermal infrared (TIR: 6000-14,500 nm) wavelength region of the electromagnetic spectrum (Hancock & Huntington, 2010;Huntington et al., 1997;Schodlok et al., 2016). High-resolution (0.1 mm pixel) imaging of the core was also acquired during the scan and compiled with reflectance spectral data in continuous 4 mm intervals along the core (Schodlok et al., 2016). The HyLogger-3 data were processed using the spectral geologist (tsg) software and the in-built algorithm, The Spectral Assistant (TSA), for automated mineral identification and estimates of relative abundance (Berman et al., 1999). The TSA algorithm identifies minerals by comparing the fit of measured spectra to spectra stored in the TSG in-built spectral reference library, which contains ~500 reference spectra for ~60 common minerals (Berman et al., 1999;Berman & Bischof, 1997a, 1997b, 1997cHuntington et al., 1997). TSA identifies between two to three spectrally dominant minerals for each wavelength region with relative proportions (weights) for each spectrum (Hancock & Huntington, 2010;Huntington et al., 1997). Therefore, lower representation of minor mineral components within mixed assemblages can be expected. Validation of TSA mineral matches was completed via visual comparison of acquired spectra to reference library spectra and manually refining the TSA matching scalars to produce a final user-guided mineral match (Schodlok et al., 2016).
Further independent validation was undertaken by comparison with XRD and scanning electron microscopy (SEM), energy dispersive X-Ray (EDX) mineral and chemical data. Additional information on the mineral identification and validation of Hylogger-3 data can be found in Appendix S1.

| X-ray diffraction and mineral quantification
Samples for XRD analysis were prepared as whole-rock pressed powders for bulk analysis and as both Ca-saturated and acetic acidtreated pressed powder and Mg-saturated and glycerol treatedoriented films for the <2 μm size fractions. The <2 μm size fractions were separated, following dispersion with the aid of NaCl, by repeated centrifuging at 570 g. Oriented clay fractions were prepared by dispersing with an ultrasonic probe and depositing the suspensions onto 0.22 μm cellulose nitrate Millipore filter membranes under suction. Samples were analysed by XRD using a PANalytical X'Pert Pro Multi-purpose Diffractometer with Fe-filtered Co Kα radiation.
The diffraction patterns were recorded from 3.8° to 80° (pressed powders) and 2.5° to 33° (oriented <2 μm fractions) 2θ interval, in steps of 0.017° and a 0.5 s counting time. Qualitative analysis was performed on the XRD data using in-house xplot and highscore plus (from PANalytical) search/match software. Quantitative analysis was performed on the XRD data from all bulk samples using the commercial package siroquant from Sietronics Pty Ltd. The results are normalised to 100%, and hence do not include estimates of unidentified or amorphous materials. The molar proportion of CaCO 3 in dolomite and Mg substitution in calcite was calculated from a and c unit cell data determined by Rietveld analysis (Bischoff et al., 1983;Reeder & Sheppard, 1984).

| Microscopy and microanalysis
Scanning electron microscopy (SEM) observations and semiquantitative elemental analysis were performed on the thin sections (carbon-coated) and freshly broken surfaces of clay-rich dolocrete samples (platinum coated) using a Verios XHR fitted with an Oxford instruments SDD-X-Max EDX spectrometer.

| Comparison of mineral data sets
The major mineral assemblages of the dolocrete profiles from reflectance spectra and XRD data, which consist primarily of carbonate, quartz, Fe oxides and a variety of clay minerals, are displayed in Figure 2. Overall, the TSA mineral identification corresponds well with mineralogy obtained by XRD and SEM-EDX, particularly for kaolinite and palygorskite. However, distinctions in the reported mineral abundance between data sets were observed for a number of minerals. Dolomite abundance is under-represented by TSA in some core sections, likely due to its ultrafine-grain size (<65 μm), which is known to result in diminished diagnostic features due to scattering effects (Gaffey, 1986). This is particularly evident in the top few metres of core 015, which is composed of ~50-75 wt% carbonate and contains abundant sub-micron-sized carbonate but few counts of carbonate are detected by TIR or SWIR ( Figure 2). The abundance of smectite is also not well matched between reflectance spectra and XRD. This may largely be explained by mineral mixtures of smectites, kaolinite and carbonates, of which the absorption features largely overlap.
The identification of specific smectite minerals is challenging for both XRD and Hylogger-3 data sets. The TSG spectral reference library contains only montmorillonite, nontronite and saponite, which may not fully represent the range of smectite minerals present.

F I G U R E 2
The XRD (bulk) and Hylogger-3 mineralogy of dolocrete profiles alongside an overview of the sedimentology of the profiles for core 010 (a), core 009 (b) and core 015 (c). Reflectance spectra TSA mineralogy is presented as the total weight (proportion) of the particular mineral detected within a 100 mm bin interval. As the data are compiled in continuous 4 mm intervals, the total possible weight for each HyLogger-3 mineral set is 25, indicated by the dashed line labelled Max set weight. The mineral abundance from HyLogger-3 data is from the thermal infrared (TIR) wavelength region for quartz and carbonate, visible-near infrared (VNIR) wavelength region for Fe oxides and shortwave infrared (SWIR) wavelength region for clay minerals.
Sepiolite is also not included within the TSG reference library, although, TSA does identify the mineral group (e.g. palygorskitesepiolite). As the non-referenced minerals are present in very minor proportions, this does not impact the construction of major mineral assemblages and trends. Observations of the position of the smectite 060 d-space from XRD patterns of the <2 μm fraction allowed for the determination of trioctahedral versus dioctahedral structures (Table 1), which is consistent with TSA smectite mineralogy where comparable. However, samples that contain few or mixed smectites were difficult to distinguish with XRD patterns and for those with a significant palygorskite content due to the overlapping palygorskite peak (at 1.5 Å). Clay chemical compositions determined from EDX point analyses confirm the expected distribution of clay mineral assemblages within the profiles identified from TSA and XRD (Appendix S1). Further details on the validation of TSA results can be found in Appendix S1.

| Mineralogy and sedimentology of profiles
The sedimentology of the Coondiner dolocrete profiles was previously described by Mather et al. (2019) and an overview is shown alongside the mineralogical data in Figure  and Ca-rich dolomite (~53-60 mol% CaCO 3 ), determined from XRD data (Reeder & Sheppard, 1984). Ordering reflections (015 and 021) were identified in XRD patterns of both stoichiometric and Ca-rich dolomite, indicating relatively well-ordered dolomite (Goldsmith & Graf, 1958; Figure 3). Dolocrete textures are primarily characterised by micritic dolomite (<10 μm diameter) matrix, although crystal size ranges from sub-micron to ~100 μm intercalated with authigenic clay minerals and varying proportions of detrital grains and clasts. Dissolution features are common resulting in elongated fissures and vugs, which are predominantly filled with dolomite or clay cement, leading to brecciation of the matrix and resistant grains.
Dolomite crystal textures range from non-planar to planar-e (Sibley & Gregg, 1987), with rhombic dolomite more common deeper in the profiles. Phreatic dolospar cements of equant crystals, ~20-70 μm across, are common in pores and along cracks, commonly with coarser crystals towards the centre of the void (drusy fabric), and as isopachous fringes along pore margins or around grains. Irregular gravitational cements lining pores are occasionally observed, more commonly in the upper 5 m of the profiles. Calcite is also present as a late-stage cement, primarily within 1 m of the surface (although extending to ~3 m in core 009), where it accounts for 17-60 wt% and contains 1-6 mol% MgCO 3 as determined from XRD data (Bischoff et al., 1983). Calcite is observed to have filled cracks and voids surrounding detrital clasts and within fractures (~0.05-1 mm in diameter) cross-cutting dolomite matrix. In the upper metre of core 009, calcite appears to have replaced dolomite where the dolomite matrix displays micro-porosity and calcite has formed in pores that commonly interconnect (Mather et al., 2019). There is no evidence of earlier formed calcite as a precursor to dolomite.        Palygorskite-sepiolite also appears as fine sheets that curve around at angles that appear to be from clay coating of pre-existing grains or minerals that have since dissolved (Figure 8a)  and SiO 2 (~12 wt%), determined by EDX point spectra, which extend outward from the cell wall (Figure 9h).

| DISCUSS ION
The results from this study provide evidence for the close associa-
D-smectite mineralogy was primarily montmorillonite, which is expected as montmorillonite is a major weathering product of common silicate minerals such as plagioclase and mica and it may form directly from Si and cation-rich solutions (Wilson, 1999). Partly T-smectite and sepiolite are primarily distributed at a similar elevation in cores 010 and 015, suggestive of formation associated with a palaeo-water table under saline and alkaline conditions (i.e. pH > 8; Birsoy, 2002;Mees & Van Ranst, 2011;Pozo & Calvo, 2018;Webster & Jones, 1994). The t-smectite is typically finely intergrown with dolomite suggesting formation associated with dolocretisation. The distribution of palygorskite is also strongly linked to the palaeo-groundwater table. Palygorskite authigenesis is commonly associated with dolocrete forming around a fluctuating water table, promoted by seasonal semi-arid to arid wetting and drying processes, as described by Colson et al. (1998) and Kadiṙ et al. (2010). Palygorskite cement surrounding dolomite matrix and the coating of dolomite rhombs indicates palygorskite formed during or shortly after the major phase of dolocretisation.
Precipitation of both palygorskite and dolomite is promoted by alkaline (optimal pH ~7.7-8.5; Birsoy, 2002) and high-Mg 2+ conditions, and the dissolution of silicates during dolocretisation at the higher pH can result in palygorskite development under the Mg-rich groundwater conditions (Galán & Pozo, 2011). Seasonal fluctuations causing minor changes in Si, Mg and Al availability, which will have altered the stability of palygorskite and dolomite, likely contributed to the formation of palygorskite cements (Calvo et al., 1999;Suárez et al., 1994).
Authigenic t-smectite and sepiolite are also associated with dolomite within the pedogenic zone, indicating a moderately saline and alkaline soil environment (Casado et al., 2014). Some of the Mg-clay in the pedogenic horizon may be associated with the earlier formation in the alkaline lacustrine and groundwater environment.
However, accumulations of neoformed stevensite and sepiolite that comprise the matrix in which carbonate nodules have developed, suggest clay formation is also associated with pedogenic processes.
Textural characteristics indicate Mg-clay in the soil zone formed by direct precipitation from pore waters as well as the replacement of EPS (Mayayo et al., 1998).

| The role of EPS in dolomite and clay mineralisation
Extracellular polymeric substances have been identified as important substrates for the precipitation of various minerals due to their capacity to attract cations, such as Ca, Mg, Si and Al (Trichet & Défarge, 1995). The accumulation of ions in the EPS matrix has been shown to provide sufficient concentration for amorphous nanopar- explain dolomite that appears to have grown in and formed from smectitic material (e.g. Figure 5f). Clay mineralisation has occurred broadly within the alveolar EPS matrix, whereas the dolomite appears to have formed from the EPS surface and grown into pore space within the alveolar structure; these fabrics resemble those documented in studies of carbonate and associated Mg-silicate in salt-lake microbialites, which demonstrate the development of an initial Mg-silicate phase within the alveolar EPS matrix, followed by nucleation of carbonate, indicating a similar process may have occurred within the Coondiner dolocrete (Bischoff et al., 2020;Burne et al., 2014;Pace et al., 2016;Souza-Egipsy et al., 2005). Once nucleated, the carbonate may further precipitate and occlude the remaining pore space as the cells that produced EPS degrade and the EPS themselves (Pace et al., 2016). Zhang et al. (2012Zhang et al. ( , 2015 proposed that hydrogen bonding between EPS (H in O-H) and surfaces of a forming carbonate (O in CO 2− 3 ) mineral is a key factor for the displacement of surface water molecules and complete desolvation of Mg 2+ ions, augmenting the incorporation of Mg 2+ into the mineral and promoting dolomite crystallisation.
The ability of EPS to attract and bind cations is controlled by the presence of certain functional groups in the EPS polymer matrix, such as carboxylic acids, that have a strongly negatively charged surface (Braissant et al., 2007;Douglas & Beveridge, 1998;Dupraz et al., 2004;Kenward et al., 2013;Krause et al., 2012;Perri et al., 2018;Souza-Egipsy et al., 2005). Changes in the physicochemical characteristics of EPS during degradation, including the re-arrangement of functional groups, is also a key factor for the accumulation of ions in the EPS substrate to reach a critical radius for mineral nucleation Dupraz et al., 2004;Trichet & Défarge, 1995). The honeycombalveolar structures, which are commonly observed encasing dolomite in the Coondiner dolocrete, may reflect a degrading EPS matrix that reorganises the acidic bonds within the EPS and provides a template for mineral nucleation and growth (Defarge et al., 1994(Defarge et al., , 1996Görgen et al., 2021;Trichet et al., 2001). The types of microbes secreting EPS, as well as differences in EPS degradation pathways, influence the volume and composition of the polymer matrix and are, therefore, important for the resultant ion-binding capacity Dupraz & Visscher, 2005). As degradation of EPS progresses HCO − 3 ions are produced, increasing the alkalinity and decreasing the surface charge, and these may lead to the release of cations and the nucleation of amorphous nanoparticles del Buey et al., 2018;Dupraz et al., 2004). For example, the release of Mg 2+ ions during degradation could promote the incorporation of Mg 2+ into the incipient clay or carbonate mineral.
Microbial activity may also directly influence mineralisation by causing chemical changes within the EPS and surrounding pore waters that promote the precipitation of carbonate and clay minerals (e.g. del Buey et al., 2021;Van Lith et al., 2003;Vasconcelos & McKenzie, 1997). For example, sulfate-reducing bacteria, which are recognised as a common driver of microbial carbonate formation, reduce the SO 2− 4 ion concentration and locally increase alkalinity; this can enhance the precipitation of carbonate and silicate minerals (Braissant et al., 2007;del Buey et al., 2021;Krause et al., 2012;Vasconcelos et al., 1995;Wright, 1999;Zhang et al., 2012). The presence of NH 4 + ions produced by several different microbial processes in soils and porewaters has also been shown to induce carbonate precipitation (Görgen et al., 2021) and they could also reduce the energy for Mg dehydration.
The processes outlined above can result in the formation of a range of carbonate and silicate minerals. A variety of authigenic Mgclays, including t-smectite (stevensite and kerolite) and palygorskitesepiolite, have been observed to have formed in association with carbonate and EPS in saline and alkaline lacustrine and marine environments (e.g. Calvo et al., 1999;Eardley, 1938;Leguey et al., 2010;Léveillé et al., 2002;Perri et al., 2022). Microbially influenced mineralisation of other silicates, such as Al and Fe silicates, has also been demonstrated in different environments (e.g. Konhauser et al., 1994;Konhauser et al., 2002;Ueshima & Tazaki, 2001), indicating that although the mechanisms of mineralisation may be the same, the minerals produced will reflect the broader hydrogeochemical conditions. The mineralogy of clays formed within the EPS matrix reflects the initial pore water chemistry as well as progressive changes that occurred during EPS degradation and incipient clay formation. Experimental work by Bontognali, Martinez-Ruiz, et al. (2014) -Zarza et al., 1992;Casado et al., 2014;Cuadros et al., 2016;Díaz-Hernández et al., 2013;Martín-Pérez et al., 2015;Wanas & Sallam, 2016). Smectite (particularly incipient) has been shown to have gel-like and viscous properties that influence the diffusion and ability for adsorption and complexation of ions Fernández-Díaz et al., 1996). The slow diffusion of ions within the clayey medium promotes the incorporation of Mg 2+ into dolomite, under dolomite-supersaturated conditions (Díaz-Hernández et al., 2013, 2018Fernández-Díaz et al., 1996. In addition, clays with a strong negative surface charge, such as montmorillonite, may also bind cations and promote dolomite precipitation by the same mechanism as described above for charged EPS surfaces (Liu et al., 2019). primarily forms as Ca-rich and disordered sub-micron crystals or VHMC (e.g. Bontognali et al., 2010;Braissant et al., 2003;Carballo et al., 1987;Daye et al., 2019;Liu et al., 2019;Roberts et al., 2004;Wright & Wacey, 2005;You et al., 2013;Zhang et al., 2015). The inferred expectation is that disordered and partially ordered dolomite (referred to as 'protodolomite', see Gaines, 1977) will mature in time to stoichiometric dolomite (which may be Ca-rich or Mg-rich).
However, a recent study by Daye et al. (2019) (Fang et al., 2022;Gregg et al., 2015;Kaczmarek et al., 2017;Kaczmarek & Sibley, 2014;Lippmann, 1973;Machel, 2004;Petrash et al., 2017;Vasconcelos & McKenzie, 1997)  This transformation into ordered dolomite crystals and larger crystal forms is most likely an inorganic process of syntaxial overgrowth and/or dissolution-reprecipitation that follows the initial nucleation of disordered dolomite within the amorphous substrate concentrated in the EPS (Kaczmarek et al., 2017;Wright, 1999). The net result of this process is the extensive dolocretisation of the sediment in the Hamersley Basin valleys close to the groundwater table.

| Spheroidal dolomite
Variation in the morphology of dolomite crystals and crystal aggregates within the Coondiner dolocrete, ranging from spheroidal to rhombic, likely reflect a range of conditions or processes, such as differences in structure and composition of the substrate (i.e. the EPS), changes in pore water chemistry as well as the influence of biological processes (Braissant et al., 2003;Fernández-Díaz et al., 2006).
Spheroid formation has been frequently linked to microbially influenced processes (e.g. Bischoff et al., 2020;Nielsen et al., 1997;Perri & Tucker, 2007;Spadafora et al., 2010;Wright, 1999;You et al., 2014) as well as abiotic processes (e.g. Casado et al., 2014;Wanas, 2002). Gunatilaka (1989) also proposed a possible mechanism of spheroid formation associated with the seepage of hydrocarbons, where gas bubbles of CO 2 could provide a nucleus for spheroid formation. The spheroidal aggregates present in the Coondiner dolocrete core 010, which consist of coalesced dolomite nanocrystals, have morphological features that are similar to coccoidal bacterial forms that may have provided a template for mineralisation (Spadafora et al., 2010).
Small sub-spherical to polyhedral clusters of nanocrystals and rings of nanocrystals with hollow centres, which comprise the spheroidal aggregates, are primarily ~500 nm to 1 μm in diameter and resemble the size and shape of coccoid cells.

| Microbial and geochemical drivers of groundwater dolomite formation
The microbial structures identified within the Coondiner dolocretes are consistent with observations from other carbonate environments worldwide, where microbes and/or associated EPS are identified as key factors in the formation of dolomite in low-temperature environments (Petrash et al., 2017). Microbial influence on dolomite mineralisation has been most frequently identified within salineevaporitic lacustrine and shallow-marine environments, commonly associated with modern or fossil microbial mats (e.g. Dupraz & Visscher, 2005;Van Lith et al., 2003;Wright, 1999). Whilst the role of microbes and microbial EPS on mineralisation in groundwater dolocrete has rarely been reported (Casado et al., 2014), similarities in hydrochemical conditions between the variety of saline-evaporitic settings in which dolomite has formed, highlight the requirement of hydrochemical drivers for dolomite formation. Favourable hydrochemical conditions for dolomite precipitation include high Mg/ Ca ratios and high CO 2− 3 activity (Morrow, 1982), typical of many saline-evaporitic and alkaline environments. In the Hamersley Basin, weathering of Proterozoic marine dolomite basement rocks provide the source of Mg 2+ , Ca 2+ and HCO − 3 ions to groundwater (Dogramaci et al., 2012). Evaporation and CO 2 -degassing from shallow groundwater were likely important for promoting dolomite formation as salinity and alkalinity increased (Mather et al., 2019).
The hydrochemical conditions are also an important factor influencing microbial communities and their activities, which may be further linked to the promotion of dolomite formation. Steep physicochemical gradients (e.g. in salinity and redox) can give rise to zones of significant biogeochemical activity (Arakel, 1986;Mann & Deutscher, 1978). For example, strong redox gradients potentially release organic compounds that may be broken down by various microbes (Castanier et al., 1999;Humphreys et al., 2009). These zones are typically associated with the mixing of waters and are common in coastal zones between seawater and fresher groundwater and also in continental settings around salt lakes and playas where regional groundwaters mix with highly saline terminal groundwater (Humphreys et al., 2009;Mather et al., 2018;Petrash et al., 2021). Stable oxygen and carbon isotope signatures of carbonates can inform on the metabolic pathways of the microbes, where microbial activity has strongly influenced carbonate precipitation. The stable isotope compositions are modified by biogeochemical processes during microbial respiration that is incorporated into the mineral.
However, analysis of stable isotope signatures of the Coondiner dolocrete by Mather et al. (2019) found that the δ 13 C and δ 18 O values (δ 18 O: −7.63 to −3.40‰ VPBD; δ 13 C: −6.27 to −4.28‰ VPBD) are within a similar range to several investigations of carbonates for both biotic and abiotic dolomite (e.g. Casado et al., 2014;Petrash et al., 2017), but that they did not provide direct evidence of microbially mediated precipitation. The ambiguous stable oxygen and carbon isotope signatures likely reflect the combined influence of broader hydrochemical processes, such as evaporation and CO 2 degassing, in addition to microbial activity.

Evidence of biological activity in groundwater systems in the
Hamersley Basin has been demonstrated in several studies (e.g. Humphreys et al., 2009;Karanovic, 2007;Leys et al., 2003;Leys & Watts, 2008;Perina et al., 2019), where shallow calcrete and dolocrete aquifers are renowned for high biodiversity and prevalence of endemic species of the stygofauna (subterranean aquatic animals; Humphreys, 2001;Leys et al., 2003). Ancient stygofauna lineages are known to have been present in the aquifers since the Late Miocene (Finston et al., 2009;Humphreys et al., 2009;Perina et al., 2019), thus overlapping with the period of dolocretisation and representative of a diverse groundwater ecosystem, which would most likely have contained a variety of microbiological communities (Deharveng & Bedos, 2000;Humphreys, 2006). Biofilms of EPS within the phreatic zone are known to provide a key food source for subterranean invertebrates, which in turn help support the ecosystem and further EPS production (Bärlocher & Murdoch, 1989;Humphreys, 2006). Microbial activity in modern groundwater has also been documented (Dogramaci et al., 2017;Fellman et al., 2014;Siebers et al., 2020), although the role of the microbes in mineralisation has not previously been established. The association of EPS within dolocrete and the presence and diversity of the stygofauna suggest that dolocrete aquifers provide a unique physicochemical environment that merits further research into the links between biological processes and mineral formation in groundwater.

| CON CLUS IONS
Our study investigated the association between microbial EPS, clay and carbonate minerals within dolocrete profiles in arid northwest Australia. The results demonstrate a close link between the microbial EPS and mineralisation of clay minerals and dolomite within both pedogenic and groundwater dolocrete horizons, indicating microbially influenced mineralisation. Replacement and replication of EPS by authigenic clay minerals have preserved the EPS alveolar structure, which is observed throughout the profiles, encasing dolomite crystals. The accumulation of ions and then the development of amorphous nanoparticles in the EPS provide seeds for mineral nucleation.
The close textural association of clay-mineralised EPS and dolomite suggest that both minerals formed from the same initial amorphous substrate in the EPS (Bontognali et al., 2010;Perri et al., 2018).
Dolomite is also observed to have formed directly within an Mg-clay matrix, including in montmorillonite and Mg-silicate (t-smectite and/ or sepiolite), more commonly within the pedogenic carbonate horizon as opposed to groundwater dolocrete. Both EPS and clay minerals, in particular smectite minerals, may provide viscous properties, as well as negatively charged surfaces, that can overcome kinetic effects and promote dolomite nucleation (Fernández-Díaz et al., 1996;Liu et al., 2019;Wanas & Sallam, 2016). Local physicochemical changes, such as small increases in alkalinity, associated with the degradation of EPS or microbial activity, were likely important for both clay and dolomite formation (Braissant et al., 2007).
High microbial activity frequently documented from groundwater systems (Griebler & Lueders, 2009;Petrash et al., 2021) and the abundance of EPS textures observed here within the dolocretes, suggest that microbial influence on mineral authigenesis is a common process that likely occurs in similar environments worldwide.
Given that Mg-bearing clays are ubiquitous alongside dolomite in alkaline Mg-rich settings, more detailed attention should also be given to the roles they may have in driving dolomite precipitation.
Further microanalysis, using high-resolution and higher magnification tools, such as transmission electron microscopy, may provide further insight into the nature of mineral nucleation upon and within the various substrates. The catalytic role of clay minerals and EPS for dolomite precipitation may help explain the distribution of dolocrete in many inland drainage systems with low hydraulic gradients and has implications for models of groundwater dolocrete formation. The occurrence of the EPS-mineral textures in dolocretes documented here has implications for other Mg-rich alkaline lacustrine and groundwater environments worldwide. The outcomes of this study demonstrating the strong biogenic influence on dolomite formation as well as clay authigenesis should lead to a better understanding of low-temperature dolomite formation generally as well as the interpretation of mineral-derived information for palaeoenvironmental research.

CO N FLI C T O F I NTER E S T S TATEM ENT
The authors declare that they have no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The majority of data supporting the findings of this study are available within the article and its Supporting Information. Core scan hyperspectral data are publicly available via the AuScope portal http://portal.ausco pe.org.au/ and can be accessed under Boreholes and National Virtual Core Library data using the borehole IDs (GD13EA0009, GD13EA0010 and GD13EA0015). Further mineralogical data are available on request to the corresponding author.