Seasonal biogeochemical variations in a modern microbialite reef under early Earth-like conditions

Abstract

Microbialites are organosedimentary structures that have existed since the Precambrian and have endured through geological time, serving as archives of Earth’s environmental evolution. Today, they persist in only a few environments markedly different from those in which they first arose. Here, we report a modern microbialite reef in Laguna Pozo Bravo (Puna region, Argentina), exposed to high radiation, low oxygen pressure, and volcanic inputs reminiscent of early Earth. Through physicochemical, mineralogical, spectroscopic, electron microscopy, and metagenomic analyses, we identified diverse microbial communities with metabolic capacities that induce mineralisation. Seasonal environmental fluctuations drive cyclical changes in community composition, producing potential mineralisation patterns. Our findings suggest that carbon fixation and the metabolic drivers of alkalinity in microbialites may have evolved over time. Moreover, the variability in prokaryotic compositions among modern microbialites demonstrates that carbonate precipitation is governed by metabolic potential rather than taxonomy, reinforcing their role as dynamic records of environmental conditions.

Introduction

The Altiplano-Puna plateau, located in the central Andes mountains of South America, is comprised of numerous wetlands and salt flats. Due to the high altitude and volcanic activity of the area, these environments present extreme conditions. These include low oxygen pressure, intense solar radiation, low rates of precipitation, high rates of evaporation, pronounced seasonal and daily temperature fluctuations, strong winds, and elevated arsenic concentration (up to 230 mg L^–1)¹. Together, these factors make the region a valuable model for primitive Earth^2,3. Consequently, the study of the microbial ecosystems inhabiting these unique environments could provide insights into the adaptations and survival strategies of early complex life, as well as uncover the geobiological processes operating in Earth’s earliest biological structures.

The microbial ecosystems of the Altiplano-Puna are complex associations of microorganisms with minerals, mainly halite (NaCl), gypsum (CaSO₄-2H₂O), calcite (CaCO₃), and aragonite (CaCO₃)⁴. These ecosystems can be classified, based on community organisation and extent of mineralisation, into biofilms, microbial mats, and microbialites, among others¹. Biofilms are microbial communities attached to a surface forming a matrix of extracellular polymeric substances (EPS)⁵ that binds the cells together, protecting them from adverse conditions and facilitating their survival⁶. Microbial mats are stratified biofilms that can grow up to tens of centimetres thick. Their layers comprise species from different microbial guilds whose activity is governed by the availability of light, oxygen, nutrients (e.g. organic carbon, nitrogen, phosphorus), and electron donors and acceptors (e.g. sulphide, sulphate, iron)⁷. The interplay of metabolic activity, molecular diffusion, and light attenuation generates steep vertical gradients of metabolites, nutrients, and light, which create distinct ecological niches at different depth intervals⁸. Microbial mats have been around for more than 3500 Ma; however, the emergence of grazing metazoans and early macrophytes, together with environmental, chemical, and tectonic changes, have potentially caused their decline from 1000 Ma⁸. Today, these ecosystems develop in a wide range of environments, from temperate freshwater lakes, intertidal flats, and coastal lagoons to hypersaline and geothermal habitats like those in the Altiplano-Puna region^4,9.

Some biofilms and microbial mats have the capacity to lithify, forming organosedimentary structures known as microbialites¹⁰. This process, described in detail by Dupraz et al.^11,12, requires two components: first, an alkalinity engine; and second, an organic matrix comprised of EPS that provides a template for carbonate nucleation. The alkalinity engine is any process that can increase carbonate alkalinity, indirectly promoting carbonate precipitation. In other words, the alkalinity engine is any process that increases the calcium carbonate saturation index (SI), favouring precipitation. When the alkalinity engine is driven by microbial metabolic activities, the mineralisation is biologically-induced¹³. On the contrary, when the alkalinity engine is driven by environmental conditions, the mineralisation is biologically-influenced¹². In the biologically-induced mineralisation, photosynthesis and sulphate reduction are the microbial metabolic activities that increase the pH promoting carbonate precipitation, whereas aerobic respiration, fermentation, and sulphide oxidation are the metabolisms that decrease the pH triggering carbonate dissolution¹¹. The net carbonate precipitation will depend on the balance between these metabolic activities as well as their temporal and spatial variations. In the biologically-influenced mineralisation, two main physicochemical processes lead to carbonate precipitation: evaporation of water and carbon dioxide degassing¹². In both types of mineralisation, the organic EPS matrix serves as a template for mineral nucleation, but only when specific intrinsic conditions are met. The matrix contains abundant functional groups (hydroxyl and/or carboxyl groups) that bind Ca²⁺ or Mg²⁺ ions, inhibiting the precipitation of carbonate minerals. Therefore, to enable calcium carbonate precipitation, the Ca²⁺-binding capacity of the EPS matrix must be diminished or neutralized. This can occur through processes such as EPS decomposition, alteration, or saturation with Ca^2+11,12. Despite our current understanding of microbially-mediated mineralisation, further data on community composition across different temporal intervals and spatial scales, as well as detailed characterisation of mineral features (e.g. mineralogy, morphology) linked to microbial metabolism, are necessary to fully uncover past and present microbe–mineral interactions.

Microbialites can be classified as stromatolites¹⁴, thrombolites¹⁵, dendrolites¹⁶, and leiolites¹⁷ according to their laminated, clotted, shrub-like, and structureless mesostructures, respectively¹⁸. In general, microbialites are carbonate-rich in composition, although some are siliceous or even evaporitic¹⁹. The different types of microbialites result from different types of biofilms and microbial mats²⁰, along with changes in environmental conditions^21,22. Microbialites extend in the geological record from the Precambrian (~3500 Ma) to the present. Throughout Earth’s history, they experienced periods of expansion (between 2800 and 1000 Ma) and distinct periods of decline at approximately 2000 Ma, 1000 Ma, and 675 Ma, driven by major environmental and ecological changes^{19,23,24,25,26}. Today, modern microbialites are distributed in few marine and continental environments. These include Hamelin Pool²⁷ and Clifton Lake²⁸ in Australia, Satonda Crater Lake²⁹ in Indonesia, Pavilion Lake³⁰ in Canada, Great Salt Lake³¹ in USA, Highborne Cay^32,33 in The Bahamas, Alchichica Lake^34,35,36 and the Trans-Mexican Volcanic Belt³⁷ in Mexico, and the Altiplano-Puna⁴ in Argentina and Bolivia.

In recent years, explorations have been conducted in the Altiplano-Puna⁴ to identify and study extreme microbial ecosystems. Although numerous microbial mats and microbialites have been reported in the region¹, most remain undiscovered or poorly studied, having been characterized either geologically or microbiologically, with few investigations integrating both approaches. Laguna Pozo Bravo, in Salar de Antofalla (Argentina), harbours modern microbial mats and microbialites under early Earth-like conditions. Lencina et al.³⁸ recently described the morphology and internal structures of the microbialites inhabiting this environment, and hypothesized about the role of environmental factors in determining these characteristics. However, a full understanding of microbialite formation also requires the study of the biological communities and their responses to environmental variability. In this work, we present an in-depth biogeochemical characterisation of the microbial ecosystems in Laguna Pozo Bravo, combining physicochemical analyses, mineralogy, bulk geochemistry, scanning micro X-ray fluorescence (µXRF), Raman spectroscopy, scanning electron microscopy (SEM), 16S rRNA (16S ribosomal RNA) amplicon sequencing, and metagenomic analyses. Unlike other studies of modern microbialites, this interdisciplinary work was conducted over several years, capturing seasonal and interannual dynamics (see “Methods”). This integrative dataset reveals how microbial communities inhabiting this unique ecosystem adapt to seasonal environmental fluctuations and drive mineral precipitation processes and biogeochemical cycles, insights that are valuable for reconstructing the geobiological processes that shaped ancient sedimentary deposits on Earth.

Results and discussion

Laguna Pozo Bravo presents physicochemical conditions reminiscent of early Earth

Laguna Pozo Bravo (Fig. 1a) is a small (500 m long, 52 m wide) shallow (~2.5 m deep) lagoon located at the foot of basaltic lava deposits in the northwest region of Salar de Antofalla (25° 30’ 49.82” S, 67° 34’ 38.92” W). Certain physicochemical conditions of this lagoon align with those of primitive Earth. Due to its high altitude (3330 m above sea level), surface pressure at Pozo Bravo remains consistently low (613.9–618.2 hPa), ~60% of average sea level pressure³⁹ (Supplementary Fig. S1). This reduced pressure leads to a partial pressure of oxygen between 128.9 and 129.8 hPa, considerably lower than the ~210 hPa at sea level, and comparable to oxygen levels estimated for the Late Paleozoic (Devonian–Carboniferous), when atmospheric oxygen may have transiently reached 60–100% Present Atmospheric Level (PAL) (~126–210 hPa)^2,40. Although the partial pressure of oxygen was much lower in the Precambrian after the Great Oxidation Event (1–10% PAL, ~2.1–21 hPa) and the Lomagundi Event (5–20% PAL, ~10.5–42 hPa), the oxygen regime at Pozo Bravo provides a useful modern analogue for studying microbial ecosystems under low atmospheric oxygen conditions. The UV index is persistently elevated, with high values (~6–9) in winter and extreme levels (~17–19) in summer (Supplementary Fig. S2). These values are among the highest measured on modern Earth, highlighting the extreme solar radiation of the region. While UV levels during the Archean (~4000–2500 Ma) would have been greater (>100) due to a negligible ozone layer⁴¹, Pozo Bravo still serves as a valuable modern analogue for studying microbial life under extreme UV conditions that may have prevailed on early Earth. Land surface temperatures at Pozo Bravo show substantial fluctuations between day and night (ΔT up to 49.8 °C in summer) and across seasons (Supplementary Fig. S3). Such pronounced diurnal and seasonal swings likely resemble those experienced on the early Earth due to a thinner atmosphere, lower greenhouse gas levels, and limited ocean coverage⁴². The lake water is alkaline (pH ~8), highly saline (over 35,000 mg L^–1), and rich in elements of volcanic origin, including sulphate (1920–1840 mg L^–1), boron (10.5–152 mg L^–1), lithium (21.3–101 mg L^–1), arsenic (0.07–0.86 mg L^–1), and manganese (0.05–0.60 mg L^–1) (Table 1). This water chemistry shares certain similarities with the metal-rich conditions of Precambrian Earth’s oceans, which likely held high concentrations of volcanic-derived elements^43,44,45,46. Overall, these extreme physicochemical conditions make Laguna Pozo Bravo a compelling model for ancient Earth, offering insights into adaptations, survival strategies, and geobiological mechanisms underlying Earth’s first life forms.

Fig. 1: Laguna Pozo Bravo harbours a modern microbialite reef analogue for ancient Earth biostructures.

a Location of Laguna Pozo Bravo in Salar de Antofalla, Catamarca (Argentina), and the sampling area. The sampling area (south-west margin of the lagoon) is highlighted with a red frame. b North view of the microbialite reef from the middle-west margin of the lagoon. c South view of the microbialite reef from the middle-west margin of the lagoon. d North view of the microbialite reef from the sampling area. e Transition of microbial structures along the intertidal zone in the sampling area. BS Biostrome, LM Lithified mats (carbonate pavement), BH Bioherms, SM Soft mats.

Table 1 Physicochemical parameters of surface water from Laguna Pozo Bravo

Laguna Pozo Bravo is a unique environment not only because it presents extreme physicochemical conditions, but also because these conditions vary between seasons (Table 1). In winter, when the water temperature is lower (9 °C) and the water level is higher (26 cm) due to reduced evaporation, the dissolved oxygen (DO) level is remarkably high at 171% air saturation. This high oxygen level corresponds with lower concentrations of ions (sodium, magnesium, chlorine, potassium, calcium) and minerals (e.g. calcite and aragonite), resulting in lower conductivity (150.4 mS cm^–1) and total hardness (8460 mg CaCO₃ L^–1). In contrast, during summer, the elevated water temperature (23 °C) and lower water level (7 cm), driven by high evaporation rates, result in reduced dissolved oxygen (27.5% air saturation). The increased concentration of ions and minerals under these conditions leads to a rise in both conductivity (175.2 mS cm^–1) and total hardness (10,250 mg CaCO₃ L^–1). Seasonal changes are also evident in biochemical oxygen demand (BOD₅) and chemical oxygen demand (COD). Both BOD₅ and COD values increase in summer to 230 mg L^–1 and 705 mg L^–1, respectively, compared to their winter levels of 161 mg L^–1 and 475 mg L^–1. This indicates a likely seasonal boost in microbial activity and biomass during the warmer months.

Microbial mats in Pozo Bravo exhibit diel vertical geochemical gradients

In Laguna Pozo Bravo, the combination of minimal physical disturbance, abundant volcanic-derived element inputs, strong light availability, and low sedimentation rates, provides ideal conditions for the growth of microbial mats, especially along the lagoon’s southwest margin (Fig. 1a, d, e), where they exhibit cerebroid- or snake-like morphologies due to biogenic gas accumulation (Fig. 2a). In the absence of considerable physical disruption, microbial mats organise into layers, each containing microorganisms with distinct metabolic activities. The microbial mats in Pozo Bravo display three main layers: a green surface layer, a reddish-pink middle layer, and a dark brown underlying layer (Fig. 2b).

Fig. 2: Pozo Bravo microbial mats show spatial and temporal heterogeneity in their chemical composition.

a Overview of Pozo Bravo soft microbial mats displaying cerebroid- or snake-like structures. b Close-up look at a cross-section of a soft microbial mat exhibiting a green surface layer (G), a reddish-pink middle layer (R), and a dark brown underlying layer (D). c Spatial distribution of elements (deconvoluted counts) in a freeze-dried soft microbial mat sample determined by µXRF spectrometry. The area of µXRF scanning (4.2 mm × 4.3 mm) is marked with a black frame, and the dotted lines highlight the green, reddish-pink, and dark brown layers. d Oxygen and hydrogen sulphide profiles (mean values, n = 3) of a soft microbial mat, measured in situ during the middle of the day (from 13:00 to 16:00; yellow symbols) and night (from 20:00 to 01:00; blue symbols).

Microbial mat layered structures are shaped by steep vertical geochemical gradients generated by environmental factors and the metabolic activities of mat organisms. In the Pozo Bravo microbial mats, oxygen is detectable in the green surface layer, where its concentration rises during the day (370.69 µM at 0.1 cm depth) and drops at night (24.00 µM at 0.1 cm depth) (Fig. 2d), suggesting oxygenic phototrophic activity. With oxygen available mainly in this layer, aerobic respiration may be confined to it. A small increase in oxygen concentration is also detected during the day (16.05 µM at 0.45 cm depth) at the interface between the reddish-pink and dark brown layers (Fig. 2d), possibly reflecting oxygenic phototrophic activity driven by far-red light. Manganese is mainly localised between the green and reddish-pink layers (Fig. 2c and Supplementary Fig. S4), which may suggest manganese oxidoreduction in this zone. Sulphur is concentrated within the reddish-pink and dark brown layers (Fig. 2c and Supplementary Fig. S4), which correlate with the observed hydrogen sulphide profiles (Fig. 2d). In the reddish-pink layer, hydrogen sulphide levels decrease during the day (4.80 µM at 0.3 cm depth) but increase again at night (11.88 µM at 0.3 cm depth), possibly reflecting anoxygenic phototrophic activity driven by sulphide oxidation. Nevertheless, since pH could not be simultaneously measured during hydrogen sulphide profiling, day-night variations in hydrogen sulphide concentrations due to pH-dependent speciation cannot be excluded. The fact that sulphur concentrates in the dark brown layer may further suggest sulphur oxidoreduction activity in this area (Fig. 2c and Supplementary Fig. S4). Arsenic also accumulates in the dark brown layer, implying potential arsenic oxidoreduction activity within this anaerobic region (Fig. 2c). The evaluation of elemental composition in the mapped area of a soft microbial mat sample can be found in Supplementary Table S1.

The observed spatial and temporal variations in chemical composition suggest distinct microbial communities within each mat layer (Fig. 2). The green surface layer is likely dominated by oxygenic phototrophs, which couple inorganic carbon assimilation to light energy, oxidising water and producing oxygen as a byproduct. This layer thereby would provide an ideal micro-environment for aerobic and facultative aerobic heterotrophs. Aerobic heterotrophs gain energy by oxidising organic carbon exudates through oxygen respiration, while facultative aerobic heterotrophs can switch to using other terminal electron acceptors when oxygen is limited. The reddish-pink layer is potentially dominated by anoxygenic phototrophs, which use sulphide as an electron donor for photosynthesis and release elemental sulphur as a byproduct. The dark brown layer is likely dominated by anaerobic heterotrophs and fermenters. Anaerobic heterotrophs, mainly sulphate reducing bacteria (SRB), oxidise organic carbon using a range of terminal electron acceptors, while fermenters use organic carbon as both electron donor and acceptor. Sulphide-oxidising bacteria, which oxidise reduced sulphur compounds with oxygen or nitrate, might also be abundant in the dark brown layer.

Both eukaryotes and prokaryotes may participate in the organomineralisation process

In Laguna Pozo Bravo, the formation of lithified mats and microbialites (Fig. 1b–e) is driven by the constantly elevated calcium carbonate saturation index (aragonite SI is 2.78 in winter and 3.36 in summer; calcite SI is 2.94 in winter and 3.54 in summer), which indicates supersaturation of this mineral throughout the year, as well as the presence of abundant EPS, which may serve as a template for carbonate nucleation (Fig. 3). The elevated calcium carbonate saturation index (Supplementary Tables S2 and S3) likely results from environmental conditions like water evaporation, reflected in the seasonal variations in water column height (Table 1). Nevertheless, the spatial and temporal separation of metabolic activities observed in the microbial mats (Fig. 2) may also be responsible for local differences in the saturation index which promote carbonate precipitation. Consequently, mineralisation in Pozo Bravo may be both biologically induced and influenced. Chemical analyses conducted on the microbial mats (Table 2) revealed increased concentrations of arsenic, calcium, magnesium, iron, boron, and manganese compared to their levels in the water (Table 1). This suggests that the EPS matrix of the mats acts as a chelator for cations. Therefore, the precipitation of calcium carbonate is likely enhanced by mechanisms that weaken the capacity of the EPS matrix to bind cations. For instance, extreme UV radiation (Supplementary Fig. S2) may contribute to partial EPS destruction through Maillard browning reactions (involving the reaction of sugars and aminoacids)⁴⁷, whereas the activity of heterotrophic microorganisms may lead to microbial decomposition of the EPS. Additionally, the notably higher concentrations of calcium and magnesium in the mats (Table 2) compared to the water (Table 1) suggest that saturation of the EPS cation-binding capacity might also facilitate calcium carbonate precipitation¹².

Fig. 3: Diatoms, cyanobacteria, and other prokaryotic organisms may participate in the lithification process of Pozo Bravo microbial mats.

a–p Scanning electron microscopy images from microbial mat samples. Different recognized structures are marked with an arrow: Orange, EPS serving as a template for mineral precipitation; Red, mineral structure; Green, filamentous cyanobacterium; Yellow, bacillus; Purple, spirochete; Blue, diatom (genera identified morphologically: Navicula in a and b; Halamphora in c, d and k; Nitzschia in e).

Table 2 Physicochemical parameters of a soft microbial mat from Laguna Pozo Bravo

The organomineralisation in Pozo Bravo has been further analysed through SEM of a microbial mat sample (Fig. 3). The SEM images obtained revealed the presence of diatoms (Bacillariophyceae), filamentous cyanobacteria, and other prokaryotic cells (cocci, bacilli, and spirilla) forming aggregates embedded in the EPS matrix, where abundant mineral precipitation takes place. These observations suggest the potential contributions from both prokaryotic and eukaryotic microorganisms in the mineralisation processes that lead to the formation of a lithified microbial mat or microbialite. Nevertheless, as outlined by Petryshyn et al.⁴⁸, it is important to acknowledge that the microorganisms observed could play different roles, ranging from builders (e.g. diatoms and cyanobacteria) that are responsible for the construction of the structure, trapping and binding particles or driving mineral precipitation via their metabolism, to tenants that reside within the structure but are not directly responsible for its accretion or lithification, or squatters that are passively incorporated to the structure post-lithification. Based on morphological characteristics, we could identify at least three genera of diatoms: Navicula (Fig. 3a, b), Halamphora (Fig. 3c, d, and k), and Nitzschia (Fig. 3e), previously reported to compose microbial mats and microbialites in the central Andes region⁴.

Environmental conditions and microbial activity shape microbialite structure and composition

In Laguna Pozo Bravo, modern microbialites extend along the entire margin, forming a reef analogue for early Earth biostructures (Fig. 1b–e). These microbialites vary in height from 0.05 to 0.65 m and display a wide range of macrostructures, including domical, discoidal, and tabular shapes (Fig. 4a). However, the most distinctive feature of these microbialites is their internal organization, which transitions gradually from a thrombolite core to a dendrolite middle layer and, finally, to a stromatolite surface³⁸ (Fig. 4b). Thrombolite and dendrolite mesostructures are characterized by lighter colouration, whereas stromatolite mesostructures appear darker (Fig. 4b, c). Consequently, we have divided the internal organization of the microbialites into two distinct zones: (I) an internal light area and (II) an external black rim. Notably, these two zones are also present in the lithified microbial mats, suggesting that both mats and microbialites underwent similar mineralisation processes.

Fig. 4: Pozo Bravo microbialites are morphologically, microstructurally and mineralogically heterogeneous.

a Overview of Pozo Bravo bioherms displaying diverse external morphologies (macrostructure). DM, domical shape; DS, discoidal shape; TB, tabular shape. b Cross-section of domical bioherm, showing an external dark rim followed by an internal light area. The division between the two areas is indicated by a green dotted line. c On the left, a polished hand specimen of a domical bioherm showing the dark external rim (above) and the light internal area (below). On the top-right, a photomicrograph of the dark external rim (plain-polarised light) with thin layers of dark reddish oxides, corresponding to manganese oxides intergrown with calcite. On the bottom-right, a photomicrograph of the light internal area (cross polars) dominated by calcite with detrital grains of plagioclase (Pl) and volcanic rock lithic fragments (Lv). d Representative Raman spectra in the dark external rim characterised by ramsdellite and calcite peaks. e Representative Raman spectra in the light internal area characterised by calcite bands.

A Raman spectroscopy analysis of a microbialite sample (Fig. 4d, e) revealed that both distinct internal zones are primarily composed of calcite (characteristic Raman modes at 282, 711, and 1085 cm⁻¹). However, the dark external rim also appears to contain ramsdellite (bands at 573–577 and 647–654 cm⁻¹)⁴⁹, a black Mn(IV) oxide mineral that probably explains the colour difference between zones and imparts the distinctive outer colouration of lithified mats and microbialites. The presence of ramsdellite in the outermost zone could be attributed to microbial activity, as microorganisms are well-known for mediating the precipitation and transformation of Mn(IV) oxide minerals^50,51,52. The spatial association of manganese accumulation at the oxic–anoxic transition between the green and reddish-pink mat layers (Fig. 2c and Supplementary Fig. S4) supports a biotic origin for ramsdellite in Pozo Bravo. In this region of the mats, microbial oxidation of dissolved Mn(II) may proceed by oxygen-dependent multi‑copper oxidases, superoxide-mediated pathways, sulphur-coupled oxidation, nitrate reduction, ferric iron-mediated oxidation, or anoxygenic photosynthesis, leading to Mn(IV) oxide precipitation⁵³. Abiotic oxidation of Mn(II) may also occur via molecular O₂ or UV photooxidation in Pozo Bravo⁵³. However, because abiotic Mn(II) oxidation by O₂ is far slower than biotic pathways⁵⁴ and manganese accumulates poorly in the mat’s photic surface zone (Fig. 2c and Supplementary Fig. S4), a purely abiotic origin for the ramsdellite is unlikely. In the Archean oceans, low concentrations of Mn(II) oxidants (like oxidised sulphur, nitrate, and iron oxides) and high concentrations of labile Mn reductants (like ferrous iron) would have limited Mn(II) oxidation in the absence of O₂, making Mn(II) oxidation via molecular O₂ or superoxide produced through photosynthesis the most plausible mechanisms⁵³. Consequently, ramsdellite could prove useful as a biosignature of oxygenic photosynthesis in Earth’s earliest biostructures, but interpretations must remain cautious, since abiotic oxidation of Mn(II) could have also been catalysed by UV photooxidation.

Powder X-Ray diffraction (XRD) analyses of microbial mat (soft and lithified) and microbialite samples (Supplementary Table S4 and Supplementary Figs. S5 and S6) provided further insights into their mineralogy, and the environmental and microbial processes shaping it. Both types of microbial mats are predominantly composed of magnesium calcite, with minor contributions from silicate minerals such as quartz (SiO₂) and plagioclase feldspars. In the soft mats, the plagioclase component consists of andesine ((Ca,Na)(Al,Si)₄O₈) and albite (NaAlSi₃O₈), while the lithified mats contain albite and anorthite (CaAl₂Si₂O₈). The differential abundance of magnesium calcite in the soft microbial mat layers (Supplementary Table S4) suggests that carbonate precipitation mainly occurs in the green surface and reddish-pink middle layers, where oxygenic photosynthesis and anoxygenic photosynthesis coupled with sulphate reduction, respectively, increase carbonate alkalinity, favouring precipitation. This pattern is consistent with the slight calcium accumulation observed principally in the reddish-pink middle layer of the soft microbial mat sample (Fig. 2c). Together, these observations support the idea that net carbonate precipitation depends on the balance between different metabolic activities as well as their temporal and spatial variations. Contrary to microbial mats, microbialites are primarily composed of calcite throughout their structure, with minor contributions from hydrotalcite (Mg₆Al₂CO₃(OH)₁₆·4H₂O), illite ((K,H₃O)(Al,Mg,Fe)₂(Si,Al)₄O₁₀[(OH)₂,(H₂O)]), and silicate minerals, including quartz, albite, andesine, and labradorite ((Ca,Na)(Al,Si)₄O₈). Microscopic examination of the microbialite samples (Fig. 4c) suggests that quartz and plagioclase are detrital clasts, likely sourced from adjacent volcanic rocks. This volcanic input may also account for the presence of hydrotalcite, which is commonly formed as an alteration product of basalts in alkaline water environments⁵⁵. The presence of low-magnesium calcite in the microbialites may also be the result of post-depositional alteration, where magnesium ions in magnesium calcite are replaced by calcium ions, leading to the formation of a more stable mineral. All these findings suggest that the mineralogical composition of microbial mats and microbialites reflects a combination of biogenic and detrital inputs, with microbialites further shaped by diagenetic transformations.

The carbon and nitrogen systematics of soft microbial mat and microbialite samples (Supplementary Table 5) revealed notable differences in the content of total organic carbon (TOC) and total nitrogen (TN). The soft microbial mat sample exhibited higher levels of TOC (2.1% of dry weight) and TN (0.42%), compared to the microbialite sample, which had only 0.35% TOC and no detectable TN. While this pattern could reflect greater microbial activity and organic matter inputs in the microbial mats, it may also arise from post-depositional processes in the microbialites, such as remineralisation of organic matter, diagenetic loss of nitrogen, and a decrease in the proportion of organics due to mineral precipitation during lithification. The isotopic composition of organic carbon (δ¹³C_org) measured in the soft microbial mat (−18.2‰) and microbialite (−23.4‰) samples primarily suggests carbon fixation via the reductive pentose phosphate (Calvin−Benson−Bassham) cycle, employed by oxygenic (e.g. diatoms and cyanobacteria) and anoxygenic (e.g. purple sulphur bacteria) phototrophs^56,57,58,59. Nevertheless, the notable δ¹³C_org variation (~5‰) observed between samples may reflect additional contributions from alternative carbon fixation pathways and/or post-depositional alteration, including heterotrophic remineralisation and selective preservation processes. Minor contributions from the reductive acetyl-CoA (Wood–Ljungdahl) pathway would have shifted δ¹³C_org towards more negative values, whereas small inputs from the reductive tricarboxylic acid, 3-hydroxypropionate, 3-hydroxypropionate-4-hydroxybutyrate, or dicarboxylate-4-hydroxybutyrate cycles would have yielded less negative δ¹³C values, collectively modifying the overall isotopic signature of the bulk organic carbon pool^59,60,61. The bulk nitrogen isotopic composition (δ¹⁵N) of the soft microbial mat sample (0.4‰) suggests nitrogen fixation, likely mediated by cyanobacteria, through nitrogenase enzymes⁶². These results are consistent with the phototrophic activities inferred from the microelectrode measurements (Fig. 2d) and the organisms identified through SEM imaging (Fig. 3) of the microbial mats.

The microbialite reef in Pozo Bravo shows seasonal variations in microbial community composition

A molecular diversity analysis based on the small subunit rRNA gene of Bacteria and Archaea was conducted over an annual cycle to study the prokaryotic composition of the water, microbial mats, and microbialites (Fig. 5). The results revealed significant differences in community composition (PERMANOVA, F = 6.9934, R² = 0.60847, p ≤ 0.001, n_obs = 11) between the water and the benthic microbial ecosystems (Fig. 5a, c). The microbial community in the water is dominated by two phyla, Bacteroidota (60.91–88.31%) and Pseudomonadota (11.10–37.69%), largely represented by the classes Flavobacteriia (60.49–88.12%) and Gammaproteobacteria (10.40–35.83%). These classes are composed mainly of heterotrophic and halophilic bacteria from the genera Psychroflexus, Owenweeksia, and Halomonas. The prevalence of these taxa may reflect organic matter processing in the water and suggests adaptations to the hypersaline conditions and elevated organic content of the environment. In contrast, microbial mats and microbialites support more diverse and complex microbial communities (Fig. 5b). These communities are dominated by several phyla, including Pseudomonadota (24.34–63.67%), Bacteroidota (3.42–38.35%), Bacillota (0.18–23.64%), Cyanobacteriota (1.02–14.31%), Spirochaetota (0.02–12.21%), Chloroflexota (0.53–8.29%), Deinococcota (0.25–5.49%), Planctomycetota (1.44–4.46%), and Verrucomicrobiota (0.08–2.93%). Pseudomonadota is mainly represented by the classes Gammaproteobacteria (4.52–56.02%), Alphaproteobacteria (6.64–20.70%), and Deltaproteobacteria (0.64–10.87%). These groups include previously reported anoxygenic phototrophs such as purple sulphur bacteria (Thiocapsa, Thiococcus, Thiocystis, Thioflavicoccus, Thiohalocapsa, Thiorhodococcus, Thiorhodovibrio, Halochromatium, Ectothiorhodospira, and Halorhodospira) and purple non-sulphur bacteria (Rhodomicrobium, Roseospira, Rhodospira, and Rhodovibrio), heterotrophic sulphur oxidisers (Sulfurimonas, Sulfurifustis, Sedimenticola, Desulfobacula, Desulfobacter, Desulfovibrio, and Desulfuromonas), sulphur reducers (Desulfobacter, Desulfobacula, Desulfovibrio, Desulfuromonas, Desulfuromusa, Shewanella, and Geobacter), and metal/metalloid reducers potentially capable of reducing iron, arsenic, manganese, cobalt, selenium, or chromium (Desulfuromonas, Desulfuromusa, Desulfovibrio, Geobacter, and Shewanella). The phyla Bacteroidota, Bacillota, Spirochaetota, Deinococcota, Planctomycetota, and Verrucomicrobiota are primarily represented by the classes Bacteroidia (0.11–10.61%), Cytophagia (0.48–8.41%), Flavobacteriia (0.62–18.84%), Saprospiria (0.20–8.17%), Deinococci (0.25–5.48), Bacilli (0.00–23.08%), Clostridia (0.12–12.19%), Phycisphaerae (0.30–2.68%), Spirochaetia (0.02–12.19%), and Opitutae (0.00–1.24%). These groups are predominantly composed of heterotrophic bacteria, with some likely capable of sulphur reduction (Bacteroides, Marinifilum, Anaerophaga, Clostridium, Desulfotomaculum, Halanaerobium, Spirochaeta, and Sediminispirochaeta), sulphur oxidation (Cytophaga, Clostridium, Flavobacterium, and Spirochaeta), ammonification (Bacteroides and Anaerophaga), nitrite reduction (Flavobacterium and Maribacter), and denitrification or nitrogen fixation (Clostridium, Desulfotomaculum, Halanaerobium, and Bacillus). Cyanobacteriota is represented by oxygenic phototrophs from different genera, such as Halothece, Coleofasciculus, Halomicronema, and Synechococcus. Among these genera, Coleofasciculus and Synechococcus include some species potentially capable of nitrogen fixation. Meanwhile, Chloroflexota, primarily represented by Anaerolineae (0.33–5.29%), includes green non-sulphur bacteria (Candidatus Chlorothrix) which may be capable of anoxygenic photosynthesis. The prokaryotic composition observed in the microbial mats and microbialites aligns with the microbial activities inferred from the microelectrode measurements (Fig. 2d) and the bulk carbon and nitrogen isotopic compositions (Supplementary Table S5), including oxygenic and anoxygenic photosynthesis, carbon and nitrogen fixation, and sulphur, arsenic, and manganese oxidoreduction. The diversity of bacterial genera likely capable of these activities highlights the complex metabolic interactions driving the biogeochemical cycles within the benthic ecosystems. This diversity contrasts with that of the water column, which is dominated by a few potential heterotrophic taxa, reflecting the distinct ecological roles of these habitats. The functional potential of the genera identified in Laguna Pozo Bravo was inferred from the BacDive database⁶³.

Fig. 5: Prokaryotic diversity in the water, soft microbial mats, and microbialites changes throughout the seasons.

a Relative abundance of the twenty major prokaryotic phyla in the different seasons. A, Autumn; W, Winter; Sp, Spring; S, Summer. b Comparison of alpha diversity metrics (Shannon index) between samples throughout the different seasons. c Principal coordinate analysis (PCoA) based on Bray-Curtis dissimilarity between samples throughout the different seasons.

The molecular diversity analysis also revealed seasonal variations in the prokaryotic composition of microbial mats and microbialites. Cyanobacteria increase in abundance during periods of higher light availability and decrease when light levels are lower (Fig. 5 and Supplementary Fig. S7). In the microbial mats, they reach their highest relative abundances in summer, whereas in the microbialites, the peaks occur in autumn and spring. These differences may be related to the greater susceptibility of microbialites to desiccation, making summer conditions more extreme and thus less favourable for cyanobacterial proliferation. Green non-sulphur bacteria (GNSB) show a similar pattern, peaking in summer and spring, reflecting their strong dependence on light availability as well. Purple sulphur bacteria (PSB) also increase during months with higher light, but reach their maximum abundances primarily in autumn in both microbial mats and microbialites (Supplementary Fig. S7). This deviation from a mid-summer peak, as seen in cyanobacteria and GNSB, may be linked to the activity of SRB. These bacteria show greater abundances in spring and summer in the mats, and in autumn in the microbialites (Supplementary Fig. S7). Elevated levels of sulphide produced by SRB could promote the autumnal bloom of PSB. In contrast, cyanobacteria and GNSB do not depend on sulphide, which could explain why their seasonal trends align more directly with light intensity rather than the sulphur cycle. These findings suggest that multiple environmental factors, including light availability and the presence of redox-active species, drive the seasonal compositional shifts and metabolic interactions within these microbial ecosystems.

Symbiotic microorganisms regulate biogeochemical cycles that shape the structure and mineralisation of mats

A metagenomic analysis was performed on a soft microbial mat to explore the metabolic diversity of their microorganisms and to elucidate their roles in the biogeochemical cycles (Supplementary Table S6). The high relative abundance of genes encoding subunits of Photosystems I and II (psa and psb, respectively), along with the presence of puf genes associated with bacterial phototrophic reaction centres, suggests co-occurring oxygenic and anoxygenic photosynthesis within the microbial mat (Fig. 6a), as confirmed by microelectrode measurements. The presence of nif genes may indicate that the microorganisms in the mat fix atmospheric nitrogen through nitrogenase enzymes, as reflected by the bulk nitrogen isotopic composition (δ¹⁵N = 0.4‰) (Supplementary Table S5). Since vnf genes encoding vanadium-dependent nitrogenases were not identified, nitrogen fixation is likely carried out solely by molybdenum-dependent nitrogenases (nif). Regarding carbon fixation, the high relative abundance of cbb and prk genes suggests that the mat’s primary CO₂ assimilation occurs via the reductive pentose phosphate cycle, in line with the isotopic composition of organic carbon (δ¹³C_org = −18.2‰) (Supplementary Table S5), 16S rRNA data revealing abundant oxygenic phototrophs (cyanobacteria) in the mat community (Fig. 5a), and microelectrode measurements demonstrating strong oxygen production in the surface layer (Fig. 2d) consistent with active CBB-based photosynthesis. However, the presence of acl genes associated with the reductive tricarboxylic acid cycle, acs genes involved in the reductive acetyl-CoA pathway, and mcl and mcr bacterial genes related to the 3-hydroxypropionate bicycle suggests the potential operation of alternative carbon fixation routes, which could contribute to the observed isotopic range. Genes involved in the oxidation of hydrogen (hya), sulphur (sqr, fcc, sox, dsr, hdr, sor, soe, apr, sat, and tcdh), nitrogen (hao and nxr), arsenic (aio), and manganese (mnxG) compounds were also identified, indicating the capacity of the microbial mat community to engage in these processes. Some of them may even be obtaining energy through the oxidation of these compounds. Finally, the identification of genes involved in the reduction of sulphur (sat, apr, and dsr), nitrogen (narG, napA, nir, norB, and nosZ), iron (mtr) and arsenic (arr) compounds suggests that some microorganisms in the mat could use alternative electron acceptors to oxygen for respiration. In fact, the mtr system has been implicated not only in the reduction of iron but also in the reduction of manganese, cobalt, and chromium⁶⁴, expanding the possibilities for anaerobic respiration in Pozo Bravo. These findings align with the microelectrode measurements, which revealed a large anoxic section in the microbial mat where anaerobic microorganisms flourish.

Fig. 6: Microorganisms drive main biogeochemical cycles in Pozo Bravo soft microbial mats.

a Relative abundance in the metagenome (presented as reads per kilobase per million mapped reads, RPKM) of genes associated with hydrogen, carbon, nitrogen, oxygen, sulphur, manganese, iron, arsenic, and selenium cycling. b Metabolic potential of the metagenome-assembled genomes. Grey squares indicate the presence of a gene.

The reconstruction of thirteen metagenome-assembled genomes (MAGs) from the metagenomic contigs further expanded our understanding of the biogeochemical cycles in the microbial mats (Fig. 6b and Supplementary Tables S7 and S8). Of these thirteen, four are nearly complete and nine are of medium quality. However, they are estimated to have very low contamination, with seven MAGs having contamination levels less than or equal to 1% (see “Methods”). Four of these MAGs belong to the phylum Cyanobacteriota, with two (MAGs 7 and 8) identified as members of the family Geitlerinemaceae and the other two (MAGs 9 and 10) classified under the genus Halothece (Supplementary Table S8). MAGs 7 and 10 contain nif, napA, and norB genes, suggesting that cyanobacteria would play a key role in the nitrogen cycle, particularly in nitrogen fixation and denitrification. All four cyanobacterial MAGs possess genes for carbon fixation (cbb and prk) and sulphide oxidation (sqr). On the one hand, the presence of cbb and prk genes indicates that cyanobacteria drive carbon fixation through the reductive pentose phosphate cycle, which appears to be the primary carbon fixation process in the mat. On the other hand, the detection of the sqr gene suggests that cyanobacteria may use sulphide oxidation as an alternative electron source for photosynthesis, resembling some anoxygenic phototrophs. The ability of cyanobacteria to perform anoxygenic photosynthesis via sulphide oxidation is supported by the microelectrode measurements (Fig. 2d), which showed a decrease in the hydrogen sulphide levels during the day in the green surface layer, where cyanobacteria are found (Fig. 2b). Our results suggest the possibility of photosynthetic competition for sulphide between cyanobacteria and PSB. Such competition could further explain why PSB thrive in autumn, whereas cyanobacteria and GNSB are more abundant in summer. Interestingly, MAG 11 appears to be a phototroph capable of anoxygenic photosynthesis through sulphide oxidation. Its genetic potential also suggests that it fixes nitrogen and carbon via the reductive pentose phosphate cycle, complementing the activities of cyanobacteria. This MAG does not belong to the PSB but is instead classified within the Thiohalospiraceae family (Supplementary Table S8). Members of Thiohalospiraceae have been described exclusively as chemolithoautotrophs that obtain energy by oxidising inorganic sulphur compounds⁶⁵. Thus, the discovery of photosynthesis-related genes in this MAG may broaden the metabolic potential known for this family. Future isolation of an anoxygenic phototroph from Thiohalospiraceae would confirm this finding. MAGs 12 and 13, belonging to the order Desulfobacterales, are predicted to participate in both the reduction and oxidation parts of the sulphur cycle. MAG 13 also contains acs genes for carbon fixation via the reductive acetyl-CoA pathway, suggesting that non-phototrophic groups may contribute to primary production. Links between each gene and its associated metabolic pathway, EC number, and HMM profile are provided in Supplementary Table S9.

The integration of the metagenomic analysis with the biogeochemical data enabled us to construct a conceptual model of the Pozo Bravo microbial mats, detailing microbial stratification, predicted metabolic functions, and their roles in key biogeochemical cycles (Fig. 7). During the day, cyanobacteria and diatoms in the green surface layer perform oxygenic photosynthesis and assimilate inorganic carbon through the reductive pentose phosphate cycle. Certain cyanobacteria also fix atmospheric nitrogen and perform anoxygenic photosynthesis, releasing elemental sulphur instead of oxygen. The oxygen generated in the green surface layer supports aerobic and facultative aerobic organisms, including chemoorganotrophic eukaryotes (Supplementary Table S10), which consume the organic exudates. Some of these organisms oxidise reduced compounds of hydrogen, sulphur, nitrogen, arsenic, and manganese. Anoxygenic phototrophs in the deeper reddish-pink layer contribute to primary production by assimilating inorganic carbon via the reductive pentose phosphate cycle (PSB), the reductive tricarboxylic acid cycle (purple non-sulphur bacteria, PNSB), and the 3-hydroxypropionate bicycle (GNSB and PNSB). During anoxygenic photosynthesis, PSB use thiosulphate, sulphide, or elemental sulphur as electron donors, whereas GNSB and PNSB use organic compounds or hydrogen. In the bottom dark brown layer of the mat (Fig. 2b), anaerobic heterotrophs oxidise the organic exudates using iron, manganese, cobalt, chromium, or other oxidised compounds as terminal electron acceptors, while fermenters use organic carbon as both electron donor and acceptor. Once night falls, photosynthesis ceases, and the residual oxygen is rapidly depleted by respiration. As a result, the anoxic zone expands, creating favourable conditions for anaerobic heterotrophs. Among them, SRB reduce sulphate or other oxidised sulphur compounds, contributing to the regeneration of sulphide levels. Certain SRB assimilate inorganic carbon via the acetyl-CoA pathway, further linking the carbon and sulphur cycles in the mat. The metabolic activities observed in the mat play a crucial role in shaping its mineralisation. The spatial and temporal separation of these metabolisms, as observed in the mat, may create localised differences in the saturation index that influence where and when carbonate precipitation occurs. Seasonal shifts in community composition also alter the balance of microbial metabolisms, potentially leading to cyclical patterns of organomineralisation. Therefore, the dynamic interplay between microbial processes and environmental conditions ultimately determines the lithification of the mat.

Fig. 7: Schematic representation of Pozo Bravo soft microbial mat.

The illustration depicts the major microbial guilds and biogeochemical cycles in the different layers of the microbial mat.

Distinct prokaryotic compositions define modern microbialites around the world

To understand whether microbialites found in different locations around the world exhibit a uniform prokaryotic composition or whether their composition is influenced by spatial and environmental factors, we performed a geographically broad comparative analysis of modern microbialites from North America, South America, and Australia (Fig. 8). Our study reveals that these microbial structures do not share a similar prokaryotic composition; instead, their composition varies according to geographic location and environmental characteristics, a pattern also recently observed in the eukaryotic communities of microbialites worldwide⁶⁶. For instance, the freshwater microbialites from Pavilion Lake³⁰ contain 140 distinct prokaryotic genera characteristic of freshwater environments (Fig. 8b). Some members of these genera contribute to carbonate precipitation processes, including freshwater cyanobacteria (Aphanocapsa, Chroococcus, Oscillatoria, Cyanomargarita, and Tolypothrix) and sulphate reducers (Sulfuricaulis, Sulfurisoma, Sulfurivirga, Desulforhabdus, and Bilophila). The microbialites from Clifton Lake²⁸, which develop under poikilosaline conditions (i.e. widely fluctuating salinity), host 24 distinct prokaryotic genera, whose members are adapted to salinity variation. These include halophilic or halotolerant cyanobacteria (Gloeocapsa, Scytonema, and Chondrocystis), anoxygenic phototrophs (Chloracidobacterium), and sulphate reducers (Candidatus Allobeggiato, Desulfocella, Desulfospira, and Dissulfuribacter) that drive carbonate precipitation. Microbialites from Pozo Bravo and Socompa⁶⁷ lakes, which are exposed to similar conditions of hypersalinity, high radiation, low oxygen pressure, and slight alkalinity, share 8 distinct prokaryotic genera adapted to these extreme conditions (Fig. 8b). These genera include polyextremophilic anoxygenic phototrophs (Thiorhodovibrio) and sulphate reducers (Dethiosulfatibacter), which modify carbonate alkalinity and drive mineralisation. Despite the shared extreme conditions, the microbialites from Pozo Bravo host 47 distinct prokaryotic genera not found in Socompa with the data currently available. These include polyextremophilic anoxygenic phototrophs (Thiococcus, Thioflavicoccus, and Thiorhodococcus) and sulphate reducers (Pseudodesulfovibrio) involved in carbonate precipitation, as well as genera involved in organic matter decomposition or fermentation (Acetohalobium, Halanaerobaculum, Natronorubrum, Salinirubrum, Salimicrobium, Sporohalobacter, Anaerostipes, Faecalibacterium, Holdemanella, Intestinimonas, Succiniclasticum, and Agathobacter). While the absence of these genera in Socompa reflects the influence of geographic location in shaping the prokaryotic community composition, interestingly, 14 prokaryotic genera are shared by all modern microbialites. These include genera involved in photosynthesis (Thiohalocapsa and Candidatus Alysiosphaera), sulphur reduction (Desulfonema and Desulfovibrio), and organic matter decomposition (Sedimenticola, Sediminispirochaeta, and Peredibacter), reflecting core ecological processes critical for microbialite formation and stability. The variability in prokaryotic compositions among modern microbialites demonstrates that the process of organomineralisation depends not on the taxonomy of the organisms within the community but rather on their metabolic potential and the prevailing environmental conditions.

Fig. 8: Modern microbialites around the world share some prokaryotic genera, but the degree of similarity varies depending on specific environmental conditions and locations.

a Worldwide distribution of modern microbialites. b UpSet plot depicting common prokaryotic genera among modern microbialites around the globe. Vertical bars show the number of prokaryotic genera in each intersection, while horizontal bars display the number of prokaryotic genera in each set. The taxonomic classification of genera found only in Pozo Bravo microbialites and genera common to all microbialites is shown in the form of a pie chart.

Conclusions

Pozo Bravo harbours a modern microbialite reef that develops under environmental conditions partially resembling those of the Precambrian era, making these microbialites one of the best models for the earliest biostructures on Earth. Therefore, the geobiological processes operating in this environment could provide valuable insights into the mechanisms that shaped ancient microbialites. The formation of lithified mats and microbialites in Pozo Bravo is driven by microbial metabolic activities and environmental conditions that collectively determine their macro-, meso-, and microstructures, as well as their mineralogical composition. Ancient microbialites may have shared some similar underlying mechanisms of formation, including microbially mediated precipitation of Mn(IV) oxides. Consequently, Mn(IV) oxides like ramsdellite could serve as valuable biosignatures for investigating early photosynthetic oxygen production on Earth; however, their interpretation must be approached with caution, accounting for potential abiotic formation pathways and ideally supported by multiple lines of evidence. Microbial mats and microbialites in Pozo Bravo experience environmental fluctuations throughout the year, leading to seasonal variations in the microbial community composition. Both environmental changes and microbial successions could influence organomineralisation. In the warmer months, increased evaporation rates and higher abundances of photoautotrophs raise saturation indices, favouring carbonate precipitation. UV radiation also intensifies during this period, potentially destroying the EPS and lifting constraints on mineral precipitation. As a result, these seasonal dynamics would create a cyclical pattern of organomineralisation. Given that environmental cycles, including fluctuations in temperature, radiation levels, and evaporation rates, also existed in Precambrian Earth, it is probable that the earliest microbialites experienced similar periodic changes in microbial composition and organomineralisation.

Although many geobiological processes that shape modern microbialites also played a fundamental role in the formation of ancient microbialites, some processes have changed in relevance over the course of Earth’s history. The metabolic repertoire of microbial communities, which is central to organomineralisation, differed in ancient microbialites compared to most modern systems. Oxygenic photosynthesis carried out by cyanobacteria and unicellular eukaryotes, for instance, is a key driver of organomineralisation in many modern microbialites. However, in the earliest microbialites, before the evolution of eukaryotic photoautotrophs, oxygenic photosynthesis may not have been the primary driver. Other processes, such as anoxygenic photosynthesis coupled to the oxidation of reduced sulphur or arsenic compounds, as observed in Pozo Bravo, could have been crucial alkaline engines in ancient Earth environments. Additionally, the evidence of carbon fixation processes beyond the reductive pentose phosphate cycle in Pozo Bravo suggests that ancient microbialites may have assimilated inorganic carbon via a more diverse suite of biochemical routes than is typical of most modern systems.

In summary, while many processes shaping ancient microbialites persist in modern systems, some key drivers of microbialite formation may have evolved over time. Understanding both past and present geobiological processes is crucial for guiding our search for life beyond our planet and for interpreting the fossil record of Earth’s earliest biostructures. However, using modern analogues to reconstruct ancient Earth environments remains challenging due to differences in physicochemical conditions, geological settings, evolutionary pathways, and other unknown variables accumulated over geological time.

Methods

Site description

In the southwest region of Laguna Pozo Bravo, soft (non-lithified) microbial mats, lithified microbial mats, and microbialites coexist along the intertidal zone (Fig. 1). Microbial mats extend over the lagoon substrate, sometimes covering part of the microbialite surfaces, and exhibit their greatest lithification between the lagoon margin and the shoreline. Microbialites occur as biostromes¹⁸ (the minimum width is more than one hundred times its maximum thickness) near the lagoon margin or bioherms¹⁸ (the minimum width is less than or equal to one hundred times its maximum thickness) from the shoreline towards the lagoon interior. When water levels rise and submerge the microbialites, biofilms develop on their surfaces; conversely, when water levels drop and the microbialites become exposed and desiccated, those biofilms disappear.

Sample collection

Field campaigns to Laguna Pozo Bravo were carried out in May 2017 (fall), August 2017 (winter), November 2017 (spring), January 2018 (summer), January 2019 (summer), and January 2023 (summer) to collect samples and perform in situ analyses. The collected samples were selected based on a preliminary inspection of the sedimentary structures present throughout the lagoon. Water, soft microbial mat, and microbialite samples were collected in May 2017, August 2017, November 2017 and January 2018 for a seasonal 16S rRNA diversity analysis. Surface water and soft microbial mat samples were taken in August 2017 and January 2018 for chemical analyses. Additional soft microbial mat samples were collected in January 2023 for µXRF, SEM, metagenomic, and bulk geochemistry analyses. Lithified microbial mats and microbialites were sampled in January 2019 and January 2023 for mineralogy and Raman spectroscopy analyses, respectively.

Samples for 16S rRNA diversity and metagenomic analyses were collected employing autoclaved stainless steel tweezers and spoon spatulas, and transferred to sterile 50 ml tubes or polyethylene containers. These samples were stored in RNAlater® solution (Thermo Fisher Scientific, United States) at 4 °C in the dark and processed within a week. Samples for µXRF, SEM, mineralogy, Raman spectroscopy, chemical, and bulk geochemistry analyses were collected employing solvent-cleaned (dichloromethane and methanol) stainless steel tweezers and spoon spatulas, and transferred to polyethylene containers. These samples were kept at 4 °C in the dark until analysis in the laboratory.

Physicochemical analysis

Environmental parameters (pH, conductivity, temperature, and dissolved oxygen) of both surface water and soft microbial mats were measured in situ in August 2017 and January 2018 using a portable Hanna HI 9829 multiparameter probe. The instrument was calibrated according to the manufacturer’s instructions prior to measurement. Water-level fluctuations along the shoreline were monitored by placing PVC pipes at the lagoon banks and measuring the water column height. Chemical parameters of the water and soft microbial mats were analysed by Grupo Induser S.R.L (Salta, Argentina), and the corresponding values are presented in Tables 1 and 2. Using the physicochemical parameters of the water, mineral saturation indices were modelled with PHREEQC v3.0 geochemical software⁶⁸. The saturation index is defined as SI=log(IAP K_sp^-1), where IAP represents the ion activity product, and K_sp the solubility product constant of the corresponding mineral. If SI < 0 the solution is undersaturated with respect to the mineral, if SI = 0 the solution is at equilibrium with the mineral, and if SI > 0 the solution is supersaturated with respect to the mineral.

Satellite data acquisition and analysis

Surface pressure (monthly) [v5.12.4 STD], UV index (local noon) [standard v003], and land surface temperature (L3, monthly, day/night, TES algorithm) [v6.1 STD] data for the Laguna Pozo Bravo region were obtained from NASA Earthdata (https://earthdata.nasa.gov/). Data processing was conducted using the R packages rhdf5 and ncdf4. The rhdf5 package was used to extract data from HDF5 format files, while the ncdf4 package facilitated handling netCDF format files. Data visualisation was carried out using the R package ggplot2.

Microelectrode measurements

Oxygen and hydrogen sulphide profiles covering the upper 5–10 mm of the soft microbial mats were measured in situ in triplicate during both daytime (from 13:00 to 16:00) and nighttime (from 20:00 to 01:00) in January 2019. Clark-type oxygen and amperometric hydrogen sulphide microsensors with a tip width of <50 µm were employed for these measurements, constructed, calibrated, and operated as previously described^69,70,71,72. The microsensors were placed on a multisensor holder with a tip distance of ~0.5 cm for simultaneous measurements, with electricity supplied by a portable generator. Oxygen concentration was adjusted for salinity and temperature, according to specifications (Unisense), and values were determined assuming an atmospheric pressure of 1 bar⁷³.

Scanning micro X-ray fluorescence analysis

The distribution of elements within a freeze-dried soft microbial mat sample was determined by μXRF spectrometry employing an M4 Tornado μXRF spectrometer (Bruker Nano GmbH, Germany). The instrument was equipped with a Rh X-ray source and polycapillary optics (20 µm spot size). Scanning was operated at 50 kV and 600 µA tube setting under vacuum condition of 20 mb. The pixel size was set to 25 µm, and scan time was 20 ms pix^–1. Elemental distributions were analysed as net intensities (deconvoluted counts) using the Bruker M4 Esprit software package.

Raman spectroscopy

Raman spectroscopy analysis was performed on a microbialite sample to assess mineralogical variations through both macro- and microscopic observations. Spectra were acquired using backscattering geometry with a LabRam HR Evolution Raman microspectrometer, set to a resolution of 0.4 cm⁻¹ and equipped with a He-Ne laser line (633 nm) at INQUIMAE (UBA-CONICET), Buenos Aires, Argentina. Beam power and acquisition times were optimised for each sample to obtain informative spectra without causing sample alteration. Spectra were recorded using a 50× microscope objective (spatial resolution of ~1 μm), with an exposure time of 10–30 s and three accumulations. For band identification, we consulted the RRUFF database (https://rruff.info) and complemented this data with detailed mode assignments from Post et al.⁴⁹.

Mineralogy and bulk geochemistry analyses

The mineral composition of freeze-dried (soft and lithified) microbial mat and microbialite samples was determined through powder XRD analysis using a Bruker X-ray diffractometer (Eco D8 Advance, XRD), equipped with a Cu X-ray source (Cu Kα1,2, λ = 1.54056 Å), operated at 40 kV and 25 mA. The freeze-dried and finely ground (<20  μm) samples were scanned from 5° to 60° in the 2·ϴ-diffraction angle, with a scanning step size of 0.05° and a time per step of 1 s, using the Bragg–Brentano geometry. The phase identification was performed with the DIFFRAC.EVA software by comparing the measured diffractograms to reference entries from the integrated PDF database. The accuracy and reliability of the crystal structure determinations were evaluated using the figure of merit, which quantitatively measures the fit between the proposed crystal model and the experimental XRD data. Semi-quantification of the minerals was performed based on the intensity of the main peaks in each XRD pattern^74,75. The mineralogy analysis of the microbialite sample was further complemented with microscopic observations using conventional optical microscopy.

The isotopic composition of organic carbon (δ¹³C_org) and total nitrogen (δ¹⁵N) in bulk soft microbial mat and microbialite samples was determined using isotope-ratio mass spectrometry (IRMS), following the method outlined by the United States Geological Survey (USGS)⁷⁶. Samples were first homogenised by manual grinding in a mortar and then decarbonated with HCl (3 N). After 24 h of equilibration, the samples were adjusted to neutral pH using ultrapure water and subsequently dried in an oven at 50 °C until a constant weight was achieved. The δ¹³C_org and δ¹⁵N ratios were measured using a MAT 253 IRMS (Thermo Fisher Scientific, Waltham, Massachusetts, USA) and reported in standard per mil notation (‰) relative to international reference standards (δ¹³C_org vs. Vienna Pee Dee Belemnite and δ¹⁵N vs. air). Three certified standards (USGS41, IAEA-600, and USGS40) were employed with an analytical precision of 0.1‰. Total organic carbon (TOC%) and total nitrogen (TN%) content were determined using an elemental analyser (HT Flash, Thermo Fisher Scientific, Waltham, Massachusetts, USA) during the stable isotope measurements. The mineralogical and isotopic analysis have been done at the Centro de Astrobiología (CAB), in Madrid (Spain).

Scanning electron microscopy analysis

SEM images of the soft microbial mats were obtained under vacuum conditions using a Zeiss Supra 55 VP scanning electron microscope (Carl Zeiss NTS GmbH, Germany) at Centro Integral de Microscopía Electrónica (CIME) in Tucumán, Argentina. Prior to observation, samples were fixed overnight at 4 °C in modified Karnovsky fixative, composed of 8% (v/v) formaldehyde, 16% (v/v) glutaraldehyde, and phosphate-buffered saline (200 mM PBS, pH 7.4). Samples were then washed three times with PBS and calcium chloride (CaCl₂) for 10 min each, followed by fixation with 2% (v/v) osmium tetroxide overnight. After two ethanol washes (30% v/v) for 10 min each, the samples were dried at the critical point and sputtered with gold.

Diatom taxa were identified using SEM images and identification keys⁷⁷, based on morphological characteristics of their frustules. The analysed morphological features included valve size, shape, and symmetry; the presence, location and structure of the axial area; the presence, location, and structure of the raphe; the presence of septa; the presence and structure of costae; the location, structure, and density of striae; and the presence of stauros, among others.

DNA extraction and sequencing

Different procedures were employed for DNA extraction depending on the type of sample. Water samples were filtered through a 0.22-µm pore size membrane (DURAPORE GV) to concentrate suspended microorganisms. Soft microbial mat samples were homogenised to ensure representativeness. While microbialite samples were immersed in phosphate-buffered saline (50 mM PBS, pH 7.5) and sonicated three times to detach microorganisms from the surface. The supernatants were then collected and centrifuged at 13,000 × g for 15 min. The resulting cell pellets served as starting material for DNA extraction.

Total genomic DNA was isolated using the FastDNA® SPIN Kit for Soil (MP Biomedicals, United States) following the manufacturer’s protocol. DNA quality and concentration were assessed using a NanoDrop 2000c Spectrophotometer (Thermo Fisher Scientific, United States). The hypervariable V3 and V4 regions of the bacterial and archaeal 16S rRNA gene were amplified using the primers Bakt_341F (5-CCTACGGGNGGCWGCAG) and Bakt_805R (5-GACTACHVGGGTATCTAATCC). Amplicons obtained from water, soft microbial mat and microbialite samples, along with total DNA isolated from a soft microbial mat sample, were sequenced using an Illumina MiSeq platform. Raw sequences were deposited in the ENA Project database under the accession numbers ERP120954 (amplicon sequences) and ERP113390 (metagenomic sequences).

Amplicon sequence analysis

The prokaryotic composition of water, soft microbial mat and microbialite samples (ENA accession number: ERP120954) was determined with the MGnify metagenomic analysis pipeline v.5.0⁷⁸ (MGnify accession number: MGYS00005312) through the taxonomic assignment of the 16S rRNA gene amplicon sequences (SILVA release 132)⁷⁹. Alpha diversity metrics (Shannon index)⁸⁰ were calculated using the R package phyloseq⁸¹. The phyloseq package was also employed to perform a principal coordinate analysis (PCoA) based on Bray-Curtis dissimilarity⁸². One-sided permutational ANOVA (PERMANOVA) tests were carried out using the R package vegan⁸³ to evaluate the PCoA result. To compare the genera of prokaryotes found in Pozo Bravo microbialites with those in modern microbialites from other locations around the world - Alchichica Lake, Puebla, Mexico (SRP061655, SRP072547)^34,35,36; Great Salt Lake, Utah, United States (SRP067068)³¹; Highborne Cay, Bahamas (SRP004035)³³; Pavilion Lake, British Columbia, Canada (SRP035880, ERP020021)³⁰; Hamelin Pool, Shark Bay, Australia (SRP055055)²⁷; Clifton Lake, Yalgorup National Park, Australia (SRP072185)²⁸; and Socompa Lake, Salta, Argentina (SRP007748, SRP072938, ERP021393)^67,84 - an UpSet plot was generated using the R package UpSetR⁸⁵. The prokaryotic composition of these modern microbialites was also determined with the MGnify metagenomic analysis pipeline v.5.0 using raw sequence data downloaded from the SRA (Sequence Read Archive) projects specified above.

Metagenomic analysis

Quality control checks on the raw paired-end sequence data (ENA accession number: ERP113390) were first conducted using FastQC v.0.11.9 (Babraham Bioinformatics website). The results from this analysis were later used to determine the necessary trimming steps (ILLUMINACLIP = TruSeq2-PE.fa:2:30:5, SLIDINGWINDOW = 4:20, AVGQUAL = 20, and MINLEN = 50), which were carried out with Trimmomatic v.0.39⁸⁶. Taxonomic and functional annotations of the cleaned paired reads were performed using the MGnify metagenomic analysis pipeline v.4.1⁷⁸ (MGnify accession number: MGYS00004549). The cleaned paired reads were further assembled into contigs using MetaSPAdes v.3.14.1⁸⁷, and the quality of the assembly (Supplementary Table S6) was assessed with QUAST v.4.6⁸⁸ and Bowtie2 v.2.4.2⁸⁹. MAGs were reconstructed from the contigs employing MetaWRAP v.1.1⁹⁰. Quality control checks on the MAGs were conducted using CheckM v.1.0.13⁹¹ (Supplementary Table S7). Using the quality control metrics, dereplication of MAGs was conducted with dRep v.1.4.3⁹². This tool identifies groups of MAGs with high similarity (ANI, Average Nucleotide Identity >95%) and selects the representative MAG from each group. In this way, redundant MAGs are filtered out. Taxonomic assignment of MAGs was performed by comparison against the Genome Database Taxonomy (GTDB) release 220⁹³ (Supplementary Table S8). The prediction and functional annotation of protein-coding genes in the assembled metagenome and MAGs were performed with Prodigal v.2.6.3⁹⁴ and InterProScan v.5.34-73.0⁹⁵, respectively. After the functional annotation of the genes, an analysis of the metabolic pathways present in the MAGs was carried out with Genome Properties v.2.0⁹⁶. Genes involved in metabolic pathways/cycles of interest were additionally annotated with HMMER-3.1 (http://hmmer.org/) using hidden Markov model (HMM) profiles for KEGG/KO with predefined score thresholds⁹⁷ (Supplementary Table S9). Cleaned paired reads were mapped to the identified genes using Bowtie2 v.2.4.2⁸⁹, and the total number of mapped reads obtained was used to determine the abundance of each gene, normalised by gene length and library size [Reads per kilobase per million mapped reads (RPKM)].

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.