Contrasting Modes of Carbonate Precipitation in a Hypersaline Microbial Mat and Their Influence on Biomarker Preservation (Kiritimati, Central Pacific)

Abstract

Microbial mats represented the earliest complex ecosystems on Earth, since fossil mineralized examples (i.e., microbialites) date back to the Archean Eon. Some microbialites contain putative remains of organic matter (OM), however the processes and pathways that lead to the preservation of OM within microbialite minerals are still poorly understood. Here, a multidisciplinary study is presented (including petrographic, mineralogical and organic geochemical analyses), focusing on a modern calcifying mat from a hypersaline lake in the Kiritimati atoll (Central Pacific). The results show that this mat has a complex history, with two main growth phases under hypersaline conditions, separated by an interruption caused by desiccation and/or freshening of the lake. The mineral precipitates of the mat are predominantly aragonitic and two contrasting precipitation modes are observed: the main growth phases of the mat were characterized by the slow formation of irregular micritic particles with micropeloidal textures and subspherical particles, linked to the degradation of the exopolymer (EPS) matrix of the mat; whereas the interruption period was characterized by the rapid development of a thin but laterally continuous crust composed of superposed fibrous aragonite botryoids that entombed their contemporaneous benthic microbial community. These two precipitation modes triggered different preservation pathways for the OM of the mat as the thin crust shows a particular lipid biomarker signature, different from that of other layers and the relatively rapid precipitation of the crust protecting the underlying lipids from degradation, causing them to show a preservation equivalent to that of a modern active microbial community, despite them being >1100 years old. Equivalent thin mineral crusts occur in other microbialite examples and, thus, this study highlights them as excellent targets for the search of well-preserved biomarker signatures in fossil microbialites. Nevertheless, the results of this work warn for extreme caution when interpreting complex microbialite biomarker signatures, advising combined petrographic, mineralogical and geochemical investigations for the different microbialite layers and mineral microfabrics.

1. Introduction

Microbial mats (benthic, laminated, mm- to cm-thick microbial communities) probably represented the earliest complex ecosystems on Earth and they are still ubiquitous nowadays, thriving in many different contexts, from freshwater to hypersaline settings, including hot springs and other extreme environments [1,2,3,4,5,6,7]. Microbial mats consist of several specialized biotic consortia, and each layer contains different microorganisms with distinct metabolic activities [8,9]. The fossilized counterparts of microbial mats are microbialites [10], which have an extremely long geological record of ~3.5 Ga [11,12]. Microbialites are most commonly composed of carbonate minerals (with participation of other minerals, such as phosphates, sulphates, silicates, iron minerals and oxides) and putative remains of the organic matter (OM) of their original microbial mats have also been found within some of them [13,14,15]. Nevertheless, the interpretation of biomarkers and signatures of OM within microbialites is always challenging, given the wide variety of biotic and abiotic factors involved in their complex history, ranging from the paleoenvironment and origin of the OM to its taphonomy and final fossilization [16,17,18,19,20,21]. In addition, many different organic and inorganic processes can contribute to carbonate precipitation in microbial mats (see reviews in [22,23]), which further complicates the preservation pathways of OM within microbialites and the interpretation of their organic geochemical signatures.

In this work, a modern and actively mineralizing microbial mat is studied, from a hypersaline lake in Kiritimati Island (Central Pacific), the largest atoll in the world. This detailed geomicrobiological study encompasses different petrographic techniques for both organic and mineral components of the microbial mat, as well as mineralogical and organic geochemical analyses, to unravel the complex history of the mat. This allows one to interpret the different ways in which carbonate precipitated within the mat and to understand how precipitation modes controlled the different biomarker signatures observed, providing an excellent example of the integration of organic geochemical, petrographic and mineralogical analyses, and indicating that these techniques should be always coupled to elucidate the origin and taphonomy of OM within microbialites.

2. Materials and Methods

2.1. Site Description and Sample Collection

Kiritimati (formerly Christmas Island) is the world’s largest atoll, with a surface area c. 360 km². It is located in the central Pacific and close to the Equator (1°55′ N, 157°25′ W). The island harbors approximately 500 lakes with salinities ranging from brackish to hypersaline [24,25], in many of which, well-developed microbial mats occur [26,27]. These microbial mats are often developed on top of subfossil reef-structures and microbialites that represent older and almost fully mineralized generations of microbial mats [5,27,28].

In general, Kiritimati has an arid climate and is mostly controlled by the El Niño-Southern Oscillation (ENSO). Recharge of lakes occurs as a result of heavy rains during El Niño wet events, leading to a decrease in lake salinities. During La Niña dry events, reduced precipitation and higher evaporation induce an increase in lake salinities [24,25,29].

Sampling was carried out during an expedition to Kiritimati in March 2011, with the required research permit issued by the Environment and Conservation Division of the Ministry of Environment, Lands and Agriculture Developments (MELAD) of the Republic of Kiribati. The microbial mat studied in this work is 10 cm thick and it was collected from the center of Lake 2 (Figure 1; 01°50′49.35″ N, 157°21′36.32″ W) at water depth c. 4 m [30]. The salinity of Lake 2 was 97‰ in 2002 and 125‰ in 2011 (unpublished own data,). Blumenberg et al. [30] analyzed some aspects of the same mat studied here, and divided it in five layers. Nevertheless, given the more comprehensive and detailed analysis conducted here, a different layer division has been made (see Section 3 Results, below).

In order to compare the mineralization of the studied modern mat with that of older counterparts, we also conducted petrographic studies of thin sections of subfossil microbialites from another lake on Kiritimati, sampled in the same campaign (Lake 21 01°57′45.63″ N, 157°19′53.03″ W). The studied subfossil microbialite from Lake 21 occurs immediately below the modern active microbial mat at the bottom of the lake [28].

2.2. Previous ¹⁴C Dating of the Studied Material

Blumenberg et al. [30] provided ¹⁴C ages of the studied mat, shown in Figure 1. The downwards increase in age suggests that the dating is consistent. However, the location in an atoll with a substrate of coral reef limestone [27] suggests that the ¹⁴C signal could be affected by a marine reservoir and local hardwater reservoir effects, as typically occurs in atoll settings [31,32], which was not taken into account by Blumenberg et al. [30]. Studies that considered the marine reservoir effect in Kiritimati showed that a ¹⁴C correction factor of 39 ± 56 years produces good correlations with U/Th ages [33,34]. In addition, the distance of the studied Kiritimati Lake 2 from the coast (Figure 1) and its persistent hypersalinity indicate disconnection from seawater, allowing equilibration of C reservoir in lake water with atmospheric CO₂ [29]. A hardwater reservoir effect could be another potential modifier of the ¹⁴C ages of Kiritimati mats, if groundwaters from the underlying coral reefs seeped into the mat-bearing lakes. Hydrological data led Arp et al. [27] to suggest that surface or groundwater influx could have been possible in Kiritimati hypersaline lakes, however their stable O and C isotope values also showed that the water of these lakes mainly derives from highly evaporated marine water. Additional ¹⁴C analyses were obtained for corals from the surroundings of Lake 2, obtaining ages of ca. 2500 years BP (unpublished own data), significantly older than the ages obtained by Blumenberg et al. [30] for the mat studied here, supporting that the ¹⁴C signature of the studied microbial mats is not strongly influenced by that of old underground water. Therefore, although a minor hardwater effect is possible, we will use here the ¹⁴C ages obtained by Blumenberg et al. [30], as for the aim of this study the absolute age values are not directly relevant, but rather the downwards increase in age and the relative differences between ages of different mat layers.

2.3. Preparation for Histological Thin Sections

For the preparation of histological thin sections, samples were dehydrated with a graded ethanol series. Afterwards, mat samples were embedded in LR White resin (medium grade, London Resin Company Ltd., Reading, UK). The embedded samples were cut with a microtome saw (Leica SP1600) to a thickness of c. 100 µm, mounted on glass slides using Biomount mounting medium (Electron Microscopy Sciences, Hatfield, PA, USA). Thin sections were studied under petrographic and fluorescence microscope (Zeiss Axiophot, Jena, Germany).

2.4. Preparation for the Electron Microscope

To observe the mineral fraction of the microbial mat under field-emission scanning electron microscopy (FE-SEM), the organic matter of the samples was removed through oxidation with NaOCl [35]. Each sample from individual mat layer was immersed in 6% NaOCl, changing the solution every 12 h for several days, until no traces of organic matter were visible. The remaining mineral particles were washed with distilled H₂O until pH was neutral, and then dried. Mineral particles were mounted on SEM sample holders and sputtered with Pt/Pd (14.1 nm for 5 min, using a Leica EM QSG100) and then observed in a FE-SEM (Zeiss Leo Gemini 1530).

2.5. Mineralogical and Elemental Analyses

Various non-destructive methods were used for investigating the composition of the microbial mat minerals. For detailed elemental mapping and point analyses, the FE-SEM microscope was used, coupled with an Oxford Instruments INCA x-act energy dispersive X-ray spectrometry (EDX) detector. In addition, a micro X-ray fluorescence (μ-XRF) device was applied to obtain elemental distribution images of thin sections. The analyses were conducted with a Bruker M4 Tornado instrument equipped with an XFlash 430 Silicon Drift Detector. Measurements (spatial resolution = 25–50 µm, pixel time = 8–25 ms) were performed at 50 kV and 400 µA at a chamber pressure of 20 mbar.

Mineralogical identification was conducted using Raman analyses. Raman spectra were collected using a WITec alpha300R coupled ultra-high throughput spectrometer. Before analysis, the system was calibrated using an integrated light source. The experimental setup included a 405 nm laser, 10 mW laser power, a 100× long working distance objective with a numerical aperture of 0.75, and a 1200 g mm⁻¹ grating. This setup had a spectral resolution of 2.6 cm⁻¹. The spectrometer was centered at 1530 cm⁻¹, covering a spectral range from 122 to 2759 cm⁻¹. The 405 nm laser was chosen to reduce fluorescence effects [36]. Each spectrum was collected by two accumulations, with an acquisition time of 60 s. Raman spectra were processed with the WITec project software.

2.6. Extraction and Derivatization

Two different types of lipids were extracted and analyzed separately in this study: the lipids present in the organic matrix of the mat (here termed ‘freely extractable lipids’) and those present within the carbonate precipitates of the mat (here termed ‘carbonate-bound lipids’). Both lipid types were analyzed for each layer of the microbial mat. A small portion (5–20 g) of each layer was sampled and freeze dried. After freeze-drying, samples were homogenized (not powdered) and extracted by 4 × 50 mL dichloromethane/methanol (3:1; v:v) (10 min ultrasonication) to obtain the freely extractable lipids. The resulting extraction residue was decalcified using 37% HCl and was again extracted as described above to obtain the carbonate-bound lipids.

To make carboxylic acids GC-amenable, a mixture of trimethylchlorosilane (TMCS)/MeOH (1:9, v:v) was added to lipid extracts (5% aliquot of the lipids from the previous layer division and 25% aliquot from layers 2b, 4a and 4b). Both freely extractable and carbonate-bound lipids were reacted at 80 °C for 60 min. The resulting carboxylic acid methyl esters and the neutral lipids (including n-alkane and hopanoid hydrocarbons) were extracted from the reaction mixture by vigorous shaking with 3 × 1 mL n-hexane. The extracts were combined and evaporated to near-dryness under a gentle N₂ flow, re-dissolved in n-hexane, and analyzed by gas chromatography-mass spectrometry (GC-MS).

To make alcohols (including sterols and hopanols) GC-amenable, all aliquots (identical amounts were used as showed above) of lipid fractions were silylated using BSTFA (N,O-bis(trimethylsilyl)trifluoroacetamide) containing 5% (v:v) trimethylchlorosilane (TMCS) as a catalyzer (heated at 70 °C for 60 min). The resulting trimethylsilyl (TMS-) derivatives were dried, and re-dissolved in n-hexane, and analyzed by gas chromatography-mass spectrometry (GC-MS).

GC-MS analyses were conducted using a Thermo Fisher Trace 1310 GC coupled to a Thermo Fisher Quantum XLS Ultra MS as described elsewhere [37]. N-Icosane-D₄₂ was used as an internal standard for quantification.

2.7. Compound-Specific Stable Carbon Isotopes (GC-C-IRMS)

Compound-specific stable carbon isotope ratios were measured for microbial lipids (i.e., FAs and hopanoids), hydrocarbons and sterols (including freely extractable and carbonate-bound lipid fractions) in the studied microbial mat. In order to interpret the general trend of isotope signatures, layers 1, 3, 5 and 6 were chosen and analyzed, thus including data from the upper, middle and lower parts of the mat, respectively (see description of the mat in Section 3 Results section). Analyses were conducted using a Thermo Scientific Trace gas chromatograph (GC) coupled to a Delta Plus isotope ratio mass spectrometer (IRMS) as described elsewhere [38].

3. Results

3.1. Macroscopic Description of the Mat Layers

The studied microbial mat has a thickness of c. 10 cm and it is divided in two main parts (composed of mineral particles within an organic matrix, Figure 1c and Figure 2) and separated by a thin yet conspicuous and continuous mineral crust (Figure 1c and Figure 2). The upper part is composed of two layers (1–2, c. 3 cm thick, Figure 1c and Figure 2), which have a cohesive appearance, sticking together when handled, due to abundant and relatively fresh organic material. This organic matrix shows bright orange, green and brown colors with transparent or whitish mineral particles, rare and scattered in the top layer 1 and more abundant in layer 2. The lower part of layer 2, closer to the underlying mineral crust has been termed layer 2b and analyzed separately using organic geochemical techniques (Figure 2). The mineral crust that separates the two main parts of the mat is here termed layer 3 and it is thin (1–2 mm) although laterally very continuous (Figure 1c, Figure 2, Figure 6 and Figure 7). The lower part of the mat (c. 7 cm thick) is more friable, having a crumbly appearance, due to the rarer and poorly preserved organic material and more abundant mineral particles, producing brown and beige colors (Figure 1c and Figure 2). This lower part is subdivided in three main layers (4–6, Figure 1c and Figure 2). The top parts of layer 4, closer to the mineral crust, have been also analyzed separately using organic geochemical techniques, as layers 4a (dark brown color and immediately below the crust) and 4b (below 4a and with a lighter beige color) (Figure 2). The ¹⁴C ages of the mat [30] show that its upper (and thinner) part was formed approximately in the last 1100 years (62 ± 40 years BP for layer 1, 551 ± 40 years BP for layer 2 and 1111 ± 40 years BP for layer 3; Figure 1c), whereas the lower (and thicker) part was formed approximately in the previous 350 years (1331 ± 40 years BP for layer 5 and 1440 ± 40 years BP for layer 6; Figure 1c).

3.2. Microscopic Observation of Mat Layers

The cohesive appearance of the upper part of the mat (layers 1–2) is due to its abundant and fresh organic matter. Under the microscope, this organic matter is mainly seen as a dense reticulate microstructure formed by a net of intertwined delicate fibers of extracellular polymeric substances (EPS, Figure 3a,b). Within this network of EPS fibers, the dominant microorganisms observed are colonies of coccoid cyanobacteria and less abundant filamentous cyanobacteria (Figure 8a), as is typical in hypersaline Kiritimati mats [27,28,39]. Locally, miliolid foraminifera are also observed in layer 1. Scattered mineral particles occur within these layers. Local idiomorphic gypsum crystals occur in layer 1 (Figure 2 and Figure 4a), however the predominant particles are composed of aragonite [27,40]. Two different types of aragonitic particles are observed in layers 1–2: (a) relatively abundant small (<2 mm in diameter) subspherical particles with a crystalline fibrous-radial internal structure and typically growing around a micritic nucleus (Figure 3a,b and Figure 4a,d,e; cf. ‘spherulites’ of Arp et al. [27], or ‘non-laminated ooids’ of Suarez-Gonzalez and Reitner, [40]); (b) less abundant irregular micritic particles with a micropeloidal internal texture (Figure 4c). When various subspherical particles are very close to each other, they tend to coalesce together during their growth, forming composite particles (Figure 3a,b and Figure 4a,e). These two types of aragonitic particles, subspherical and irregular micritic, are also the main mineral components of the lower layers 4–6 (Figure 3d, Figure 4b,f and Figure 5), but layer 3 shows a markedly different composition.

Layer 3 is a thin but laterally continuous mineral crust with features not common in other microbial mats previously described from Kiritimati [26,27,41]. With the naked eye, this crust shows white-beige colors with darker inclusions, and it generally has a flat bottom and domed top (Figure 2 and Figure 6). Microscopically, the crust is composed of up to seven superposed laminae with a fibrous-radial internal texture, formed by long mineral needles (Figure 6, Figure 7, Figure 8, Figure 9, Figure 10, Figure 11 and Figure 12). The two lowermost laminae are flat, however the rest are botryoidal in shape (Figure 6, Figure 7, Figure 8 and Figure 9). The lower laminae are thin (100–500 µm) and laterally discontinuous (Figure 6c and Figure 7a). The upper laminae are thicker (up to 1 mm) and laterally more continuous (Figure 6c and Figure 7a). The mineralogical and elemental composition of the crust has been studied in detail. Raman spectra of the long mineral needles that compose it exhibit wave numbers (206, 702 and 1086 cm⁻¹) characteristic of aragonite (Figure 6b). No organic remains were detected within these aragonite crystals. In addition, μ-XRF elemental maps of the crust show that the main cations present are Ca and Sr (Figure 7). Minor and patchy occurrence of Mg is related with local miliolid foraminifers at the base of the crust (Figure 7, see further description below). EDX point measurements of the crust were compared with an aragonite standard, showing Sr values calculated between 8000–10,000 ppm, which is characteristic for aragonite precipitated in seawater equilibrium.

The layer 3 crust includes abundant large pores filled with residual organic matter (mainly EPS, Figure 6, Figure 8 and Figure 12), which generate the darker areas observed macroscopically within the crust (Figure 2 and Figure 6). Some of the inclusions are filled by entombed colonies of coccoid cyanobacteria (Figure 8a,c). Raman spectra of these colonies show that they are mainly organic, with a typical broad D-band with many subordinate peaks (main peak 1278) and G-band with 1579 as the main wave number (Figure 6b). The wave number 1046 could due to nahcolite (NaHCO₃), which is characteristic of hypersaline environments, however this should be tested with further analyses. Other organic remains within layer 3 are molds of filamentous microbes and diatoms, which are ubiquitous throughout the crust, enclosed within the fibrous crystalline structure of the botryoids (Figure 8, Figure 9, Figure 10, Figure 11 and Figure 12). Miliolid foraminifera have also locally been observed within the lower part of the crust, which contrast in elemental composition with the rest of the crust, given their high-Mg-calcite mineralogy (Figure 7, Figure 9d and Figure 12b). The molds of filamentous microbes are long (up to 200 µm) but thin (<5 µm, and typically ~1 µm) and very abundant, thus being easily observed under the petrographic microscope (Figure 8, Figure 9, Figure 10 and Figure 11). In some of the botryoids they are so abundant that they provide a cloudy appearance to the aragonite (Figure 8a, Figure 9a and Figure 11a). Generally, the filaments are erected and curved, forming tufted structures that mimic the shape of the aragonite botryoids (Figure 8a,d and Figure 9a–c) and some of them seem to have branching structure (Figure 10b–d). These filaments show bright fluorescence (Figure 11). Locally, at the top part of the crust, larger filaments (up to 500 µm long and 40 µm thick), with brownish color in thin section and no fluorescence, are also observed, showing irregular orientation, from vertical to horizontal, apparently not associated with the botryoid structure (Figure 6c and Figure 11). Contrarily to the filament molds, diatom molds are much smaller (up to 50 µm long and 10 µm wide) and thus not observable in the thin section, however are ubiquitously observed under SEM (Figure 12d). Both filament and diatom molds are either empty (preserved as porosity) or filled by carbonate with only local preservation of the original EPS (Figure 12d,f).

Similar thin mineral crusts have not been observed in other modern Kiritimati microbial mats, however they occur in some of the subfossil mineralized mats, preserved as microbialites and occurring below the modern mats (Figure 13). Interestingly, the crusts in these examples also occur as thin yet laterally-continuous interruptions in accretion, around the middle part of the microbialites (Figure 13a,b), and they show the same layered botryoidal structure (Figure 13b–d). Furthermore, most botryoids of the crusts of these subfossil microbialites also have a cloudy appearance in thin section, due to very abundant filament molds, identical to those described in layer 3 of the studied microbial mat of Lake 2 (Figure 13c–f). Some of the filaments within the microbialite crusts are so well preserved that they even show the branching and segmented shaped of the original microbe (Figure 13f).

The layers below the crust (layers 4–6) have a crumbly appearance due to the larger size and higher abundance of mineral precipitates and to the less abundance of organic matter between them (Figure 3c,d). When seen under the microscope, the organic matrix of these layers is much less dense than in layers 1–2. It has a porous laminated microstructure with coarser EPS fibers, which are parallel oriented and bundled together, and include many nano- to micro-scale precipitates (Figure 3c,d). Unlike fresher EPS fibers of the upper layers, these older more degraded ones show birefringence under cross-polarized light (Figure 3d and Figure 4b; cf. [40,44,45,46]). Very few remains of microbes have been observed within this organic matrix, showing only scattered empty and partially degraded sheaths of filamentous microbes. Miliolid foraminifera are also scarcely preserved in these layers. Mineral precipitates of layers 4–6 are also aragonitic and of the same types as those of layers 1–2: (a) subspherical fibrous-radial crystalline particles (Figure 5); and (b) irregular micropeloidal micritic particles (Figure 3d, Figure 4b and Figure 5). However, they are much larger and much more abundant than in layers 1–2. Also, they are typically merged with each other forming complex irregular aggregates of particles, which are larger downwards in the mat, reaching up to 3 cm of width (Figure 4f). In the thin section, these aggregates are composed of individual or coalesced subspherical particles, some of them with a laminated cortex (cf. ‘ooids’of Suarez-Gonzalez and Reitner. [40]), which are merged together typically by micropeloidal micrite, although crystalline areas are also observed between the merged particles (Figure 5). In layer 4, where the aggregates are smaller and in earlier stages of development, the micritic bridges merging particles together commonly have a meniscus morphology (Figure 5).

3.3. Extractable Lipids within the Studied Mat

In the freely extractable portion, a variety of steroids, hopanoids and fatty acids (FAs) were detected in the studied mat (Figure 14). C₂₇-C₂₉ conventional sterols and distinct C₃₁ sterols were observed in the different mat layers, and low amounts of C₃₀ dinosterols were additionally observed in some of the layers (

Table 1. (a) Concentrations of sterols in the freely extractable lipids of microbial mat layers (μg/g dry mat; n.d. = not detected, modified after Shen et al., [38]). (b) Concentrations of sterols in the carbonate-bound lipids of microbial mat layers (μg/g dry mat; n.d. = not detected, modified after Shen et al., [38]).

(a)
Trivial Names		Cholesterol	Cholestanol	Brassicasterol	Campesterol	Ergostanol	Stigmasterol	β-Sitosterol	Stigmastanol	Dinosterol	4α-Methylgorgosterol	4α-Methylgorgostanol
	Compound	Cholest-5-en-3β-ol	5α-cholestan-3β-ol	24-methylcholesta-5,22-dien-3β-ol	24-methylcholest-5-en-3β-ol	5α-24-methylcholestan-3β-ol	24-ethylcholesta-5,22-dien-3β-ol	24-ethylcholest-5-en-3β-ol	5α-24-ethylcholestan-3β-ol	4α,23,24-trimethylcholest-22-en-3β-ol	22,23-methylene-4α, 23,24-trimethylcholest-5-en-3β-ol	22,23-methylene-4α, 23,24-trimethylcholestan-3β-ol
Layers
1		5.20	0.86	4.93	4.91	0.20	3.74	5.44	0.77	n.d.	1.10	5.21
2		0.24	0.19	n.d.	0.34	0.06	0.43	0.77	0.19	n.d.	0.36	1.55
2b		0.89	1.96	0.52	0.54	0.99	0.89	0.94	1.85	1.05	1.07	4.71
3		0.01	0.04	n.d.	0.02	0.02	0.02	0.03	0.05	n.d.	0.09	0.37
4a		1.15	2.01	0.51	0.52	1.19	2.03	2.78	3.37	0.82	0.39	2.72
4b		0.06	0.05	0.03	0.02	0.04	0.09	0.09	0.09	0.04	0.04	0.18
4		0.02	<0.01	n.d.	0.02	0.01	0.05	0.02	0.02	n.d.	0.12	0.40
5		0.03	0.02	n.d.	0.07	0.04	0.16	0.12	0.13	n.d.	0.95	2.72
6		0.08	<0.01	n.d.	n.d.	n.d.	0.08	0.02	0.01	n.d.	0.10	0.73
(b)
Trivial Names		Cholesterol	Cholestanol	Brassicasterol	Campesterol	Ergostanol	Stigmasterol	β-Sitosterol	Stigmastanol	Dinosterol	4α-Methylgorgosterol	4α-Methylgorgostanol
	Compound	Cholest-5-en-3β-ol	5α-cholestan-3β-ol	24-methylcholesta-5,22-dien-3β-ol	24-methylcholest-5-en-3β-ol	5α-24-methylcholestan-3β-ol	24-ethylcholesta-5,22-dien-3β-ol	24-ethylcholest-5-en-3β-ol	5α-24-ethylcholestan-3β-ol	4α,23,24-trimethylcholest-22-en-3β-ol	22,23-methylene-4α, 23,24-trimethylcholest-5-en-3β-ol	22,23-methylene-4α, 23,24-trimethylcholestan-3β-ol
Layers
1		n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
2		0.02	<0.01	n.d.	0.08	<0.01	0.08	0.07	0.01	n.d.	n.d.	0.34
2b		1.49	0.61	n.d.	n.d.	0.73	n.d.	0.65	0.92	n.d.	n.d.	2.13
3		0.04	0.05	n.d.	0.08	0.02	0.03	0.06	0.04	n.d.	n.d.	0.40
4a		n.d.	0.36	n.d.	n.d.	0.20	n.d.	n.d.	0.35	n.d.	n.d.	n.d.
4b		0.09	0.05	n.d.	n.d.	0.17	0.08	0.04	0.08	n.d.	0.04	0.18
4		0.11	0.07	n.d.	0.12	0.03	0.25	0.12	0.09	n.d.	n.d.	0.45
5		0.02	<0.01	n.d.	n.d.	n.d.	0.06	0.01	<0.01	n.d.	n.d.	0.22
6		0.02	<0.01	n.d.	0.03	<0.01	0.04	0.03	0.01	n.d.	n.d.	0.59

a, Figure 15a and Figure 16a). Total sterol concentrations range from 0.65 μg/g dry mat to 32.36 μg/g dry mat, with the highest amounts in the topmost layer 1 (Figure 15a,

Table 2. Concentrations of summed sterols, major hopanoids and fatty acids (FAs) in freely extractable and carbonate-bound lipids of the microbial mat layers (μg/g dry mat; n.d. = not detected).

Layer	Freely Extractable Lipids			Carbonate-Bound Lipids
	Sterols *	Hopanoids	FAs	Sterols *	Hopanoids	FAs
1	32.36	16.67	212.72	n.d.	n.d.	19.63
2	4.13	15.76	36.73	0.61	0.94	15.49
2b	15.41	28.77	9.24	6.53	26.98	6.28
3	0.65	3.98	3.34	0.72	3.22	80.45
4a	17.50	70.65	230.79	0.91	27.87	473.27
4b	0.73	5.25	9.89	0.73	1.86	1.2
4	0.67	3.34	4.74	1.24	2.24	21.22
5	4.24	9.76	11.90	0.32	0.22	6.87
6	1.02	5.70	12.50	0.73	0.72	8.64

* Including C₃₀ and C₃₁ sterols.

). Layer 4a, immediately below the thin mineral crust, shows an enhancement in sterol abundance, being the second highest concentration in the mat. Hopanoids, i.e., hop-22(29)-ene (diploptene), ββ-bishomohopanoic acid, ββ-bishomohopanol, have been observed in all mat layers. The summed hopanoids remarkably increase in layer 4a, up to 70.65 μg/g dry mat (Table S1, Figure 15a). For the FAs (i.e., iso-/anteiso-C₁₅-C₁₇ and unsaturated C_18:1 and C_19:1 FAs), their summed concentrations show ranges between 10⁰–10² μg/g dry mat. Like sterols and hopanoids, the abundance of FAs exhibited a significant increase in layer 4a, reaching 230.79 μg/g dry mat (see Table 2 and Table S1). In all freely extractable lipids, concentrations in layer 4a were in the same order as in the topmost and biotically active layer 1. Consequently, two declining trends for these three groups of compounds occur in the studied mat: a first one from layer 1 to the thin mineral crust of layer 3, and a second one from layer 4a downwards.

Carbonate-bound lipids also include various steroids, hopanoids and fatty acids as in the freely extractable lipid fraction. For sterols, the summed concentrations are one or two orders of magnitude less abundant than freely extractable sterols, ranging from 0.32 μg/g dry mat to 6.53 μg/g dry mat (Table 1b; Figure 15b). Besides, these carbonate-bound sterols were not detected in the topmost layer 1. Hopanoids in the carbonate-bound fraction were also not detected in layer 1, and in other layers their major summed hopanoids (i.e., hop-17(21)-ene and ββ-bishomohopanoic acid) show lower abundance, an order of magnitude less abundant, than in the freely extractable pool. The highest value is observed in layer 4a (27.87 μg/g dry mat), from which the abundance decreases downwards in the mat (Table 2; Figure 15b). Regarding FAs, similar major groups (i.e., iso-/anteiso- C₁₅- C₁₇ and unsaturated C_18:1 and C_19:1 FAs) were observed, and the summed concentrations are of the same order as in the freely extractable fraction (10⁰~10² µg/g dry mat range). The exception is found in layer 1, where FAs are an order of magnitude less abundant than in the freely extractable lipids (10² µg/g dry mat range, see Table 2 and Table S2). Interestingly, the summed FAs show a significant increase in abundance in the thin mineral crust (layer 3), and even more enriched in layer 4a (highest concentration: 473.27 µg/g dry mat; Table 2). Below, their abundances decrease sharply, as observed for layer 4b (1.2 μg/g dry mat; Table 2), then similar low abundances are sustained in the deeper mat layers. In both lipid fractions, the relative abundance of C₃₁ sterols shows general increase in depth, with a local peak in the thin crust (layer 3), while the other C₂₇-C₃₀ sterols decrease downwards in the mat (Figure 16).

3.4. Carbon Isotopic Compositions of Microbial Lipids and Hydrocarbons

FAs (including C₁₄-C₂₆ homologues), and hopanoids (including C₃₀ hopene and C₃₂ hopanol) as well as hydrocarbon (C₁₇ n-alkane) were identified. GC-C-IRMS measurements revealed relatively high δ¹³C values (i.e., heavier than −20‰) in both lipid fractions of the studied mat, ranging from −15.4 to −20.7‰ for C₁₇ n-alkane, −4.4 to −16.4‰ for FAs, −17.6 to −20.2‰ for hopanoids, and −7.2 to −13.3‰ for sterols (

Table 3. Carbon isotopic signatures of lipid biomarkers observed in the studied mat (unit: ‰ relative to V-PDB; n.d. = not detected).

	Layer	1		3		5		6
Compounds		Freely Extractable Lipids	Carbonate-Bound Lipids	Freely Extractable Lipids	Carbonate-Bound Lipids	Freely Extractable Lipids	Carbonate-Bound Lipids	Freely Extractable Lipids	Carbonate-Bound Lipids
n-alkane
17		−15.4	−16.1	−18.7	−20.7	−17.3	−17.4	−15.8	−15.7
FAs
14:0		−8.5	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
iso-15		n.d.	n.d.	n.d.	−11.8	n.d.	−16.4	n.d.	−12.1
15:0		−4.4	−8.6	n.d.	−4.7	n.d.	n.d.	n.d.	−11.0
16:1		−9.4	−8.7	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
iso-16		n.d.	−9.3	n.d.	−12.6	n.d.	−12.4	n.d.	−12.9
16:0		−6.8	−12.8	n.d.	−16.4	n.d.	−14.3	n.d.	−13.7
iso-17		n.d.	n.d.	n.d.	n.d.	n.d.	−9.3	n.d.	−12.8
anteiso-17		n.d.	n.d.	n.d.	n.d.	n.d.	−10.5	n.d.	−12.0
17:0		−4.8	n.d.	n.d.	−12.9	n.d.	−13.8	n.d.	−13.3
18:1		−11.6	−11.4	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
18:0		n.d.	−11.4	n.d.	−14.3	n.d.	−12.5	n.d.	−12.9
19:1		−11.8	−10.7	n.d.	−16.3	−15.0	−13.8	n.d.	−12.9
24:0		n.d.	n.d.	n.d.	−14.8	−8.5	−14.4	−7.8	−15.1
26:0		n.d.	n.d.	n.d.	−14.5	n.d.	n.d.	n.d.	−15.9
Sterols
C31 ∆0		−7.2	n.d.	n.d.	n.d.	−12.5	n.d.	−13.3	n.d.
Hopanoids
C30 hopene		n.d.	n.d.	n.d.	n.d.	−20.2	n.d.	−18.5	n.d.
C32 hopanol		n.d.	n.d.	n.d.	n.d.	−17.6	n.d.	−17.8	n.d.

).

Of all the identifiable isotopic signals, short-chain FAs (including C₁₄-C₁₇) showed highly enriched δ¹³C values (ranging from −4.4 to −14.3‰), while ¹³C-depletion was observed for their long-chain homologues (including C₁₈-C₂₆; ranging from −10.7 to −16.3‰). δ¹³C values for FAs in the deeper, older layers of the studied mat were consistently depleted by c. 0.2–11.5‰ as compared with the top layer 1 (see Table 3). This depletion trend is even stronger in the mineral crust of layer 3, as evident in the long-chain FAs as well as C₁₇ n-alkane (see Table 3). Another characteristic feature of carbon isotopic distribution in the studied mat is that much lighter ¹³C contents observed for the hopanoids compared with the other lipids (see Table 3), ranging from −17.6 to −20.2‰.

4. Discussion

4.1. Interpretation of the Mat Growth History and Significance of the Mineral Crust

The detailed petrographic study of the microbial mat presented here, together with the previously published ¹⁴C dates (Figure 1c; Blumenberg et al. [30]) indicate that the mat records two distinct phases of development with dissimilar growth rates, separated by the mineral crust of layer 3 (Figure 1c and Figure 2). The older phase is represented by the three lower layers (4–6, c. 7 cm thick, Figure 1c and Figure 2), which formed in less than 350 years (from 1440 ± 40 to 1111 ± 40 years BP). The younger phase is represented by the two upper layers (1–2, c. 3 cm thick, Figure 1c and Figure 2), which formed in more than 550 years (from 551 ± 40 to 62 ± 40 years BP) and maybe up to 1000 years (if the age of the underlying layer 3, 1111 ± 40 years BP, is considered). Therefore, according to these dates, the younger and thinner phase seems to have accreted at a much slower rate than the older phase.

Despite their different thickness and accretion rate, both mat phases show similar mineral composition, being mainly formed by irregular micritic aggregates with micropeloidal texture and sub-spherical particles with crystalline fibrous-radial texture. These are the typical mineral precipitates of microbial mats from Kiritimati [26,27,40,41], and are also very common in many other examples of modern and fossil microbialites [47,48,49,50,51]. The main difference between the mineral precipitates of the two growth phases is their abundance, size and development stage. In the upper, younger layers, where the EPS matrix is fresher and denser (Figure 3a,b), mineral precipitates are less abundant and developed: irregular micritic particles with micropeloidal texture are few and small, and subspherical fibrous-radial particles are also small and typically not merged with each other (Figure 2 and Figure 4d). On the contrary, in the lower, older layers, where the EPS matrix is scarce and more degraded (Figure 3c,d), subspherical particles are more abundant and merged in large aggregates with micropeloidal micrite between them (Figure 4f and Figure 5). Thus, the taphonomic evolution of EPS downwards in the mat is consistent with the downwards increase in mineral size and abundance, supporting the commonly interpreted role of the degradation of microbial EPS as a trigger to mineral precipitation within microbial communities and to the formation of microbialites [23,44,52,53]. Therefore, the upper and lower parts of the studied sample may be interpreted as two different growth phases of a microbial mat, formed probably under very similar hypersaline environmental conditions. Their difference in accretion rate could have been caused by slight hydrochemical differences, which in turn could have influenced microbial activity and mat accretion. In any case, these two phases represent different development stages: the upper part being the earliest steps of mineralization and the lower part characterizing a more evolved proto-microbialite, after ~1400 years of degradation of EPS and of mineral precipitation.

Interestingly, the boundary between those two development stages of the mat is marked by the mineral crust of layer 3, which shows strong contrasts with the other layers (Figure 1c and Figure 2) and with most mineral precipitates previously described from hypersaline mats of Kiritimati [26,27,41]. The ¹⁴C dates of the mat (Blumenberg et al. [30]) seem to indicate that the crust was formed during an interruption in microbial mat accretion, c. 1100 years BP. In addition, the strong contrast in mineral precipitation style between the crust and the other mat layers suggests that a significant environmental (i.e., hydrological) change occurred during the interruption in accretion. The µm-scale superposition of fibrous-radial botryoidal laminae (Figure 6c) indicates that precipitation was episodic, yet occurred repeatedly. Similar botryoidal or fan-like fibrous-radial fabrics have been described in fossil microbialites and interpreted to form under different hydrological settings, ranging from freshwater to hypersaline conditions [54,55]. In Kiritimati, the only similar microfabric has been described locally as fibrous aragonite veneers interpreted to precipitate at low CaCO₃ supersaturation [27]. The fact that Kiritimati lakes may experience inflow of fresh groundwater [25] makes it plausible that the interruption in the hypersaline mat accretion, and the subsequent formation of the crust of layer 3, were related to a hydrochemical change due to freshwater input, which would reduce salinity and alkalinity. A complementary interpretation would be that the environmental change was triggered by a partial or total desiccation of Lake 2. This could be supported by the presence of meniscus precipitates between subspherical particles in layer 4 (the top mat layer of the first development phase), which are typically interpreted as formed in vadose conditions [56,57,58]. After desiccation, the beginning of the new filling stage of the lake may have had different hydrochemical conditions (e.g., salinity and alkalinity), which could explain the different precipitation mode of the crust. In the context of Lake 2, desiccation episodes are plausible, given the small size of the lake and the strong climatic variability of Kiritimati island, mainly due to ENSO cycles, which strongly influence the water level and salinities of the lakes [24,25,29,40].

Regardless of the exact parameters and conditions that triggered the environmental change between the first hypersaline mat phase (layers 4–6) and the mineral crust of layer 3, that change is further supported by the biotic remains present within the crust (Figure 8, Figure 9, Figure 10, Figure 11 and Figure 12). Coccoid cyanobacteria are dominant in the current active mat (as in most hypersaline Kiritimati mats [27,28,39], however are only locally preserved within the crust (Figure 6b and Figure 8). Diatom molds are ubiquitous in the crust, although diatoms are absent or very rare in the current hypersaline lake 2 and in the topmost active part of the mat (layer 1), and in addition they are only scarcely preserved in the lower layers 4–6. Similarly, the high abundance of filamentous microbes preserved as molds within the crust contrasts with the currently active mat, dominated by coccoid cyanobacteria. Filamentous cyanobacteria occur in modern hypersaline mats of Kiritimati [27,28], however are morphologically different from the molds observed within the crust, whose size and morphology (Figure 8, Figure 9, Figure 10, Figure 11 and Figure 12) make them similar to Actinobacteria. Actinobacteria are Gram-positive bacteria with high G+C content and they look like filamentous fungi, forming a mycelium-like network of very thin fibers non-typical for cyanobacteria [43]. The fibers have a thickness of less than a micron and they exhibit various spore-types and sporangia [43]. In addition, the filaments are branching, which is not observed in filamentous cyanobacteria (compare Figure 10b,d with figures in Li et al. [43]). Actinobacteria are common in soil biocrusts [59], which exhibit structural similarities to aquatic stromatolites-microbialites. In Kiritimati, only small amounts of Actinobacteria have been detected within the lower older portions of microbial mats from Lake 21 [28], and they have never been observed in the younger and active parts of hypersaline microbial mats. To our knowledge, there are no previous studies about the possible mineralization behavior of Actinobacteria, and there are very scarce examples of fossilized specimens [60,61,62]. Therefore, this might be one of the first studies showing their mineralized preservation in carbonate settings. Nevertheless, the fact that their mycelium-like structures are so abundantly and pristinely preserved, not only in the thin crust (layer 3) of the studied microbial mat of Lake 2, however also in similar crusts of older subfossil microbialites from other lakes of Kiritimati (Figure 13), suggests that they have a significant preservation potential in the fossil record, at least under some circumstances, e.g., rapid precipitation of aragonite crusts. The subfossil microbial carbonates of Lakes 21 and 22 are interpreted as the result of a mineralized vein systems, through which shallow groundwater is transported, and which forms a large meter-sized polygonal pattern in the lake [63].

Another interpretation for the filament molds observed within the crust could be that they represent microboring microorganisms (e.g., cyanobacteria, fungi) penetrating the crust botryoids after their precipitation, as occurs in other modern microbial mats and biofilms [64,65]. This interpretation could be supported by the fact that in SEM the filaments are observed to cross through several aragonite crystals (Figure 12). However, the filaments are not observed to penetrate the aragonite botryoids from above and SEM images of the upper surface of the crust do not show circular holes consistent with the size of the filaments. In addition, some of the botryoids are covered by a thin crust of short aragonite crystals, which are not penetrated by these thin filaments (Figure 8a,d and Figure 9a). Only some of the scarcer brown-colored and larger filaments seem to penetrate the upper part of the crust (Figure 11a), and maybe their lack of fluorescence (Figure 11b) could indicate that they are microborings performed by endoliths that have eliminated, with their perforation, the organic matter present within the crust. In contrast, the highly abundant and smaller filaments do not seem to have been produced by microborers, seeming more like original mat microbes, due to their erected position and tufted grouping, sometimes equivalent to the botryoid shape and size (Figure 8a,d and Figure 9a–c).

Thus, it is interpreted here that both diatoms and Actinobacteria were original microbes of the mat then, and were entombed by a relatively rapid growth of the aragonite botryoids, as observed in other similar fibrous-radial microbial crusts [54]. In fact, some of the aragonite botryoids mimic the shape of filament tufts (Figure 8a,d and Figure 9a–c), indicating that the tufts were probably used as a template for mineral growth. This entombment of original microbes is further supported by the abundant pores of the crust filled by EPS remains (Figure 12b,c,e), suggesting that diatoms and Actinobacteria were part of a microbial community held together by the cohesive action of EPS. Rapid and episodic precipitation of fibrous-radial aragonite botryoidal laminae within the biofilm caused the fossilization of its microbes and EPS within the mineral crust. The contrast in biotic composition of that community with the two phases of hypersaline microbial mat development further suggests that an abrupt modification of environmental parameters (probably a desiccation and/or freshening of the lake) caused the interruption in mat accretion and the subsequent biotic change. In fact, it is noteworthy that Actinobacteria do not occur in modern active microbial mats of the hypersaline lakes of Kiritimati [28,39,66], which contrasts with their extreme abundance within the thin crust of layer 3. The fact that Actinobacteria thrive in special and extreme environments (e.g., Qin et al. [67], and references therein) and that some of them produce antimicrobial substances [68,69,70] is consistent with a significant environmental change in lake 2, which led to the interruption in accretion of the mat and to a radical modification of its microbial community.

In summary, the evolution of the studied mat is here envisaged as consisting of three stages: (i) a first phase of microbial mat accretion, developed under hypersaline conditions probably very similar to the current ones in Lake 2; (ii) interruption of mat accretion due to an environmental change (probably desiccation and/or freshening of the lake waters), and development of a thin microbial community, which was periodically fossilized by rapid precipitation of fibrous-radial botryoidal aragonite (probably during particular periods of increased alkalinity); (iii) reinstatement of hypersaline conditions in Lake 2, triggering development of a new generation of microbial mat with similar features to phase i. Although equivalent thin fibrous-radial botryoidal crusts have not been previously described in modern mats from Kiritimati, the fact that they occur in several subfossil microbialites (Figure 13), which represent older generations of microbial mats, shows that the formation processes of the crust are relatively common in those environments. In addition, similar fabrics have been described in other microbialite examples [54,55], further pointing to their common occurrence in the geological record.

4.2. Lipid Preservation and Carbonate Precipitation within the Mat

A variety of C₂₇-C₃₁ sterols have been observed in the studied mat, suggesting multiple sources mainly from algae, animals, fungi, and terrestrial plants [38,71,72,73,74]. The major portion of GC-amenable hopanoids observed in the mat might be formed as the diagenetic products of bacteriohopanepolys (BHPs; [75,76,77]) through microbially-driven, progressive defunctionalisation. The predominance of FAs observed are saturated and monounsaturated C₁₆ and C₁₈ together with a cyclopropyl-C₁₉ homologues, implying the primary contributors are bacteria [78,79,80,81]. Highly enriched δ¹³C values have been reported for the C₃₁ sterols and the fatty acids from layer 1, indicating their autochthonous input and derived from the CO₂-limited ecosystem of hypersaline Lake 2 [38,82].

Figure 16 shows the relative abundance of the observed sterols in both lipid pools, and the general increase downwards of C₃₁ against C₂₇-C₃₀ sterols (with a particular increase in the thin crust of layer 3) indicates that C₃₁ sterols experienced different degradation patterns compared with the other conventional sterols [38]. Figure 15 shows depth distribution of the summed GC-amenable sterols, hopanoids, and FAs in both lipid fractions. We observe these three group compounds exhibits significant increase in layer 4a, and their summed concentrations of each group are of the same order as the topmost layer 1 (Table 2; Figure 15). In particular, FAs shows very similar abundances between layers 1 and 4a.

These observations support the interpretation that two growth cycles may have occurred in this mat, and that they were interrupted by the formation of the mineral crust of layer 3. Layer 4a, located right below the crust, might represent the once active surface mat community of the first growth phase. Due to the interruption of growth of that first mat (probably due to desiccation and/or freshening of the lake), all microorganisms thriving in the mat and adapted to hypersaline conditions (including the heterotrophic degraders), may have died away relatively fast, at least before its organic matter could be degraded. Subsequently, the surface of that first mat, with its abundant lipids, was sealed by the thin mineral crust, which accreted in relatively fast episodes (as reflected by the pristinely entombed biotic remains). Thus, the mineral crust could isolate the former mat surface and its lipids from further degradation, which would explain the unusual lipid abundance, only comparable to the current surface of the modern hypersaline mat.

In conclusion, it is proposed here that the stable and sustained hypersaline conditions that predominated during both growth phases of the studied mat are very unlikely to produce a significant preservation of biomarkers within the carbonate matrix of the resulting microbialites, as during those periods most lipids would be continuously digested by heterotrophs in the deeper mat layers. Only a special event, such as the drastic environmental change occurred between both growth phases and the subsequent precipitation of the thin mineral crust formed that eliminate the degradation, and rapidly lithify the surface mat community and its organic compounds.

4.3. Relationship between the Mineral Crust and the Carbon Isotope Signature of Lipids

Highly enriched δ¹³C values (weighted average c. −10‰) were observed for the short-chain FAs in the studied mat, clearly implying their carbon sources were autochthonous and derived from the hypersaline, CO₂-limited ecosystem of Lake 2 [82]. This ¹³C-enrichment in microbial lipids is also consistent with high δ¹³C values for eukaryotic sterols observed at the top layer 1 of the studied lake 2 mat [38]. Furthermore, the long-chain FAs were generally depleted (by weighted average 0.1–8.2 ‰) than the short-chain homologues, indicating minor contributions from terrestrial higher plants [82]. Hopanoids showed lower δ¹³C values than the other lipids, suggesting less amounts of ¹³C contents that originate from methanogens or anaerobic methane-oxidizing archaea [83]. In addition, van der Meer et al. [84] reported isoprenoid lipids were 7 to 9 ‰ in ¹³C-depletion relative to the straight-chain lipids, due to the utilization of the reversed tricarboxylic acid cycle. The observed hopanoids results (pentacyclic isoprenoids) in the studied mat are well within the range of the previously published isotope data of isoprenoids [84].

Throughout the mat profile, slightly lighter δ¹³C values were observed in the deeper, older growth phase of the mat (layers 5 and 6) than the topmost layer 1, suggesting a decrease in suppression effect caused by CO₂-limitation of local Lake 2 ecosystem. This is also observed in the mineral crust of layer 3, which shows more ¹³C-depleted values. Furthermore, FAs and hydrocarbon (i.e., C₁₇ n-alkane) show stronger ¹³C-depletions in layer 3, which further supports the interpretation proposed here for the formation of this layer: strong hydrological changes at the end of the first growth cycle of the mat resulting in a significant change of biota, and in the depleted δ¹³C value of its lipids, recorded within the thin mineral crust of layer 3.

5. Concluding Remarks

The detailed geomicrobiological work presented here (encompassing different petrographic techniques for both organic and mineral components of the mat, as well as mineralogical and organic geochemical analyses) shows that the studied microbial mat, from a hypersaline lake of the Kiritimati atoll, had a complex development in its ~1500 years of history. Two main growth phases occurred under hypersaline conditions similar to the current ones. The first growth phase was interrupted by a significant environmental change (probably desiccation and/or freshening of the lake), which led to a modification of the microbiota and to the precipitation of a thin mineral crust. Resumption of the hypersaline conditions produced the second (and actual) growth phase of the mat. During this history, two different types of aragonite precipitation have been recorded. The main growth phases were dominated by the relatively slow formation of subspherical particles and irregular micritic particles with micropeloidal texture, both quite common in microbialites and linked to the degradation of the EPS matrix of the mat. In contrast, the interruption stage between the phases is recorded by a thin and laterally continuous crust composed of superposed fibrous aragonite botryoids, which precipitated in relatively rapid episodes, entombing the microbiota present at the surface.

These two contrasting types of microbially-influenced carbonate precipitation occur within the same microbial mat and they induce very different pathways for the preservation of its lipid biomarkers, both in their abundance and in their carbon isotopes. The lipids present within the thin mineral crust show a particular isotopic signature, different from that of other layers. Furthermore, the relatively rapid precipitation of the crust acted as a seal that protected the lipids immediately below from degradation, causing an abundance and preservation pattern of lipids equivalent to that of a modern active microbial community, despite being >1100 years old. The fact that similar thin, although laterally continuous, crusts occur in subfossil microbialites of Kiritimati and in other microbialite examples indicates that the complex preservation pathways and patterns of biomarkers shown by the studied example have the potential to be more a rule than an exception, and they can be tested in future works on lithifying microbial mats or fossil microbialites. Therefore, this modern example can be used as an analogy for other biomarker investigations, especially since it clearly shows that care must be taken when analyzing microbialite biomarkers: the different microbialite layers and fabrics should be analyzed separately and compared meticulously, and organic geochemistry should always be coupled with thorough petrographic and mineralogical techniques.