Unveiling microbial preservation under hyperacidic and oxidizing conditions in the Oligocene Rio Tinto deposit

The preservation of biosignatures on Mars is largely associated with extensive deposits of clays formed under mild early Noachian conditions (> 3.9 Ga). They were followed by widespread precipitation of acidic sulfates considered adverse for biomolecule preservation. In this paper, an exhaustive mass spectrometry investigation of ferric subsurface materials in the Rio Tinto gossan deposit (~ 25 Ma) provides evidence of well-preserved molecular biosignatures under oxidative and acidic conditions. Time of flight secondary ion mass spectrometry (ToF–SIMS) analysis shows a direct association between physical-templating biological structures and molecular biosignatures. This relation implies that the quality of molecular preservation is exceptional and provides information on microbial life formerly operating in the shallow regions of the Rio Tinto subsurface. Consequently, low-pH oxidative environments on Mars could also record molecular information about ancient life in the same way as the Noachian clay-rich deposits.

www.nature.com/scientificreports/ activity itself through direct or indirect mechanisms that can trigger or even increase mineral precipitation on the cell surface 14,15 . Iron oxides have been claimed to preserve molecular biosignatures with implications for detecting traces of life on Mars 16 . However, acidic and oxidizing environments have currently been ruled out as ideal conditions for the preservation of molecular biosignatures. This is also the case for Martian materials that have resulted from alteration or precipitation due to the activity of such acidic and oxidizing fluids 17,18 . The reason lies in the fact that these solutions inherit a high oxidizing capacity that can be released through different biotic or abiotic pathways. If life arose on Mars, it could have been a very active agent for recycling inorganic and organic compounds by using inorganic electron acceptors (e.g., Fe 3+ and SO 4 2− ) that are abundant in acidic fluids 19 . Although the solar X-ray and UV irradiation rates occurring in the Hesperian are unknown, different oxidants that would surely have been produced by incident radiation on the atmosphere and hydrosphere 7,20 would have oxidized biomolecules and organic compounds without the participation of microbes. As the preservation potential of molecular biosignatures under acidic conditions is considered very low, the planetary community has ruled out the list of potential targets to search for molecular evidence of extinct life on Mars 17 . Therefore, deposits formed under mild conditions (neutral pH, low oxidizing to reducing conditions) have become first-class targets to find traces of living organisms on early Mars 21 .
On Earth, records of ancient life activity can occur in the form of physical structures replicated by mineral precipitation 22 . Indeed, as the final diagenetic byproducts of acidic deposits are goethite (FeOOH) and hematite (Fe 2 O 3 ), the preservation potential of mineralized structures is extremely high. The diagenetic processes affecting acidic materials consist of a combination of loss of sulfate ions and dehydration under oxic conditions that are controlled by the inflow of meteoric solutions after exhumation of materials from the initial acidic conditions 23 . Very recently, a diverse association of lipids has been detected in modern to recent materials formed under acidic and oxidizing conditions 24 , which suggests that organics can be preserved in the first stage of diagenesis. Indeed, the preservation potential of acidic deposits for organic compounds has been wrongly dismissed, as different organic compounds 25 , including evidence of peptide chains 26 , have been found in the 2-Ma upper terrace of Rio Tinto.
In this paper, long-term preservation of complex molecular biosignatures in ancient Rio Tinto subsurface ferruginous deposits (~ 25 Ma) is targeted. This task was accomplished by analyzing samples through state-ofthe-art time of flight secondary ion spectrometry (ToF-SIMS) with Fourier transform ion cyclotron resonance mass spectrometry (FTICR-MS) and gas chromatograph mass spectrometry (G-MS). The identification of large molecular species suggesting that the acidic ferruginous materials have a high preservation potential agrees with the finding of peptide chains 26 in the oldest Rio Tinto terrace, which also requires exceptional preservation conditions for its persistence. In this regard, preservation is favored in the Rio Tinto acidic deposits despite high water availability for mineral transformation. As this was not the case for sediments formed under low-pH conditions on Mars with limited water 23 , high preservation should be expected in those Hesperian materials. In this context, the extensive ferruginous deposits formed in acidic and oxidizing Hesperian environments should be considered first-class targets to look for ancient traces of life if any biosphere emerged at some time on the red planet.
Geobiological settings. Gossan deposits occur in the Peña de Hierro area (Fig. 1), which is considered one of the sources of the acidic waters of Rio Tinto 12 . Indirect rock dating using geomagnetic techniques 27 suggests that the gossan materials in this location were formed through the acidic weathering of Carboniferous sulfide orebodies 28 during the late Oligocene 27 . Such alteration coincides with a thermal climatic event 29 that likely enhanced the acidic weathering of the orebodies. This was followed by a drop in the water table in response to subsequent climatic evolution during the Messinian saline crisis starting by 6 Ma 27 .
where O 2 is used as an electron acceptor for the oxidation of FeS 2 and free Fe 2+ .
where M = Na, K or H 3 O, and. 8·Fe 3+ (aq) + SO 4 2− (aq) + 14·H 2 O (l) ↔ Fe 8 O 8 (OH) 6 (SO 4 ) (s) + 22·H + (aq). The iron oxides that are the main mineral components of gossan and other underground ferric deposits in Peña del Hierro 13 come from the diagenesis of primary acidic sulfates 31 (Fig. 2) when they are exposed to meteoric solutions with a circumneutral pH (~ 6-7). The process involves a combination of hydrolysis and dehydration that results in removing the sulfate groups to produce oxyhydroxides such as goethite (FeOOH) after dissolution-reprecipitation 32 . In addition, hematite (Fe 2 O 3 ) can be directly formed from acidic sulfates by dehydration, which is accelerated by temperature 31 . Furthermore, hematite can also be formed by dehydration of nanocrystalline goethite 32 or nanophase iron oxides 23 whose composition is unclear. The diagenesis of materials formed under acidic conditions (Fig. 2) is much more complex than a simple transformation of ferric sulfate to  10,32 . In the Peña del Hierro gossan, the abundance of hematite over other iron oxide species such as goethite 13,23 suggests that the protolith was mainly composed of ferric sulfates that evolved to Fe 2 O 3 during early diagenesis. Such a complex diagenesis that involves an interplay among dramatic pH and redox changes, hydrolysis and dehydration should have prevented the preservation of molecular compounds with a biological origin. However, the detection of abundant and diverse biomolecules in the Rio Tinto gossan shows that different preservation  Fig. 7) showing a laminated fabric not recognized through observation by the naked eye; this sample contains different 30-micron ovoid elements of likely fungal origin; because they correspond with the preservation of filamentous structures, the ovoid sections suggest that they have the same orientation. www.nature.com/scientificreports/ mechanisms have played essential roles in the fossilization of organic compounds that were produced by microbial activity in the subsurface of Peña de Hierro.

Results
To determine the preservation potential of ferruginous deposits that formed under acidic and oxidizing conditions, sample 010,109-1 ( Fig. 1b- [34][35][36] . The FTICR mass spectrum of the extracted Upper Gossan fraction is shown in Fig. 3a. All mass signals detected correspond to singly charged ions and were assigned and classified as O, S, N-functionalized CHO, CHNO and CHOS molecular compositions ( Fig. 3b-f). In fact, these molecular series are visualized in a van Krevelen plot (Fig. 3c), which is common to illustrate the molecular space of most complex natural samples 34,35,37 . The advantage of this plot is that it displays the assigned molecular species present in a sample rather than the computed average values. As shown in Fig. 3c, the van Krevelen plot of the Upper Gossan extract exhibits regular smooth and extensive distributions over large ranges of unsaturation (high H/C ratio), down to high degrees of oxygenation (high O/C ratio). Condensed compounds were relevant (Fig. 3c), had double bond equivalents (DBEs) higher than 3 with at least one or more phenyl rings, and were more abundant than saturated and unsaturated compounds with no phenyl ring (DBE ≤ 3) (Fig. S1). Moreover, the extracted gossan fraction revealed significant chemodiversity mainly attributed to CHO (39.0%), CHNO (51.0%), and CHOS (10.0%) molecular compositions (Fig. 3d-f). While CHO-bearing molecules have shown the presence of saturated/unsaturated compounds (e.g., fatty acid-like, glycerins and oxygenated hydrocarbons CHO n>2 ), carbohydrate-like and carboxylic-rich alicyclic molecules (CRAM), including phytochemicals such as terpenes, flavonoids, lignins, polyphenol compounds and derivatives, as significant tracers of terrestrial organic matter (Fig. 3d), CHNO-bearing molecules have shown enrichment in oxygenated compounds, peptide-like compounds and aliphatics (Fig. 3e). Fewer CHNO-bearing molecules are localized . The diagram shows that mineral maturation toward iron oxides requires hydrolysis and dehydration, which are controlled by changes in pH and Eh. The red arrows trace weathering pathways in the clays forming sedimentary deposits at the surface and subsurface of Rio Tinto. www.nature.com/scientificreports/ in the CRAM region (Fig. 3e). Such enrichment in CHNO-bearing molecules suggests more preserved bioavailable organic matter and its derivatives, which are known as drivers of microbial community structure and fundamental metabolic strategies [38][39][40] . Conversely, CHOS-bearing molecules display a narrow range of hydrogen saturation H/C (0.55-2.2) evocative of reduced CHOS organic compounds and more oxygenated molecular species bounded by approximate O/C (0.25-0.75) and H/C (0.6-1.4) values characteristic of CRAM (Fig. 3f). The appearance of CRAM and derivative compounds in the lignin-like and polyphenol-like regions on the van Krevelen plot could possibly arise from vegetation (i.e., vascular/higher plants and/or algal polyphenols). Indeed, the formation of organics by transitional organic matter involving monosaccharides, amino acids or fatty acids produces long-term humic organics via abiotic and/or humification reactions 41 . In parallel, the generation of insoluble, nonhydrolyzable and nonbiodegradable organics by various organisms may also be involved in the production and preservation processes of organic matter 41,42 . Overall, this large molecular diversity found in the Upper Gossan extract within and across the assigned molecular compositions is an example of the diverse biogeochemical availability of organic compound signatures, which is an acknowledged aspect of progressive organic matter preservation in the environment.

GC-MS molecular screening.
The analysis of the total lipid extract (i.e., including nonpolar, polar and acidic fractions) revealed the presence of diverse lipid families (i.e., normal and branched alkanes, including isoprenoids; alkenes; branched alkanols; steroids and hopanoids; and normal, branched, and monounsaturated fatty acids) ( Table S1). The nonpolar fraction was mostly composed of straight chain or normal alkanes (n-alkanes) with 11 to 31 carbons that showed a bimodal distribution with maxima at C 18 -C 21 , C 27 and C 29 . In conjunction with the n-alkanes, the isoprenoids pristane and phytane were also found ( Table S1). The acidic fraction was composed of n-fatty acids with 11 to 30 carbons, monounsaturated fatty acids with 16, 18 and 20 carbons, branched fatty acids of iso/anteiso configurations (i.e., methyl groups at ultimate or penultimate positions) with 13 to 17 carbons (from i-/a-C 13:0 to i/a-C 17:0 ), and other minor compounds (Table S1). The n-fatty acid series showed a dominance of even-over-odd carbons, with maximum peaks at C 24 , C 26 and C 28 . Meanwhile, the polar fraction was primarily composed of n-alkanols with 11 to 24 carbons and even-over-odd preference, glycerol compounds suggesting the occurrence of glycolipids, a lupane derivative (lupan-3-ol, acetate), and a series of sterols, including cholesterol β-sitosterol, stigmasterol, and an ergostanol derivate (Table S1). The observed lipid distribution patterns provide information about possible sources of organic matter in this sample, which are mainly related to bacteria, algae/phytoplankton and plants. The latter source is inferred from the detection in the sample of odd and high molecular weight (HMW) n-alkanes such as C 25 for macrophytes www.nature.com/scientificreports/ and C 27 , C 29 , and C 31 for terrestrial plants 43,44 . In contrast, the detection of low molecular weight (LMW) n-alkanes (< C 22 ) is associated with either bacterial (mostly even LMW n-alkanes) or phytoplanktonic (odd LMW n-alkanes) sources 45 . Pristane and phytane have as a primary source the degradation of chlorophyll pigments. Both isoprenoids mainly originate from phytol, the esterifying alcohol of phototrophic chlorophylls 46 , whose degradation gives rise to pristane or phytane in the presence or absence of oxygen, respectively 47 . Vegetal and microbial biomarkers are also found in the acidic and polar fractions, where LMW n-fatty acids and n-alkanols (i.e., < C 20 ), as well as the hopanoid lupan-3-ol and acetate, are most likely associated with bacterial sources 48 , whereas HMW n-fatty acids and n-alkanols, as well as phytosterols (i.e., β-sitosterol or stigmasterol) can be related instead to macrophytes or land plants 43,49 . Sterols are ubiquitous constituents in all eukaryotic organisms, where stigmastanol or β-sitosterol is produced almost exclusively by higher plants 50 , whereas cholesterol is primarily produced by animals 51 . The presence of ergosterol can be related to fungal species 52 .
ToF-SIMS molecular mapping. ToF-SIMS molecular imaging has allowed the characterization of different groups of molecules that are preferentially associated with distinctive morphologies, the sizes of which are generally smaller than 500 μm (see Table 1 where groups and morphologies are associated with compounds). In sample 010,109-1, TA fabric and texture fit the distribution of inorganic and organic ions well (Figs. 4, 5, 6, and 7; Figs. S2-S4), which mainly correspond to filamentous networks and isolated elements associated with organic compounds of biological origin (e.g., see Figs. 4, 5, 6, and 7). In some cases, ToF-SIMS ion mapping is the only technique for recognizing the microstructures of samples when this is not possible using different microscopic techniques (e.g., light or scanning electron microscopes). Furthermore, the TOF-SIMS ioninduced secondary electron images also provided additional information on the microstructure of the sample  ) that is affecting some microstructures in G2 and G5. As discussed in the main text, the composition and structure of groups G1 to G8 suggest that the groups could come from the preservation of different biological structures such as filamentous fungi (G1 and G3), or plant fragments (G5 and G6).  (Fig. 4) than filament networks found in TA2, which are thinner, overlap more tightly, and show 20-micron nodular ends (Fig. 6). In TA3, the distribution of positive and negative ions outlines an elliptic microstructure that is perpendicular to straight laminar projections (Fig. 7). Group 1 shows maximum intensities in different molecular groups, including a series of N-bearing compounds (e.g., m/z peaks at 268. 30   Interestingly, G4 appears as a compositional variation in the G3 structure. The second RGB association (b) could include three different groups, namely, G1, G3 and G8 that could correspond to a microbial structure such as a biofilm in which the biomass is dominated by microbial forms biomineralizing sulfate. G3 structures interlock empty complementary areas in G1, suggesting that these areas could be the remains of different biological entities, while G8 branched microstructures merge with some filamentous units in G1, suggesting compositional variations in some specific zones of the filamentous mass of G1. www.nature.com/scientificreports/ Figure 6. Characterization of possible microbial structures preserved in Target Area 2 (TA2) by using the ToF-SIMS molecular mapping capabilities. The image analysis follows the same morphological groups G1 to G8 adding one new group G9. Two different main structures defined by G1 and G9 are limited by a sharp boundary between them. The G1 microstructure is characterized by a strong intensity of different N-and S-bearing compounds that outline a diffuse filament network, whose primary composition would be large lipids such as ceramides. However, G9 shows a more massive filamentous mass that is traced by a high intensity in acylglycerides and FA fragments. G3 occurs as an independent group of microstructures that interlock both G1 and G9 microstructures. The RGB overlay shows two main microbial structures (G1 and G9) in which empty cavities would be occupied by a third microbial unit defined by G3. Interestingly, the G1 microbial structure is strongly mineralized by sulfate while G3 and G9 microstructures are mineralized by ferruginous materials, suggesting that they were formed in two separate mineralization episodes. The microbial structure and composition of G1 and G9 suggest that these groups could correspond to mineralized biofilms.  . Groups G1, G3, G4, and G9 outline different oval elements (< 80 microns) that are associated with parallel straight structures longer than 500 microns. Interestingly, groups G2, G5, and G6, which have been associated with phytochemicals as well as PAHs, occur in the mineral matrix that encloses the biological structure. Such a distribution suggests that the microbial structure had been transported or was thriving in a water mass that was enriched in refractory organic compounds such as PAHs and some phytochemicals. The RGB overlay shows that the biological structure has compositional changes that represent sulfate biomineralization (Na 3 SO 4 + , and K 2 NaSO 4 + ), as well as sphingolipid (C 4 H 10 NO + ) and acylglyceride fragments (C 16   . As discussed in the ToF-SIMS spectral characterization, the fragmentation pattern of positive and negative ions in Group 1 suggests that the group results from sulfate mineralization on a lipidic surface composed of sphingolipids ( Fig. S6; Table S2).
Group 2 A couplet of parallel linear elements (Figs. 4a) with an average thickness of > 80 microns is not associated with any structure defined by inorganic ions (Figs. S2-S3). In negative ions, the average morphology varies in thicker microstructures with internal triangle-like patterns and external irregular boundaries ( Fig. 4b;  Fig. S7). In some cases, the internal microstructure defined by the negative ions shows a more complex design that partially resembles the Group 1 pattern in TA1 (e.g., m/zpeak at 58.01; see Fig. S7a). Furthermore, in TA3, the anions defining Group 1 occur in the mineral matrix as a negative cast of the main structure (Fig. 7). The group shows a maximum in positive peaks at m/z 31 (Fig. S7b), which reflect anions This group is a set of different ovoid microstructures averaging 60 microns. In TA1, the positive ions define nodular microelements that follow a similar orientation (Figs. 4 and 5). In contrast, the negative ions outline a structure following a square-like shape limited by < 40-micron filamentous elements (Figs. 4 and 5). In TA2, the positive and negative ions outline nodulous rows that are closely parallel to the matrix fabric (Fig. 6) and are characterized by a sharp boundary separating two subzones with different microtextures (Fig. S3). In TA3, positive peaks show the same pattern as observed in Group 2, which has a maximum in the rock matrix, leaving the main microtexture as an empty cast (Fig. 7). The area is defined by m/z peaks at 28 (Fig. 4). One of the micronodules fits well with the distribution of a couple of micronodules found in Group 3. At the same time, the occurrence of the second micronodule complements the microelements of Groups 3 and 6 ( Figs. 4 and 5). This group corresponds to Group 3 of TA2 (Fig. 6), suggesting that it results from spatial variations in the molecular composition of organic N-bearing compounds with the same structure preserved (Fig. 5). In TA3, the group shows a pattern similar to that observed in Groups 2 and 3, which outline the main microstructure preserved as an empty cast (Fig. 7).
Group 5 This group is characterized by a 600 micron-long elongated triangle-like morphology crosscutting the main structure fabric of Groups 1 and 2 (Figs. 4 and 5). While the molecular fragments in TA2 are not associated with any specific morphology, in TA3, they follow the same pattern as observed in Groups 2 to 4, which outline the main microstructure as an empty cast. In fact, Group  . Such a distribution of fragments is consistent with the occurrence of piperonal derivatives that are common in phytochemicals 56 .
Group 6 Three 120-micron subcircular morphologies show an empty central void. Such subcircular microstructures are equidistant from each other at a distance of ~ 350 microns in TA1 (Fig. 4). The inner empty area of each edge is filled by ions of Groups 3 and 4 (Fig. 5). While there is no morphology associated with this group in TA2, the area TA3 follows the same pattern as observed in Groups 2 to 5 (Fig. 7). In fact, Group 57 .
Group 7 The occurrence of 10-to 50-micron micronodules follows the main pattern appearing in Groups 3 and 4 (Fig. 4). This group is found in only TA1 through the positive ion at 57.04 (C 3 H 5 O + ), as well as by negative fragments at 279. 16  . The positive and negative ion fragmentation pattern suggests that the source compound is a terpenoid-like structure occurring in different plant tissues, such as dioic acids found in the cuticular wax or terpene-like fungal metabolites 58 (e.g., pestaloporonin derivatives).
Group 8 In this group, microstructures characterized by a few negative ions arranged in two sets (at m/z 666.06, 667.07, 668.07, 736.14, and 737.15) occur only in TA1 (Fig. 4). Such fragments outline structures with short branches with an average length of 200 microns that are transverse to the fabric of Group 1 (Fig. 5). Thus, the anion mass fits different fragments with high oxygen moieties (i.e., C 27  www.nature.com/scientificreports/ which appear in TA2 along the sharp boundary that contacts Group 1 ( Fig. 6; Fig. S3). The distribution of positive and negative ions defining Group 9 fits the primary microstructure found in TA3 (Fig. 7), while in the TA1 matrix, the positive ions of this group show a more homogeneous occurrence.
ToF-SIMS spectral analysis: compound classification. The spectral analysis of the sample 010,109-1 TAs provides the identification of different biomolecular fragments and organic compounds. Although the spectral fragmentation patterns are very intricate as a result of secondary ionization between the C-bearing compounds and the Fe-and S-bearing mineral matrix, it is possible to identify diverse groups of organic compounds, which are characterized in detail in the Supplementary Information. In short, the ToF-SIMS spectral analysis identifies the following molecular groups.
1. n-Alkanes and other straight hydrocarbons with low degrees of unsaturation The occurrence of saturated and monounsaturated straight chain hydrocarbons (i.e., n-alkanes and alkenes) is supported by the detection of low mass peaks (< 120 Da). The peak intensity and distribution of the two sets of ions suggest that alkenes are the dominant forms among the straight hydrocarbons. This feature implies the presence of abundant large lipid compounds in the Upper Gossan unit 61 that should be composed of saturated and monounsaturated chains. The TA cation fragmentation pattern also occurs as straight hydrocarbons that form NH 4 + -bearing adducts 62 (Fig. S6). The fragment composition and distribution strongly suggest that the N-bearing compounds and adducts are produced by the ionization of a matrix enriched in different sphingolipids, which are associated with filamentous fungal structures 45,63,64 (Fig. 8).

Branched hydrocarbons and isoprenoids
The occurrence of high-intensity peaks at 43.06 and 57.07 is also consistent with the preservation of different branched hydrocarbons, including isoprenoids, that are detected using GC-MS (Table S1).

Polycyclic aromatic hydrocarbons (PAHs)
A set of TA1 and TA3 peaks suggests the occurrence of PAHs 65,66 ( Table S3). The PAH cation distribution is homogeneous in the mineral matrix (Fig. 7), suggesting that it was produced during organic mineralization in the Upper Gossan unit. Such a process resulted from the combination of fungal and microbial degradation of cellulose, lignin, pectin, and other polysaccharides 67,68 , as well as other sources such as woodland fires 69 . 4. Fatty acids The mass spectra of negative fragments obtained through ToF-SIMS record a diverse set of fatty acids (FAs) ranging from C 11 to C 30 (Fig. 9). These FAs include saturated, unsaturated, and hydroxylated compounds (Table S4), the distribution of which follows the TA fabric (Figs. 4, 6, and 7; Figs. S2 to S4). A rough estimation of FA abundance shows a higher concentration of fatty acids in TA2 than in TA1, agreeing with a higher concentration in glycerolipids 70 (Fig. 6). The anion intensity distribution shows that the concentration of FA chains with fewer than 22 carbon atoms is higher than that of > n-C 21:0 FAs (Fig. 9). The total sum of the [M-H]anion intensity shows that in TA1 and TA2, the saturated and monounsaturated ≤ n-C 20 FAs have a higher intensity than the > n-C 20 FAs, while the total intensity even FAs exceeds that of odd FAs in ≤ n-C 20 chains (Fig. 9). Such a distribution pattern is also observed in the spectral data obtained from mineralized fungi that are used as a reference for FA composition 71 . The mineralized fungal reference (Fig. 9) shows the occurrence of ≤ n-C 18 saturated and monounsaturated chains, while saturated and monounsaturated FAs longer than 18 carbon atoms have negligible concentrations. In the ferruginized fungi 71 , the ≤ n-C 20 even fraction exceeds approximately three times the total intensity of ≤ n-C 20 odd FAs, as is also observed in TA1 and TA2 (Fig. 9). The ToF-SIMS spectral data also show the occurrence of polyunsaturated fatty acids (PUFAs), such as C 20:5 and C 20:4 , corresponding to eicosapentaenoic and arachidonic acids ( Fig. 9; Table S4). The TA1 and TA2 fatty acid distributions suggest that the record of organic compounds in the Upper Gossan unit has resulted in intermixing from heterogeneous biological sources, including microbial communities and higher plants 72,73 .
However, both even and odd FAs found in sample 010109-1 are associated with fungal and nonfungal microbial structures (see Fig. 4, 5, 6, and 7), which suggests that in the Rio Tinto subsurface, FAs are not mainly sourced from higher plants, although higher plants can be a secondary source. The occurrence of C 20:5 and C 20:4 PUFAs has been reported in different fungi 72,74 , as has been found in different higher plants. This is also the case for the long-chain hydroxylated FAs (≥ C 24 ) considered molecular markers of vascular plants derived from suberin 75,76 . Alternatively, they are also found in fungi 77 that are produced through the enzymatic oxidation of alcohols (e.g., lipoxygenase) and other FAs 78 .

Glycerolipids and glycerophospholipids
A diverse set of odd cations occurring in a wide range of masses > 300 Da agrees with the fragmentation pattern of different glycerolipids, including monoacylglycerides (MAGs), diacylglicerids (DAGs), and glycerophospholipids (GPLs) (Figs. S8, S9, S10, and S11; Table S5) 60,70 . The detection of diglyceride structures is additionally supported by the occurrence of different carbonyl-bearing fragments 79 (Table S5). Furthermore, different peaks found in TA1 to TA3 match cations that are well known as fragments of head groups for choline and phosphatidylcholine 79 . The presence of glycerophospholipids is supported by different peaks that occur at higher intensity in TA2 79 . 6. Sphingolipids (ceramides) The presence of different even peaks in TA1, TA2 and TA3 (Figs. 4, 5, 6, and 7) can be ascribed to fragmented sphingolipid backbone chains 80 such as C 20 H 40 NO + . These chains can be produced through two main dehydration pathways 80  www.nature.com/scientificreports/ , and TA3 (c). The peak distribution shows the occurrence of two different sets of fragments that include even versus odd masses corresponding to acylglycerides (red), NH 4 + -adducts (blue), and sphingolipid fragments (orange) (see Supplementary Tables 5 and 6). The ToF-SIMS molecular images show a high affinity between sphingolipid fragments and NH 4 + -adducts (a), suggesting that both are sourced from the same lipidic structures. Furthermore, the peak intensities of the different cations also support a change in the molecular composition for each target area. While TA1 (a) shows high intensity for NH 4 + -adducts (1200 cps at 550 Da), it is doubled by the acylglyceride intensity in TA2 (e.g., > 2000 cps at 551.53). TA3 (c) acylglyceride intensity is three times lower than that in TA2 (c). Such intensity and composition changes strongly suggest that TA2 (b) is the target area recording the highest biomass, which is dominated by acylglycerides, while sphingolipids would be the main compounds in TA1 (a) in terms of biomass. TA3 (c) appears as an intermediate area in terms of acylglycerides and sphingolipids which are mainly concentrated in the microbial structures (Fig. 7).  www.nature.com/scientificreports/ 6 oxygen atoms suggests the presence of different types of sphingolipids with chain lengths greater than C 38 . The mass distribution of positive ions above 530 Da agrees with the preservation of diverse ceramides, including cerebrosides and sphingomyelins 80 . Although the identification of different ceramide fragments with the same chain length but the number of oxygens varying between 1 and 6 is consistent with different dehydration pathways in the ionization process 80 , these fragments could also originate from the loss of ceramide hydroxyls during early mineralization. Furthermore, the decrease in oxygen and carbon numbers could also result from fragmentation of the glycosidic head in cerebrosides during the ionization process 80,82 (Fig. S12). However, microbial degradation cannot be excluded as a cause for the loss of glycosidic groups. 7. Peptide fragments and amino acids TA1, TA2 and TA3 have positive and negative peaks (< 150 Da) that match well with a diverse series of N-bearing fragments (Table S7) 83 . They could correspond to fragments of multiple amino acids (e.g., C 3 H 6 N + , C 5 H 10 N + , and C 5 H 12 N + ) and specific fragments for some amino acids, including lysine (C 5 H 10 N + ), asparagine (C 3 H 7 N 2 O + ), and aspartic acid (C 4 H 5 O 4 -) 84,85 . The potential preservation of protein fragments and amino acids in the oldest deposits of the acidic Rio Tinto basin, such as the Upper Gossan unit, is suggested by the detection of peptidic chains in the 2.1-Ma Rio Tinto terrace 26 . 8. Steroids and hopanoids Sample 010,109-1 shows m/z peaks that are characteristic of different cyclic triterpenoids, such as steroids. However, some of the diagnostic cations either have weak peaks or occur as shoulders on the sides of larger peaks, which prevents characterizing compounds at the level that has been achieved by GC-MS (Table S1). Diagnostic fragment masses for steroids 86 (Table S1). The detection of different peaks as fragments of different sterols shows that eukaryotic organisms such as fungi and plants were the main contributors to the gossan biolipid inventory. However, the most abundant sterol in terms of intensity, which is cholesteryl sulfate, suggests that the mineralization process driven by enriched sulfate solution produced secondary organic complexes composed of lipid structures bonded to inorganic compounds. 9. Polyphenols, lignin derivatives and other phytochemicals The ion mass distribution of some distinctive cation and anion peaks defining Groups 5 and 6 (Figs. 4, 5, 6, and 7) suggests the occurrence of compounds derived from phytochemicals such as terpenoids and polyphenols (Table S9) Furthermore, the spectral analysis provides a set of negative ions of unknown origin occurring at m/z 650.07, 666.06, 668.08, 736.15, 737.15, and 738.14 that occur in only TA1 as Group 8 (Figs. 4,5,6,7,8,9,and 10). This suggests the presence of compounds with a high number of O and/or N, which is compatible with the preservation of larger fragments of polysaccharides or polyketides.

Discussion
The analysis of the Upper Gossan sample through the different spectral approaches provides complementary evidence of the high degree of preservation of the acidic and oxic ancient deposits. The analysis of high-resolution FTICR-MS results yields a data source to recognize the occurrence of different major groups of biomolecules, including peptide-like and oxygen-rich compounds (Fig. 3), which are signs of a well-preserved molecular association 90 . The ToF-SIMS imaging analysis also suggests that the molecular association results from the combination of microbial activity and the presence of phytochemicals. Such results are accompanied by high chemodiversity and abundant molecular information, which is constrained by the extractive technique (Fig. S13) to masses lower than < 600 Da. However, the ToF-SIMS analysis shows that the molecular weights of the compounds in the Upper Gossan sample 010,109-1 are much higher than 1,000 Da, suggesting that the preservation of the biological materials was remarkably high.
The GC-MS analysis of solvent-extractable lipids from the gossan sample provides many compounds, including a very diverse set of FAs, hydrocarbons, isoprenoids, alkanols, hopanoids and steroids (Table S1). As previously described in the GC-MS results, FAs consist of an association of very different compounds, including straight chains with even-over-odd dominance (from 11:0 to 30:0), branched FAs (including iso-/anteiso-pairs from 13:0 to 25:0), monounsaturated FAs (16:1 ω7, 18:1 ω9, 18:1 ω10, and 20:1 ω8) and minor HMW hydroxyl FAs (24:0, 25:0, and 26:0). The triterpenoid structures have different cholesterol derivatives that concur with other triterpenoid structures, such as β-sitosterol, stigmasterol, ergostanol, lanostanol, and lupanol (Table GC-MS). Such a triterpenoid association likely resulted from a very heterogeneous source that included different eukaryotic forms, such as plantae, animalia, and fungi 51,52,91 . Therefore, the biological information recovered by the GC-MS www.nature.com/scientificreports/ analysis of the gossan sample strongly supports organic production from multiple sources that include aquatic (i.e., macrophytes) and land plants, bacteria, algae/phytoplankton and fungi. Both FTICR-MS and GC-MS results are highlighted by the ToF-SIMS imaging capabilities, which have allowed the characterization of microbial communities through the distribution of distinctive biomolecular fragments. Such distributions distinguish different imaging groups (from 1 to 9), whose morphologies suggest that they correspond to parts of living organisms, microbial communities and/or biofilms 92 . These morphologies are observed in the molecular imaging of the different target areas that captures the occurrence of several biological structures. In fact, the G1 group in TA1 (Fig. 4) shows thick filamentous structures with a high content of N-bearing positive fragments that can be attributed to sphingolipids. The ToF-SIMS spectral analysis of TA1 also shows high enrichments in different sets of positive ions that have been assigned to plain ceramides and glucosylceramides (e.g., cerebrosides and sphingomyelins). As such, an association of biomolecules is currently found in microbial fungi 82,93 that occur as filament clusters (hyphae) displaying a magnitude comparable to that of the filamentous structures defined by N-bearing fragments as G1 (Fig. 4). The filamentous structures are also well defined by high intensities of different inorganic SO n (2 ≤ n ≤ 4) anions and organic sulfate adducts, which strongly suggests hyphae biomineralization, as commonly occurs in some modern Rio Tinto fungi 14 . The detection of high contents of sulfate adducts likely results from the binding between SO 4 2 and HSO 4 anions with a positively charged fungal hyphae surface, which is partly formed by lipids. In fact, the occurrence of cholesteryl sulfate defined by a large peak (I > 400 to 1050 cps) at m/z 465.30 in the different target areas (Figs. 4, 5, and 6) strongly suggests that the sulfate organic adducts result from the ionization of the biomineralized fungal surface by sulfate anions. Organic sulfate-bearing adducts are not found in biological structures other than the filamentous fungal structures, suggesting that the sulfate biomineralization is highly specific to some fungal microorganisms 14 .
As discussed above, the different biological structures defined by morphological groups 1 to 9 are emphasized by the highly distinctive molecular composition distribution. The red/green/blue (RGB) overlay of the different morphological groups (Figs. 5, 6, and 7) shows that most of the biological structures interlock with each other to form larger biological assemblages. Although the origin of some of the preserved structures is unknown, the morphology and composition of most of them strongly suggest that they correspond to an association of different fungal structures (e.g., G1, G3 and G7) that are associated with parts of other living forms, such as plants represented in both Groups 5 and 6 (Figs. 4 and 5). As shown in the results, G1, G3 and G7 are formed by N-bearing fragments and sulfur-bearing adducts sourced in sphingolipids and biomineralized bodies by sulfate ( Fig. 2; Table S2), which corresponds to the preservation of fungal communities 14,80,93 , while G5 and G6 are defined by different ions fitting the fragmentation pattern of some phytochemicals, such as vanillic acid and piperonal 56,57 . In terms of biomass, the main fungal structure corresponds to G1, which is characterized by dozens of fragments (Table S2) that usually have a high intensity as well. This feature is very common in the different N-bearing m/z cations that are characterized as sphingolipid fragments and secondary anions (e.g., G8) that could have been produced by the fragmentation of large polysaccharides forming fungal biofilms 94 .
The occurrence of exceptionally preserved biofilms with two different microbial forms is observed in TA2 (Figs. 6 and 8; Figs. S10 and S11), where two different biological structures are found. They are distinguished by the molecular composition that defines G1, which is enriched in ceramides and sulfur adducts, versus G9, which is characterized by a high abundance of diacylglycerides, through a sharp boundary between the two groups (Fig. 6). G3 interlocks G1 (Fig. 6) in the same way as in TA1, which suggests that there are compositional changes in the fungal biofilm. Furthermore, in TA3, ToF-SIMS image analysis shows additional biological structures whose morphology and molecular composition suggest a fungal origin but also contain glycerolipids in addition to sphingolipids and SO 4 -mineralized compounds (Fig. 7). As discussed above, the fungal structures defined in TA1 and TA2 show high concentrations of sphingolipids and SO 4 -mineralized compounds but depletion in www.nature.com/scientificreports/ glycerolipids (Figs. 4 and 6), which suggests that the biological structure found in TA3 corresponds to a different biological origin. The phytochemicals (e.g., vanillic acid and piperonal derivatives) that characterize some biological structures in TA1, such as G5 and G6 (Fig. 4), occur in TA3 within the mineral matrix in which the biological structures are embedded (Figs. 1 and 7). Such a distribution suggests that they were transported in the acidic solution after being released from plant fragments in more proximal and/or surficial areas where the plants were growing. They come together with a set of different PAHs ( Fig. 7; Table S3), which are recalcitrant compounds resulting from the photochemical process and are common in continental waters. This agrees with the parallel arrangement of the fungal structures (Fig. 1), which suggests that they were oriented along an acidic flow with a high concentration in ferric oxysulfate polymers 95 . Indeed, the ToF-SIMS image analysis provides enough evidence in the form of chemical compositions to gain insight into the mineralization processes that led to the preservation of biological structures and species. This preservation was accomplished by means of different mineralizing episodes that started with sulfate biomineralization by the activity of some fungal and bacterial communities 14 . Sulfate biomineralization was followed by ferric-bearing mineralization, whose composition could have been mediated through different phases, including schwertmannite, hydronium jarosite, and nanophase goethite 23 . The mineralization in this case was a passive process in which biological structures were coated and infilled with ferric-bearing compounds whose composition was controlled by the solution hydrochemistry. The rock maturation ended in partial replacement of the primary phases by goethite and hematite (Fig. 2). Such limited maturation is evidenced by high concentrations of sulfate and organic carbon, which were determined by ToF-SIMS analysis (Figs. 4, 5, 6, and 7). In this context, the precipitation of the ferric and sulfate phases played an essential role in the preservation of the structure and composition of the biological elements in three ways: (1) building rigid mineralized structures resistant to sediment load or tectonic stress and chemical attack 96,97 , (2)

Conclusions
The spectral and ToF-SIMS image analyses of the Upper Gossan sample show that hyperacidic and oxic conditions leading to ferric deposits have remarkable preservation potential in terms of structure and composition. Such a preservation potential is also supported by the occurrence of peptidic chains in 6-to 2.2-Ma Rio Tinto terrace deposits 26 . The same low-pH and oxidizing conditions have been hypothesized for some regions on Mars during the Noachian-Hesperian transition 6,7 . In this period, different mineral phases, such as jarosite, schwertmannite, nanophase goethite, and other unknown Fe 2+ /Fe 3+ -bearing sulfate or oxide phases 9,102,103 , could have played essential roles in the mineralization of microbial remains if they were ever produced on Mars. The precipitation of ferric phases has been reported to follow fast saturation 7 , which is observed in the Rio Tinto Mars analog. The saturation of acidic solutions on Mars would have increased the concentrations of different cations, including toxic metals 98,99 , and decreased the water activity 100 , both of which greatly inhibit microbial activity. Indeed, the water activity decreases with temperature 104 , so the net preservation of biomolecular compounds would be greatly favored at the low temperatures on Mars. The deposit maturation to goethite and hematite would confer on them a higher resistance to deformation and weathering. High concentrations of ferric ions in rock and solutions 105 would also have provided protection against UV radiation once the microbial remains were eventually produced on Mars.
Although preservation in clay-rich deposits is currently associated with organic preservation 21 , the preservation potential can be greatly reduced when clay deposits are exposed to different maturation and alteration processes that can degrade organic compounds in later stages of rock diagenesis on Mars, such as those associated with enhanced weathering (Fig. S14), which likely occurred during the Noachian to Hesperian transition. The solutions involved in clay formation would have not had high concentrations of cations that could eventually mineralize the biological structures and protect them against UV radiation 105 and lithostatic compression 97 . When mineralization does not occur, biomolecular compounds can be broken into smaller compounds by lithostatic compression. Indeed, such mild environments are associated with unfavorable mineralizing conditions, which implies low potential for preserving structures. Assuming that Mars was exposed to acidic episodes, there is a high possibility that if life emerged on the planet, its molecular preservation might have followed the same mechanisms to record traces of ancient life on the surface and in the interior of the planet. In fact, the search for life on Mars could be expanded to large and extensive acidic deposits formed billions of years ago.

Material and methods
Sample collection. A field campaign was conducted for surface sampling of the upper part of the Cenozoic gossan 13 topping the Peña de Hierro Carboniferous basement (Fig. 1a). This unit corresponds with the oldest alteration episode that formed the Upper Gossan materials through sulfide weathering by a water table that remained relatively stable during the late Oligocene-early Miocene 27 .
Sample 010109-1. was collected in Peña de Hierro (GPS coordinates 37.724918°, 6.553717°) from the uppermost part of the iron oxide deposits corresponding to the Upper Gossan unit aged late Oligocene 27 . The collected materials consist of massive iron oxide strata-like deposits forming ~ 1-m-thick lenses that are associated with the central and highest part of the gossan (Fig. 1a). Such materials currently vary downward or laterally to brecciated or tuffaceous oxide deposits (Fig. 1b-c), showing that the gossan formation involves a complex interplay between surface and subsurface processes. In this regard, such massive deposits are currently delimited at the top by efflorescent pseudomorphs and coated filamentous structures (Fig. 1c)  www.nature.com/scientificreports/ an outcrop showing few signs of alteration, and a geological hammer was used to prevent contamination that can be produced through drilling, which employs water as a coolant. The sample was carefully collected using nitrile gloves, covered by aluminum foil and introduced into a sterilized sampling bag to prevent contamination. Then, the sample was split into three parts; subsequently, each subsample surface was sterilized by igniting mass spectrometry-grade ethanol (sigma). Then, the three subsamples were delivered for spectrometry analysis.
Observations under an optical microscope and a scanning electron microscope (SEM) of thin sections of the massive unit provided evidence for the occurrence of abundant biological fabrics that were templated by coating and mineralization (Fig. 1c). Furthermore, the sample observations under the different microscope techniques also evaluated the origin of the biological traces that were found in the rock matrix (Figs. S14 and S15). Interestingly, scanning electron and energy dispersive spectroscopy (SEM-EDS) analysis suggested very slight, if any, alteration of the mineral matrix of the sample. Indeed, the lack of different cations, such as Ca, Mg and K, and the low content of Si (Figs. S15 and S16) indicate that no alteration occurred 12 . In this regard, Fernandez-Remolar et al. 106 reported the formation of silica laminations. Such laminations might have formed if the gossan material had been exposed to subsequent alteration associated with the occurrence of secondary biomolecules that were younger than the rock. Indeed, the SEM-EDS images show the presence of colomorphic goethite (Figs. S15 and S16) covered by laminated hematite, which is the result of long sedimentary maturation of the primary iron oxysulfates and/or oxyhydroxides 13,23 . Furthermore, the EDS analysis implies that goethite is associated with traces of S (see spectrum 3 in both Figs. S15 and S16) that likely came from a primary mineral composition of sulfates. Interestingly, SEM-EDS also targeted goethite with microstructures ( Fig. S16) that were very similar to the filamentous microstructures shown in Figs. 4 and 5. The SEM-EDS image shows that the filamentous goethite is enclosed inside a hematite-rich lamina (Fig. S16), which indicates an old age and supports a primary origin for the preserved biological structures. Sample preparation. Sample analysis was performed using different state-ofthe-art analytical techniques, including time of flight secondary ion mass spectrometry (ToF-SIMS) and Fourier transform ion cyclotron resonance mass spectrometry (FTICR-MS) 26,36 . These methods were supported by the more conventional technique of gas chromatography coupled with mass spectrometry (GC-MS), all of which have access to molecular databases with many compounds. In this regard, GC-MS is a technique that provides essential information to confirm compound identification. The preparation for the sample analyses using these three distinctive spectrometry techniques followed two different processes: simple thin sectioning and organic extraction from powdered samples. Thin sections of sample 010109-1 were prepared for mineral and textural analysis under an optical microscope and a Scanning Electron Microscope (SEM) coupled with an Electron Dispersive Spectroscopy (EDS) probe, and molecular surface analysis by ToF-SIMS 107 . They were polished an extra time by using a 0.3 μm alumina paste to reduce surface imperfections that produce frequent interference during surface analysis. On the other hand, as FTICR-MS and GC-MS analyses of compounds require organic extraction, sample 010,109-1 was initially crushed under sterile conditions to prevent contamination.

Microscope analysis.
The inner structure of sample thin sections was observed under an upright trinocular microscope Motic BA400 coupled to a MoticamPro 282a camera. Optical observations were complemented with backscattered electron imagery using a Jeol JSM-5600 LV Scanning Electron Microscope (SEM). In addition, semiquantitative chemical microanalysis was performed with a Cambridge INCAx SIGHT EDS (Electron Dispersive Spectroscopy) to associate microstructure and mineral substrate with elemental composition.

GC-MS and FTICR-MS analyses of organic extracts.
Before lipid extraction, the sample was externally cleaned, and the outer layer was removed to avoid cross contamination with allochthonous material from the surface. In the first step, an external layer approximately 0.5 cm thick was removed with a solvent-cleaned (methanol and dichloromethane) saw and discarded. Subsequently, the remaining inner sample was externally cleaned with a solution of approximately 100 ml of hydroxide potassium (6% in Milli-Q water) while being gently shaken for 15 min in a glass beaker. After rinsing with Milli-Q water, the sample was then cleaned with nitric acid (6%) under the same conditions (shaking for 15 min). Finally, the sample was rinsed with Milli-Q water and ethanol (> 99.5% purity) and then dried in an oven (45 °C). Once dried, the sample was ground with a solvent-cleaned mortar and pestle for lipid analysis. Approximately 100 g of sample 010,109-1 was prepared for organic-solvent extraction for GC-MS analysis (Fig. S13). The extraction was performed with a mixture of dichloromethane/methanol (DCM/MeOH, 3:1, v/v) for 24 h with a Soxhlet apparatus. Internal standards (tetracosane-D 50 , myristic acid-D 27 , and 2-hexadecanol) were added prior to the extraction. The total lipid extract was concentrated by rotary evaporation to approximately 2 ml. After this process, activated copper was added, and the sample was allowed to stand overnight for elemental sulfur removal. The extract was separated into three fractions of different polarities (nonpolar, polar, and acidic). In the first step, a Bond-elute column (bond phase NH 2 , 500 mg, 40 µm particle size) was used to separate the neutral and acidic fractions by eluting first with 15 ml DCM/2-propanol (2:1, v/v) and subsequently with 15 ml diethyl ether with 2% acetic acid. In the second step, the neutral fraction was further separated into nonpolar and polar fractions by using an Al 2 O 3 column (activated, neutral, 0.05-0.15 mm particle size) and eluting with 5 ml of hexane/DCM (9:1, v/v) and with 4 ml of DCM/methanol (1:1, v/v), respectively. The acidic fraction was derivatized with BF 3 in methanol, and the polar fraction was derivatized with N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA). Together with the sample, a blank was extracted (i.e., an empty cellulose thimble), concentrated, fractionated and analyzed in parallel as a negative control. All glassware and sorbents were heated in a furnace at 550 °C for 24 h. All solvents were provided by Sigma Aldrich (Spain) and were of analytical grade (> 99.5% purity).
GC-MS analysis was performed to characterize the different nonpolar, acidic, and polar fractions. This analysis used a 6850 GC system and a 5975 VL MSD with a triple axis detector (Agilent Technologies www.nature.com/scientificreports/ of spectral data was obtained using full scan mode (m/z 50 to 650) that operated with electron ionization at 70 eV. The analytes were injected (1 µl) into an HP-5MS column (30 m × 0.25 mm i.d. × 0.25 µm film thickness) with He as a carrier gas (1.1 ml min -1 ). For the nonpolar fraction, the oven temperature was programmed from 50 to 130 °C at 20 °C min −1 and then to 300 °C at 6 °C min -1 (held for 20 min). For the acidic fraction, the oven temperature was programmed from 70 to 130 °C at 20 °C min −1 and then to 300 °C at 10 °C min −1 (held for 10 min). For the polar fraction, the oven temperature program was the same as that for the acidic fraction, except that the oven was held for 15 min at 300 °C. The injector temperature was 290 °C, the transfer line was 300 °C, and the MS source was 240 °C. The identification of compounds was based on the comparison of retention time and mass spectra with the National Institute of Standards and Technology (NIST) mass spectral database and reference materials: n-alkanes (C 10 to C 40 ), fatty acid methyl esters (FAMEs; C 8 to C 24 ), n-alkanols (C 10 , C 14 , C 18 , and C 20 ), and branched isoprenoids (2,6,10-trimethyl-docosane, crocetane, pristane, phytane, squalane, and squalene). All chemicals and standards were supplied by Sigma Aldrich and Chiron. The GC-MS data were acquired and processed by Agilent MSD ChemStation software, which performed molecular identification though NIST library matches of peak spectra.
To obtain the molecular information for sample 010,109-1, a solid-phase extraction was performed on the supernatant as described elsewhere 36,37 , and the extract was used for FTICR-MS analysis. This analysis was conducted with a SolariX FTICR mass spectrometer (Bruker Daltonik GmbH, Germany) equipped with a 12 T superconducting magnet (Magnex Scientific Inc., GB) and an APOLLO II electrospray ionization source (Bruker Daltonik GmbH) in negative ionization mode. The injection was performed using a microliter pump at a liquid flow rate of 120 ml h -1 . Both sheath gas and curtain gas consisted of nitrogen. A source heater temperature of 200 °C was maintained to ensure rapid solvent evaporation in the ionized droplets. The FTICR mass spectrometer was precalibrated using known arginine clusters. The mass spectrum was zero filled to a processing size of four megawords using 750 scans within a mass range of m/z 147.4 to 1000 and internal calibration with an appropriate reference mass list, reaching accuracy values lower than 100 ppb. A blank sample was run before sample injection and did not show any clusters of contamination. Data treatment was performed using Compass Data Analysis 4.0 (Bruker Daltonics, Bremen, Germany). The exact masses of the detected m/z ions were defined, and their molecular formulas were calculated using a software tool written in-house 109 . The final formulas were branched into groups containing CHO, CHNO and CHOS molecular compositions, which were used to reconstruct the group-selective mass spectrum.
ToF-SIMS analysis. ToF-SIMS is an analytic surface technique that characterizes the distribution of positive and negative ions in a sample. For this reason, it can provide essential information about the preservation of organic compounds, establishing a direct association between the mineral matrix and the occurrence of molecular biosignatures in the sample 108 . ToF-SIMS has a high detection level for different compounds, including contaminants that can potentially be attached to the thin section surface. To prevent this problem, the thin sections of sample 010109-1 were cleaned by surface sputtering with a 100 nA 500 eV oxygen ion beam for 3 s. The molecular surface analyses were performed by using a ToF-SIMS IV (ION-TOF, Münster, Germany) system under a pressure of 5·10 -9 mbar. The thin section was sputtered with a pulsed bismuth liquid metal ion source (Bi 3 2+ ) at 50 keV. The ion beam was operated with a 20 ns pulse width and 0.3 pA pulsed ion current for a dosage lower than 5·10 9 ions·cm -2 , below the threshold level of the static SIMS conditions (1·10 11 ions·cm −2 ). The released secondary ions were detected with a reflectron time-of-flight analyzer, a multichannel plate (MCP), and a time-to-digital converter (TDC). Charge neutralization was achieved with a low energy (20 eV) electron flood gun. Secondary ion spectra were obtained from a randomly rasterized square surface area with a 500-μm side within the thin section surface. Secondary ions were extracted with a 2 kV accelerating voltage and were postaccelerated to 10 keV kinetic energy just before hitting the detector.
Furthermore, the observations of microfabric and microtextural features of the inspected areas for each sample were accomplished by the ToF-SIMS ability to collect ion-induced secondary electron images. This was the first step that was taken during the ToF-SIMS analytic process that searched for differences in fabric and textural properties along the polished surface of the sample. As a result, different target areas showing characteristic fabrics and textures were selected in sample 010,109-1 (Figs. 2, 3, 4, 5, 6, and 7; Figs. S2 to S3). The characterization of the fabric and texture through the imaging capabilities of ToF-SIMS was essential to recognize the association among fabric/texture, mineralized microstructures, and distribution of preserved biological compounds.
Mass spectral acquisition and image analysis, including the overload of red, green and blue (RGB) images for molecular mapping, were performed with ION-TOF Surface Lab software (version 6.8). The analysis of the molecular fragmentation pattern was completed through the Chemical online tool 109 and the open source mass spectrometry tool Mmass 110 . Chemspider, METLIN and LIPID MAPS Structure databases [111][112][113] were used as sources of information to identify molecular fragments and compounds.