Variability of carbonate isotope signatures in a hydrothermally influenced system: insights from the Pastos Grandes caldera (Bolivia).

: Laguna Pastos Grandes (Bolivia), nesting in a volcanic caldera, is a large, palustrine-to-lacustrine system fed by meteoric and hydrothermal calco–carbonic ﬂuids. These di ﬀ erent ﬂuid inputs favor a complex mosaic of depositional environments, including hydrothermal springs, pools, and an ephemeral lake, producing abundant present-day carbonates developing over a Holocene carbonate crust dated by U–Th. Present-day carbonates (muds, concretions, and microbialites) recorded a large range of isotope variations, reaching 13.9% (cid:24) in δ 13 C and 11.1% (cid:24) in δ 18 O. Sedimentological and geochemical data indicated that the main processes inﬂuencing the isotope record were: (i) rapid CO 2 degassing and temperature decreases along hydrothermal discharges; (ii) strong evaporation favored by the arid high-altitude Andean climate, locally enhanced by capillary water rise within microbial mats or by wind-induced spray falling on vadose concretions. Unlike past or present perennial lake systems in Central Andes, the short residence time of brine waters in the ephemeral central lake prevents enrichment of lacustrine carbonates in 13 C and 18 O. The very low fraction modern F 14 C in these present-day carbonates demonstrates that incorporation of fossil magmatic carbon related to the volcanic context also prevents any radiocarbon dating. The use of isotopes for the interpretation of ancient continental series should always be accompanied by a thorough characterization of the environmental setting. di ﬀ raction (XRD). Samples were crushed and ﬂattened without orienting crystals. Whole rocks were analyzed at the Biogeosciences Laboratory of Burgundy, Dijon, France) with a Bruker D4 Endeavour di ﬀ ractometer (Bruker, Billerica, MA, USA), with CuK α radiations, LynxEye detector, and Ni ﬁlter under 40 kV voltage and 25 mA intensity. Scanning from the goniometer ranged from 2 ◦ to 65 ◦ for each analysis. water–surface oscillation, and capillary rise in the microbial mat.


Regional Background, Volcanic Setting, and Carbonate Facies of the Pastos Grandes Area
The Andean Cordillera (South America) is a mountain range 7000 km long resulting from the subduction of the oceanic Nazca plate beneath the South American continental plate. This mountain range is also associated with a magmatic arc that spread all along the range. However, the distribution of this magmatism is heterogeneous due to the variation of the dip angle. Only locations above a high dip angle (~30 • ) present volcanic activity [17]. In the Central Andes, the volcanic activity is referred to as the Central Volcanic Zone (between 16 • S and 28 • S, Figure 1A), characterized by numerous andesitic volcanoes and calderas (e.g., [18]), and located at the west of the Altiplano and Puna Plateaus, large intermontane basins [19]. Additionally, these plateaus are separated by a large volcanic region, the Altiplano-Puna Volcanic Complex (APVC), covering 70,000 km 2 between 21 • S and 24 • S [20,21]. The APVC is formed by several calderas and stratovolcanoes, identified by approximately thirty ignimbrite deposits, composed of dacitic to rhyolitic rocks [21]. The activity of this volcanic complex spanned from 11 to 0.7 Ma [20][21][22], and its origin is thought to result from the steepening of the dip angle during the Late Miocene, favoring mantle melting and intense delamination of the lower continental crust [23,24]. This leads to the formation of a large magmatic body, the Altiplano-Puna Magmatic Body, identified by a 500,000 km 3 low-velocity zone between 4 and 25 km below sea level [25,26]. This magmatic body fed several shallower pre-eruptive chambers beneath the Altiplano-Puna Volcanic Complex [22,24].
The Laguna Pastos Grandes (−21.64 • N; −67.79 • E; 4450 m above sea level) is an extensive NW-SE elliptical playa (120 km 2 ) in the South Lipez region of Bolivia ( Figure 1A). The laguna lies in the northeastern part of one of its calderas, the Pastos Grandes Caldera Complex [20], fed by one of the pre-eruptive chambers of the APVC, and shaped by two main eruptions: the Chuchuilla ignimbrite (5.45 ± 0.02 Ma) that ejected~1200 km 3 of dense rock equivalent (DRE) of crystal-rich andesitic-to-dacitic rocks; the Pastos Grandes ignimbrite (2.89 ± 0.01 Ma) that ejected more than 1500 km 3 of DRE of crystal-rich dacites [21]. This eruption is thought to have resulted from a single, long-lived, monotonous intermediate magma flow, where near-eutectic conditions were maintained over~1.1 Myr before and after the eruption by recurrent andesitic recharge, identified by chronochemical investigations [27]. This high-altitude area is characterized by a cold, dry climate [28]. Most precipitation occurs as summer rainfall, from December to March, with an annual mean of 100 mm/y, while dry conditions dominate during the rest of the year, with an evaporation rate of 1400 mm/y [29,30]. Air temperature can rise to +25 • C during the austral summer and fall to −30 • C during the austral winter [31].
The central and eastern parts of the playa are partly covered by an ephemeral shallow lake (maximum depth < 2 m) during the wet season, while the western part is a palustrine platform with hydrothermal springs and pools ( Figure 1B). The ephemeral lake is fed by cool, fresh streams flowing from the Chuchuilla scarp on the eastern side, while the western and northern parts are characterized by piedmont springs, either cool and fresh, or warm and saline [14]. Lacustrine deposits are identified as mud containing gypsum, diatoms, volcanoclastics, and clay [10]. In the central part of the playa, the gypsiferous mud contains ulexite ore (NaCaB 5 O 9 ·8H 2 O) [12].
The palustrine area is characterized by a carbonate-rich platform (approximately 40 km 2 , up to 30 cm thick), fragmented by frost action [12], which is the largest carbonate platform in the Altiplano, where carbonate deposits are rare and restricted to smaller areas. As recently demonstrated, it results from hydrothermal fluids that deliver large quantities of CO 2 and Ca ++ originating from the hydrothermal alteration of the volcanic basement in the presence of CO 2 [16]. Established models suggest that CO 2 comes exclusively from magmatism, deriving partly from the mantle (47%), indicating a deep origin for this gas. The recharge in chemical elements is thought to occur at 225 • C, based on geothermometers, corresponding to a depth of about 5 km [16], and may be triggered by the volcanic chamber under the laguna [22]. These hydrothermal fluids then rise to the surface through fractures [16]. The exceptional production of carbonates in this purely volcanic area could thus be explained by hypogean boosting of the sedimentary system. In fact, the CO 2 delivered by hydrothermal springs is from 60 to 389 mbar, equal to or higher than the amount at other Andean sites (see Reference [16]).
Previous sedimentary studies on Pastos Grandes have mainly documented diatom assemblages in submerged areas (e.g., [30]), and modern carbonates precipitated on some parts of the carbonate platform [12][13][14]. They report microbialites, rich in Ca and Si, which develop where moderately saline hot springs emerge, while pisolites (sub-spherical concretions up to 20 cm in diameter) form in small ponds farther from the springs. The other types of modern carbonate deposits occurring on the platform remain poorly described to date, both in terms of facies and geochemical characteristics.

Materials and Methods
Rock and water samples were collected during field investigations at Laguna Pastos Grandes in January 2016 (dry conditions) and March 2017 (wet conditions). Two sets of carbonate samples were distinguished: one from the vast, dried-up carbonate platform, and the second composed of present-day carbonates actively developing over the carbonate platform in ephemeral or perennial water conditions [12][13][14], associated with microbial mats that induce ongoing carbonate precipitation (see Reference [14]).
Carbonate deposits forming the carbonate platform were collected to constrain its approximate age of formation. Dating was carried out using U-series determination on six samples collected from the carbonate crust (n = 4), a pisolite core (n = 1), and calcitic mud (n = 1). They were analyzed at the GEOTOP research center, University of Quebec at Montreal, Canada (see details in Reference [32]). The calcite samples were cut using a diamond-abrading disc device (Dremel®rotary tool), and the external layers of the sample were removed to reduce the risk of contamination by 230 Th-bearing detrital particles. Each sample was dissolved with 7N HNO 3 in Teflon beakers, and a known amount of spike ( 233 U, 236 U, and 229 Th) was added to determine U and Th isotopes by the isotope dilution technique. The methods used for separation and purification of U and Th isotopes were based on Edwards et al. [33]. The purified U and Th fractions were dissolved in 2% HNO 3 and measured by MC-ICP-MS (multi-collector inductively coupled plasma mass spectrometry) with a NU II Instrument (Ametek, Berwyn, IL, USA). After mechanical cleaning, some detrital material was still present in the samples, as indicated by the low 230 Th/ 232 Th activity ratios (see Table 1). One of the main biases in U-Th dating is the presence of detrital elements. Correction for detrital contamination was performed following Ludwig and Paces [34], using an average crustal model ( 232 Th/ 238 U = 1.21 ± 0.65, 230 Th/ 238 U = 1 ± 0.1 and 234 U/ 238 U = 1 ± 0.1).
Microfacies were observed on polished slabs and polished thin sections. Thin sections were partially stained with alizarin Red-S to identify calcite, and potassium ferricyanide to identify ferrous carbonates. Thin sections were observed with a Nikon AZ100 microscope (Nikon, Tokyo, Japan) coupled with a Zeiss MrC5 camera (Carl Zeiss, Oberkochen, Germany). Scanning electron microscopy (SEM) was performed on soft sediments. Soft sediments were gold-coated and then observed with a Zeiss Auriga FEG-FIB microscope (Carl Zeiss, Oberkochen, Germany) at the IPGP (Institut de physique du globe de Paris, France) at an accelerating voltage of 15 kV, and a high current (up to 1 nA), to determine crystal morphology. Crystal chemical composition was determined with a Bruker Quanta 2000 EDX (energy-dispersive X-ray spectrometer; Bruker, Billerica, MA, USA) coupled to the microscope with the same voltage.
Whole rock mineralogy was determined by X-ray diffraction (XRD). Samples were crushed and flattened without orienting crystals. Whole rocks were analyzed at the Biogeosciences Laboratory (University of Burgundy, Dijon, France) with a Bruker D4 Endeavour diffractometer (Bruker, Billerica, MA, USA), with CuKα radiations, LynxEye detector, and Ni filter under 40 kV voltage and 25 mA intensity. Scanning from the goniometer ranged from 2 • to 65 • for each analysis.
Stable isotope analyses (δ 18 O carb and δ 13 C carb ) were performed at the Biogeosciences Laboratory (University of Burgundy, Dijon, France). Rock samples were cleaned with ethanol, and 30-50 µg of powder was retrieved from the most recent laminae using a Dremel drill with a diamond bit. Powder samples were dried at 50 • C for 12 h in a stove and then reacted with purified orthophosphoric acid at 70 • C in a ThermoFisher Kiel IV carbonate device connected to a ThermoFisher Delta V PLUS mass spectrometer (ThermoFisher Scientific, Waltham, MA, USA). The δ 18 O carb and δ 13 C carb values thus obtained were normalized to Vienna Pee Dee Belemnite (V-PDB), using international standard NBS-19 (see Table 3). Replicates were made to evaluate reproducibility at approximately 0.04% for δ 13 C carb and 0.08% for δ 18 O carb , or better.
Measurements of the fraction modern carbon (F 14 C) were obtained from the A.E. Lalonde Laboratory (Ottawa, ON, Canada) on 5 modern calcite samples (i.e., mushroom-like microbialite, shrub, pisolite, platystromatolite, and palmatoid), following the analytic protocol of St-Jean et al. [35]. The most recent laminae in these samples were abraded with a Dremel drill and diamond bit to remove impurities, and pre-etched with 0.2 N HCl to clean the outer surface. The cleaned samples were then completely dissolved in anhydrous H 3 PO 4 and reacted overnight at room temperature. The CO 2 thus produced was extracted, cryogenically purified, and then converted into pure carbon by reacting with hydrogen and iron at 550 • C. The 12,13,14 C 3+ ions were measured at 2.5 MV terminal voltage with Ar stripping, using a 3 MV tandem accelerator mass spectrometer (high voltage engineering). The fraction modern carbon F 14 C was calculated using the ratio of the sample 14 C/ 12 C ratio to the standard 14 C/ 12 C ratio (A.E. Lalonde Laboratory standard: Ox-II).
The carbon and oxygen isotope compositions of the carbonates (δ 18 O carb , δ 13 C carb , and 14 C) were acquired specifically for this study, while the data for water chemistry and stable isotopes (δ 13 C gas , δ 13 C DIC -dissolved inorganic carbon, and δ 18 O water ) from hydrothermal springs, piedmont sources, ephemeral streams, and the lake itself were taken from Muller et al. [16]. As no data were available for the depositional environments where the pisolites, palmatoid concretions, and platystromatolites formed, additional physicochemical measurements from the pools (i.e., temperature, pH, and conductivity) were acquired in situ with a Hach HQ40D Multi-instrument (Hach Company, Loveland, CO, USA). Water was also collected from several points and filtered in the field with a 1 µm filter. Filtrate composition was analyzed at the BRGM Laboratory (French Geological Survey, Orléans, France) and Total CSTJF Laboratory (Pau, France) laboratories using ion chromatography, atomic absorption spectrophotometry, inductively coupled plasma mass spectrometry (ICP-MS), colorimetry, the ion electrode method, and titration.

Carbonates of Laguna Pastos Grandes and Dating
Laguna Pastos Grandes can be divided into two main domains with: (i) the upper and lower palustrine carbonate platforms covered by hydrothermal springs and pools [14]; (ii) a shallow, ephemeral saline lake (Figure 2A; Figure S1). The carbonates of the first domain corresponding to the palustrine platforms, and associated carbonates currently forming over it in springs and pools were dated by the U-Th and 14 C methods, respectively.
The U-Th method was used to determine the average age of the carbonate platform (Table 1; Supplementary Materials, Table S1). The ages obtained on the upper platform came from a carbonate crust composed of an assemblage of cemented calcitic grainy facies, with pisolites, intraclasts, peloids, and ooids ( Figure 3A) [14]. They ranged from 1 ka (kiloannum) to 2 ka, indicating the recent development of this platform. The lower platform was mostly formed of white calcite mud ( Figure 2C), with pluricentimetric calcitic clasts (resulting from active cryoturbation; Figure 2C). The range of ages obtained from the clasts and muds was much greater than for the upper carbonate platform: botryoidal cement clast ( Figure 3B) approximately 3.5 ka; peloidal grainstone clast approximately 34 ka, and mud approximately 13 ka. Nevertheless, the mud was sampled in a reworked location (cryoturbation), probably removing several underlying layers, which would prevent precise dating. The clast composed of peloidal grainstone was found in the southern area where the alluvial fan intrudes into the carbonate platform. This region is characterized by a major influx of volcanoclastics mixing and reworking carbonates, which could come from lower layers of the palustrine area ( Figure 2D). Overall, the stability of the facies preserved on the upper platform made the ages obtained more robust, and a Holocene estimation was therefore proposed for the deposition of the platform. Nevertheless, a greater number of ages would be useful to ensure that the variability of ages recorded did not result from a re-opening of the carbonate system [36].
Minerals 2020, 10, x FOR PEER REVIEW 6 of 26 southern area where the alluvial fan intrudes into the carbonate platform. This region is characterized by a major influx of volcanoclastics mixing and reworking carbonates, which could come from lower layers of the palustrine area ( Figure 2D). Overall, the stability of the facies preserved on the upper platform made the ages obtained more robust, and a Holocene estimation was therefore proposed for the deposition of the platform. Nevertheless, a greater number of ages would be useful to ensure that the variability of ages recorded did not result from a re-opening of the carbonate system [36].  Figure  1A). Zone 1 to Zone 4 (Z1-Z4) are depositional zones associated with the main hydrothermal springs, described in Bougeault et al. [14]: Z1, the hydrothermal feeding system; Z2, the apron; Z3, the proximal-distal transitional belt; Z4, the distal zone.   Figure 1A). Zone 1 to Zone 4 (Z1-Z4) are depositional zones associated with the main hydrothermal springs, described in Bougeault et al. [14]: Z1, the hydrothermal feeding system; Z2, the apron; Z3, the proximal-distal transitional belt; Z4, the distal zone.   (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017)(2018)(2019) show that the central lake evaporates almost completely every year ( Figure S2), triggering gypsum and ulexite sedimentation in the playa [10]. Here, in contrast with the methods used for the carbonate platform, the age of these modern carbonate deposits associated with hydrothermal springs and pools was obtained using 14 C dating (see Section 4.3). The facies corresponding to the hydrothermal springs have been described in Reference [14], and the ones from pools are described below (Section 4.2.).

Modern Calcite Facies and Parent Waters
Above recent facies of the carbonate platform, five groups of facies (including one or more sub-facies) and their associated parent waters were identified along a W-E transect from the volcanic glacis to the lake ( Figure 2A).

Microbialites
Microbialites develop in association with hydrothermal springs, mainly found in the northern part of the upper platform ( Figure S1). Hydrothermal springs emerge at the top of large flat mounds, composed of brownish-to-reddish sediments containing diatoms, calcite mud, calcite intraclasts, and iron oxide [14]. Waters at the vents are hot (20 to 75 °C) and slightly acidic (pH ~5.8-6.9) [12,16]. They are moderately saline (10-17 mg.L −1 ; electrical conductivity: 19-30 mS.cm −1 ), and enriched in Na and Cl ions with K, Mg, Li, Ca, and Si. The main gas emitted was CO2 with an elevated partial pressure reaching 309 mbar. Expelled fluids fed successive depositional zones, upstream to downstream: a central pool, a channel, an apron, a proximal-distal transitional belt, and a distal zone ( Figure 4A). In this distal zone, hydrothermal waters evaporate or sink beneath the Holocene carbonate cap, feeding the underlying water table. The Holocene deposits of the upper carbonate platform are locally covered by hydrothermal springs, discharging hot acidic waters, but also ephemeral pools filled during the wet season, triggered by a water-table rise. Several large perennial "mid-platform pools" have formed just below the edge of the upper platform, which is approximately 30 cm higher than the lower platform ( Figure 2A; Figure S1). The limit between the lower platform and the ephemeral central lake ( Figure 2D) fluctuates due to water-level variation, coinciding with numerous "distal springs" caused by hydrothermal seepage. Observations and quantitative analyses of the area covered by the lake based on satellite views (2000-2019) show that the central lake evaporates almost completely every year ( Figure S2), triggering gypsum and ulexite sedimentation in the playa [10]. Here, in contrast with the methods used for the carbonate platform, the age of these modern carbonate deposits associated with hydrothermal springs and pools was obtained using 14 C dating (see Section 4.3). The facies corresponding to the hydrothermal springs have been described in Reference [14], and the ones from pools are described below (Section 4.2.).

Modern Calcite Facies and Parent Waters
Above recent facies of the carbonate platform, five groups of facies (including one or more sub-facies) and their associated parent waters were identified along a W-E transect from the volcanic glacis to the lake ( Figure 2A).

Microbialites
Microbialites develop in association with hydrothermal springs, mainly found in the northern part of the upper platform ( Figure S1). Hydrothermal springs emerge at the top of large flat mounds, composed of brownish-to-reddish sediments containing diatoms, calcite mud, calcite intraclasts, and iron oxide [14]. Waters at the vents are hot (20 to 75 • C) and slightly acidic (pH~5.8-6.9) [12,16]. They are moderately saline (10-17 mg·L −1 ; electrical conductivity: 19-30 mS·cm −1 ), and enriched in Na and Cl ions with K, Mg, Li, Ca, and Si. The main gas emitted was CO 2 with an elevated partial pressure reaching 309 mbar. Expelled fluids fed successive depositional zones, upstream to downstream: a central pool, a channel, an apron, a proximal-distal transitional belt, and a distal zone ( Figure 4A). In this distal zone, hydrothermal waters evaporate or sink beneath the Holocene carbonate cap, feeding the underlying water table.
Along the hydrothermal pathway, several microbialite morphologies were observed: ledge, mushroom-like, cerebroid, snake-like, isolated shrub, and planar shrub structures (Figure 4B-D; Table 2). These microbialites were mainly composed of micritic laminae, bundles of microbial sheaths, diatom accumulations, and/or silica laminae. (Further details of macro-to microscale microbialite analyses and associated depositional environments can be found in Bougeault et al. [14]).   Figure S1) and associated facies [14]: (A) air drone image of the hydrothermal discharge of La Salsa, showing depositional zones (Z1-Z4); (B) outflow channel (Z1a-Z1b) with well-developed ledge microbialites. Dense microbial mats cover the channel floor; (C) apron (Z2), with numerous cerebroid microbialites, developing on mud composed of diatoms and aggregated anhedral calcite crystals; (D) calcite shrubs in association with black microbial mats in the distal zone (Z4). Pisolites are non-fixed, spherical-to-ovoid carbonate concretions from a few millimeters to 20 cm in diameter, found in ephemeral or perennial pools (Figures 2A and 5A). Rainfall during the wet season triggers a rise in the water table, filling depressions in the upper carbonate platform, thus forming shallow ephemeral pools (<20 cm deep). Some pools are associated with hydrothermal seepage through the fractured carbonate cap, associated with iron oxide precipitation ( Figure 5B). At the transition between the upper and lower platforms, deeper perennial pools (<50 cm deep) are fed by the water table (Table 2). In these perennial pools, pisolites are often cemented to the floor, commonly fusing together to form large slabs (20-30 cm thick) which can cover several square meters ( Figure 5C). Physicochemical parameters for both ephemeral and perennial pools recorded during the wet season indicate pH values between 7.3 and 7.8, and electrical conductivity between 39 and 73 mS·cm −1 . The water isotope composition measured at one of these pools during the wet season (March 2017) was 4.87% for δ 13 C DIC and −3.4% for δ 18 O H2O . Great seasonal variation is nevertheless to be expected.  At the microscale, pisolites are composed of alternations of sparitic, micritic, and amorphous silica laminae ( Figure 5D), often developing around a nucleus (e.g., calcite clast or volcanic mineral) ( Table 1). Their concentric spherical form probably results from freeze/thaw processes [11,12] but also from wind action [11], which could occasionally trigger displacements, thus grouping small pisolites into symmetric ripples, for example.

Palmatoid Concretions
Palmatoid concretions are fixed concretions developing in perennial pools, at the upper-lower platform transition (Figure 2A, Table 2). These "arborescent" concretions were first mentioned by Risacher and Eugster [12]. Here, they are called palmatoid concretions, because they are similar in appearance to Acropora palmata corals. Some grow from pool floors, forming~10 cm columnar concretions, divided into a submerged part ( Figure 5E,F), composed of continuous laminae with regular alternations of micrite and sparite, and an aerially exposed part at the air-water interface ( Figure 5E-G), composed of millimetric branches with discontinuous laminae of micrite, sparite, and amorphous silica. These aerial branching structures can also form on top of pisolites, growing horizontally rather than vertically ( Figure 5C,G). During windy conditions, considerable water oscillation and spray affect the upper part of palmatoid concretions [12].

Platystromatolites
Platystromatolites are stromatolites with a plate-shaped morphology, developing in the distal part of the lower carbonate platform, in very shallow "distal" pools (<10 cm deep), surrounded by soft sediment (Figure 6A,B). They are frequently associated with small springs (seepages), indicated by a brown-to-reddish color due to Fe-oxide impregnation in the surrounding sediments. This area can be temporarily flooded by the saline lake during the wet season. Physicochemical parameters measured in March 2017 indicated a pH of 7.5 and electrical conductivity of 70 mS·cm −1 , rising to 127 mS·cm −1 for pools close to the lake.
Thin sections of these structures show that they start to grow on rock fragments (mostly calcite clasts) in the mud and gradually acquire their plate-shape at the air-water interface, like micro-atolls. They are composed of successive micritic laminae, incorporating detrital grains and microbial filaments ( Figure 6C, Table 2).

Calcite Mud
Mud containing calcite crystals covers large areas at Laguna Pastos Grandes. Two main types of mud can be identified, based on depositional setting, micro-texture, and elemental composition ( Table 2).
The first type, anhedral mud, is composed of small anhedral crystals of calcite a few µm in size, formed in proximal-to-intermediate depositional zones of hydrothermal systems (Z1-Z2, see Figure 4A-C), as observed by Bougeault et al. [14]. Anhedral mud is often associated with diatom frustules, forming micritic aggregates ( Figure 5D).
The second type, fusoid mud, is composed of fusoid crystals of calcite, approximately 3 to 100 µm in length along the c-axis, with a length-to-width ratio of 4 to 5. Fusoid calcite crystals exhibit a scalenohedral morphology, with faces delimited by curved edges (Figure 6E,F). The XRD analysis identifies this calcite as a low magnesian calcite (main diffraction peak d 001 at 3.03 Å) but EDX analysis indicates a non-negligible percentage of magnesium (~3.4 mol% MgCO 3 ), reaching the range of intermediate magnesium calcite noted by Rucker and Carver [37]. Similar morphologies have been observed in several Chilean lakes, determined as ikaite or calcite pseudomorphs after ikaite (e.g., Laguna Lejía [38]; Laguna Potrok Aike [39]; Laguna Cháltel [40]).
Fusoid mud precipitates in many places, but not in the purely hydrothermal systems, and can be divided into two subgroups based on calcite crystal content. The first subgroup, containing 90% to 95% of calcite, was observed in both the upper and lower platforms. In the upper platform, fusoid calcite mud can be observed in pools, occasionally associated with pisolites. In the lower platform, fusoid mud was more frequent, and was characterized by thin whitish deposits, mostly in very shallow depressions (See Figure 2C). The second type, containing less than 20% of fusoid calcite, was termed fusoid-bearing mud. It was found in the eastern part of the playa, covered by the ephemeral lake, and associated with gypsum, volcanoclastic grains, and diatoms ( Figure 6G).

Carbon and Oxygen Stable Isotopes
A broad range of low-to-high δ 13 C carb and δ 18 O carb values was recorded in these modern carbonate facies, along a transect from the western margin to the lake, exclusively composed of calcite ( Figure 7A; Table 3). Calcite from microbialites in the hydrothermal La Salsa system showed the greatest amplitude, with values ranging from −1.4% to +10.8% for δ 13 C carb , and −12.6% to −2.0% for δ 18 O carb , with the lowest value recorded in the central pool, and the highest in the distal zone ( Figure 7A). Downstream along the transect ( Figure 7B), three trends were observed for δ 13 C carb : (1) a moderate increase in Z1 (−1.4% to +4.0% ); (2) a relative stability from Z2 to Z3 (+4.0% to +3.1% ); and (3) a stronger increase in Z4 (+3.1% to +10.8% ). For δ 18 O carb values, three trends were also distinguished: (1) an increase from Z1 to Z2 (−11.5% to −6.6% ); (2) a relative stability in Z2; −6.6% to −7.3% ); and (3) a further increase from Z3 to Z4 (−7.3% to −2.0% ).     [41], ages considered before 1950 (yr BP = year before present), after correction of the fractionation produced by the spectrometer. 14 C ages cannot be used for dating due to the incorporation of fossil carbon (see Section 5-Discussion).
Anhedral calcite, in mud precipitating in the hydrothermal La Salsa system recorded lower isotope values, from −1.4% to +2.1% for δ 13 C carb and −11.5% to −5.8% for δ 18 O carb . Pisolites, collected from ephemeral and perennial pools on the upper platform, recorded intermediate values compared to La Salsa, ranging from +1.3% to +5.4% for δ 13 C carb and from −9.3% to −3.7% for δ 18 O carb . Fusoid mud from the lower and upper platforms can be distinguished from pisolites, with slightly lower values for δ 13 C carb (from +1.3% to +3.5% ), but higher for δ 18 O carb (from −5.9% to −3.1% ). Platystromatolites presented similar values, ranging from +2.6% to +4.0% for δ 13 C carb , and from −5.3% to −2.6% for δ 18 O carb . Palmatoid concretions recorded a much broader isotope amplitude, ranging from +0.6% to +11.5% for δ 13 C carb and from −8.1% to −2% for δ 18 O carb , close to the amplitude observed over the entire La Salsa transect. Two groups of values can nevertheless be identified: the submerged part, with lower values (+0.6% to +2.3% for δ 13 C carb and −8.1% to −7.2% for δ 18 O carb ), and the subaerially exposed part, with higher values (+6.1% to +11.5% for δ 13 C carb and −5.1% to −2.0% for δ 18 O carb ). Fusoid-bearing mud from the saline lake appeared very different from the other facies, with δ 13 C carb values ranging from −2.5% to −0.6% , and δ 18 O carb values ranging from −3.4% to −1.5% .

Fraction Modern Carbon (F 14 C) from Modern Carbonates
In five of the facies analyzed, the fraction modern (F 14 C) was very low (Table 3). In La Salsa, the F 14 C value was 0.0469 for a proximal microbialite (Z1-mushroom-like), increasing to 0.1068 (2.3 times higher) in a distal microbialite (Z4-shrub). The fraction modern measured for concretions growing in pools was also low, with an F 14 C value of 0.0639 for a pisolite and 0.0333 for a palmatoid concretion. A higher value of 0.1038 was recorded for a platystromatolite.

Using Isotopes for Carbonate Dating in Volcanic Settings
Comparison between the U-Th ages from the carbonate platform and the 14 C ages from the modern deposits showed that the 14 C ages were globally older than the U-Th ages, even though these modern carbonate deposits are stratigraphically higher than the carbonate platform.
In the uppermost layers of the actively precipitating microbialites at the edge of the central pool of La Salsa spring, the recorded fraction modern was very low (F 14 C = 0.0469). The age calculated from this value (24.5 ka) would be inconsistent, as modern carbonates are associated with the modern development of microbial mats and ongoing carbonate precipitation. In addition, modern carbonates develop above the Holocene carbonate platform, dated at 1-2 ka by U-Th, and so they should be younger. The anomalous 14 C ages were coherent with the incorporation of "dead carbon", as the pCO 2 estimated in these systems was approximately 60.3 to 389 mbar [16], that is to say 260 to 1676 times greater than atmospheric pCO 2 (considering an atmospheric pCO 2 of 407.18 ppm at 0 m, and so 232 ppm (0.232 mbar) at 4500 m on March 2017, source: NOAA). Such high hypogean pCO 2 would trigger major dilution of atmospheric CO 2 in the laguna water system (e.g., [42]). This overestimate results from the incorporation of metamorphic or magmatic CO 2 or possibly of CO 2 from the recycling of older carbonates in the laguna [16,42,43]. Caution is therefore recommended in settings where hypogean CO 2 emissions are present.

Factors Controlling the Isotope Record in Each Hydrological Setting
Each part of the complex Pastos Grandes hydrological system (i.e., piedmont springs, ephemeral streams, hydrothermal springs, ephemeral or perennial pools, and an ephemeral saline lake) has a specific chemical composition [16], leading to different types of mineralization ( Figure 8A). The origin of chemical elements and gas in the system, particularly calcium and CO 2 , were investigated by Muller et al. [16]. They determined that the main driver for the precipitation of modern carbonates at Pastos Grandes was the deeply sourced CO 2 surging from hydrothermal springs and partly deriving from the mantle (47%). Additionally, the modeling provided by their study showed that fluid circulation under high PCO 2 (20 bar) and high temperature (~225 • C) resulted in the alteration of the volcanic bedrock, which contains Ca-rich minerals (mainly feldspars). Therefore, hydrothermal springs can be considered as the main providers of solutes, and thus as reference water compositions. Carbonate mineralization is mainly restricted to the primary solute providers (i.e., hydrothermal springs) and receivers (i.e., pools and the lake). These three hydrological environments ( Figure 8B) have similar chemical compositions with different levels of concentration, and some temporal variability related to meteoric input and evaporation.

Pure Hydrothermal Systems
Pure hydrothermal systems are places where hydrothermal inputs rise to the surface, characterized by hot, acidic waters, high pCO2, and low isotope compositions ( 18 OH2O approximately -12‰ and  13 CCO2(gas) approximately -11.4‰ [16]). In the system under study, La Salsa, the amplitude of isotope values recorded from microbialites was large (12.2‰ for  13 C; 10.6‰ for  18 O), over a short distance (~280 m), and close to the full range of values measured in Pastos Grandes ( Figure 8C). A previous study showed that microbialite morphology and distribution in La Salsa was primarily controlled by CO2 degassing, temperature decrease, microbial photosynthesis, and substrate nature [14].
The  13 Ccarb values for calcite precipitating as ledge microbialites on the edges of the central pool were -0.8‰, higher than the  13 Cgas (about -11.2‰ PDB) and  13 CDIC (about -6‰ PDB) values [16]. The abundance of CO2 in the pool favors degassing, triggering isotope fractionation between the gas-liquid phases and liquid-mineral phases, thus explaining the higher values recorded by the calcite in the ledge microbialites [44][45][46].
The  18 Ocarb value measured at the edge of the central pool (-12.5‰ PDB) was controlled by the  18 Owater. Applying the equation of Friedman and O'Neil [47], with a  18 Owater of -11.4‰ SMOW (Standard Mean Ocean Water) [16] and a precipitation temperature of 43 °C, the  18 Ocarb value calculated for the same microbialites would be lower (-16.7‰ PDB). This 4.2‰ difference corresponds to the non-equilibrium calcite-water fractionation observed in most other hydrothermal systems around the world, due to kinetic effects, particularly in conjunction with CO2 Compared with hydrothermal profiles, the profiles for both the pools and the lake show the same relative decrease in concentration for two solutes, Ca ++ and HCO 3 − , implying that these solutes have been consumed by carbonate precipitation.

Pure Hydrothermal Systems
Pure hydrothermal systems are places where hydrothermal inputs rise to the surface, characterized by hot, acidic waters, high pCO 2 , and low isotope compositions (δ 18 O H2O approximately −12% and δ 13 C CO2(gas) approximately −11.4% [16]). In the system under study, La Salsa, the amplitude of isotope values recorded from microbialites was large (12.2% for δ 13 C; 10.6% for δ 18 O), over a short distance (~280 m), and close to the full range of values measured in Pastos Grandes ( Figure 8C). A previous study showed that microbialite morphology and distribution in La Salsa was primarily controlled by CO 2 degassing, temperature decrease, microbial photosynthesis, and substrate nature [14].
The δ 13 C carb values for calcite precipitating as ledge microbialites on the edges of the central pool were −0.8% , higher than the δ 13 C gas (about −11.2% PDB) and δ 13 C DIC (about −6% PDB) values [16]. The abundance of CO 2 in the pool favors degassing, triggering isotope fractionation between the gas-liquid phases and liquid-mineral phases, thus explaining the higher values recorded by the calcite in the ledge microbialites [44][45][46].
The δ 18 O carb value measured at the edge of the central pool (−12.5% PDB) was controlled by the δ 18 O water . Applying the equation of Friedman and O'Neil [47], with a δ 18 O water of −11.4% SMOW (Standard Mean Ocean Water) [16] and a precipitation temperature of 43 • C, the δ 18 O carb value calculated for the same microbialites would be lower (−16.7% PDB). This 4.2% difference corresponds to the non-equilibrium calcite-water fractionation observed in most other hydrothermal systems around the world, due to kinetic effects, particularly in conjunction with CO 2 degassing [45,46]. This observation confirms the difficulty of using δ 18 O carb for paleotemperature calculation in hydrothermal contexts with abundant CO 2 .
From Zone 1 (proximal) to Zone 4 (distal) of the hydrothermal system, an increase was observed in δ 13 C carb and δ 18 O carb values measured in microbialites. Nevertheless, the increase in δ 13 C was relatively low and may have resulted from a rapid transfer of water from the channel to the distal area, as the water flow was relatively high. A similar increase was also observed in many travertine-producing hydrothermal springs, usually triggered by CO 2 degassing and temperature decrease (e.g., [44,46,48]). Zone 4 showed an abrupt increase in δ 13 C carb values (+7.1% to +10.7% PDB), which were uncommonly elevated in comparison with other hydrothermal systems. These high δ 13 C carb values were associated with the presence of millimetric branches at the top of the shrubs. This branching mineralization resulted from subaerial exposure, as water was drawn upward by frequent water-surface oscillations, and capillary rise through microbial mats [14,49]. Subaerial exposure and strong evaporation could trigger 13 C enrichment at the top of shrubby morphologies. Isotope enrichment has seldom been documented in shrub morphologies in hydrothermal settings. Although similar morphologies in Bahía Concepción (Mexico), with high δ 13 C carb values, have been explained by CO 2 degassing [50], it is possible that subaerial exposure and evaporation may have also played a role in this intertidal context. For the fraction modern, the value (F( 14 C): 0.1068) measured in a shrub in Zone 4 was 2.28 times higher than the value measured in a ledge in Zone 1 (F( 14 C): 0.0469). This increase suggests greater incorporation of atmospheric carbon in the carbonate precipitated in Zone 4. Despite this increase, the fraction modern remains relatively low when compared to the atmospheric value measured in 2010 (F 14 C: 1.05; [51]), indicating that "dead carbon" remains the main source of CO 2 incorporated in carbonates.

Combined Hydrothermal and Meteoric Inputs
The pools between the hydrothermal springs and the saline lake are fed by hydrothermal water, more or less diluted by meteoric inputs ( Figure 8A). The amplitude of isotope values (7.3% PDB for δ 18 O carb and 10.6% PDB for δ 13 C carb ) for the main carbonate macrostructures found on the platform (pisolites, palmatoids, platystromatolites, and fusoid muds) is lower than for the pure hydrothermal system, with higher minimum values ( Figure 8C). The broad range of δ 18 O carb values measured in pisolites (from −9.3% to −3.7% PDB), which precipitated during the same period, could originate from microenvironmental variations, with different inputs from meteoric and hydrothermal waters, but also from the stability of the pool, which depends on whether and how much it is fed by waters, but also variations in the evaporation rate among the pools, which consequently induces fluctuations in δ 18 O carb [52]. The δ 18 O carb values measured in platystromatolites and fusoid muds were more positive than those observed in pisolites, suggesting a higher rate of evaporation.
Palmatoid concretions only developed in persistent pools lying between the upper and the lower platforms, in association with pisolites. The values for δ 18 O carb and δ 13 C carb measured in the submerged part of these palmatoids were similar to the most negative values measured in pisolites ( Figure 7A), suggesting subaqueous precipitation for these carbonates. By contrast, the values for δ 18 O carb and δ 13 C carb measured from the subaerially exposed part of the concretions were much higher than even the highest values for pisolites ( Figure 7A). The branching morphology of these palmatoid concretions is very similar to that of the shrub microbialites from hydrothermal springs, with equally high δ 18 O carb and δ 13 C carb values. In both cases, water was drawn upward by water-surface oscillations, but the associated process for shrubs was capillary rise through microbial mats [14], while for palmatoid concretions, the process was wind-induced spray [12]. Such positive shifts have already been described for other carbonate deposits, including vadose crusts and cements near marine coastlines [53], stalagmites [54], and carbonate precipitated in mountain streams [55]. Coupled with strong evaporation, spray-induced processes accentuate the release of light CO 2 in strongly alkaline waters, greatly increasing both δ 13 C carb and δ 18 O carb values.

The Saline Lake, Mainly Fed by Meteoric Waters
Slightly negative δ 13 C carb and δ 18 O carb values were recorded in the fusoid-bearing mud in the eastern part of the laguna ( Figure 8C), with an ephemeral central lake that is strongly affected by the arid climate, triggering evaporation over the year. The slightly negative δ 18 O carb values observed here were similar to the highest δ 18 O carb values recorded on the carbonate platform (i.e., fusoid mud, shrubs, palmatoid concretions, and platystromatolites), which are interpreted as having precipitated from strongly evaporated waters. This result suggests that the lake water was also strongly evaporated when the fusoid-bearing mud precipitated. The lake is mainly fed by meteoric inputs with a very low δ 18 O water value (−17% ), supplemented by underground, non-evaporated, hydrothermal waters, with a slightly higher, but still negative δ 18 O water value (−11.4% ). The slightly negative δ 13 C carb values observed here were similar to the most negative δ 13 C carb values found at the outlet of hydrothermal springs. They are best explained by two factors: (i) the extremely efficient C-isotope equilibrium of the dissolved inorganic carbon in the lake water with the atmospheric CO 2 because of the shallowness of the lake, which evaporates completely every year, and (ii) the addition of low δ 13 C DIC in the lake by underground hydrothermal inputs and organic matter mineralization.

Laguna Pastos Grandes in Comparison with Other Andean Contexts
Laguna Pastos Grandes carbonates record a great range of values for both δ 13 C (13.9% ) and δ 18 O (11.1% ), associated with three main depositional environments. This dispersion is among the highest known in the Central Andes ( Figure 9) and elsewhere ( Figure S3). A greater dispersion of isotope values has only been observed at Laguna Negra in Argentina (17.3% in δ 13 C; 11.3% in δ 18 O) where the amplitude results not only from temporal variations in moisture, which are recorded in microbialites (temporal amplitude varying from 4% to 11% for δ 13 C, and 3% to 8% for δ 18 O) [9], but also from spatial controls driven by the diversity of depositional environments (inlet mixing with lake, or lake alone) [9,56]. Differences in depositional environments strongly affect the isotope record, even over short distances.
At Pastos Grandes, depositional environments fed by pure hydrothermal or mixed hydrothermal-meteoric waters record markedly negative δ 18 O values, contrasting strongly with the higher values recorded in carbonates at other hydrothermal Andean sites (i.e., Las Peladas and Las Coladas, Argentina) [4]. Higher δ 18 O values indicate either a difference in initial isotope composition at the spring outlet, or enrichment in heavy isotopes further from the outlet, caused by decreasing temperatures and increased evaporation. The most positive δ 13 C values, found in carbonates deriving from hydrothermal fluids at Pastos Grandes, were close to the most positive values obtained in other Andean hydrothermal settings [4]. At both Las Coladas and Las Peladas, the origin of these high values, recorded in laminated travertines, was attributed to intense CO 2 release. In contrast, at Pastos Grandes, the high δ 13 C values were measured in isolated, partially emerged objects (palmatoid and shrub concretions), far from the spring outlet. In this subaerial depositional context, the high δ 18 O values resulted from different processes, recorded exclusively at the facies scale: the evaporation of alkaline waters drawn upward by frequent water-surface oscillation, and capillary rise in the microbial mat.
In the ephemeral central lake of Pastos Grandes, the slightly negative values (between −2.5% and −0.6% for δ 13 C; −3.4% and −1.5% for δ 18 O) were low compared to values for other Andean lacustrine carbonates (including mud and microbialites), particularly the values for δ 13 C. In fact, other Andean lakes frequently record higher values, with some almost reaching +10% for both δ 13 C and δ 18 O (e.g., Laguna Negra and El Peinado in Argentina, and Laguna Seca and Laguna del Negro Francisco in Chile; Figure 9). Even though some latitudinal variations in precipitation exist, which may induce some changes in the initial δ 18 O coming from rainfall, the strong evaporation ruling over the Central Andes affects the isotope record, inducing very positive δ 18 O values, (e.g., [56]), enhanced by the perennial character of the lakes with a relatively long water residence time [4,52,57]. In such lakes, partial evaporation favors the escape of light oxygen isotopes ( 16 O), so that heavy oxygen isotopes ( 18 O) accumulate in lake water over the years. This 18 O enrichment is then recorded in the carbonates. In Andean systems, the elevated groundwater CO 2 concentration, common in volcanic areas, can reach lakes, even if superficial markers of hypogean input (e.g., travertine) are not always observed [11,42]. The major degassing of light CO 2 of the lake has been shown to trigger 13 C enrichment in water and carbonates [43,56]. The strong enrichment in heavy carbon isotopes in carbonates may track the rise of CO 2 in systems influenced by volcanism, even in the quiescent stage. These abiotic processes are also enhanced by the long water residence time in perennial lakes that favors storage of heavy isotopes such as oxygen [9,43]. These values can be further amplified by photosynthesis and microbial methanogenesis activities [11]. In the case of Pastos Grandes, the origin of CO 2 in the lake results either from hydrothermal inputs or from the air-water equilibrium. However, as the lake water completely evaporates every year, the short residence time prevents significant enrichment in 13 C and 18 O, thus explaining the lower δ 13 C carb -δ 18 O carb values observed. Moreover, as Pastos Grandes is an endoreic lake, the short water residence time would also explain the absence of covariance between δ 13 C carb and δ 18 O carb , often used to identify closed-basin settings [58].

Conclusions
Laguna Pastos Grandes is a palustrine-to-lacustrine site developing in a quiescent volcanic caldera with abundant ongoing carbonate sedimentation, diversified macrofacies, and several parent waters. Seven different carbonate facies were identified along a platform-lake transect. The sedimentological and hydrological diversity is accompanied by great amplitude in δ 13 C carb and δ 18 O carb values, unusual for modern continental environments, whether in the Altiplano or elsewhere in the world. This study shows that several extrinsic factors play an important role in the resulting isotope record in carbonates:

•
The initial isotope composition of parent waters (purely hydrothermal, mixed hydrothermalmeteoric, or meteoric-dominated) is the main extrinsic factor triggering differences in carbon and oxygen isotope values recorded in carbonates from different depositional environments; • By moving away from hydrothermal inputs, isotopes are increasingly influenced by aridity, causing evaporation, enhanced by wind action, and by capillary rise through the microbial mat, thus triggering major carbon and oxygen isotope enrichments; • The short water residence time explains the main difference between Laguna Pastos Grandes (with a slightly negative isotope signature of carbonates) and the other Andean lakes (with positive isotope signatures of carbonates), which are all under the influence of an arid climate. The isotope convergence toward high δ 13 C carb from Pastos Grandes and several other Andean lakes is triggered by volcanic CO 2 input, and so carbon isotopes can be particularly useful to identify ancient carbonate systems affected by volcanic activity. However, controlling factors triggering 13 C enrichment in carbonates corresponded to large-scale processes for Andean lakes (long residence time), and facies-scale processes for Pastos Grandes (evaporation of drawn-upward waters); • The fraction modern F 14 C was modified in volcanic settings by the injection of "dead carbon" into the system. This makes 14 C dating of carbonates particularly unreliable in Andean lakes enriched in volcanic CO 2 .
Laguna Pastos Grandes is an interesting natural laboratory, where isotope variations can be observed in an active system, in a limited area, over a very short period (compared to geological time scales). The extrinsic processes observed here could well have affected carbonate isotope signatures in ancient subaerial carbonate deposits in continental settings under volcanic influence.
Supplementary Materials: The following are available online at http://www.mdpi.com/2075-163X/10/11/989/s1, Figure S1: Laguna Pastos Grandes sedimentary system, showing sampling sites (modern carbonates) used for isotope measurements; Table S1: Location of samples from the carbonate platform used for U-Th dating; Figure S2: Lake level fluctuations from 2000 to 2019; Figure S3: δ 18 O/δ 13 C values, recorded in modern-to-recent carbonates, precipitating in hydrothermal to lacustrine settings in Africa (in red), Europe (in yellow), and North America (in green).