Geothermal Fluid Variation Recorded by Banded Ca-Carbonate Veins in a Fault-Related, Fissure Ridge-Type Travertine Depositional System (Iano, southern Tuscany, Italy)

Banded Ca-carbonate veins in travertine deposits are efficient recorders of the compositional fluctuations of geothermal fluids flowing (or flowed) from deep reservoirs up to the surface, within fault zones. In this view, these veins represent key tools for decoding those factors that influenced the geochemical variations. We have analyzed veins developed in fractures channeling geothermal fluids forming travertine deposits. The studied veins cut a fossil travertine fissure ridge, near the Larderello geothermal area (Iano area, southern Tuscany) where geothermal fluid circulation is favored by NE-trending strike-to-oblique-slip faults and their intersections with NW-trending normal ones. U-Th dating indicates that fluid circulation occurred from (at least) 172 ka to 21 ka. In this time span, the geothermal fluid changed in composition, and the banded Ca-carbonate veins recorded these variations in terms of mineralogical and stable isotope composition and temperature (
 
 T
 
 ) of deposition. We also documented for the first time the occurrence of Mn-rich black tree-shaped structures within the veins. Mineralogy coupled with stable and clumped isotope measurements allows the reconstruction of some features (i.e., crystal texture, temperature, and CO2 origin) and the inference of the processes (i.e., pH, T, and pCO2 variations) that have controlled the fluid evolution through time. Multiple-stage and one-stage deposition processes have played an important role in modifying the stable isotope composition of banded Ca-carbonate veins; temperature coupled with pCO2 also influenced their mineralogical composition. Interpreted in the context of the tectonic setting, the data show that the NW-trending faults have mainly controlled travertine deposition. Their intersection with NE-trending faults, interpreted as transfer faults, highlights the important role of transfer zones in channeling the geothermal fluids.

The interplay between fault activity and travertine deposition was addressed by several authors in the last decades [3,16,18,19,[22][23][24][25]. Travertine fissure ridges [26] are considered the most useful deposits for neotectonic investigations in geothermal areas. A fissure ridge-type travertine deposit consists of a ridge made of bedded travertine, from gently to steeply dipping away from the apical part, where the central fissure is located, corresponding to the fault trace intersecting the substratum. From such a fissure, the geothermal fluids flow out, and travertine deposition takes place, growing the ridge.
The relationships among fluid flow, travertine deposits, and tectonic activity have been discussed by several authors, especially focusing on fissure ridges (e.g., [3,12,18,19,[27][28][29][30][31]). Variations in the chemistry and isotope composition of the travertine forming the ridge and, especially, of the banded Ca-carbonate veins (i.e., banded travertine in [22]) filling the fissure can be linked to different processes influencing hydrogeochemistry, including climate [8,32], volcanism [33], fluid-rock interaction, and tectonics [3,7,34,35], with particular concern on the palaeoseismicity (cf. [8,13,15]). This attempt encounters several difficulties, because there are many parameters that influence the geochemical characteristics of this kind of carbonate deposit. In particular, the tectonic activity controls the permeability of the fault rocks and, as a consequence, the hydrothermal fluid pathways within the crust. Such tectonic effects likely exert stronger control than climatic effects on hydrothermal fluid property variations within hypogean conduits, such as faults and fractures. In order to constrain the possible impact of the tectonic activity on the geochemical and isotopic composition of the geothermal fluids, we analyzed banded Ca-carbonate veins formed in a late Pleistocene fissure ridge-type travertine depositional system cropping out nearby the Larderello geothermal area (southern Tuscany, Italy; [36]). This is an ideal region for this kind of study, since it records a long evolution of the geothermal fluid flow, veins, and related travertine deposits that are exposed in the area (Iano area, Figure 1) located in the shoulder of a broad Neo-gene structural depression (known as the Volterra Basin, [37,38]). Here, late Quaternary tectonics gave rise to a shallow hydrothermal system and travertine deposition and still ensures active thermal springs. The analyzed veins formed during several episodes show different fabrics and colors suggesting compositional changes through time. We discuss the variation of the mineralogy (i.e., aragonite and calcite) and stable isotopic composition in relation with the tectonic activity. Furthermore, we document the occurrence of Mnrich black three-shaped structures forming some bands within the banded Ca-carbonate veins. We interpret these observations as evidence of change in geothermal fluid properties, possibly induced by the tectonic activity, promoting a variation in the fluid-rock interaction.
These Neogene-Quaternary basins, including the Volterra Basin, were filled by Pliocene-Quaternary marine and continental sediments [37,52]. Their deposition was controlled by the activity of NNW-SSE-striking normal faults and coeval NE-trending transfer zones [69,70], along which 2 Geofluids most of the late Miocene-Pleistocene magmatic activity is concentrated [24,68,[71][72][73][74][75]. Geothermal fluid circulation in the whole inner Northern Apennines also appears to be controlled by transfer zones [19,[75][76][77][78]. The Iano area is located within a transfer zone (Figure 1(b)) dissecting a NNW-SSE trending normal fault system that bounds the Volterra Basin to the east. The interplay between the two fault systems (i.e., transfer and normal faults) gave rise to favorable conditions for hydrothermal circulation and thermal springs occurrence ( Figure 2 [19] and (b) geological map of the Larderello geothermal area and Volterra Basin. The study area is also indicated. Redrawn after Brogi et al. [134] and Liotta and Brogi [19].

Geofluids
Iano village where low temperature thermal springs (T = 22°C) are still discharging Ca(Mg)-HCO 3 fluids with high content of Sr and SO 4 [11].
The Iano area was investigated during Hg and coal mining exploration since eighteenth and nineteenth centuries, and detailed information about its geological setting can be found in Barberi [79] and Costantini et al. [80].

Methods
We conducted detailed description and analysis of the banded Ca-carbonate veins crossing travertine deposits, to investigate their (i) geometrical setting in relation to the main structures, (ii) mineralogy and textural features, (iii) stableand clumped isotopic compositions related to the fluid geochemistry and temperature, and (iv) depositional ages as inferred by U-Th geochronology.
3.1. Geometrical Setting, Mineralogy and Textural Features. A detailed study of travertine stratigraphic setting and its structural features was carried out applying the classical methods of sedimentary and structural geology. The collected data were interpreted in the context of the area's tectonic setting, which we have represented in a new geological map at 1 : 10 000 scale ( Figure 2). This map was produced via integration of preexisting geological maps (geological map of Tuscany 1 : 10.000 scale and geological map of [80]) and new field data, covering an area of about 11 km 2 . Structural and kinematic data were collected in favorable exposures through structural stations, as indicated in Figure 2. The attitude of

Geofluids
banded Ca-carbonate veins was also analyzed and compared with the fault striking.
To visualize textural and mineralogical changes, a petrographic investigation of the samples was performed by polarized light optical microscopy on thin sections from several samples of banded Ca-carbonate veins using a Zeiss-Axioplane polarizing microscope. Based on this investigation, three key samples, labelled PAM1, PAM13 and ET3, were selected for SEM-EDS, X-ray powder diffraction, micro-Raman analyses, stable isotope measurements, and U-Th dating. Several aliquots of one sample (IA1) were selected for clumped isotope thermometry.
Scanning electron microscope (SEM) observations and microanalysis (EDS) of selected bands were performed at the Department of Earth and GeoEnvironmental Sciences (University of Bari), using a ZEISS EVO 50XVP instrument, coupled with energy dispersive X-ray Spectroscopy (EDS, Oxford-Link). Selected areas were observed on thin sections of samples PAM1 and PAM13 after graphite sputtercoating (30 nm thick). Elemental distributions of Mn were also mapped.
Mineralogical identification of PAM1 and PAM13 samples was determined by means of X-ray powder diffraction using an Empyrean, Malvern Panalytical Diffractometer at the Department of Earth and Geo-Environmental Sciences (University of Bari). X-ray powder diffractogram (XRPD) data were collected in the 6-80 2θ range using Ni-filtered CuKα radiation obtained at 40 kV and 40 mA. For this purpose, polished sample fragments and randomly oriented powder mounts were used. The analysis of the powder XRPD patterns was carried out using Highscore Plus software. The lattice parameter determination was carried out by means of Rietveld refinement using the software BGMN [81]. In addition to XRPD analysis, confocal Raman microscopy was used for a more detailed mineralogical characterization. Micro-Raman spectra were recorded with a LabRAM HR Evolution Horiba spectrometer at the Department of Chemistry (University of Bari). This spectrometer is equipped with an Ultra-Low Frequency (ULF) notch filter and an electrically cooled CCD detector. Raman spectra were excited by a He/Ne (λ = 632:8 nm) laser source using a 50× long working distance objective. The instrument was calibrated using the 520 cm −1 line of a Si wafer. Raman spectra were collected on polished thin section of PAM1 and PAM13 samples by using LabSpec6-Horiba software and reworked by using Origin software. Repeated acquisitions were accumulated in the 100-3800 cm -1 range to improve the signal-to-noise ratio. Micro-Raman spectra were recorded on different colored bands for each sample. Spectral scanning was also performed in the sample areas where black tree-shaped structures were clearly identified.

Vein Isotopic Composition and Geochronology.
Stable carbon and oxygen isotope measurements were performed on single bands composing each vein (PAM1, PAM13, and ET3). Analyses on microdrilled powders were carried out at the Institute for Geological and Geochemical Research, Research Centre for Astronomy and Earth Sciences (Budapest, Hungary). Powdered sample analyses were carried out using the continuous flow technique with the H 3 PO 4 digestion method [82]. 13 C/ 12 C and 18 O/ 16 O ratios were measured in CO 2 gases using a Finnigan Thermo delta plus XP mass spectrometer equipped with an automated GasBench II. The results are expressed in δ-notation [δ = ðR1/R2 − 1Þ × 1000, where R1 is the 13 C/ 12 C or 18 O/ 16 O ratio of the sample, and R2 is the corresponding ratio of the Vienna Pee Dee Belemnite (V-PDB) standard] in parts per thousand. Isotopic compositions are expressed in the standard δ-notation in parts per thousand (‰) relative to V-PDB (δ 13 C and δ 18 O) and Vienna Standard Mean Ocean Water (V-SMOW; δ 18 O). Reproducibility was better than ±0.1‰ for carbonate δ 13 C and δ 18 O.
Clumped isotope thermometry [83][84][85] was applied to determine the precipitation temperature of carbonate minerals in order to reconstruct the palaeotemperature of the hydrothermal fluids channeled through the faults. The powdered sample (IA1) was analyzed at the University of Washington's IsoLab, Seattle, WA, USA, using the procedures of Burgener et al. [86]. For each replicate analysis, 6-8 mg of carbonate powder was digested in a common acid bath at 90°C, and the resulting sample CO 2 was purified cryogenically using an automated system. Measurements were performed using a Thermo Scientific MAT 253 mass spectrometer with dual inlet configured to measure m/z 44, 45, 46, 47, 48, and 49. Data were corrected following Schauer et al. [87], using the 17 O interference correction parameters recommended by Brand et al. [88] as well as a pressure baseline correction [89] to calculate δ 13 C, δ 18 O, and Δ 47 . δ 13 C and δ 18 O values were referred to the V-PDB and V-SMOW scales using internal laboratory standards calibrated against international standards NBS-18 and NBS-19 (International Atomic Energy Agency, Vienna, Austria) and LSVEC (National Institute of Standards and Technology, Gaithersburg, MD, USA). A calcium carbonate reference material (ETH 1-4, [90]) or in-house standards were analyzed every 4 unknown samples. Δ 47 values were referred to the carbon dioxide equilibrium scale (CDES) using CO 2 heated to 1000°C and CO 2 equilibrated with water at 60°C and 4°C. Palaeotemperatures were calculated from the sample average corrected mean Δ 47 value (reported in the CDES, 25°C acid digestion scale using an acid fractionation factor of +0.088‰ for 90°C acid digestion) and using the calibration of Petersen et al. [91], which is the most up-to-date available for carbonates of different mineralogies and sample types (synthetic, natural, or biogenic). This calibration includes the calibration dataset of Kele et al. [92], which is travertine/tufa-specific, and the synthetic carbonate calibration data of Kelson et al. [93], which was produced in IsoLab using the same methods as the sample data presented here.
The carbonate clumped isotope values measured at ATOMKI (Institute for Nuclear Research, Debrecen, Hungary) were performed using Thermo Scientific Kiel IV Automatic Carbonate Device coupled to a Thermo Scientific MAT253 Plus isotope ratio mass spectrometer. The measurement system was installed in 2018 in Isotope Climatology and Environmental Research Center (ICER). Each carbonate sample consisted of 10-12 replicate analyses of 100 μg aliquots. ETH1, ETH2, and ETH3 were used as normalization 5 Geofluids standards, and IAEA-C2 was used as quality control sample with the values represented in Table S1 [90] of the supplementary material section. Data evaluation was carried out with Easotope application [94], using CO 2 clumped ETH PBL replicate analysis method and Brand parameters (Table S2, in supplementary material section). Δ 47 is reported in the CDES (carbon-dioxide equivalent scale) [90], using an acid fractionation fractor of +0.066‰ to place data generated with 70°C acid digestion on the 25°C digestion scale, and temperatures were calculated using the calibration of Petersen et al. [91], to be comparable to the University of Washington data. For reference, we also report the temperature values calculated using the Kele et al. [92] calibration, recalculated with brand parameters: Δ 47 = 0:0449 ð±0:001Þ × ð10 6 /T 2 Þ + 0:167 ð±0:01Þ, where T is the temperature in Kelvin, because this calibration was produced using instrumentation similar to the ATOMKI instrument.
U-Th dating was performed to set up a reliable chronological frame for these banded carbonate veins, which are important records of fluid circulation. To this purpose, multicollector inductively coupled mass spectrometry (MC-ICP-MS) [95] was applied to two samples (PAM1 and PAM13) at the High-Precision Mass Spectrometry and Environment Change Laboratory (HISPEC) of the National Taiwan University. About 0.05 g of each sample was dissolved in a concentrated HNO 3 solution. The sample solution was spiked with a 229 Th-233 U-236 U tracer [96], and 0.5 ml of HClO 4 was added to decompose the organic matter. The final separation of uranium and thorium for MC-ICP-MS analysis was achieved by chemical methods [96], including Fe coprecipitation and anion-exchange chromatography. Single secondary electron multiplier protocols on a Thermo-Fisher NEPTUNE multicollection inductively coupled plasma mass spectrometer was used for determining the U-Th isotopic compositions and contents [95]. Offline data reduction methods followed Shen et al. [97], and half-lives of U-Th nuclides used for age determination are reported in Cheng et al. [98]. Analytical errors are 2σ of the mean-age corrections, relative to 1950 AD, were calculated using an estimated atomic 230 Th/ 232 Th ratio of 4 ð±2Þ × 10 −6 .

Results
In the study area, several patches of fossil travertine deposits are exposed. In a few cases, deposition is still active. These deposits unconformably overlie Neogene-Quaternary and pre-Neogene units (i.e., Ligurian Units).
In the following sections, we illustrate the main stratigraphic and structural features of the travertine substratum. Then, the main stratigraphic and sedimentological features of the travertine deposits are reported, followed by a detailed description of the textural, mineralogical, geochemical and isotopic features of the banded Ca-carbonate veins, crossing the travertine deposits and filling fractures. These are from the bottom (i) Palaeozoic-Mesozoic metamorphic rocks (Monticiano-Roccastrada Unit, CP, VER, and TCC in Figure 2); (ii) late Triassic-Oligocene sedimentary rocks (Tuscan Nappe, Ev, Di, Sc, and Mac in Figure 2); (iii) Cretaceous sandy-marl turbiditic succession (external Ligurian Unit); and (iv) Jurassic-Cretaceous ophiolite succession overlain by the marine shale and limestone cover (inner Ligurian Unit, Jc, and APA, respectively, in Figure 2).
The contractional tectonics that gave rise to the orogenic belt produced superimposed structures already described in several papers [80,99,100]. For the goals of this work, we focus on the Neogene-Quaternary extensional tectonics, which overprinted all the previously developed structures. Thus, low-angle normal faults (LANFs) caused the direct superimposition of the successions belonging to the uppermost unit (inner Ligurian Unit) either above the metamorphic rocks of the lowermost unit (Monticiano-Roccastrada Unit) or above the evaporite level forming the basal part of the Tuscan Nappe ( Figure 2). This setting implies the lateral segmentation of several geological bodies as well as the partial or complete tectonic omission of the previously stacked units. Later, high-angle normal to oblique-slip faults, NWand NE-striking, dissected the LANFs and determined the eastern shoulder of the Volterra Basin, where the study area is located ( Figure 2).
The Neogene sediments are mostly made up of sand, clay and sandy-clay with some conglomerate embedded at different levels. These sediments were deposited in the eastern margin of the Volterra Basin, recording its tectonic evolution, as documented in detail by Brogi and Liotta [61].
Travertine deposits are mainly located along the NE-and NW-striking faults, overall at their intersection, which juxtaposes the Neogene sediments with the pre-Neogene units (e.g., nearby Iano village, Figure 2). In the southern part of the study area (e.g., near Il Palagio locality, Figure 2), the travertine deposits are in part still active and appear to be mainly related to NW-striking faults. NE-and NW-striking faults dissect Zanclean-Piacenzian sediments and, in a few cases, the travertine deposits, suggesting Quaternary activity, as discussed below.
The major faults are never exposed but their occurrence can be inferred by NE-and NW-striking scarps recognized across the whole area. Nevertheless, secondary faults are exposed in abandoned quarries or along road fresh cuts. These fault zones are characterized by apparent offsets of few meters and consist of damaged rock volumes exceeding 30 m in thickness with pervasive subvertical fractures and minor faults (Figures 3(a) and 3(b)). In some cases, hydrothermal alteration and/or localized mineralization ( Figure 3) characterize fault planes and their damage zones. Fault slip planes are surrounded by fault cores, up to 10 cm thick, formed by brecciated rocks which, in some cases, show yellowish coatings as an effect of the circulation of previous geothermal fluids (Figure 3(a)).
Structural and kinematic data were collected in structural stations ( Figure 2), in damaged rock volumes exposed in abandoned quarries where exposures and lithotypes are favorable to record and preserve striation on the slip surfaces. 6 Geofluids The results ( Figure 4) indicate that the orientation of the analyzed mesofaults is in agreement with the regional-scale structures ( Figure 2), thus displaying dominant NW-and NE-striking fault systems. Kinematic indicators mainly consist of mechanical striations (Figures 3(c) and 3(d)), chatter marks, lunate structures, tool marker and, in some cases, quartz and calcite slickensides, often accompanied by Fehydroxides crusts and patinas (Figures 3(e) and 3(f)). Kinematic indicators on the NE-striking faults show right-and left-lateral strike to oblique-slip movements, as well as normal ones ( Figure 4). Their chronological relation was not defined, although it is likely that the normal movements are probably the last ones, based on chronologies that have been already described in the surrounding areas (cf. [68]). The NW-striking faults show dominant normal component, though left-and right-lateral oblique-slip movements have been locally observed ( Figure 4). These measurements from faults were used to derive the palaeostress field, assuming that the measured slickenlines are parallel to the main shear stress. Figure 4 shows the fault plane solutions obtained for each fault trend and their dominant kinematics. For the NE-striking faults, the maximum (compressional) and minimum axes are close to horizontal, while the intermediate axis is close to vertical, during the and 5(c)), suggesting a slope depositional system that passed to distal zones by subhorizontal morphology and low-energy   8 Geofluids environments. Facies distribution and bedding attitudes suggest cone-shaped mounds, indicating the location of the palaeothermal springs at their top. In this scenario, the travertine deposits were fed by thermal springs which deposited laminated abiotic crystalline crusts, interrupted by small ponds of microbial mats, and formed where fluids run along slopes, close to the palaeothermal springs. In the distal parts, fluids fed palustrine environments, where travertine deposited in vegetated areas suggest the cooling of the flowing waters.
The travertine patches in the Il Palagio area consist of two main bodies (Figures 2 and 6(a)). The smaller is only a few hundred square meters wide and consists of a partly dismantled fossil deposit, developed in the surroundings of a thermal water discharge area. The larger body is characterized by fossil deposits (Figure 6(b)) as well as an active thermal spring, causing present travertine deposition ( Figure 5(d)) in its southern portion. From the active springs, a 22°C water is flowing, with a flow rate minor than 1 liter per second [102]. The water is characterized by pH = 6:30 and high TDS values (6665 mg/L), Ca(Mg)-HCO 3 , and high Sr and SO 4 contents, accompanied by CO 2 and H 2 S leakage [11].
The fossil deposits were quarried and today, the abandoned saw cut walls are covered by thick vegetation. The fossil travertine body is defined by NW-trending fissure ridge (Figures 6(a) and 6(b)) and aligned cone-shape mounds, revealing the main trend of the structures that controlled the hydrothermal fluid flow. Fissure ridges and travertine bodies are crossed by steeply dipping cm-to meters-thick Ca-carbonate veins, parallel to the main fault trends (Figure 6(a)), as described in detail in Section 4.3 in the next paragraph.
The main fissure ridge was deeply quarried, especially for the exploitation of banded Ca-carbonate veins crossing the central and lateral part of the body and corresponding to the feeding conduits of the palaeothermal springs. Moreover, small quarries are distributed along the fissure ridge, mainly following the meters thick central vein and the minor veins that cross the bedded travertine in the flanks of the ridge (Figure 6(b)).   b)) mostly in the NW direction, accounting for secondary conduits. The veins are arranged in parallel bands, locally with internal unconformities crosscutting the previously formed bands; the unconformable contact relationships range from parallel to angular, thus recording an interruption in the crystallization. Some veins include black bands that look like tree-shaped structures (Figure 7(d)) characterized by groups of branches in a white matrix. Their shape is very similar to the black "shrubs" identified by previous authors in travertine deposits in several localities such as Morocco (Africa), Bagni di Tivoli (Italy), Mammoth Hot Springs (Yellowstone National Park, USA) and Hierapolis (Turkey), [103][104][105]. Their thickness ranges from submillimeter to a few centimeters, forming thin laminae between vein bands (Figure 8(a)).
Each band is thought to correspond to a crack-and-seal event ( [17], and references therein). Crack episodes mostly caused reopening of previously formed fractures. The new fracture developed parallel to slightly oblique with respect to the previously deposited bands, thus giving rise to a neoformational mm-to cm-thick, geometrical irregular, syn-taxial veins (Figure 7(c)). In some cases, the geothermal fluid permeated the mechanical discontinuities of the bedded travertine, thus producing cm-thick syntaxial vein packages extending for meters, parallel to the bedded travertine strata ( Figure 5(b)). These strata, mainly formed by abiotic crystalline crusts and made up of feather-like crystals and bedded porous laminae, are gently to steeply dipping (from 5°to 50°).
The sample location is indicated in Figures 6(a) and 6(c) and described in Table 1 (Figure 8(a)) and PAM13 (Figure 9(a)) samples was analyzed by XRD to identify the carbonate minerals. XRPD analysis reveals that the V1 diffractogram (Figure 8 Figure 8(b)), likely due to a poorly crystalline impurity phase that could not be indexed with calcite or aragonite phases. The cell parameters of calcite are slightly smaller than those present in the Joint Committee on Powder Diffraction Standards (JCPDS 47-1443), while the aragonite unit cell parameters deviate slightly from those for pure aragonite causing a small enlargement of cell volume (JCPDS 41-1475). Although calcite has a rhombohedral unit cell and aragonite is orthorhombic, both structures show similar Ca lattice but different CO 3 2ion arrangements. Carbonate structures can accommodate the trace element by exchange of Ca with Mg, Mn, Fe, or Sr, which is governed by the flexibility of the Ca site surrounding structure. The substitution causes dilation or contraction of the cell volume. Therefore, the small variation of cell volumes is associated with small amounts of larger and smaller sized ions substituting for Ca in the structure of aragonite and calcite, respectively [106][107][108]. To substantiate XRPD results, SEM/EDX analysis of calcite particles detected Mg, Ca, C, and O, whilst SEM/EDX of aragonite crystals detected Sr, Ca, and O. Confocal Raman analyses also support XRPD results. Indeed, the Raman spectra (Figure 8(i), points 2 and 3) of vein V1 (PAM1 sample) showed the typical vibration modes ν 1 (1085 cm -1 calcite and 1086 cm -1 aragonite) and ν 4 (711 cm -1 calcite and 705 cm -1 aragonite) for the CO 3 group in a carbonate crystal lattice. Additional lattice mode peaks were located at 156 cm -1 , 282 cm -1 (calcite), and 207 cm -1 (aragonite). In addition, we used EDS to determine various elements at spots forming the black tree-shaped phase. The elemental mapping (Figure 8(f)) shows high intensities for manganese oxide, varying between 77.87% and 83.22%. Micro-Raman spectra were also acquired in the sample PAM1, where black spots were clearly identified. All the Raman spectra show a strong wide band at about 620 cm -1 with shoulders at about 492 and 530 cm -1 (Figure 8(j), point 4). The spectra show the characteristic bands ascribed to natural manganese dioxide minerals [109]. Based on SEM/EDX and Raman analyses, the black material appears to consist of poorly defined manganese minerals (Figures 8(c) and 8(e)). Minor percentages of MgO, FeO, and NiO have been measured in the black deposits, with values around 2%, 1% (with some values higher than 5%), and 5%, respectively (see Table 2 for data).
For sample PAM13 (Figure 9(a)), the X-ray diffraction pattern (Figure 9(c)) reveals that the carbonate bands are made up predominantly of aragonite. Also, in this case, the cell parameters calculated from diffraction peaks (a = 4:9637ð4Þ, b = 7:9715ð7Þ, c = 5:7473ð5Þ) indicate a slight dilation of the cell volume with respect to the ideal pure aragonite phase. Raman analyses (Figure 9(b)) confirm that the sample PAM13 is made up of aragonite, showing the typical aragonite peaks of the lattice modes located at 154 cm −1 , 207 cm −1 , and 1086 cm −1 . In the polished hand specimen of sample PAM1, calcite bands appear translucent and defined by different shades of gray. Calcite crystals consist of elongate-blocky crystals (Figure 10(a)), 150-300 μm wide, and from few millimeters to 5 mm long. Crystals enlarge in the growth direction, and their terminations are characterized by pointed curved faces (Figure 10(a)), suggesting that the crystals had the time to develop. Under crossed polars, these crystals show wavy extinction. Some calcite bands include black points, aligned perpendicularly to the growth direction (Figure 10(b)). These are made up of Mn-oxides, determining a dark color to the band, at the specimen scale (see Figure 8(a)).
Aragonite crystals confer a white, near opaque aspect to the bands in the polished hand specimen. The acicular aragonite crystals are made up of needle-like crystals extending along their c-axis direction (Figure 8(g)).
Calcite bands appear slightly darker gray on the backscattered images compared to the aragonite bands (Figure 10(c)). This is because calcite bands contain small percentages of Mg (3.25%). In contrast, aragonite bands do not contain Mg, but small percentages of Sr (3.08%), as also confirmed by the mineralogical analyses (cf. 4.3.1 section).
Vein V2 shows a syntaxial fabric developed perpendicular to the V1 band. This crystal texture is well represented in sample ET3 (Figure 10 Vein V3 developed parallel to the V1 and interrupted the continuity of the V2 (Figure 8(a)). The bands forming the  (Figure 10(f)). The crystals are densely packed forming thin fans of ray crystals that develop from an apical point. Their ending faces form pointed terminations, highlighting that some crystals grew faster than others. Mn-oxide-rich lamina retrace the ending crystal faces (Figure 8(h)), separating vein V1 from vein V3.
Similar features are observed in sample PAM13, formed by parallel whitish bands (Figures 7(a) and 9(a)). This vein is completely made of aragonite crystals. Aragonite bands are made up of elongated polycrystals ranging in length from 14 Geofluids 0.2 to 1.3 cm and resulting from the packing of needle thin crystals ( Figure 10(g)). The latter shows wavy extinction caused by the systematic variation of the c-axis orientation with respect to the substratum of needle-like crystals [110] and very high interference colors ( Figure 10(g)).
Two samples were selected for U-Th dating [95]: (i) V1 and V3 from sample PAM1, in order to get information on the time interval between the different veining events and (ii) sample PAM13. All samples are representative of a single band. The resulting data are summarized in Table 3. These highlight that V3 and V1 developed in a restricted time span encompassed between 26 ± 1 and 172 ± 3 thousand years ago (ka, before 1950 AD). In contrast, a radiometric age ranging between 21:1 ± 0:1 and 27:1 ± 0:1 ka was computed for the sample PAM13.

Stable and Clumped Isotope Analyses.
The results of the δ 13 C and δ 18 O analyses of the veins (i.e., PAM1, ET3, and PAM13) indicate values ranging from -1.2‰ to 4.2‰ (V-PDB) and from 24.3‰ to 26.8‰ (V-SMOW), respectively. These ranges of values can be ascribed to the group of thermogenic travertine (see [5]).
The dataset is reported in Table 4 and displayed in the diagram of Figure 11  On the basis of δ 13 C values, data can be separated in two distinct groups. The veins crossing the flank of the fissure ridge ( Figure 6(c)) are characterized by lower δ 13 C values with respect to those defining the vein from the feeder conduit. δ 18 O values are very similar for the two vein types, although PAM13 and V3 samples show slightly higher values than those from V1 and V2 veins. Stable isotope values of vein V3 fall between the two groups ( Figure 11(a)).
Clumped isotope composition (Δ 47 ) was measured on a band of aragonitic sample (IA1) from the central part of the     17 Geofluids ridge (in the proximity of PAM13 location), in order to reconstruct the palaeotemperature of the fluids that deposited the Ca-carbonate veins. The calculation gave a temperature of 67°C ± 6°C. Clumped isotope composition was determined at ATOMKI in order to calculate palaeotemperature on two other bands of the IA1 sample, which gave values of 84°C ± 5°C and 111°C ± 5°C, respectively. The data are reported in Table 5. In addition, we also report the temperature values calculated using Kele et al.'s [92] calibration for these last two values, for the reasons explained i Section 3.2. The calculation gave temperatures of 68°± 4°C and 85°± 3°C, which are more suitable for the geothermal system of   Figure 11: (a) Graph showing the distribution of stable isotope data for the analyzed samples. Blue rhombus refers to the PAM13 data; yellow square refers to the vein V1 data from PAM1 and ET3; the symbol "x" refers to the vein V2 from ET3; pale blue square refers to the vein V3 data from the PAM1 sample. (b) PAM1 isotope data distinguished on the basis of their mineralogy. See the good correlation (R 2 = 0:8527) between stable isotopes of aragonite crystals.

Discussion
The discussion focuses on three main points: (i) fluid origin, (ii) tectonic control on the fluid flow, and (iii) role of geochemistry in reconstructing the geothermal fluid property variations and the possible triggering factors.

Fluid Origin.
Geothermal fluids that formed the travertine deposits interacted with carbonate and sulfate rocks, as suggested by their high potential to deposit Ca-carbonate and by their Sr content [11]. This implies that the geothermal fluids flowed within the late Triassic evaporite and Jurassic carbonate located at the base of the Tuscan Nappe [36,111]. It is known that in the neighboring Larderello area, the geothermal fluids are mostly of meteoric origin (>95%) with a minor inflow of thermometamorphic and magmatic fluids [112][113][114]. It is therefore reasonable to assume that the parent fluids of the travertine would be related to meteoric waters, too.
In our case, we computed δ 13 C CO2 values from the carbonate δ 13 C values of the study veins, using the empirical equation (δ 13 C CO2 = 1:2δ 13 C trav: -10:5) of Panichi and Tongiorgi [118], resulting from the comparison between the Italian thermal springs and associated travertine deposits. The δ 13 C CO2 calculations for V1 and V2 veins give values ranging from -12‰ to -9‰ (V-PDB). In contrast, the computed δ 13 C CO2 values for V3 vein and for banded veins from the central fissure of the ridge are slightly higher, ranging from -7‰ to -5‰ (V-PDB). Interestingly, the computed δ 13 C CO2 values are significantly lower than most of the CO 2 gas emissions in central Italy, which vary from -2‰ to +2‰ (V-PDB, [124]).
As previously said, banded Ca-carbonate veins are the results of crack-and-seal events, likely triggered by the tectonic activity. This process could account for the variation of pCO 2 (probably due to tectonic activity) during the multistep precipitation of banded Ca-carbonate veins, from fluids that migrated from depth to the surface. According to Minissale et al. [115], this process could produce a large isotopic fractionation, causing the deposition of banded Ca-carbonate veins that are enriched in 13 C at depth. As a consequence, the rising fluid, depleted in 13 C, could deposit banded Ca-carbonate veins close to the surface, resulting in lower δ 13 C values, as we observe inV1 and V2 from PAM1 and ET3 samples. In contrast, δ 13 C CO2 calculations for V3 and PAM13 overlap with the δ 13 C CO2 values recorded in the northern part of the Larderello geothermal area, where the study area is located (Figure 1). Here, most of the gas emissions are characterized by δ 13 C CO2 ranging between -7‰ and -4‰ (V-PDB), with few values lower than -7‰ (V-PDB), [115]. These values have been interpreted as the results of primary CO 2 rising up from the mantle [116] or, alternatively, a mixing between isotopically heavy and light CO 2 coming from the geothermal reservoir and organic sediments produced in the Neogene basins, respectively [124]. Therefore, V3 and PAM13 veins could be the result of a one-stage deposition from a fluid directly coming from depth.
Temperature of the hydrothermal fluid circulated within the main fissure of the travertine ridge was determined by clumped isotope thermometry to be 67°C ± 6°C for the band analyzed at the University of Washington and 68°C ± 4°C and 85°C ± 3°C for the bands analyzed at ATOMKI. These values exceed the temperature of the modern fluids (22°C, [11,125]) of the active thermal spring located in the Il Palagio area (Figure 2), confirming that the temperature of the fluid has changed through time.
We calculated the δ 18 O w values from the measured carbonate δ 18 O values, using the clumped isotope derived temperature for IA1 and the Kim et al. [126] and Chacko and Deines [127] equations for the temperature dependence of aragonite-water oxygen isotope fractionation, assuming equilibrium conditions during aragonite deposition. The results using the two equations are indistinguishable (2.6‰ V-SMOW). The δ 18 O w values are out of the range of present-day regional rainwater δ 18 O w values (δ 18 O w = −7 to −6‰) that feed the shallow hydrothermal system of southern Tuscany [11,124]. This result could suggest that (a) the fluids were possibly influenced by magmatic source water (in Larderello geothermal field δ 18 O of magmatic waters ranges from +3‰ to +8‰, in [128]), causing a heavier composition of the O isotopes, or that (b) fossil meteoric water extensively interacted with the Tuscan Nappe carbonate rocks that are characterized by a positive signature of δ 18 O (~+25‰, [129]).

The Tectonic Control on the Fluid Flow.
Mesostructural data indicate that the fault-slip surfaces and damage zones of the two main NE-and NW-trending fault systems are often characterized by Fe-hydroxides and banded Cacarbonate veins (Figure 3). This observation implies that both systems channeled geothermal fluids. Comparable evidence comes from the field mapping (Figure 2), which indicates that both the fossil and active travertine deposition is mainly located along the NW-striking faults, at the intersection with the NE-striking major faults. Thus, we conclude that in the study area, permeability and hydraulic conductivity of deep fluids are controlled by fractures.
In this view, it is noteworthy to underline that the main fissure ridge (from which the analyzed samples were collected) displays a NW trend, parallel to the main fault zones characterizing the whole study area. Furthermore, in the proximity of Iano village, travertine bodies are associated 19 Geofluids with the interplay between NE-and NW-trending faults. Similar trends, parallel to the main NW-trending faults, characterize the minor banded Ca-carbonate veins cutting both the travertine bodies and units forming the travertine substratum. The main banded Ca-carbonate vein developed in the central part of the NW-trending fissure ridge (Figure 6(a)) suggests that this travertine deposit developed along a permeable sector of a NW-striking fault. At least part of the fault activity (from 172 ± 3 to 21:2 ± 0:1 ka) is recorded by the banded Ca-carbonate veins development. Similarly, a small number of thin veins are NE-oriented, such as vein V2 from PAM1 and ET3 samples. This superposed veining (V1, V2, and V3) is indicative of repeated deformation episodes, followed by progressive fracture sealing by Cacarbonate deposition.
Kinematic indicators on the NE-trending faults indicate strike-slip movements. This is significant since, in agreement with Sibson [130], the intermediate axis attitude indicates the preferential direction of fluid flow. Thus, an oblique, near vertical, intermediate axis can explain the up flow migration of deep fluids and therefore the occurrence of CO 2 of mantle origin ( Figure 4). This suggests that the NE-striking faultzones play a role in channeling upward fluid migration, as already suggested for the Larderello area [68].
Interaction with NW-trending faults is then significant for two related reasons: (a) the intersection of NW-and with NE-trending faults increases fracture distribution, promoting hydraulic conductivity and (b) the intermediate axis of the NW-trending faults being close to horizontal enables, geothermal fluids to flow laterally, interacting with the host rocks (hence, enriching their content in 18 O, for instance), and arriving to the surface when the dilatation component of the normal fault is efficient.

Factors Influencing Fluid Property Variation.
Deformation of the crust in geothermal areas promotes the mobilization of large volumes of saline and CO 2 -rich waters [8,11,143]. These fluids are involved in fault zones during sudden strain that can promote CO 2 degassing and consequent carbonate precipitation, especially within fractures that act as conduits [7,13,17,34,144]. Carbonate precipitation, in terms of mineralogy and isotopic composition, is the result of combined effects such as fluid composition, degassing rate, pH variation and fluid temperature. As a consequence, each mineralogical and isotopic variation recorded by the banded Ca-carbonate veins can potentially reveal important tectonic episodes.
Our samples show repeated changes of the mineralogical composition and isotopic signatures, which we interpret in this context. Vein V1 of the PAM1 sample (vein crossing the flank of the ridge, Figures 7(b) and 8(a)) is characterized by the alter-nation of (Sr)-aragonite and calcite bands (see Figures 7(g) and 7(h)), implying that the depositional conditions (fluid composition, temperature and pCO 2 ) changed through time. On the other hand, vein V3 and PAM13 (vein filling the central part of the ridge, Figure 6(b)) are completely composed of bands formed by acicular aragonite crystals.
Several studies were aimed at defining the main factors that favor the aragonite formation instead of calcite in depositing carbonate. The most significant factors for aragonite formation are (i) the high temperature of water [145][146][147] and (ii) the high Mg content [145,148,149]; nevertheless, several studies underlined that other factors have to be taken into consideration. Malesani and Vannucci [150] proposed that high Sr levels also favor aragonite deposition. The high Sr concentration in the fluid plays an essential role in modifying the surface energy term, therefore favoring Sr in the aragonite lattice [151]. Lippman [147] suggests that aragonite precipitates preferentially when the rate of CO 2 degassing is the greatest, implying that the kinetic effect controls the mineralogy. Similar conclusions were reached by Duchi et al. [125] who investigated springs depositing travertine characterized by different temperature and geochemical features. They argued that in addition to the role of the temperature and Mg content, the partial pressure of CO 2 can contribute to the deposition of aragonite instead of calcite. This is due to the rapid CO 2 degassing inducing high level of supersaturation and pH increase, both triggering the rapid deposition of aragonite (Duchi et al. [125]). Therefore, aragonite deposition can be favored by (i) thermal springs characterized by lower temperatures (below 30°C, [145,147]) and pCO 2 significantly higher than the atmospheric pCO 2 (which is the case of the study area; Duchi et al. [125]; [11]) or (ii) thermal springs characterized by higher temperature (above 30°C; [5]).
The rate and amount of CO 2 discharge can be influenced by the tectonic activity [15,29,142,152] and/or climatic variations [8,32,153]. Climate can control the amount of CO 2 dissolved in the water (cf. [8]). However, considering the fact that the parent fluids we analyzed came from hypogean conduits (faults and fractures) and reached the surface (i.e., fluids with pCO 2 equilibrating with the pCO 2 of the atmosphere), other factors are likely needed to explain the (sudden) CO 2 degassing. In our opinion, pulses of the tectonic activity are reasonable factors. In particular, for samples PAM1 and ET3, it can be assumed that the high CO 2 degassing rate could have caused mineralogical differentiation between calcite and aragonite. This implies that the CO 2 partial pressure has changed through time. This is supported by the stable isotope data of our banded Ca-carbonate veins (Figure 11(a)) which display larger δ 13 C variation than δ 18 O. In particular, aragonite bands are characterized by the highest δ 13 C values, suggesting that fast CO 2 degassing has strongly influenced the deposition of these bands, causing disequilibrium due to the kinetic effect. In addition, δ 18 O and δ 13 C values correlate (R 2 = 0:85, Figure 11(b)), implying a deposition under disequilibrium conditions. In fact, during degassing, isotopically light carbon dioxide preferentially escapes and, as a consequence, the 13 C and 18 O contents in depositing CaCO 3 increase in the same direction [5,7,8,34,115,[154][155][156]. For this reason, a positive correlation is recorded. Furthermore, the texture of these crystals shows ending faces with pointed terminations (cfr. 4.3.3), implying that some crystals grew faster than others. Therefore, fast degassing could have increased the depositional rate of aragonite crystals.
In contrast, isotope values of calcite crystals of sample PAM1 do not show any correlation and δ 13 C values are all around 0‰, with some values slightly negative. This evidence implies that the isotopic composition of fluids remained constant during calcite crystals formation, and (possibly) crystals formed under isotopic equilibrium. The texture of calcite crystals validate this hypothesis because crystals enlarge in the growth direction and their terminations are characterized by pointed curved faces, suggesting that the crystals grew slowly and had the time to develop. Therefore, if CO 2 degassing rate was a factor, it would imply that the veins record CO 2 partial pressure changes through time.
In contrast, sample PAM13 is completely made up of aragonite (Figure 9(a)), and stable isotope values do not show any correlation (Figure 11(a)). Clumped isotopes highlight a depositional temperature between 67°and 85°C ( Table 5), suggesting that temperature played a more important role than pCO 2 in depositing aragonite, an interpretation that is also supported by the lack of correlation between stable isotope values.
Finally, we discuss the tree-shaped black structures described for the vein V1 (Figures 7(d) and 8(a)). These structures have never been described in banded Cacarbonate veins; they have only been documented for travertine deposits [103-105, 157, 158] and speleothems [159].
From a mineralogical point of view, the black tree-shaped structures contain manganese dioxides, which are based on [MnO 6 ] octahedra units linked in chains to form a framework structure containing monodimensional tunnels with specific sizes. Tunnels can accommodate different chemical species, such as Mg, Cu, Ni, Co, and Fe in their ionic state [105]. For this reason, in some cases, high percentages of Ni (5.01% on average) have been measured. The Raman results suggest that the black tree-shaped structures consist of todorokite, although the possible presence of cryptomelane or hollandite cannot be ruled out. Indeed, our data are in good agreement with Bernardini et al. [109] who report for todorokite a broad and strong band at 630 cm −1 with shoulders at 507 and 575 cm −1 . However, the possible occurrence of cryptomelame or hollandite cannot be ruled out. Although spectroscopic data from the literature are conflicting, some recent studies [109] report for cryptomelane a position of the strong band centered between 623 and 655 cm −1 and a shoulder at~580 cm −1 . Similar positions are reported for hollandite.
In speleothems and travertine deposits, tree-shaped structures are strictly associated with bacterial activity inducing the precipitation of Mn-oxides. Precipitation produce aggregates with high Mn concentration (Mn-oxides), even from water with low Mn 2+ concentration (parts per million range, [103]). This process was very likely promoted by specific environmental parameters: in surficial flowing waters, pH changes are favored by temperature and CO 2 degassing rate variations. In contrast, these factors cannot abruptly change in deep conduits. As a consequence, the occurrence of Mn-rich deposits is indicative of changes of the

Geofluids
hydrothermal fluid conditions. Changes in pH within the circulating waters suggest fluid compositional variation. Values of pH above 5 [103] and oxidizing environments promote the precipitation of Mn-oxides [160]. In the case of Iano, these tree-shaped black structures are characterized by our lowest measured δ 13 C values (-1.2‰ and -0.4‰ V-PDB), suggesting significant mixing between hydrothermal and meteoric waters. In these conditions, microorganisms could thrive, locally modifying the pH conditions and discharging metabolic products able to oxidize Mn 2+ [105,161]. As long as these conditions persist, Mn-oxides continue to deposit due to microorganism metabolic processes. As a consequence, tree-shaped black structures are indicative of a lowtemperature fluid depleted in CO 2 . These conditions can develop after rapid CO 2 degassing in waters that remain in surficial levels, favoring mixing with meteoric waters. These structures are absent in the banded Ca-carbonate veins filling the central fissure of the ridge, suggesting that the fluids in the central fissure maintained high temperature through time, favoring aragonite deposition and did not interact with surficial waters.
In this view, the tree-shaped black structures can suggest a phase of transition after a fast CO 2 degassing event, during which hydrothermal waters mix with more surficial ones, causing temperature decrease. Thus, tree-shaped black structures can be useful indicators for a stasis phase of the hydrothermal fluid circulation and a mixing of hydrothermal and meteoric waters.

Conclusions
Banded Ca-carbonate veins represent key tools for decoding variations in geothermal fluid properties though time. Veins forming in the Iano travertine depositional system preserve compositional fluctuations of the geothermal fluids (red arrows in Figure 12) that flowed from the deep reservoir up to the surface, along fault zones ( Figure 12). These fault zones are characterized by changes in their kinematics, which combined with their orthogonal arrays, and gave rise to localized permeable volumes able to channel deep fluids up to the surface, therefore mixing with shallow (i.e., meteoric) origin fluids (blue arrows in Figure 12). Fluids move following the intermediate axis along the transfer faults indicated by orange arrows in Figure 12. Our record started in the middle Pleistocene, where the fault permeability induced the geothermal fluids to flow at the surface. From 172 ± 3 ka until (at least) 27:1 ± 0:1 ka, the circulating fluids deposited banded Ca-carbonate veins following a multistep process. During its rise along faults and fractures, the fluid interacted with carbonate deposits, becoming enriched in δ 18 O ( Figure 12); as a consequence, the stable isotopic composition of the fluid changed. Banded Ca-carbonate veins at different depths developed within faults and fractures, probably due to pulses of tectonic activity. Starting by at least 27:1 ± 0:1 ka, the fluid preserved the initial stable isotope imprint, from depth to the surface, following a one-stage deposition process.
Travertine deposits mostly formed from thermal springs aligned along NW-trending faults and at their intersection with NE-trending faults. Most of the banded-Ca carbonate veins follow the NW direction, even if in some cases, some very thin veins cut the previously deposited veins in the NE direction.
Our observations from the Iano area represent the first documented tree-shaped black structures, made up by Mnoxides, in banded Ca-carbonate veins. The mineralogy is mostly likely todorokite and their deposition can be indicative of a stasis phase following a rapid CO 2 degassing that favors the mixing of deep origin hydrothermal fluids with shallow meteoric waters. As a consequence, fluid temperature and CO 2 content decrease, allowing the development of microbial communities in oxidizing and acid environment (pH > 5). Thus, Mn-rich deposits could have a key role in decoding periods of low fluid flux and mixing with more surficial waters.

Data Availability
The data support the results of this study and are reported in the Tables.

Conflicts of Interest
The authors declare that there is no conflict of interest regarding the publication of this paper.

Supplementary Materials
The carbonate clumped isotope values measured at ATOMKI (Institute for Nuclear Research, Debrecen, Hungary) were performed using Thermo Scientific Kiel IV Automatic Carbonate Device coupled to a Thermo Scientific MAT253 Plus isotope ratio mass spectrometer (as previously said in Section 3.2). The normalization standards used were ETH1, ETH2, and 22 Geofluids ETH3 coupled with IAEA-C2, used as quality control sample with the values reported in Table S1 [91]. Δ 47 is reported in the CDES (carbon-dioxide equivalent scale) [91], using an acid fractionation factor of +0.066‰ to set down data generated with 70°C acid digestion on the 25°C digestion scale. Easotope application [95] was used for data evaluation, using CO 2 clumped ETH PBL replicate analyses method and Brand parameters (Table S2). (Supplementary Materials)