Combined use of conventional and clumped carbonate stable isotopes to identify hydrothermal isotopic alteration in cave walls

Abstract

Alteration of conventional carbonate stable isotopes (δ¹⁸O, δ¹³C) in cave walls has been shown to be a useful tool to identify cave formation driven by deep-seated processes, i.e., hypogene karstification. If combined with a prior information on the paleowater stable isotope composition, further insights can be obtained on the temperature and the source of the paleowater. Clumped isotope composition (Δ₄₇) of carbonates is an independent measurement of temperature, and if combined with the conventional stable isotopes, can provide information on the paleowater stable isotope composition. On the example of Provalata Cave (N. Macedonia), we apply for the first time, both conventional and clumped stable isotope analysis, and identify two different isotope alteration trends, reflecting two distinct hydrothermal events: an older, hotter one, where isotope alteration was likely related to isotope diffusion, lowering the δ¹⁸O values of the carbonate; and a younger one, related to the cave formation by low-temperature CO₂-rich thermal waters, with dissolution-reprecipitation as the alteration mechanism, causing decrease in δ¹⁸O values, and unexpected increase in δ¹³C values. The findings are further corroborated by additional insight from optical petrography and cathodoluminescence microscopy, as well as fluid inclusion analysis of secondary calcite crystals related to the cave forming phase.

Introduction

Hydrothermal caves are one of the main genetic type of caves, a sub-type of hypogene caves, whose origin is related to deep-seated processes (source of acidity and recharge to the soluble rock formation is from depth), contrary to the most commonly found caves whose origin is connected to surface-related processes (i.e., epigene caves; source of acidity and recharge is from the surface)^1,2,3. Hydrothermal caves are formed by thermal waters that have generally higher CO₂ concentrations, and dissolve carbonate bedrock as they cool down along their rising flow path due to inverse relationship of calcite solubility and temperature⁴. At shallower levels of the aquifer, decrease in pressure facilitates CO₂ degassing, leading to supersaturation and precipitation of calcite⁵. They are generally strongly structure controlled and develop a characteristic suite of cave morphologies and deposits^2,6,7.

While some active thermal caves are accessible and offer possibility to study some of the related processes occurring in shallower conditions^8,9, most of the known hydrothermal caves are fossil, and became accessible only after being intercepted by surface erosion¹⁰. Identification of characteristic morphologies has been the most common approach to identify the hydrothermal origin of caves¹¹, however equifinality of some morphological features (e.g., convectional forms such as cupolas, pockets etc.) makes this approach not resolutive. Study of secondary minerals has been proven to be a more reliable approach, with microthermometry and stable isotope composition used to confirm hydrothermal origin and identify the sources of fluids as well as their formation temperature^{12,13,14,15,16}.

Although secondary minerals might have been deposited from thermal waters, their hydrothermal origin is not a proof of the primary hydrothermal origin of the cave¹⁴, unless their relationship to the cave origin can be unambiguously demonstrated⁴. Likewise, the absence of hydrothermal minerals does not exclude a hydrothermal origin of a cave, as they might not precipitate at all, and later modification in shallower settings can modify and overprint the primary hydrothermal features. Isotope alteration of bedrock, an approach developed for and most commonly used in hydrothermal ore-related studies^17,18,19, has also been successfully applied to cave wall bedrock. Recent studies, based on alteration of conventional carbonate stable isotope compositions (expressed as δ¹⁸O and δ¹³C values) in cave walls, have demonstrated a hypogene origin in the presence or absence of related cave minerals^15,20,21. Furthermore, it has been found that alteration halos do not develop in epigene caves²² and this approach can be used to confirm or dismiss hypogene-related cave origin²³. The most commonly identified alteration in cave walls is lowering of the bedrock δ¹⁸O values, as a result of the interaction with low-δ¹⁸O high-temperature meteoric waters, as well as lowering of bedrock δ¹³C values due to interaction with dissolved inorganic carbon (DIC) with lower δ¹³C values^12,14,15,22.

Recent developments in carbonate clumped isotope thermometry²⁴ allow an independent estimate of the mineral formation temperature, that, if combined with the conventional oxygen isotope composition, can be used to reconstruct the paleofluid oxygen composition²⁵. Carbonate clumped isotope thermometry has found wide applications, such as in reconstructing metamorphic and exhumation histories of marble formations^26,27, sediment diagenesis^28,29, as well as speleothem studies^30,31,32. The application of clumped isotope thermometry to subaerially formed speleothems has proven challenging, due to disequilibrium effects related to various controls (e.g., rapid CO₂ degassing), leading to overestimation of formation temperatures^33,34. However, subaqueous speleothems, especially slowly precipitating ones, were found most likely to have precipitated in equilibrium, thus providing more reliable results^{31,32,35,36,37}.

Here we expand on the cave wall isotope alteration approach²² by applying both clumped and conventional stable isotope analyses on carbonates. On the example of Provalata Cave (N. Macedonia), we show for the first time, that the characteristics of two distinct hydrothermal events causing different cave wall isotope alterations can be identified, the second one responsible for cave formation. We support our findings by additional insight obtained by optical petrography and cathodoluminescence microscopy, as well as fluid inclusion analysis of secondary calcite minerals, and reconstruct the subsequent evolution of the cave.

Study area

Hypogene karst in Mariovo (N. Macedonia) is found at several localities, at the intersection of low topography and major faults, and is developed mainly in calcite and dolomite marble, that are part of NNW-SSE oriented nappes along the eastern part the Pelagonian massif³⁸ (Supplementary Fig. S1). Melnica locality in Mariovo is an output zone of a hypogene karst system, where caves and cave remnants developed in calcite marble, dolomite marble and carbonate breccia, and a small lukewarm Melnica Spring, are found (Supplementary Fig. S2). Provalata Cave is a small fossil cave and the most remarkable example of this system. It was formed in calcite marble in two successive phases, first by CO₂-rich thermal waters in a phreatic setting, and then by sulfuric acid in a dominantly vadose setting, both leaving characteristic suites of cave morphologies and associated minerals³⁹. Sulfuric acid speleogenesis (SAS), that was active at ca. 1.6 Ma, left abundant gypsum deposits formed as a replacement of carbonate bedrock and calcite spar^39,40. The calcite spar is found as coatings, up to 0.5 m thick, that cap the hydrothermal carbonic phase.

From Provalata Cave we collected two wall-drilled cores⁴¹ (C2 and C3) that cover cave wall bedrock and calcite coatings, and two additional samples cut from calcite coatings covering the cave wall (PROV03 and PR20). Furthermore, we collected samples along the NNW-SSE nappes to better characterize the primary isotopic composition of the calcite marble. Details of the sampled material are given in the Supplementary Information.

Results and discussion

Macroscopic observations

The two cores, C2 and C3, have length of ~ 30 and ~ 6 cm, with calcite marble representing ~ 15 and ~ 4 cm, respectively. In addition to primary grey-colored marble, two distinct types of color alteration are visible: a narrow rim (< 2 cm) with intense white discoloration at the contact of the cave wall and the overlying calcite coating, and a dispersed pale grey discoloration in the inner part (Fig. 1, Supplementary Fig. S5). The thickness of the calcite coatings is significantly different (~ 15 cm and ~ 2 cm), with the one at C3 likely largely dissolved in the subsequent SAS phase.

Figure 1

Conventional and clumped stable isotope composition of cores C2 (a) and C3 (b) from Provalata Cave. Average values and ranges (± 1SD) for calcite marble in the wider area are given by brown lines and brown bands, respectively.

Conventional stable isotope composition

The grey calcite marble sections in both cores have similar conventional stable isotope composition, with calcite oxygen (δ¹⁸O_cc) and carbon (δ¹³C_cc) isotope values that are within the range of values obtained for samples along the NNW-SSE nappes (Figs. 1, 2). This range of values is essentially the same as what was found from a quarry ~ 30 km to the north⁴², indicating that the grey sections in the cores are isotopically unaltered.

Figure 2

Conventional and clumped stable isotope composition of the studied carbonate samples. (a) Conventional stable isotope δ¹⁸O_cc − δ¹³C_cc plot. Black lines indicate the two trends of covariation of δ¹⁸O_cc and δ¹³C_cc values from Provalata Cave calcite coatings. (b) Clumped isotope composition (Δ₄₇) vs δ¹⁸O_cc. Equilibrium curves based on the equation of Daëron et al.³¹ for a given water stable isotope composition (δ¹⁸O_w) are shown in purple. (c) Clumped isotope composition vs δ¹³C_cc. Apparent clumped isotope-based temperatures are also indicated in (b) and (c). Shaded brown and grey areas show range of values for the unaltered calcite marble and the narrow white rim, respectively.

The two discolored sections of the rock wall have altered δ¹⁸O_cc and δ¹³C_cc values that show two distinctive trends. The pale grey parts show no change in δ¹³C_cc and decrease in δ¹⁸O_cc, with maximum shift of up to − 5.4‰ Δδ¹⁸O and only + 0.1‰ Δδ¹³C (where ΔX = X_alteredMAX − X_{unalteredMEAN}, and X is δ¹⁸O_cc or δ¹³C_cc). The white narrow rim shows change in both δ¹³C_cc and δ¹⁸O_cc, with Δδ¹⁸O of − 6.7‰ to − 8.3‰ and Δδ¹³C of + 3.3‰ to + 3.2‰, for C2 and C3, respectively, with values trending towards the isotopic composition of the overlying calcite coatings (Figs. 1, 2). The observed shift to lower δ¹⁸O_cc is commonly found in hypogene caves with reported Δδ¹⁸O ranging from − 15‰ to − 2‰, reflecting interaction with thermal water of predominantly meteoric origin^{14,20,22,23,43,44,45} (Fig. 3). However, the δ¹³C_cc values are commonly reported to be either unchanged or shifted to lower values, with Δδ¹³C from − 12 to + 1‰, due to interaction with fluid with low δ¹³C DIC values (δ¹³C_DIC), reflecting predominantly organic-derived carbon^{14,20,22,23,43,44,45} (Fig. 3). The shift to higher δ¹³C_cc values observed in the narrow rims in cores C2 and C3, suggests that the paleowater at Provalata Cave had high δ¹³C_DIC. This is in agreement with recent geochemical data from nearby lukewarm Melnica Spring, where it was found that up to 54% of the carbon in the DIC is from metamorphic CO₂ with δ¹³C of + 4.5‰⁴⁶.

Figure 3

Extent of observed stable isotope alteration in carbonate bedrock walls of hypogene caves. (a) Alteration vectors in δ¹⁸O_cc − δ¹³C_cc space. (b) Magnitude of alteration (Δδ¹⁸O_cc − Δδ¹³C_cc). 1 and 2—Provalata Cave (this study); 3 to 8—Hypogene caves in Austria²² (3—Mitterschneidkar-Durchgangshohle, 4—Entrische Kirche, 5—Stegbachgraben, 6—Obir Caves, 7—Güntherhohle, 8—Marchenhohle); 9—F10 Cave, Dolomites, Italy²³; 10—Cupp-Coutunn/Promeszutochnaya, Turkmenistan¹⁴; 11 and 12—Hypogene caves in Crimea, Ukraine²⁰; 13 and 14—Movile Cave, Romania⁴⁴, 15—Plavecká jaskyňa, Slovakia⁴⁵; 16 and 17—Carlsbad Cavern, USA⁴³.

The calcite coatings have low δ¹⁸O_cc (mean − 11.8‰) and very high δ¹³C_cc (mean + 7.3‰), with a relatively small range of values (Table 1). They show covariation in δ¹⁸O_cc–δ¹³C_cc space along two distinct trends, one with positive and one with negative slope (Fig. 2).

Table 1 Summary statistics of the stable isotope composition of wall bedrock and calcite coatings from Provalata Cave.

Clumped isotope composition

The clumped isotope compositions follow a similar pattern. The grey calcite marble has the lowest Δ₄₇ values (mean 0.323 ± 0.016‰), reflecting highest apparent temperatures (mean 206 ± 22 °C) (Figs. 1, 2, Table 1). Similar values were also found for the calcite marble along the NNW-SSE nappes (Figs. 1, 2). They are in the range of equilibrium blocking temperatures for the carbonate clumped thermometer^24,27,47, and considering their unaltered δ¹⁸O_cc and δ¹³C_cc, likely reflect blocking temperatures from the last metamorphic event. Such temperatures are in the range of zircon fission track (FT) closure temperatures, and FT ages from the eastern part of the Pelagonian massif relate them to Late Cretaceous-Paleocene thrusting^48,49. The altered sections have increased Δ₄₇ values showing two distinct clumped isotope compositions, that reflect mean apparent temperatures of 147 ± 23 °C in the pale grey section and 54 ± 13 °C in the white rim (Table 1). The apparent temperatures of the two altered sections, combined with their distinct conventional stable isotope composition (e.g., δ¹³C_cc), imply that the two alterations were related to two distinct hydrothermal events.

The calcite coatings have the highest Δ₄₇ values, reflecting apparent temperatures of 11–34 °C (mean 22 ± 6 °C; Table 1). The onset of calcite deposition in C2 is at 12 °C and in C3 at 33 °C (Table S5), and all calcite coating samples show variation in temperature along their growth (Fig. 1). This variation shows that the evolution of the hydrothermal cave after the onset of calcite deposition was not following a simple cooling and degassing due to shift to shallower conditions, as generally suggested for such systems⁴, but experienced pulsations in temperature, with at least one subsequent heating event of up to ~ 34 °C.

We calculated the water oxygen isotope composition (δ¹⁸O_w) of the fluid in equilibrium with calcite for all of the identified sections, using the calcite-water oxygen equilibrium equation³¹, the measured δ¹⁸O_cc and the clumped isotope-based apparent temperature (see also fluid inclusion section). The unaltered calcite marble has δ¹⁸O_w values that are within the range for metamorphic fluids⁵⁰. However, as the apparent temperatures of the unaltered marble represent equilibrium blocking temperatures, their calculated δ¹⁸O_w values are not considered as representative. The pale grey section has lower δ¹⁸O_w values, which are within the range for magmatic fluids⁵⁰. The secondary calcite minerals have the lowest values, reflecting largely meteoric water composition, while the narrow rim has intermediate δ¹⁸O_w values. The apparent temperatures for the pale grey section are within the range of temperatures (100 to 200 °C) of the ore mineralization at the nearby Carlin-type-like Allchar deposit⁵¹ (Supplementary Fig. S1), that is related to circulation of magmatic fluids with variable contribution of meteoric water, and associated with a ca. 5 Ma volcano-plutonic center of the Kožuf-Voras volcanic system^52,53,54.

Microscopic petrography and alteration mechanisms

The macroscopic and stable isotope observations of the alteration are also supported by the microscopic petrography in both plane-polarized transmitted and cathodoluminescent light (Fig. 4). The grey marble section contains large (up to few mm) calcite crystals, with dark blue luminescence. The pale grey section shows similar size calcite crystals having violet luminescence, found in different positions: throughout the crystal; diminishing inwards from crystal boundaries; or along crystal defects. The narrow white rim section shows remnants of large crystals with dark blue or violet luminescence that are surrounded by a large number of smaller size (up to 50 μm) crystals with orange luminescence, similar to the overlying calcite coating (Fig. 4, Supplementary Fig. S6).

Figure 4

Photomicrographs of C2 core showing different sections of the calcite marble cave wall in plane-polarized transmitted and cathodoluminescent light. (a) Overview of the thin section from the outer part of the core with location of the closer examined sections. (b) View of the unaltered section of the marble having large (mm size) crystals with dark blue luminescence. (c) A cross-section from the unaltered (left) through pale grey (middle) to white rim section, showing dominantly dark blue, violet and orange luminescence, respectively. (d) Part of the white rim, with remnants of a large calcite crystal with dark blue luminescence that changes outwards to violet, surrounded by smaller crystals with orange luminescence. (e, f) Close-up view of the white altered rim section, showing very small calcite crystals filling up pore spaces. An extended version of this figure is found in Supplementary Fig. S6.

The blue luminescence in the grey marble reflects intrinsic luminescence^55,56, and supports the primary (unaltered) character of this section as also indicated by the isotopic composition. The similar luminescence of the small calcite crystals in the narrow rim to the one of the overlying calcite coatings suggests deposition from the same (or similar) fluid. This is also supported by the white rim δ¹³C_cc and δ¹⁸O_cc values, as they are shifted towards the composition of the overlying calcite coating. A continuous transition between blue, violet and orange cathodoluminescence colors can occur due to the relative and absolute intensity of the intrinsic and orange luminescence⁵⁵. However, in the absence of additional elemental analyses, it is not possible to gain more insight from the cathodoluminescence on the conditions under which calcite precipitation or alteration occurred.

Carbon and oxygen isotopes in calcite can exchange with fluids by either dissolution-reprecipitation, when mineral and fluid are far from chemical equilibrium, or by diffusion, when mineral and fluid are at or close to chemical equilibrium⁵⁷. The exchange might also first proceed by dissolution-reprecipitation until chemical equilibrium is achieved and then by diffusion⁵⁸. Rate of diffusion of carbon and oxygen in minerals is strongly temperature-controlled⁵⁷, but in the presence of water, oxygen diffusion is greatly increased, while carbon diffusion is largely unaffected^59,60,61. The alteration of oxygen isotopes with no change in carbon isotopes in the pale grey section, as well as the variation in cathodoluminescence towards the crystal surface, suggest hydrothermal alteration by diffusion. Experimental studies on oxygen isotope diffusion in calcite^61,62 suggest that the water effect to oxygen diffusion in calcite is primarily a surface process, facilitating oxygen exchange between calcite surface and fluids. However, while water enhances oxygen isotope diffusion at the calcite crystal surface, to alter the clumped isotope composition, reordering of C-O bonds requires bulk mineral increase in oxygen diffusivity^62,63, which is achieved only at higher temperatures and longer times^25,47. This implies that dissolution-reprecipitation is the more likely process to explain the alteration of the clumped isotope composition at lower temperatures. Additionally, a combination of increased diffusivity, dissolution-reprecipitation, as well as pressure-induced structural changes can increase the reordering rates⁶². Although our stable isotope data is not detailed enough to investigate variation along a single crystal, the luminescence halo seen along boundaries of some crystal grains, and no apparent sign of dissolution-reprecipitation along the crystals, support the oxygen diffusion interpretation. Thus, the variation of δ¹⁸O_cc and Δ₄₇ in the pale grey section most likely reflects the degree of isotopic exchange that occurred between the hydrothermal fluid and the calcite crystals, with alteration progressing further in smaller size crystals and crystals with defects, while creating a halo in larger crystals. The lowest apparent temperature in the pale grey section (~ 110 °C) is found in the sample with the lowest δ¹⁸O_cc values, that likely reflects the temperature of the hydrothermal fluid.

For the narrow rim, the microscopic observations (both under transmitted and cathodoluminescent light) clearly point to dissolution-reprecipitation as the alteration mechanism, where previously formed porosity is filled with secondary calcite that continues further from the cave wall as calcite coating. Thus, the stable isotope composition along the white rim reflects the frequency of secondary calcite minerals, i.e., the linear trends seen in δ¹⁸O_cc − δ¹³C_cc, Δ₄₇ − δ¹³C_cc, and Δ₄₇ − δ¹⁸O_cc space (Fig. 2) reflect the mixing of the isotopic compositions of the secondary calcite and the calcite marble, the latter varying between the compositions of the grey and pale grey sections. Based on the lowest apparent temperature in the white rim, this alteration was caused by a low-temperature (< 35 °C) hydrothermal fluid.

Fluid inclusions in secondary calcite minerals

Provalata calcite samples are rich in fluid inclusions that are dominated by primary (intra-crystalline) inclusions, mostly as large size inclusions in palisade crystals, but also as clouds of small ‘thorn’ shaped inclusions (Supplementary Fig. S7). Their water stable isotope composition indicates meteoric origin, with δ¹⁸O_w values ranging between − 13.9‰ and − 11.6‰ and δ²H_w values between − 87.0‰ and − 76.0‰ (Supplementary Table S1). The data fall between the local and Mediterranean meteoric water lines, with lower values than what was found for the cold and lukewarm springs in the area⁴⁶ (Fig. 5).

Figure 5

Isotope composition of carbonate and fluid inclusions from Provalata Cave calcite. (a) Water stable isotope composition (δ¹⁸O_w, δ²H_w). Global (GMWL), Mediterranean (MMWL) and local (LMWL⁴⁶) meteoric water lines are included. Samples for which both carbonate and water stable isotope data are available have black outlines. (b) Carbonate stable isotope composition (δ¹⁸O_cc, Δ₄₇) and clumped based apparent temperatures. (c) Calcite-water oxygen fractionation factors for Provalata Cave calcite compared to equilibrium curves of Kim and O’Neil⁶⁷ and Daëron et al.³¹. (d) Helium isotope composition (R/R_A, where R and R_A are the ³He/⁴He ratio of the sample and air respectively) in a three-component helium mixing plot (atmospheric He: R/R_A = 1, ⁴He/²⁰Ne = 0.318, crustal He: R/R_A = 0.02, ⁴He/²⁰Ne = 1000, mantle He: R/R_A = 8, ⁴He/²⁰Ne = 1000^64,65,66). Numbers indicate stratigraphic position of the subsample within the calcite coating sample.

Noble gas data show predominance of crustal He in the fluid inclusions, with R/R_A values (where R and R_A are the ³He/⁴He ratio of the sample and air, respectively) of 0.37 to 0.50 and ⁴He/²⁰Ne values of 1.5 to 2.4. The three-component (air, crust, mantle^64,65,66) helium mixing model (Fig. 5, Supplementary Table S2) shows that the mantle component is low (2–4%) and the atmospheric component ranges from 13 to 22%.

For the samples with combined conventional and clumped carbonate stable isotope data, and fluid inclusion water stable isotope data (i.e., formation T, δ¹⁸O_w and δ¹⁸O_cc; n = 4), the calcite-water oxygen isotope fractionation was calculated (Fig. 5). The data fall above the experimental equilibrium line of Kim and O’Neil⁶⁷, and exactly on the empirical equilibrium curve based on slow growing mammillary calcites from Devils Hole and Corchia Cave³¹. The latter is almost identical to the one proposed by Coplen⁶⁸ and to the theoretical prediction by Watkins et al.⁶⁹, indicating that Provalata calcite formed under conditions at or close to thermodynamic equilibrium. Considering this, the paleowater δ¹⁸O_w was calculated for the samples for which no fluid inclusion data were available, using δ¹⁸O_cc, Δ₄₇-based apparent temperature and the calcite-water equilibrium equation³¹. The calculated δ¹⁸O_w values range from − 13.4‰ to − 9.6‰, with uncertainties of 0.5–1.5‰, due to propagation of larger uncertainties in apparent temperatures.

Timing of the calcite coating deposition

The calcite coating has been estimated to form during the end of Pliocene or Early Pleistocene³⁹. The low U concentrations in C2 samples suggested that the samples are not readily suitable for U–Pb chronology. Three uranium-series analyses were used to suggest that the coating formed during the Early to Middle Pleistocene. If Pliocene in age, both the ²³⁰Th/²³⁸U activity and δ²³⁴U value would measure 1 and 0‰, respectively, which is secular equilibrium in well preserved, unaltered calcite. C2 core exhibits densely crystalline calcite that yielded ²³⁰Th/²³⁸U activities of 1.012, 1.028, and 1.027, and δ²³⁴U values of 10, 22, and 8‰, respectively for samples C2-1, C2-2a, and C2-2b (Supplementary Table S3). The U-series results indicate, given no alteration, that the calcite coating formed after 2.5 Ma. During the subsequent SAS phase, when the cave reached the water table, in addition to gypsum replacing calcite minerals, potassium sulfate minerals (e.g., alunite, jarosite) formed at the contact of sulfuric acid and clay deposits³⁹. The 1.6 Ma Ar–Ar age of alunite formed during this phase³⁹ thus provides the minimum age limit for the deposition of the calcite coating.

The evolution of the secondary calcite minerals

The calcite coatings in Provalata Cave show small variation in their δ¹⁸O_cc and δ¹³C_cc values and a strong covariation that follows two lines: one with positive slope and another with negative slope (Fig. 6). For the samples with positive δ¹⁸O_cc − δ¹³C_cc slope, the T-δ¹⁸O_w relationship also follows a positive slope (~ 0.2‰/°C), indicating that the δ¹⁸O_w values are a result of mixing of two fluids: one cold (< 10 °C) with low δ¹⁸O_w (< − 13‰) and another hot (> 35 °C) with higher δ¹⁸O_w (> − 9.5‰). The cold one clearly represents a relatively shallow, meteoric water, while the hot one indicates a deep-seated origin. Magmatic waters have δ¹⁸O_w in the range of + 6‰ to + 10‰ and metamorphic waters in the range of + 5‰ to + 25‰⁵⁰, thus based on the δ¹⁸O_w/T slope seen in the samples, and considering magmatic or metamorphic origin, the estimated temperatures of the hot component are between 100 and 200 °C, respectively.

Figure 6

Stable isotope composition of calcite coatings from Provalata Cave. (a) Covariation of δ¹⁸O_cc and δ¹³C_cc values following generally two lines, one with positive, and one with negative slope. (b) Variation of δ¹⁸O_w (from fluid inclusions or calculated) with temperature. Dashed purple lines show equilibrium lines for a given δ¹⁸O_cc. (c) Variation of δ¹³C_cc with temperature. Green and purple full lines show curves of modeled variation of δ¹³C_cc with temperature change, with δ¹³C_DIC at + 4‰ and + 1‰, respectively, for a selected fH₂CO₃ (indicated for each curve) and 0.1 mol fraction of carbon in the degassed CO₂ (X_c). Closed system calcite precipitation (no degassing) is shown with dashed lines. See Supplementary Information for explanation of calculation and additional modeled curves. (d) Variation of δ¹⁸O_w with δ¹³C_cc. For each sample the stratigraphic sequence of stable isotope data is indicated to observe the trends in stable isotope change. Encircled symbols show noble gas fluid inclusion samples and associated values indicate R/R_A values. See text for details.

While the water composition is clearly a mixture of two components, the positive correlation in δ¹⁸O_cc − δ¹³C_cc space cannot be explained by calcite precipitation due to mixing of two fluids. First, the two fluids will need to be highly saturated in calcite, as mixing of two fluids saturated in calcite at different equilibrium concentrations of Ca²⁺ produces a mixture with lower calcite saturation due to the curvature of the calcite equilibrium line, that will lead to calcite dissolution (i.e., mixing corrosion)^70,71,72. Second, the very small change observed in the δ¹³C_cc values would require two fluids of such contrasting origin, such as meteoric and deep-seated, to have similar carbon isotopic composition, which is highly unlikely. Furthermore, while mixing of two fluids with similar isotopic composition and contrasting temperatures can produce correlation in δ¹⁸O_cc − δ¹³C_cc space, it is unlikely that two fluids will have similar isotopic composition, indicating similar origin, but have contrasting temperature⁷³. Thus, it is more reasonable to assume calcite precipitation from one fluid under a temperature effect, then due to mixing of two fluids⁷³. This implies that the water mixing occurred prior to calcite precipitation, and likely contributed to the cave passage formation by mixing corrosion.

The calcite that marks the onset of deposition at C2 core has very high δ¹³C_cc for a DIC dominated by a meteoric (soil) sourced carbon. The δ¹³C_cc depends on the δ¹³C_DIC, T and pH (distribution of carbon species, mainly H₂CO₃ and HCO₃, having significantly different fractionation with calcite⁷⁴). For such high δ¹³C_cc value (+ 7.9‰) at low temperature (12 °C), if the calcite precipitates in isotopic equilibrium from fluid with pH > 6, both δ¹³C_DIC and δ¹³C of the CO₂ (δ¹³C_ext) should be higher than 0‰, suggesting a deep-seated origin of the CO₂. At pH above 7 the δ¹³C_DIC should be >  + 5.4‰, and δ¹³C_ext >  + 7.5‰, with more than 41% of the carbon in DIC coming from carbonate dissolution (see Supplementary Information). However, in such case the alkalinity will be low (Supplementary Fig. S10), and if some soil CO₂ is present (as expected in shallow meteoric waters), the deep-seated CO₂ must have even higher δ¹³C, suggesting that the combination of higher pH with such high δ¹³C_DIC values is not likely. Thus, it is most likely that the onset of calcite deposition at C2 was from low temperature, slightly acidic fluid, where the water was of meteoric origin, but the CO₂ was mostly of deep-seated origin. As the subsequent mixing with deep-seated hotter fluid occurred prior to calcite precipitation, this supports the interpretation⁴⁶ that meteoric and deep-seated fluids converged at depth along a fault structure with dominantly metamorphic CO₂ flux.

Since calcite precipitation is accompanied with temperature change in our samples (Figs. 1, 2, 6), the δ¹³C_cc values can be modeled in terms of change in temperature with or without CO₂ degassing⁷⁵. The modeled curves show characteristic slopes of T-δ¹³C_cc for a range of H₂CO₃ fractions (fH₂CO₃) in DIC (Fig. 6; Supplementary Information). At C2, the calcite precipitation follows two lines in T-δ¹³C_cc space with similar slightly negative slope, that based on our modeling (Fig. 6), indicates low degassing and lower pH of the fluid. The shift to lower intercept with same slope suggests either lowering of δ¹³C_DIC or some variation in the degassing. Similar pattern is also seen in the first part of C3 core. The larger vertical shift in the rest of the samples can be due to δ¹³C_DIC change, higher degassing, general shift to higher pH (e.g., lowering of pCO₂ in the fluid), or a combination. These samples show also larger shift in δ¹⁸O_w (< 2‰). While increase in pH has no direct effect on oxygen isotopes, concurrent increase in pH and δ¹⁸O_w can occur by increased water–rock interaction. However, this is unlikely as the low temperature setting does not favor such large water–rock oxygen isotope exchange.

Another possibility is a change in the primary isotopic composition of both DIC and water. Lowering of δ¹³C_DIC of the deep component is a possible explanation (e.g., increase in mantle-derived CO₂), but increase in mantle-derived fluids cannot explain the observed increase in δ¹⁸O_w, and noble gas fluid inclusion data show minor and relatively constant mantle He component. Variation in the δ¹⁸O_w of the shallow component is very likely, and such increase in values can be due to climate variations (e.g., higher values in warmer and/or drier conditions and vice-versa⁷⁶). Although the effect of δ¹⁸O_w change in the shallow component can be seen in the mixture δ¹⁸O_w, the δ¹³C_DIC change in the shallow component (carrying a soil δ¹³C signal, e.g., − 25 to − 10‰⁷⁶) will be overprinted by the large fraction of deep-derived CO₂ (with endogenic δ¹³C signal, e.g., − 7‰ to − 4‰ for mantle, or > 0‰ for metamorphic CO₂^50,77. If the deep-sourced CO₂ flux decreases, pCO₂ will decrease, leading to an increase in pH and lowering of the δ¹³C_cc, but also relative increase in the shallow component carbon fraction, and further lowering of δ¹³C_DIC and δ¹³C_cc. This is supported by the noble gas fluid inclusion data, where δ¹³C_cc decrease is accompanied with R/R_a increase and ⁴He/²⁰Ne decrease (Fig. 6), that reflects concurrent increase of the atmospheric He component (from 13 to 22%) and decrease of the crustal He component (from 84 to 76%), with constant minor mantle He component (2–4%).

In other words, climatically controlled variation in the isotopic composition of the cold end-member (δ¹⁸O_w, δ¹³C_DIC) is seen in the mixture δ¹⁸O_w but generally masked in the mixture δ¹³C_DIC by the high amount of deep-sourced CO₂. Thus, the offset from the positive δ¹⁸O_cc - δ¹³C_cc covariation in Provalata calcite reflects a phase of lowering of deep CO₂ contribution.

The evolution of the hydrothermal cave system: a brief summary

The grey calcite marble of the cave walls reflects primary isotopic composition, that was first altered in color and isotopic composition by hydrothermal water (~ 110 °C) of magmatic or mixed magmatic-meteoric origin (Fig. 7). It was likely close to chemical equilibrium with the rock, and caused only lowering of δ¹⁸O_cc, probably by oxygen diffusion between water and rock. The alteration was pervasive, and appearing not bound to a major fissure, although the hydrogeological setting under which it operated is unclear. This event probably corresponds to an earlier volcanic activity of the Kožuf-Voras system (6.5–1.8 Ma^52,53,54). The second alteration (narrow rim along cave walls), is related to the cave-forming phase, when convectional features developed along high-angle fractures of a NW–SE oriented fault zone³⁹, related to Late Miocene-Quaternary extension^48,49. Cave passages formed by dissolution and enlargement of fractures during cooling of rising low-temperature (< 35 °C) thermal waters (of mixed meteoric and deep-seated origin), rich in metamorphic CO₂, likely enhanced by mixing corrosion. As dissolution progressed perpendicular to cave walls, porosity was increased in the first few centimeters. Cooling continued, but the system shifted to shallower settings, forcing CO₂ degassing and calcite precipitation, first filling up pore spaces, and progressing into thick calcite coatings. The thermal evolution and δ¹³C_cc values of the coatings suggest that after the initial cooling, another heat pulse followed, and variation in pCO₂ controlled the variation in δ¹³C_cc. The calcite deposition took place between 2.5 Ma and 1.6 Ma and is likely connected to the youngest volcanic activity of the Kožuf-Voras system (3.0–1.8 Ma), focused in its southwestern part^52,53,54, close to the study site (Supplementary Fig. S1). It is possible that both alteration phases are part of the long-term evolution of the same hydrothermal system, that first commenced at deeper settings with pervasive alteration, and then evolved along newly formed extension-related high-angle faults with high CO₂ flux, creating cave passages.

Figure 7

Conceptual model of the evolution of the studied hydrothermal phases in Provalata Cave with associated alteration mechanisms and calcite deposits presented schematically at three scales: hydrogeological setting, cave and cave wall. The thickness of the blue and red groundwater flowlines in each phase indicates the relative influence of the shallow epigene and deep hypogene component, respectively. (a) Pervasive alteration of color and δ¹⁸O_cc in calcite marble by hydrothermal water (~ 110 °C) with oxygen diffusion along crystal boundaries as the most likely alteration mechanism. The hydrogeological setting is unclear. (b) Development of cave passages along high-angle fractures due to cooling of low-temperature (< 35 °C) thermal waters rich in metamorphic CO₂, likely enhanced by mixing corrosion. Porosity increases perpendicularly to the cave wall. (c) Shift to shallower depth leads to CO₂ degassing and calcite precipitation, filling up pore spaces. (d) Large decrease in temperature and δ¹⁸O_w, indicating increased influence of the shallow groundwater component. Calcite with high δ¹³C_cc values precipitates outward from the cave wall. (e) Increase in temperature and δ¹⁸O_w, likely related to nearby volcanic activity, leads to precipitation of calcite with lower δ¹³C_cc. (f) Temperature and δ¹⁸O_w decrease. Lowering of pCO₂ leads to decrease in calcite δ¹³C_cc.

Conclusions

We show that by combined use of conventional and clumped stable isotope analysis of carbonates in the study of cave wall alteration by thermal waters we can better constrain the character of the isotopic alteration, as well as unambiguously identify different alteration events.

On the example of Provalata Cave we demonstrate the primary isotopic character of the unaltered calcite marble, and identify two distinct hydrothermal phases causing different alterations. The older one was pervasive with flow along smaller structures and crystal boundaries, causing lowering of δ¹⁸O_cc values likely due to diffusion-controlled isotope exchange with fluid at or close to chemical equilibrium with the rock, at higher temperature (~ 110 °C). The second one was related to cave formation by fluid flow along a fracture, i.e., progressed perpendicularly inward from the fracture (cave wall), and proceeded with dissolution-reprecipitation-controlled alteration by lower-temperature fluids (< 35 °C) that transitioned from being calcite-undersaturated (dissolution) to calcite-saturated (precipitation).

While most of the examples of cave wall isotope alteration reported in the literature show lowering of δ¹⁸O_cc and either unchanged or lowered δ¹³C_cc, our findings show a rare example of isotope alteration with lowered δ¹⁸O_cc and significantly increased δ¹³C_cc. We expect that such case is not a unique one and can be expected at sites with high metamorphic CO₂ flux. We also show a non-linear thermal evolution of the hydrothermal cave system, suggesting that simple cooling, as usually assumed for such systems, is likely simplifying the evolution, especially for karst systems related to volcanic areas.

Methods

Stable isotope analysis of carbonates

Conventional and clumped stable isotope analyses of carbonates were carried out at the Isotope Climatology and Environmental Research Centre (ICER), Institute for Nuclear Research (ATOMKI), Debrecen.

Conventional carbon and oxygen stable isotope analyses were carried out on hand-drilled carbonate powders with an automated GASBENCH II sample preparation device (phosphoric acid digestion at 72 °C) attached to a Thermo Finnigan Delta^PLUS XP isotope ratio mass spectrometer (IRMS)⁷⁸. Conventional stable isotope results are expressed as δ¹⁸O and δ¹³C values relative to Vienna Pee-Dee Belemnite (VPDB). The precision of the measurements is ≤ 0.08‰ for δ¹³C and ≤ 0.1‰ for δ¹⁸O.

Clumped isotope analysis was done on a Thermo Scientific 253 Plus IRMS, after phosphoric acid digestion at 70 °C using a Thermo Scientific Kiel IV automatic carbonate device. Each carbonate sample measurement consisted of at least 10 replicate analyses of 100–120 μg aliquots that were measured alongside carbonate standard samples (ETH1, ETH2, ETH3, ETH4, and IAEA-C2) with assigned values^79,80. Data evaluation was done on the Easotope application⁸¹, with the CO₂ clumped ETH PBL replicate analyses method and IUPAC parameters^82,83,84,85. Δ₄₇ results are given in the I-CDES90 scale⁸⁰, and apparent temperatures in °C were calculated based on the Δ₄₇-temperature calibration from Anderson et al.⁸⁶, with temperature uncertainties propagated from the 1σ standard error (SE) of the Δ₄₇ value. Simultaneously during clumped isotope analysis, on the same samples, conventional carbonate stable isotope composition was also determined.

Petrography

Double-polished sections for petrographic observations were prepared employing low-speed precision sawing and polishing, which minimize thermal and mechanical stresses on the samples. Fluid inclusion observations of secondary calcite were carried out on a transmitted-light microscope (Nikon Eclipse E 400 POL) at the Institute of Geology, University of Innsbruck and petrographic observations of calcite marble from cave walls were carried out on a transmitted-light microscope (Olympus BX53 equipped with Olympus DP23 camera) at ICER, ATOMKI, Debrecen. Additionally, cathodoluminescence studies were carried out at the Karst Research Institute ZRC SAZU, Postojna, on a Technosyn® cold cathodoluminescence luminoscope (CITL CL8200 MK4) with a 14 kV electron beam and an electron current of 350–400 μA, mounted on a Nikon Eclipse E600 biological petrographic microscope. The thin sections, examined in transmitted light and cathodoluminescent light, were photographed using a Nikon Coolpix 990 digital camera.

Fluid inclusion

Fluid inclusion analyses were carried out at three laboratories. Water stable isotope composition was determined at the Institute of Geology, University of Innsbruck, and at the Institute for Geological and Geochemical Research (IGGR), Budapest. Noble gas isotope composition (³He, ⁴He, ²⁰Ne, ²²Ne) was determined at ICER, ATOMKI.

At Innsbruck sample PR20 was crushed in a heated crushing cell⁸⁷, and after cryogenic focusing, the water was transported by He flow into the high-temperature reactor of the TC/EA unit (ThermoFisher) and pyrolyzed into H₂ and CO at 1400 °C. The evolved gases were separated in a chromatographic column and analyzed using a Thermo Fisher Delta V Advantage IRMS. Detailed description of the method can be found in Dublyansky and Spötl⁸⁸. The remaining samples were analyzed at the IGGR in Budapest. Sample chips of about 0.5–1 g were crushed under vacuum in stainless steel tubes, after which the released water was purified by vacuum distillation and introduced into a Los Gatos liquid water isotope analyzer (LWIA-24d). Details of the method are described in Demény et al.⁸⁹. Results are expressed as δ²H and δ¹⁸O relative to Vienna Standard Mean Ocean Water (VSMOW). Analytical precision is 0.5‰ for δ¹⁸O and better than 1.5‰ (Innsbruck) or 2‰ (IGGR) for δ²H.

For noble gas isotope analyses, 1.3 to 2.3 g of calcite samples were loaded into stainless-steel holders with a magnetic ball and baked at ~ 60 °C for 10–12 h in vacuum before the measurements. Gas was extracted by single-step crushing (~ 100 strokes) at room temperature (22 °C). The relatively low number of strokes was applied for each measurement in order to avoid significant contribution of an in-situ component. Helium isotope abundances and ratios were determined by a HELIX- SFT mass spectrometer, whereas a VG-5400 mass spectrometer was used for neon. The analytical procedures are described in more detail in Papp et al.⁹⁰ and Molnár et al.⁹¹.

Uranium-series

Three small pieces of densely crystalline calcite coating (core C2) were dissolved in 15 N HNO₃, doped with a ²²⁹Th, ²³³U, and ²³⁶U mixed spike, and cleaned and separated U and Th in anion exchange resin column chemistry following the methods of Asmerom et al.⁹². Measurements were made on a Thermo-Neptune multicollector inductively coupled plasma mass spectrometer at the Department of Earth and Planetary Sciences, University of New Mexico. U and Th were measured separately as static runs where all isotopes are measured in faraday cups except for ²³⁰Th and ²³⁴U, which were measured in a secondary electron multiplier. Efficiency between the cups and multiplier were determined using an in-house Th standard and the U standard NBL-112. Decay constants and a detailed description of the uranium-series methods are in Cheng et al.⁹³. The measured δ²³⁴U value for standard NBL-112 (CRM-112a) is -38.5‰, where δ²³⁴U = [(²³⁴U_A/²³⁸U_A) – 1] × 1000, and A refers to activity.

Modeling of carbon isotope composition

To model the δ¹³C_cc values in terms of calcite precipitation due to change in temperature with or without CO₂ degassing, we used the equation of Zheng⁷⁵ with a modification, that instead of considering two extreme cases of HCO₃-dominant or H₂CO₃-dominant fluid, for which only fractionation of calcite-HCO₃ or calcite-H₂CO₃ is used, respectively, we select a set of fractions of H₂CO₃ in the DIC of the fluid (where DIC = fH₂CO₃ + (1-f) × HCO₃), that itself reflects fluid pH at a given T. For a selected δ¹³C_DIC, range of temperatures and fH₂CO₃, first we calculate the δ¹³C of the CO₂ in equilibrium with the fluid (δ¹³C_CO2) and then we calculate δ¹³C_cc values using the Rayleigh model equation of Zheng⁷⁵.

To model the fluid composition from which calcite can precipitate in isotopic equilibrium recording temperature and δ¹³C_cc as found for the onset of calcite deposition in C2 core, first we calculate the fractions of H₂CO₃ and HCO₃ in PHREEQC Version 3 software⁹⁴ for a range of pH values, with a fixed temperature and calcium and DIC concentrations determined by forcing charge balance to the solution at fixed calcite saturation index. Then we calculate the δ¹³C_CO2 and δ¹³C_DIC, and using the modeled chemical composition for dissolution of only calcite or calcite and dolomite, we determine the isotopic composition of the carbon from CO₂ not accounted for by dissolution of carbonate rocks (external CO₂; δ¹³C_ext).

Detailed explanation of the calculations is given in the Supplementary Information.

Consent to publish

Consent was obtained from the identifiable person in Fig. S4b (Marjan Temovski) to publish image in an online open access publication.