Creation of Secondary Porosity in Dolostones by Upwelling Basement Water in the Foreland of the Alpine Orogen

Creation of secondary dissolution porosity in carbonate rocks during deep burial has the potential to improve reservoir properties for hydrocarbons, gas storage, and geothermal applications. However, the occurrence and mechanisms of such porosity enhancement are controversial. Here, we present compelling evidence for generation of deep burial porosity from the Swiss Molasse Basin, where dissolution of eogenetic anhydrite nodules in dolostones of the Middle Triassic Muschelkalk increased the matrix porosity by up to 15 vol.%. We reconstruct the genesis and evolution of the anhydrite-dissolution porosity based on petrography, porosity determinations, analyses of stable and radiogenic isotopes ( δ 2 H, δ 18 O, δ 34 S, and 87 Sr/ 86 Sr), ﬂ uid inclusion studies, and laser U – Pb geochronology of secondary calcite. The results show that modi ﬁ ed meteoric waters derived from the Variscan crystalline basement ascended via basement – cover cross-formational faults into the overlying Muschelkalk, where they dissolved the anhydrite nodules throughout an area of at least 55km 2 at 700 – 2300m depth and 40 – 160 ° C. Secondary calcite in anhydrite moulds yields Late Eocene to Middle Miocene U – Pb ages, which coincide with the timing of basement uplift in the foreland bulge of the Swiss Alpine Orogen. This uplift provided the hydraulic gradients to drive meteoric water deep into the adjacent Molasse basin. Similar enhancement of reservoir properties can be expected in dolostones in other foreland basins that are bordered by a foreland bulge in which fractured basement rocks are exhumed.


Introduction
Carbonate rocks are being investigated worldwide owing to their economic importance as hydrocarbon reservoirs, as potential storage sites for injected gases (seasonal methane and anthropogenic CO 2 ) and as sources of geothermal heat. However, prediction of their porosity at depth is difficult because it depends on many factors, such as primary features inherited from a variety of possible depositional environments, as well as features due to any of various eo-, meso-, and telogenetic modifications that either reduce porosity or create new, secondary porosity by carbonate dissolution. Such secondary dissolution porosity is of great interest for exploration as it has the potential to significantly increase the storage capacity and permeability of carbonate reservoirs. The development of secondary dissolution porosity has been traditionally ascribed to early, near-surface diagenetic processes, commonly those tied to subaerial exposure and infiltration of freshwater, for example [1,2]. More recent studies claim that significant volumes of pore space can be created by dissolution of carbonate during deep burial [3][4][5][6] (and other references reviewed by Ehrenberg et al. [7]). However, Ehrenberg et al. [7] argue that the model of carbonate dissolution during deep burial lacks a rigorous observational basis and that it is inconsistent with solubility constraints.
In contrast, dissolution of calcium sulphate (gypsum or anhydrite) within carbonate rocks is a process cited by Ehrenberg et al. [7] that could create significant porosity during burial but that has received almost no attention in the literature. Here, we present an instructive case of this process from the Middle Triassic Muschelkalk in the Swiss Molasse Basin, in which dissolution of eogenetic anhydrite nodules has notably increased the porosity of the dolostone matrix. We reconstruct the genesis and evolution of the secondary porosity based on analyses of pore-filling minerals in drill core samples, including petrography, analyses of stable and radiogenic isotopes (i.e., δ 2 H, δ 18 O, δ 34 S, and 87 Sr/ 86 Sr), fluid inclusion studies, and U-Pb geochronology. The results tightly constrain the provenance and age of the anhydrite-dissolving solutions. When interpreted within the framework of existing knowledge on the burial history and hydrogeological evolution of the Muschelkalk rocks, all these new observations lead to a consistent explanation of the origin and geographical distribution of the anhydritedissolution porosity, and they provide compelling evidence for the generation of porosity in dolostones during deep burial.  (Figure 1(a)). The basement includes Variscan granites and polymetamorphic gneisses as well as troughs of Permocarboniferous terrestrial sedimentary rocks, all capped by a thin sandstone (Buntsandstein) formed by erosion of the basement rocks in the Early Triassic. The Mesozoic cover comprises a sequence of Triassic-Cretaceous sediments that dip beneath a wedge of Cenozoic Molasse deposits at an angle of approximately 3-5°towards the SE (Figure 1(b)). Thus, the top of the Muschelkalk is exposed along the northern margin of the SMB but it reaches over 5000 m depth in the SE beneath the Alpine Front.

Geological Setting
The Muschelkalk is known for its regional aquifer properties, which are largely due to the matrix porosity and permeability of two dolostone subunits, the Trigonodus Dolomit and the Dolomit der Anhydritgruppe [10]. Our investigations focus on the northeastern part of the Swiss Molasse Basin where the anhydrite-dissolution porosity has been observed in drill core and where the density of wells and thus information on the dolostones is highest (Figure 1(a)). There, the Trigonodus Dolomit (TD) is 12-42 m thick and lies at depths of 60-2270 m. Below this unit is the 24-46 m thick Hauptmuschelkalk (HMK) (Figure 1(c)), a series of low-porosity, low-permeability, partly dolomitized limestones, followed below by the Dolomit der Anhydritgruppe (DAG), which is 7-17 m thick and lies at depths of 136-2300 m (   [8]. (c) Litho-and hydrostratigraphy of the Mesozoic and Cenozoic sediments at the Benken (BEN) well (TD: Trigonodus Dolomit; HMK: Hautpmuschelkalk; DAG: Dolomit der Anhydritgruppe; modified after Nagra [9]). 2 Geofluids and halite) and by the Gipskeuper above (~70 m thick, containing anhydrite). Further above lie the clay-rich aquitard sequences of the Jurassic (~300 m thick; Figure 1(c)).

Burial
History. The burial and thermal history of NE Switzerland is rather complex (Figure 2). The Mesozoic-Early Cenozoic period (~252-66 Ma) was characterized by regional subsidence with little tectonic activity. However, from the late Cretaceous onwards, NW-SE-directed crustal shortening inverted the basin and culminated in the Eocene collision of the European and Adriatic continents and the development of the Alpine orogen [11]. Flexural loading by the Alps induced a second stage of subsidence of the North Alpine Foreland during Oligocene-Middle Miocene times (~34-10 Ma), and shallow-marine and continental Molasse sediments were deposited within the developing basin [12]. Maximum burial occurred at around 10 Ma [13] (Figure 2; [9]). Importantly for the hydrogeological development of the SMB, this foreland subsidence was accompanied by uplift along its northern margin in the area of the future Black Forest Highlands (Figures 1(a) and 3). The early stage of this process occurred during the Late Eocene-Oligocene (~38-23 Ma) and amounted to only a few hundreds of meters of uplift as shoulder horsts, concurrent with subsidence and rifting along the Upper Rhine Graben situated to the west of the Black Forest Highlands (Figure 1(a)). More extensive uplift of the Black Forest Highlands began during Middle Miocene times (~14 Ma), eventually resulting in the erosional exposure of the Variscan gneiss basement and regional tilting of the flanking cover strata towards the SSE [12] (Figures 1  and 3(c)). The extensional tectonic regime in the upper crust reactivated numerous faults and formed flexures in the SMB, primarily oriented along old Paleozoic structures in the basement [12,14,15].

Hydrogeological Evolution.
The Muschelkalk unit was deposited during Early Anisian-Early Ladinian times in shallow marine, partly lagoonal and intertidal settings [16,17]. Dolomitization to form the Trigonodus Dolomit and its anhydrite nodules occurred soon after deposition via mixing of refluxing evaporative brine with meteoric runoff from the Variscan granite-gneiss hinterland [18] (some of the same types of Variscan rocks occur today in the Black Forest Highlands and in the basement of the SMB). The runoff had become saline through interaction with near-shore sabkhas prior to laterally infiltrating the marine sediments. The Dolomit der Anhydritgruppe, on the other hand, formed directly in a sabkha environment. Therefore, during subsequent burial of the dolostones through to at least the Paleogene (Figure 2), the residual pore water in both dolostones would have been a Ca-Na-Cl brine of >20 wt.% salinity, in equilibrium with anhydrite. It is less certain what the δ 18 O values of the pore waters would have been, but presumably they were similar to modern sabkha fluids between 3 and 8‰ VSMOW [19].
In contrast to the residual dolomitizing fluid, the present-day groundwater in the Upper Muschelkalk at the shallow northern margin of the SMB has low salinities (<0.2 wt.%) and low δ 18 O values of −9 to −12‰ VSMOW [20,21]. This mostly Ca-SO 4 -type groundwater was originally meteoric runoff from the Black Forest Highlands that interacted with the SE-dipping Mesozoic sediments (acquiring SO 4 from the evaporites) and laterally infiltrated the Muschelkalk during the Pleistocene (Figure 3(c)) [21].
Deeper within the SMB towards the south (i.e., at the wells of Schafisheim, Pfaffnau, and Berlingen; Figure 1(a)), as well as at the well of Riniken, the salinity of the present-day groundwater increases to 5-11.5 wt.% and NaCl dominates its chemistry [21]. Although the δ 18 Figure 2: Reconstructed temperature evolution at the top of the Trigonodus Dolomit in the NE part of the SMB, based on data in Nagra [9]. 3 Geofluids other hydrochemical indicators (particularly high 40 Ar/ 36 Ar and He contents) led Biehler et al. [22] and Waber et al. [21] to suggest that such waters near Baden (i.e., in the area of Riniken; Figure 1(a)) may have ascended into the Muschelkalk along faults from the underlying crystalline basement. Numerous faults that link the basement with the Muschelkalk or that even rise up through the overlying Jurassic and Cretaceous sediments are documented in the region by seismic surveys [15,23,24]. For the age of the NaCl groundwaters, isotopic constraints point to at least several million years [21]. Thus, at least at the wells where samples have been taken, the present-day groundwaters in the Muschelkalk are mixtures of dominantly Pleistocene meteoric water and minor amounts of most likely older basement water, with no apparent trace of the original Triassic dolomitizing pore water.    [32]). Sulphur isotope ratios of anhydrite were measured at the Institute of Earth Surface Dynamics, University of Lausanne (method described by Spangenberg et al. [33]). Uranium-lead geochronology of secondary calcites was performed at the Institute of Earth Sciences, University of Toronto, using laser ablation inductively coupled mass spectrometry (LA-ICP-MS; method described by Davis [34,35]). Numerous spots on polished slabs were first briefly analysed for uranium and common Pb in order to select the most radiogenic targets for dating.   Figure 1(a)) [36]. The dissolution of anhydrite nodules was found to have increased bulk porosity by up to 15 vol. %.

Methods
This type of porosity is heterogeneously distributed across the SMB. At the shallow wells along the northeastern margin of the basin, the anhydrite nodules are completely

Geofluids
dissolved, whereas at the deeper, southern wells the anhydrite is only partially dissolved or it remains entirely unaffected by dissolution ( Figure 5(a)).
In the shallow wells (Bözberg, Böttstein, and Siblingen) where no pressure solution between dolomite crystals has occurred, the matrix dolomite retains smooth crystal faces and sharp edges, attesting to its chemical equilibrium with the pore waters, irrespective of whether anhydrite is dissolved or not. Where the anhydrite is dissolved, the resulting pore space is often partly filled by secondary minerals ( Figure 5(b)).
Anhydrite moulds are not the only macroscopic pores in the studied dolostones. Vertical to subvertical fractures with apertures up to 0.5 cm occasionally traverse the rocks, and they are locally filled or partly filled with secondary minerals. The various events of secondary mineral precipitation within the dissolution cavities and fractures are described in the following and summarized in Figure 6.
(1) The first precipitation event formed assemblages of quartz, pyrite, minor sphalerite, and rarely galena within pores and fractures. The abundance of these minerals is generally very low, resulting in only minimal pore clogging. In the anhydrite moulds, the individual quartz crystals contain tiny solid inclusions of relict anhydrite. The inclusions are widely spaced relative to their sizes and are thus supported by the quartz matrix, indicating that the quartz precipitated during anhydrite dissolution (Figures 7(a) and 7(b)).
(2) A second, younger event of pore clogging is quantitatively more important and involved precipitation of assemblages of calcite, kaolinite, and minor barite and fluorite, leading locally to complete clogging of the available pore space. Calcite-kaolinite intergrowths indicate that the two minerals precipitated at least in part simultaneously and they both occur in within anhydrite moulds, as well as fracture fillings (Figures 7(c)-7(f)).
(3) A third event involved precipitation of secondary anhydrite, often as a replacement of calcite within the moulds and the fractures, indicating that it postdates calcite and its associated minerals (Figure 7(g)).
(4) Another event involved precipitation of saddle dolomite, also leading to significant clogging of fractures and of anhydrite moulds. Whereas overgrowth textures show that saddle dolomite postdates quartz and its associated minerals (Figure 7(h)), the relative timing of dolomite with respect to calcite precipitation is unclear, as they are spatially separated:  6 Geofluids secondary calcite and associated minerals are restricted to the shallow northern margin of the SMB, whereas saddle dolomite occurs in the southern, deeper wells ( Figure 5(b)). Figure 8 (data in Table A.2) shows the strontium isotope ratios of the rock matrix, anhydrite nodules, and secondary pore-and fracture-filling minerals in the Muschelkalk carbonates. In order to trace and identify potential mineral parent-waters, strontium isotope ratios of primary anhydrite and secondary calcite pore and fracture fillings from underlying and overlying units, as well as of the recent groundwater, were collated from the literature and compared in the same diagram. Calcite pore and fracture fillings in the Muschelkalk both display a wide range of strontium isotope ratios (0.70798-0.71629), which partly overlaps the Sr signatures of the rock matrix and the early diagenetic anhydrite nodules. However, a large portion of the calcites show distinctly higher values. Saddle dolomite also shows high 87 Sr/ 86 Sr ratios, although the scatter in its data is distinctly smaller than that of the calcite. Importantly for tracing the fluids that transported the above secondary minerals, several rocks and groundwaters within the local stratigraphy are characterized by high 87 Sr/ 86 Sr ratios: the rock matrix of the underlying Variscan gneiss basement and the Buntsandstein (values of 0.714-0.728, partly off-axis in Figure 8), the pore-and fracturefilling calcite in the latter two units, the present-day groundwaters in these units, and the shallow groundwater and meteoric river runoff from the crystalline rocks exposed in the Black Forest Highlands.

Fluid Inclusion Studies.
Our fluid inclusion studies have focussed on secondary pore and fracture fillings in the Muschelkalk carbonates and fracture fillings in the underlying crystalline basement. Many of the host minerals contain fluid inclusions arrayed along healed fractures that intersect or crosscut crystal boundaries. These inclusions were not studied in detail because they evidently postdate the precipitation of the host mineral. All the fluid inclusions analysed in this study are either arrayed along former growth horizons of their host crystals, or they occur as dense, three-dimensional clouds within the cores of their host crystals, mantled by inclusion-free crystal rims (Figures 9(a)-9(c)). Thus, they are by definition primary inclusions. It follows that they were trapped during growth of their host crystals and hence the    8 Geofluids data obtained from the analyses are representative of the free parent-fluids at the time of mineral precipitation. At room temperature, all the primary inclusions in quartz and calcite contain a bubble of methane surrounded by aqueous liquid, although the volumetric ratios of the two phases vary strongly between inclusions within the same coeval group (i.e., within the same fluid inclusion "assemblage"; Figures 9(d) and 9(e)). This variation is interpreted to mean that both fluid phases were present at mutual saturation during entrapment of the inclusions [41]; i.e., entrapment was heterogeneous. Methane in the inclusions was unequivocally identified by crushing-stage experiments on samples immersed in naphtha, but it proved to be undetectable by laser Raman spectroscopy, indicating that its pressure within the inclusions is low. Primary fluid inclusions in saddle dolomite display uniform volume fractions of liquid and vapour within individual assemblages, demonstrating that they were trapped from a single-phase aqueous liquid (i.e., homogeneously; Figure 9(f)).
In primary fluid inclusions in quartz and calcite, only two types of equilibrium phase transitions could be measured in the inclusions by microthermometry: (1) the final melting of ice in the presence of liquid and vapour, T m ice and (2) homogenization, either via a bubble-point transition to the liquid state, T h LV → L , or via a dew-point transition to the vapour state T h LV → V . In each inclusion, the phase transitions are reproducible upon repeated heating and cooling cycles, although the dew-point measurements are typically underestimates of the true values, owing to difficulties in observing the final disappearance of the liquid wetting the inclusion walls. Fluid inclusion trapping temperatures, T trap , were therefore taken as being equal to the lowest T h LV → L value measured in each assemblage, following the approach by Diamond [41,42]. Due to the small sizes of the inclusions (i.e., in the range of 10 μm), no reliable eutectic temperatures could be determined. No methane clathrate was observed, consistent with the low internal pressure of the inclusions below the ice +  [20], 3 [38], 4 [39], 5 [21], 6 [40]). 9 Geofluids water + CH 4 -clathrate + vapour Q 1 equilibrium at 26 bar [42]. The results of the microthermometric analyses are shown in Figure 10 and Table A.3.
In the Muschelkalk, the trapping temperatures of pore-filling quartz from the wells at Benken and Schlattingen vary within the range of 41-58°C. In contrast, the paragenetically younger pore-and fracture-filling calcites were trapped at distinctly higher temperatures of 76-89°C. These latter temperatures are slightly higher than those modelled for Miocene maximum burial at these wells but lower than the temperatures modelled for Cretaceous maximum burial ( Figure 2). Primary fluid inclusions in secondary calcites from the wells at Siblingen, Böttstein, Schafisheim, and Bözberg yield trapping temperatures of 51-85°C. However, for these wells no temperature constraints from basin modelling are available for comparison.
The homogeneous trapping mode of primary fluid inclusions in saddle dolomite from the wells at Lindau and Berlingen implies that their T h values between 138 and 147°C constitute minimum trapping temperatures [41]. Assuming fluid entrapment at maximum burial depth (~2700 m; [9]) under hydrostatic conditions (i.e.,~26.5 MPa), the isochores of these low-salinity inclusions (see below for salinities) yield corrected trapping temperatures between  Figure 2). Salinities in terms of equivalent NaCl concentrations were calculated from the T m ice values using the calibration of Hall et al. [44] and are illustrated in Figure 10(b). In the Muschelkalk pore-filling quartz, primary inclusions from the Benken and Schlattingen wells show highly variable salinities of 1.2-23.9 wt.% NaCl eq (salinity variations within individual fluid inclusion assemblages are up to 17 wt.% NaCl eq ). In contrast, paragenetically younger pore-and fracture-filling calcites from these wells contain primary inclusions with relatively uniform and distinctly lower salinities in the order of 0.2-6.5 wt.%. Similarly low salinities of calcite parent-water are observed at the well of Siblingen (0.2-2.9 wt.% NaCl eq ). In contrast, primary fluid inclusions in secondary calcite from the wells at Böttstein, Bözberg, and Schafisheim show higher and variable salinities ranging from 6.5 to 24.9 wt.% NaCl eq. (at the Schafisheim well, salinity variations within individual inclusion assemblages are up to 16 wt.% NaCl eq ). Primary fluid inclusions in saddle dolomite from the wells at Lindau and Berlingen show salinities within a relatively narrow range of 3.6-7.6 wt.% NaCl eq .
In fracture-filling calcite from the crystalline basement, two generations of primary fluid inclusions were identified. One has low salinity (Siblingen: 0.2-0.5 wt.% NaCl eq. ) and trapping temperatures of 65-75°C, whereas the other shows high salinities (Leuggern, Benken: 22.5-27.8 wt.% NaCl eq ) and trapping temperatures of 55-95°C. The two generations occur in spatially separated host crystals and thus their relative timing cannot be determined. However, Mullis [43] investigated predominantly secondary inclusions in secondary calcites from the crystalline basement and the Buntsandstein and found a transition from early highsalinity inclusions (12 to >23.3 wt.% NaCl eq. and T h of 70-149°C) to late, low-to intermediate-salinity inclusions (1.2-6.4 wt.% NaCl eq. and T h of 57-108°C; data reported in Matter et al. [45] and Peters et al. [46,47]). These salinity values are in good agreement with the data obtained in this study ( Figure 10). Accordingly, it is assumed that the relative timing of the fluid generations in the basement calcites  11 Geofluids investigated in this study is the same as that deduced by Mullis [43], i.e., early high-salinity and later low-salinity fluid generations.   [48] and Zheng [49]. Analyses of carbon isotopes in the secondary carbonates revealed no sensitivity to the mixing processes described in Section 5 and so the data are not presented herein.
In the Muschelkalk, the calculated parent-water of the saddle dolomite at the well of Lindau shows the highest δ 18 Figure 11: Measured and calculated oxygen isotopic compositions of parent-waters of secondary carbonates and quartz, as well as literature data for the recent groundwater in the crystalline basement (grey area; [20]). The ranges of calculated δ 18 O of the parent waters result from the ranges of fluid inclusion trapping temperatures shown in Figure 10 and Table A.3. 12 Geofluids kaolinite parent-waters were calculated based on the measured isotopic composition of the kaolinite (Table A.5), the fluid inclusion trapping temperatures of secondary calcite (Table A.3), and the isotope fractionation factors given by Sheppard and Gilg [52]. Regarding the fracturefilling kaolinite in the crystalline basement, Mazurek [53] noted that the wall rocks are often affected by kaolinite alteration and that some of the fractures contain calcite in addition to kaolinite. The fluid inclusion data in Figure 10 (SIB basement calcite; low-salinity) are from calcite in a kaolinite-free fracture, but kaolinite is present in the adjacent altered wall rock. We therefore assume that the 65-75°C trapping temperatures of the primary inclusions in calcite also apply to the precipitation of kaolinite and so the parent-water of the kaolinite was calculated from this temperature.
The parent-water of the Muschelkalk kaolinite shows oxygen and hydrogen isotopic compositions of −7.6 to −3.4‰ VSMOW and −66.5 to −55.1‰ VSMOW, respectively. These values fall to the right of the local meteoric water line, and they deviate significantly from the recent groundwater in the Muschelkalk dolostones. In contrast, they overlap with the oxygen and hydrogen isotopic compositions of the recent groundwater in the underlying Buntsandstein and in the Variscan gneiss basement (grey field in Figure 12). Furthermore, the calculated parent-water of kaolinite in the Muschelkalk (green and orange dots in Figure 12) shows very similar oxygen and hydrogen isotopic signatures to the parent-water of kaolinite in the crystalline basement (blue dots in Figure 12).  [54], 2 [55], 3 [20], 4 [38], 5 this study. 13 Geofluids in the diagrams are reported at two sigma. Note that because 204 Pb cannot be measured accurately, a correction for common lead is not feasible. Therefore, measurements of total instead of radiogenic Pb are plotted. The presence of common Pb has the effect of shifting the data to the right and upward towards an age of approximately 5000 Ma. Hence, multiple data with a similar age but different levels of common Pb will form a linear array on the concordia plot with the lower concordia intercept giving the age of the radiogenic component [34]. Thus, the age can be obtained independently of assumptions about the isotopic composition of the common Pb component by projecting the regression line of data points towards concordia.

Sulphur Isotopes of Anhydrite.
A total of 26 samples were analysed; however, only 8 provided meaningful ages. The remaining samples contained either insufficient uranium or too much common lead. All the datable calcite samples show a substantial radiogenic component, as illustrated by the near-concordant ages, and the isotope ratios generally show a small scatter. Accordingly, they align well with the corresponding radiogenic-common Pb mixing lines, thus yielding single ages. Six secondary calcites from the Muschelkalk provided mean ages between 11.5 and 16.8 Ma (Figures 14(a)-14(f)). Two secondary calcites from the crystalline basement yielded mean ages of 8.3 and 13.3 Ma (Figures 14(g) and 14(h)).

Secondary Minerals in Dissolution Moulds versus
Fractures. The anhydrite-dissolution moulds and the subvertical fractures in the Muschelkalk dolostones contain almost the same sequences of secondary minerals. Both contain early quartz + pyrite and a later paragenesis of calcite + kaolinite ± fluorite ± barite. The strontium isotope ratios of the calcite in the moulds matches those of calcite in the fracture fillings (Figure 8), and at the wells of Schlattingen and Siblingen, the parent waters of calcite in both settings show similar oxygen isotope signatures ( Figure 11). Furthermore, calcite in the two settings has the same U/Pb age (Figures 14(a)-14(f)). It follows that both pore types (i.e., dissolution cavities and subvertical fractures) were open to influx of the same sequence of parent solutions that deposited the secondary minerals. Therefore, no distinction need be made between the two pore types in the remainder of this discussion.

Fluid
Mixing as the Trigger for Anhydrite Dissolution. As outlined in Section 2.3, upon commencement of burial, the Muschelkalk dolostones contained a highly saline pore water (with δ 18 O likely between 3 and 8‰ VSMOW) that was in chemical equilibrium with anhydrite and dolomite. Later dissolution of the anhydrite must have been caused by influx of another solution, such that the new pore water was undersaturated with respect to anhydrite but saturated with respect to dolomite (dolomite crystal faces have remained smooth, without dissolution features, up to today). The nature of the fluid that dissolved the anhydrite is revealed by the primary fluid inclusions in quartz, which precipitated during dissolution of anhydrite (Section 4.1). Within groups of coeval fluid inclusions, salinities vary widely between 1 and 24 wt.% NaCl eq (Figure 10(b)). These observations imply that mixing of two waters with contrasting salinities induced the   simultaneous dissolution of anhydrite and precipitation of quartz. The highly variable salinities of coeval primary fluid inclusions in later calcite at Schafisheim and Bözberg (Figure 10(b)) are interpreted to result from the same mixing process. The saline (~24 wt.% NaCl eq ) endmember of the mixture was evidently the early postdolomitization pore water. Therefore, it was the low-salinity endmember that infiltrated the dolostones from an external source and triggered dissolution of the anhydrite. Primary fluid inclusions in late calcite in the wells at Benken, Schlattingen, and Siblingen also show variable salinities indicative of mixing, but the values and ranges are much smaller (Figure 10(b)), suggesting that the mixtures were dominated by the dilute endmember. The salinity of the dilute endmember is considered to be equal to that of the least saline of the analysed inclusions at those three wells, namely, 0.2 wt.% NaCl eq. .
The secondary saddle dolomites in the deeper wells (Figure 7(h)) contain primary fluid inclusions with smaller but still distinct ranges of salinity between 3.6 and 7.6 wt.% NaCl eq. . These ranges may also be due to mixing, but the interpretation is not so obvious as in the cases above. In any case, it is clear that precipitation of the saddle dolomites was triggered by incursion of an external fluid, because their Sr-isotope ratios are much higher than those of the Muschelkalk rock matrix (Figure 8) and because they precipitated at 30-40°C above the temperature of their dolostone wall rocks (Section 4.4). We therefore attribute the precipitation of the saddle dolomites to the same kind of fluid mixing process as recorded by the secondary minerals at shallower wells to the NW.

Isotope
Signatures of the Endmember Fluids. The isotope signatures of the endmember fluids may now be defined to aid tracing of their provenance. Since Sr isotopes do not fractionate at water-mineral equilibrium, the 87 Sr/ 86 Sr ratio of the saline endmember is assumed to be equal to that of the calcites that precipitated from the high-salinity mixtures in the Muschelkalk at Schafisheim, Böttstein, and Bözberg, i.e., 0.7084-0.7097. These values fall within the range of 87 Sr/ 86 Sr ratios of the surrounding dolostone matrix (0.7080-0.7106, Figure 8), consistent with our conclusion above that the saline endmember is the inherited pore water in the dolostone. The δ 18 O signature of this endmember can be calculated from the isotope composition of the same calcites, since the temperature of water-mineral fractionation is known from the fluid inclusion analyses. The resulting values of ≤4.8‰ VSMOW (Figure 11) identify the pore water as having originally been an evaporative brine, in accord with the likely 3-8‰ range of values outlined in Section 2.3. Such brines typically have high and variable strontium contents of 35-900 mg/L [57].
Using the same approach as above, the isotope signatures of the dilute endmember can be obtained from analyses of the secondary calcites that precipitated from the lowsalinity fluid in the Muschelkalk at the Benken and Schlattingen wells. The highest 87 Sr/ 86 Sr ratio is 0.717 ( Figure 8) and the δ 18 O values calculated for the parent-water are negative, the lowest value being −7.0‰ VSMOW (Figure 11). The Sr content in the fluid is unknown.

Provenance of the Dilute Endmember Fluid.
According to the hydrogeological evolution of the Muschelkalk outlined in Section 2.3, three scenarios for the provenance of the dilute endmember fluid are conceivable: (1) meteoric runoff from the Black Forest Highlands that laterally infiltrated the Muschelkalk from north to south (Figures 3(c) and 3(d)), (2) infiltration via cross-formational faults from overlying units, and (3) infiltration of formation water via crossformational faults from underlying units.
Regarding scenario 1, it has been pointed out above that meteoric recharge from the Black Forest Highlands dominates the present-day groundwaters in the Muschelkalk  16 Geofluids dolostones. This recharge has high 87 Sr/ 86 Sr ratios (0.7087-0.7169; Figure 8), whereas its strontium concentration and salinity are rather low (0.02-0.2 mg/L and <0.12 wt.%, respectively [40]). Its δ 18 O ratios range from −11.2 to −8.6‰ VSMOW (modern recharge in NE-Jura Mountains and Black Forest Highlands) [20]. Using these data and those of the endmember evaporative brine (with an average Sr concentration of 260 mg/L [57]), mixing trends have been constructed for δ 18 O versus 87 Sr/ 86 Sr ratios (Figure 15(a)) and salinity versus 87 Sr/ 86 Sr ratios (Figure 15(b)) according to equations in Clark and Fritz [58], p., 105, and Appelo and Postma [59], p. 33. The resulting mixing bands (grey in Figures 15(a) and 15(b)) do not overlap with the low-salinity fluids that precipitated calcite at the Benken and Schlattingen wells, and hence, this mixing scenario cannot account for all the observations at hand. The main argument against the lateral infiltration scenario is the very low strontium concentration in the surface runoff and shallow groundwater from the Black Forest. Thus, rock-water interaction within the sedimentary sequence along the flow path into the Muschelkalk would have readily overprinted the radiogenic basement runoff with a nonradiogenic 87 Sr/ 86 Sr ratio. Scenario 1 is therefore discarded. Regarding scenario 2, Figure 8 shows that there is no source of radiogenic strontium in the Keuper-Jurassic units overlying the Muschelkalk that could explain the high 87 Sr/ 86 Sr ratios of secondary calcite and saddle dolomite at the wells of Benken, Schlattingen, Lindau, and Berlingen. Furthermore, any pore water from the anhydrite-rich Keuper that infiltrated the underlying Muschelkalk would have introduced sulphur with much lower δ 34 S values than those observed ( Figure 13). Finally, any Keuper water would have been saturated in anhydrite and therefore thermodynamically incapable of dissolving anhydrite nodules in the Muschelkalk. It follows that scenario 2 can be eliminated.
Regarding scenario 3, a possibility is that pore fluid from the immediately underlying Sulfatschichten ascended into the Muschelkalk dolostones. This source can also be discarded, because its pore fluid would have been saturated with respect to halite and anhydrite. Thus, the fluid would have been highly saline rather than dilute, and it would have been unable to dissolve anhydrite nodules in the Muschelkalk. Moreover, since the Sulfatschichten have essentially the same Sr isotope signature as the Muschelkalk (Figure 8), it could not have supplied the highly radiogenic Sr required for the dilute endmember fluid.
Having eliminated the Sulfatschichten, the remaining units underlying the Muschelkalk that are left to consider in scenario 3 are the Buntsandstein, the granite-gneiss basement, and the terrestrial sedimentary rocks in the spatially localized Permocarboniferous troughs. The present-day pore waters in the Permocarboniferous units have oxygen and strontium isotopic signatures that overlap with those of the crystalline basement, and their salinities vary from dilute to locally saline (1-10 wt.% NaCl eq. [20]). Without knowing their age, it is difficult to assess if these waters played a role in the mixing scenario. However, any contribution is likely to have been small, as their very low fluoride and alumina contents [20] cannot explain the pore-filling fluorite and kaolinite in the Muschelkalk. This leaves the Buntsandstein and the granite-gneiss basement, which can be treated as one fluid reservoir: the Buntsandstein has a similar mineralogy to the crystalline rocks because it is their erosional product, and the two units are also known to be hydraulically connected by the same fracture network [60]; hence, it can be assumed that they hosted the same waters. The fluidinclusion-bearing calcite within fractures in the basement can be used to define the properties of those waters. The early generation of calcite contains old brines (labelled "early" in Figure 10) which are thought to have percolated down from the overlying Sulfatschichten [20,43] and which cannot have been involved in the mixing process leading to anhydrite dissolution in the Muschelkalk. However, the characteristics of the dilute waters in the late calcite, which also precipitated fluorite and kaolinite, match those of the dilute endmember in the Muschelkalk: salinities are 0.2-6.4 wt.% NaCl eq. (labelled "late" in Figure 10), δ 18 O values are between −10.7 and −5.0‰ VSMOW (Figure 11), and 87 Sr/ 86 Sr ratios are up to 0.717 ( Figure 8). The only available indication of the strontium concentration in the water is that of the present-day groundwaters in these units, which range from 0.1 to 16.2 mg/L in the crystalline basement and from 0.7 to 17.8 mg/L in the Buntsandstein [61]. The grey mixing trends for these data in Figures 15(c) and 15(d) overlap with all the assembled constraints, confirming that the low-salinity endmember could have been derived from the underlying basement. Therefore, scenario 3 with the crystalline basement as a fluid source is geochemically the only feasible mixing model among those considered.
As a consistency test, a simple thermodynamic model has been constructed using the software PhreeqC [62]. The results (Appendix A.8) show that mixing of a dilute fluid from the basement into the saline brine in the Muschelkalk dolostones will cause spontaneous dissolution of anhydrite and precipitation of secondary calcite and dolomite over a wide range of mixing ratios, consistent with the observations. 5.5. Evidence from Secondary Kaolinite. Derivation of the anhydrite-dissolving endmember fluid from the crystalline basement is further supported by the finding that parentwaters of kaolinite in the Muschelkalk and in the crystalline basement show similar calculated oxygen and hydrogen isotopic signatures (Figure 12). During its geological history, the crystalline basement was affected by several events of hydrothermal alteration, most of which are of Paleozoic age. Only the youngest mineral paragenesis, kaolinite + calcite, postdates the Variscan orogeny but otherwise its age is unconstrained [53]. Kaolinite is present as fracture coatings but also as alteration haloes around the fractures, where it occurs as pseudomorphs after plagioclase. At some wells (e.g., Siblingen), substantial parts of the cored basement section is pervasively kaolinitized and thus this type of alteration may have been widespread. The locally extensive kaolinitic alteration indicates that there must have been huge amounts of fluid circulating through the basement. Accordingly, provided that the basement was hydraulically connected with the Muschelkalk carbonates and a driving force for fluid upwelling was available (see below), it is conceivable that    [53], kaolinitic alteration in the crystalline basement is present at the wells of Böttstein, Leuggern, Kaisten, Siblingen, Pfaffnau, and Herdern (no meaningful information is available for the wells at Benken and Schlattingen). Apart from the Pfaffnau well, the geographic occurrence of kaolinitic alteration in the crystalline basement correlates well with the geographic occurrence of complete anhydrite dissolution in the Muschelkalk dolostones ( Figure 5). This correlation is taken as further support for the scenario of basement-derived fluid having dissolved the anhydrite nodules. indicating precipitation of calcite ± kaolinite in the anhydrite moulds during the Early-Middle Miocene. This age constraint is another argument against lateral infiltration of meteoric runoff from the Black Forest Highlands, as that scenario requires that both the Triassic Muschelkalk and the crystalline basement were exhumed at the time of infiltration. Exhumation of the Muschelkalk is required to provide outcrops to recharge freshwater, whereas exhumation of the basement is required to provide meteoric runoff that acquired high strontium isotope ratios by interacting with the Variscan granites and gneisses. According to the erosional history of the Black Forest Highlands, which is recorded by the deposition of the Younger Juranagelfluh conglomerates, exhumation of the basement occurred at around 10 Ma [63], somewhat later than the precipitation of secondary calcite in the Muschelkalk dolostones. At Benken and Schlattingen, the rock temperature in the Muschelkalk dolostones was~52-67°C during the 11.5-16.8 Ma period of secondary calcite deposition (Figure 2), some 25°C cooler than the fluid that precipitated the calcite ( Figure 10). This is compatible with the scenario that the parent fluid of the calcite ascended from the hotter basement, but it seems impossible to explain by the scenario of lateral infiltration of freshwater from the Black Forest Highlands.  Figure 3). These events were accompanied by extensional tectonics in the upper crust that reactivated old, Paleozoic structures in the basement and that induced new faults and flexures in the Mesozoic sediments of the northeastern SMB (Section 2.2). Figure 5 shows the large number of faults that link the basement with the overlying Triassic units, many of which also rise up through the top of the Upper Muschelkalk. All of these faults could have acted as fluid pathways between the Variscan basement and the overlying Muschelkalk. Moreover, the newly established hydraulic gradients that resulted from the updoming of the Black Forest Highlands in the NW would have provided a driving force to mobilize groundwaters residing in the fracture network within the basement and Buntsandstein. Thus, these fluids could have been forced upwards along faults into the Muschelkalk. A critical point in this scenario is the need for exfiltration areas in the Muschelkalk to establish through-flow. There is evidence of fluvial erosion in the Mesozoic sediments along the southern flank of the Black Forest Highlands during Miocene times [15,63]. However, it is unclear whether this erosion cut deeply enough into the Muschelkalk to create exfiltration pathways.
Secondary quartz is shown in Section 4.1 to have precipitated during anhydrite dissolution. At Benken and Schlattingen, this precipitation occurred at 41-58°C ( Figure 10 and Table A.3). When plotted onto the temperature-time curve for the Benken area (Figure 2), these precipitation temperatures indicate an Oligocene age for anhydrite dissolution, a period during which extensional tectonics and cross-formational faulting began in the northeastern SMB. The age of saddle dolomite cementation in the eastern wells is not known, but assuming it also formed during the Oligocene, it can be expected that the occurrence of anhydrite-dissolution porosity in the entire Benken-Schlattingen-Berlingen-Lindau region was controlled by the occurrence of faults and fractures that hydraulically connected the crystalline basement with the overlying Muschelkalk.
In addition to all the above evidence from the SMB, there is also evidence that basement fluids have been mobilized since the end of the Oligocene in the Black Forest Highlands themselves. There, Jurassic/Tertiary Pb-Znfluorite-quartz-barite veins contain late-stage hydrothermal (~20-70°C) minerals including calcite, dolomite, siderite and Ca-sulphates [64]. Dating of these carbonates by the U-Pb method yielded robust ages between 20 and 0.6 Ma, which Burisch et al. [64] correlated with uplift and rifting of the basement along the Rhine graben (Figures 1 and 3). Mineral parageneses of secondary quartz, calcite, barite, fluorite, and minor ore minerals are also found in the crystalline basement of the SMB, and our U-Pb dating of two calcites in basement fractures at Siblingen yields ages of 8 3 ± 0 5 and 13 3 ± 8 2 Ma (Figures 14(g) and 14(h), respectively). These values match the timing of calcite precipitation in the Black Forest Highlands. Furthermore, within the uncertainty of the age determination, the 13.3 Ma basement sample matches the ages of the secondary calcites in the Muschelkalk (11.5-16.8 Ma), although the 8.3 Ma sample is distinctly younger.
While the geochemical evidence for structurally controlled infiltration of basement water into the overlying Muschelkalk is compelling, only faults with large offsets would be able to juxtapose basement and the Muschelkalk dolostones and bypass the intervening halite-bearing Sulfatschichten. Otherwise, as argued in Section 5.4, contact of the basement fluid with the Sulfatschichten would increase salinity, saturate the solution in anhydrite, and lower the 87 Sr/ 86 Sr ratio, all in contradiction to the observations. Accordingly, fluid ascent from the basement must have occurred along faults that displaced the stratigraphy by >100 m, such that the fractured basement was directly in 19 Geofluids contact with the dolostones. Numerous fault segments that satisfy these conditions occur in the areas of anhydrite dissolution shown in Figure 5 (an example is given in Figure A.9); thus, it appears that flow paths that circumvent the Sulfatschichten were indeed available. Figure 16 schematically illustrates the deduced infiltration of basement fluid into the Upper Muschelkalk.
The partial to complete removal of anhydrite from the dolostones over an area of >55 km 2 ( Figure 5(a)) implies that large amounts of basement fluid infiltrated the Muschelkalk. The resulting mixture of basement water and relict dolomitizing brine evidently remained in the Muschelkalk dolostones through to today, as evidenced by the more saline NaCl-dominated waters in the deeper wells of the SMB [21,22]. During the Pleistocene, these saline waters were in turn diluted by infiltration of meteoric water from the Black Forest Highlands directly into the exposed Muschelkalk, yielding the compositional variety in groundwaters now found in the NE of the SMB (Section 2.3).

Conclusions
Several sources of geochemical evidence (petrography, stable and radiogenic isotopes, fluid inclusion analyses, and radiometric ages) have been integrated to identify the process and timing of the anhydrite dissolution that enhanced porosity in the Muschelkalk dolostones. All these lines of evidence point to incursion of rock-equilibrated groundwater from the underlying crystalline basement and/or the Buntsandstein as the trigger for anhydrite dissolution. The fluids presumably ascended along the numerous known preexisting Paleozoic faults, which were reactivated by Oligocene-Miocene extensional tectonics in the upper crust. The concurrent subsidence of the Molasse Basin and updoming of the Black Forest Highlands to the north of the Basin generated hydraulic gradients that drove fluid through the fractured basement and up into the Muschelkalk (Figure 16). It follows that anhydrite-dissolution porosity in the Muschelkalk dolostones is likely to be spatially restricted to areas which are both (1) on the downthrown side of basement-cover crossformationalfaults (e.g., Figure 16) and (2) subjected to the hydraulic gradient from the Black Forest Highlands ( Figure 16). Thus, this study has shown that the porosity of dolostones can be significantly enhanced (e.g., gain of up to 15 vol.%) in the deep mesogenetic setting (700-2300 m depth at 40-160°C) by areally extensive (> 55 km 2 ) dissolution of eogenetic anhydrite. Similar effects can be expected in the foreland basins of other continent-continent collision belts, where anhydrite-bearing dolostones have been inherited from the pre-orogenic history. In such settings, the critical factors that must coincide to generate secondary porosity via anhydrite dissolution are the following: (1) the rocks in the foreland bulge must be exposed subaerially to establish a hydraulic head gradient versus the rocks buried in the adjacent basin; (2) the exposed rocks must contain permeable fracture networks or other aquifers that connect to permeable cross-formational structures intersecting the dolostones deep in the basin; and (3) the fluids that are thereby forced up into the dolostones must be undersaturated with respect to anhydrite.

Data Availability
The data used to support the findings of this study are included within the supplementary information file.

Disclosure
The abstract of this manuscript has been submitted for presentation at the 16th Swiss Geoscience Meeting, Bern 2018 (1st December).