Pedological and hydrogeological setting and subsurface flow structure of the carbonate-rock CZE Hainich in western Thuringia , Germany

1 Hydrogeology, Institute of Geosciences, Friedrich Schiller University Jena, Burgweg 11, 07749 Jena, Germany 2 Aquatic Geomicrobiology, Institute of Ecology, Friedrich Schiller University Jena, Dornburger Strasse 159, 07743 Jena, Germany 3 German Centre for Integrative Biodiversity Research (iDiv), Halle-Jena-Leipzig, Deutscher Platz 5d, 04103 Leipzig, Germany 4 Biogeochemical Processes, Max Planck Institute for Biogeochemistry Jena, Hans-Knöll-Str. 10, 07745 Jena, Germany


Introduction
The holistic reconstruction of the subsurface architecture of carbonate-rock landscapes is mandatory to reveal their hydrogeological functioning for sustainable management and resource protection.The hydrogeochemical and (micro)biological compositions of shallow (< 200 m) groundwater bodies may be to a large extent controlled by the biogeochemical processes and the fluid-rock/soil interactions while traveling through the overburden of the aquifers (Urich, 2002;Canora et al., 2008).The subsurface compartments that form the overburden are (from surface to subsurface) (I.) the soils sensu stricto, i.e. the rather thin (< 1-2 m), pedogenetically transformed, intensively rooted and enlivened uppermost parts of the Earth's crust, (II.) the partly water saturated, weakly weathered to almost unaltered rocks, and finally (III.), the water saturated rocks that may act as aquifers, aquicludes or aquitards.Infiltrating rain water, as the main component of groundwater recharge, first passes the soils, which are a source for mobile organic carbon (Guggenberger and Kaiser, 2003), suspended particles (Münch et al., 2002;Totsche et al. 2006Totsche et al. , 2007;;Schrumpf et al., 2013;Gleixner, 2013) and even biota (Dibbern et al., 2014) as well as for fertilizers (Köhler et al., 2006) and pesticides from agriculture (Müller et al., 2003).The residence time of the seepage water regulates the chemical and biological quality by controlling the extent to which the partial equilibrium of the rate-limited dissolution, retardation and release processes is achieved (Weigand and Totsche, 1998;Münch et al. 2002).Residence times depend on the duration and intensity of precipitation, moisture content, water retention characteristics and the presence of macropores.A number of papers present detailed studies of karst groundwater chemistry (Suk and Lee, 1999;Moore et al., 2009), of interactions between karst groundwater and springs (Barberá and Andreo, 2012), or they present structural karst aquifer models with the simulation of groundwater flow (Dafny et al., 2010), whereas the overlying soils are not considered in these studies.Other studies that focus on flow path models in the unsaturated zone (Pronk et al., 2009) or on water balance and groundwater recharge, consider land use and soil groups (Steward et al., 2011;Allocca et al., 2014) but without a detailed description of aquifer structure/stratigraphy.Some studies assess the aquifer vulnerability by considering unsaturated zone lithology and hydraulic conductivities (Witkowski et al., 2003), yet leaving groundwater chemistry out of focus.Up to now, a comprehensive characterization of the full range of surface and subsurface compartments, including the soils, the unsaturated zone and the detailed aquifer structure and stratigraphy is lacking.This holds in particular for karst landscapes that supply 25 % of the world's population with drinking water (Ford and Williams, 2007).
This study presents such a holistic analysis of the subsurface compartments of the Hainich CZE involved in subsurface water flow, connecting the surface (impacted by land use) with the groundwater bodies.This includes the reconstruction of the aquifer recharge areas, characterization and mapping of land use types, soils and surface geology (including karst and fracture zones), litho/hydrostratigraphic core characterization and geophysical borehole logs as well as the statistical analysis of the groundwater composition.The main aims of this study are (1) the characterization of subsurface architecture and flow paths, (2) the reconstruction and assessment of infiltration potential in the groundwater catchment, and (3) the influence of land use, soils, topography, lithology, geological structure and karstification on groundwater compositions.The results of this study provide the surface/subsurface framework and conceptual understanding of hydrochemistry for the Hainich CZE research site, which is unique for observing cultural landscapes on carbonate-rock aquifer systems in temperate climates (Küsel et al., 2016).The study is embedded within the collaborative research centre 1076 AquaDiva (www.aquadiva.uni-jena.de),that aims to explore the mutual dependency of the functional biodiversity of the subsurface compartments, controlled by land use, weather events, and especially on the role of the local geology for the functioning of the groundwater ecosystem (Küsel et al., 2016).

Site description
The Hainich CZE is part of the Hainich low mountain range and covers 430 km² of a hillslope subcatchment of the Unstrut River in northwestern Thuringia, Central Germany (Fig. 1).It is limited by the recipient stream (Unstrut River) and the distribution zones of the Upper Muschelkalk target formations.The NW-SE orientated Hainich ridge is a geological anticline that developed topographically as a low mountain range with a steep western, and a moderately inclined eastern flank (Jordan and Weder, 1995).The western flank of the Hainich is characterized by complex tectonics with NW-SE oriented grabens, bounded by normal faults that belong to the regionally important Eichenberg-Gotha-Saalfeld fault zone (Mempel, 1939;Patzelt, 1998).The eastern flank of the Hainich hillslope with the study area, shows a rather simple geostructural setup with NE dipping strata in the direction of the syncline of Mühlhausen-Bad Langensalza (Kaiser, 1905;König, 1930;Patzelt, 1998;Wätzel, 2007).Although little is known about the exact uplift and exhumation age, (NW-SE) fault orientations and the position between the two large horst structures Thüringer Wald and Harz (low mountain ranges), uplifted in the Late Cretaceous (Voigt et al. 2004;Kley and Voigt, 2008), point to a similar history of uplift of the Hainich ridge in the Late Cretaceous.
Groundwater flow is supposed to take place mainly in fractured hard rock aquifers and very subordinate in quaternary valley fills.The Hainich low mountain range is one of the regionally important recharge areas like the Dün, the Fahner Höhe or the Ettersberg in Thuringia (Fig. 1) and is also a typical example of peripheral groundwater supply for water deficient sedimentary basins (Rau and Unger, 1997;Hiekel, 2004).The area belongs to the Cfb climate region (C: warm temperate, f: fully humid, b: warm summer) according to the Köppen-Geiger classification (Kottek et al., 2006) and exhibits a leeward decline in areal precipitation and increasing mean air temperature from the Hainich ridge (> 900 mm/y; 7.5-8 °C) to the Unstrut valley (< 600 mm/y; 9-9.5 °C; long term average 1970-2010, Thüringer Landesanstalt für Umwelt und Geologie, 2016).The intensively investigated study area is limited to a 29 km 2 -subarea of the Hainich CZE, that surrounds the soil and groundwater monitoring transect (Küsel et al., 2016).
The AquaDiva transect consists of multiple wells at five sites for groundwater monitoring, termed H1 to H5.Each site contains 1 to 4 drilled wells labeled with a suffix, for instance H51, H52 and H53.Site H1 (wells H11 to 14) and H2 (H21 to H23) are located in the upper hillslope region of the catchment, which is covered by forest.Sites H3 (H31/32), H4 (H41 to H44) and H5 (H51 to H53) are situated in the agriculturally used, middle and lower slope regions of the Hainich hillslope (Fig. 1).

Field survey
Characterization of the geological, hydrogeological and pedological situation as well as of the infiltration properties in the Hainich CZE was based on a field survey carried out in the intensively investigated subcatchment of the Unstrut River (Fig. 1).482 outcrops were described within an area of 29 km 2 considering lithology (according to Dunham, 1962) and karst features.In the same area, a total of 117 soil profiles resulted from ramcore drillings (50 and 60 mm diameter), supplemented by 154 (Pürckhauer, 22 mm) soil profiles.Ramcore drilling was done with an electrically powered motor hammer (GSH 27,Robert Bosch GmbH,Germany)

Analysis of drill cores and borehole logs
For the determination of the stratigraphic succession, aquifer properties and mineralogical indicators for groundwater flow, 395 meters of cores from twelve drillings were investigated.Core description, based on DIN EN ISO 14689-1 (geotechnical classification of rock) comprised hydrogeological features like fractures and pores and a weathering index.We extended it to sedimentary structures, limestone classification (Dunham, 1962), pore classification (Lucia, 1983), the degree of karstification, the aquifer type, fracture angles and fracture colors, as well as secondary mineralization on fractures, which are indicative for recent and past groundwater flow pathways.Additional information about stratigraphy, clay content, fracturing and groundwater inflow was revealed by analyzing geophysical open-borehole logs of ten drillings from the upper to the lower slope (site H1/2/3/4/5, Fig. 1).This included data on caliper, passive gamma-ray radiation, sonic velocity (delay time of sound waves), specific electrical resistivity of rocks, as well as the temperature and specific electrical resistivity of well water.Gamma-ray curves are interpreted based on the graphical correlation of high gamma ray peaks (marlstones) and intersections of gamma ray curves with an empirically defined "shale line", here at 90 API (separating limestones from marlstones).The spatial correlation of marlstones presumes that basin centre marlstones are laterally more continuous than shallow water limestones (Aigner, 1985).
The two-dimensional correlation of geophysical and geological well logs was carried out with well management software GeoDIN V.8 (Fugro Consult GmbH, Germany).The scope of the correlation was to figure out lateral continuity of flow domains separated by aquitards for achieving information about potential modes of bedding-parallel, intrastratal and crossformational fluid flow.

Geospatial analysis and reconstruction of recharge areas
For the assessment of the major recharge areas, overlays of the compiled geological and soil maps with land use maps interpreted from field observations and aerial imagery, as well as a digital elevation model (DEM) with 2 m resolution were prepared using ArcGIS 10.3 (ESRI Inc., USA).The primary data sets were interpreted in terms of land use types, surface morphology and drainage patterns with special focus on karst phenomena like sinkholes, which are discussed in the literature as preferential input features for surface/rain water (Nennstiel, 1933;Mempel, 1939;Hecht, 2003).Owing to the local geology with geological strata inclined at an angle steeper than the hillslope, the area, position and land use types in the aquifer recharge zones can be reconstructed and preferential infiltration/recharge areas can be identified for each fractured limestone interval (potential aquifer).

Statistical analysis
SPSS 22 (IBM Corp., USA) and Origin Pro 2015 (OriginLab Corp., USA) were used for descriptive and multivariate statistics of the hydrochemical data.This includes hierarchical cluster analysis (HC) for distinguishing hydrogeochemical groups independently from geologically defined aquifer stratifications.Two different parameter sets were statistically examined at the same number of groundwater analysis from three hydrological years.A complete parameter set (a), including all measured parameters, was chosen to figure out control parameters influencing the hydrochemical compositions of the groundwater domains by means of Principal Component Analysis (PCA).As redox processes seems to be very distinct in the studied groundwaters (masking other processes), a limited parameter set (b) was defined that does not include ions which are strongly affected by redox processes.The complete parameter set (a) contains, EC-25, pH, dissolved oxygen, Eh, TOC, anions (NO 3 -, Cl -, SO 4 2-), cations (Ca 2+ , Mg 2+ , Mn 2 and 4+ , K + , Na + , Fe 2 and 3+ , Zn 2+ , Sr 2+ , Ba 2+ ) and silica (Si 4+ ).The limited parameter set (b) includes EC-25, pH, Cl -, Ca 2+ , Mg 2+ , K + , Na + , Zn 2+ , Sr 2+ , Ba 2+ and Si 4+ .Data preparation for HC includes a z-score normalization of concentrations.Strongly correlating variables (> 0.9) and constant values were excluded, outliers were eliminated (next neighbor method) and listwise deletion was carried out (Backhaus et al., 2016).Then, a euclidic distance measure was applied and five clusters were chosen based on the visual observation of the scree plot elbow criterion (Cattell, 1966).The phenon line was chosen at a linkage distance of about 67.Thus, samples with a linkage distance lower than 67 were grouped into the same cluster.Finally, Ward´s method for clustering was applied (Ward, 1963).

Landscape morphology, hydrography and land use
The study area covers altitude regions between 170 and 494 masl with an average slope of 35 m/km.The relief forms a stepless, gently inclined plane and is cut by more than ten straight, parallel and roughly equidistant SW-NE oriented oblique valleys with steeper upper slope angles in headwater areas.Karst phenomena like sinkholes (each of up to 80 m in diameter) are concentrated in a NW-SE oriented lineament at the middle Hainich hillslope (at about 355 masl).Sinkholes occur on local ridges and not in valleys.A second and parallel lineament of shallow elongated (uvala-like) karst depressions with a horizontal extent of up to 400 m crosses the lower Hainich hillslope (at about 270 masl).
The Hainich low mountain range is the topographic water divide with eastward discharge towards the Unstrut River and westward towards the Werra River.Only a few small contact springs occur in the study area, most of them coupled to boundaries between geological formations.Larger springs (for instance the springs coupled to the Küllstedter fault zone; and locally some cleared glades at the upper slope as well as riparian areas of small rivers.Cropland (locally used for wheat, corn and canola production) covers mainly the middle and lower slopes (Fig. 2).

Land use history
As we assume that the land use affects severely the properties of soil and the quality of the groundwater, we aimed for a thorough reconstruction of the land use history of the presumed rural region.We found that extensive forest use has been the dominant land use in all time periods with even no deforestation during the medieval ages (Otto, 2000).Mode of forest operation shifted from random selection of wood via coppice use (since the 15 th century) and planter forestry (since the 20 th century) to unmanaged woodland with the foundation of the Hainich National Park (www.nationalpark-hainich.de) in December 1997 (Otto, 2000;Röhling and Safar, 2004).Parts of the Hainich National Park, which are now unmanaged grassland/scrubland areas, had been formerly used as a military training area since 1964, particularly for tank trainings from 1980 until 1990 (Otto, 2000;Poser, 2004).Crop and pasture agriculture of the lower Hainich hillslope had been organized by agricultural production cooperatives in the former GDR time (from 1945 to 1990) and are now (since 1990) managed by privately owned famer's cooperatives.

Soil series and surface geology
The landscape of the Hainch low mountain range has been developed within the Triassic formations of the Upper Muschelkalk and Lower Keuper.As the geological strata in the study region dip NE in the direction of, but steeper than the angle of surface relief, outcrops of lower/older stratigraphic units are located at the upper slope positions only.To this end, all formations outcrop in distinct zones along the eastern slope of the Hainich ridge (Fig. 2).In addition to the general NE inclination, the upper slope includes a NW-SE oriented normal fault with offsets of about 10 m.Parallel, fault bound troughs are also found near this fault.Near-surface karstification is observed for limestones outcropping at the upper hillslope.The Triassic rocks are covered by young loess loam in sheltered, concave depressions of the east-exposed slopes.Alluvial soils and colluvium fill the valley bottoms.Both sediments are absent at the upper slope and increase in both thickness (max. 3.5 m) and areal cover from the middle to the lower slope.
Soils cover the entire landscape with major soil series developed from carbonate rocks ("carbonate soil series"; Rendzic Leptosols to Chromic Cambisols), siliceous rocks ("siliceous soil series"; Luvisols, Stagnosols) or alluvial sediments (WRB and German soil groups: Table 1).Cambisols are found at the upper and middle slope, Luvisols at the middle and lower slope (both of them mostly used as cropland/pasture) and Stagnosols are restricted to the lower slope.Chromic Cambisols are found in local depressions and old sinkholes.Typical two layer soil profiles, developed from two different substrates, are (I) Luvisols developed from loess loam overlying (II) Chromic Cambisols developed from marlstones (Fier, 2012) (Fig. 2 and Table 1).

Lithology, host rock mineralogy and lithostratigraphic framework
The Hainich CZE, is organized by bio-and lithostratigraphic marker beds (Ockert and Rein 2000;Kostic and Aigner, 2004).The Triassic strata is bounded by an erosional unconformity on top, which is overlain by aeolian loess loam developed from loess deposits of the last glacial period (Weichsel glacial in Germany; Greitzke and Fiedler, 1996) and alluvial/colluvial sediments of Holocene age (Rau and Unger, 1997).
The Diemel formation (mmDO) comprises thin (1-3 cm) yellowish dolomitic marlstones, fine crystalline dolomite and rarely cavernous dolomites without any fossils, gypsum or salt.The 7 m thick Trochitenkalk formation (moTK) with thick (5-30 cm), gray, coarse bioclastic limestones (mainly rudstones with the rock forming fossil Encrinus liliformis) forms a karst-fracture aquifer classifiable as intrastratal karst (Ford and Williams, 2007).Large scale (meter-scale) and consistent fractures, a high fracture index and dissolution-enlarged fractures and vugs (cf Lucia, 1983), as well as karst breccia and conduits at the base of the formation are common features.All fractures and pores are stained with red brownish (Munsell 10R6/6) coatings.Karst features in the Trochitenkalk formation occur from near surface to 90 m depth.Rock mineralogy is dominated by calcite and very low contents of dolomite with traces of quartz, muscovite, chlorite and feldspar.Based on a consistent (moTK) formation thickness, bed thickness and limestone types (Dunham, 1962) in all wells, lateral facies changes within the Trochitenkalk formation are negligible.

Aquifer stratigraphy and hydrostratigraphic standard section
In the Upper Muschelkalk strata, that represent the target formation of this study, the alternating limestone and marlstone beds create numerous aquifer and aquitard layers.Of these, the Trochitenkalk formation (moTK; referred to as HTL = Hainich transect lower aquifer assemblage by Küsel et al., 2016) is the most productive regional aquifer assemblage in current use (Hecht, 2003).It is of karst-fracture type without prominent marlstone interlayers and is sealed at the base by impermeable dolomitic marlstones of the Diemel Formation (mmDO).We observed an intact and unweathered base seal in all wells that were drilled to the base of the moTK.The second major aquifer assemblage is the Meissner formation (moM; referred to by Küsel et al. (2016) as HTU = Hainich transect upper aquifer assemblage), that contains limestone-fracture aquifers which are interbedded with marlstone-aquitards on the decimeter to meter scale (Table 2).
According to published examples for multi-storey subsurface architecture (Haag and Kaupenjohann, 2001;Heinz and Aigner, 2003;Klimchouk, 2005;Sharp, 2007), we use the term "aquifer storey" to emphasize our conceptualization of the fine-stratified setting by arbitrary definition of intervals that are dominated by fractured limestone beds and confined at the top and base by present, unfractured or low permeable beds, the latter with a minimum thickness of 80 cm.
Based on the lithological core description, the lower aquifer assemblage (HTL) represents one aquifer storey termed moTK-1, and the upper assemblage (HTU) is organized into nine aquifer storeys (moM-1 to moM-9; between less permeable units.The correlation of sedimentological core logs and geophysical borehole logs infers, that the subsurface is stratified and all aquifers and aquitards are presumably continuous on the scale of the research transect (5.4 km).In the case of all aquifer storeys flow conditions are confined, preferential groundwater recharge takes place in the upper slope positions (outcrop zones) and regional discharge is towards the NE, due to lithology and orientation of the strata.
Due to their different lithological and bedding properties, the two aquifer assemblages HTL and HTU show significantly different aquifer characteristics: The HTL-assemblage, characterized by thick limestone packages without considerable aquitard interlayers, tends to have strong karstification features, broad fractures as well as shallow soils in the recharge area.
By contrast, the HTU assemblage with its finely alternating aquifer-aquitard stratification exhibits a lesser intensity of karstification, fine fissures as flow paths and it recharges in mid slope areas with thicker soils that partially inhibit rain water infiltration at lower slope areas.In the lower slope both aquifer assemblages are overlain by low to impermeable cap rocks of the Warburg formation (Upper Muschelkalk) and the Erfurt formation (Lower Keuper).
As a result of the principal component analysis (PCA) using the complete parameter set (Fig. 4a), and the parameter set without redox sensitive parameters (Fig. 4b), data points fall into very similar groupings compared to the HCA.According to the PCA of the complete parameter set, the first two components (PC1 plus PC2) explain 54.9 %, and according to the parameter set without redox sensitive parameters PC1 plus PC2 explain 62.1 % of the total variability, respectively.For both PCA, the controlling factors for hydrochemical grouping are Si 4+ , K + , Na + , HCO 3 -, Mg 2+ for PC1 and Sr 2+ , SO 4 2-, EC-25 and Cl -for PC2.
In general, average chemical values of all aquifer storeys infer, that groundwater of the moTK-1 (HTL) aquifer exhibits generally higher O 2 , SO 4 2-, Sr 2+ and lower Mg 2+ , HCO 3 -, Si 4+ , K + , Na + and NO 3 -concentrations compared to the HTU (moM formation) aquifer assemblage when sampled at the same locations.Groundwater sampled from the moTK-1 (HTL) aquifer show consistent chemical trends from the recharge to the discharge area: ion sums increase, in particular in the concentrations of K + , Mg 2+ , Si 4+ , Sr 2+ , SO 4 2-and Cl -(Fig.5).

Recharge areas
Preferential recharge areas for the aquifer storeys match with the outcrop zones (Fig. 2).It is reasonable to infer that catchment sizes, travel distances and residence times are generally larger for lower slope wells (site H4/5) compared to upper slope wells (site H1/2).Valley incision forces partial exfiltration of moM-3 to 9 groundwaters via contact springs and thus reduces the effective catchment size of these aquifers storeys.Additional concentrated infiltration takes place in the lines of sinkholes, which can be tracked over more than four kilometers in the middle slope between transect locations H2/3 and H4.
Lineaments of sinkholes likely represent fracture zones (Smart and Hobbs, 1986;Worthington, 1999;Klimchouk, 2005) and are here related to the fault-related penetration of surface water and the subsequent subrosion of Middle Muschelkalk evaporites like salt and gypsum (Mempel, 1939;Malcher, 2014).

Soil distribution and soil development
Soil groups reflect geology and slope position: in the upper slope position and on steep slopes, the soil groups (Rendzic Leptosols, Calcaric Regosols and Chromic Cambisols) are determined by the carbonate-rock parent material (Greitzke and Fiedler, 1996;Brandtner, 1997;Rau and Unger, 1997).Chromic Cambisols are relict soils in Germany (AG Boden, 2005) and represent the final stage of decarbonatization (Rau and Unger, 1997).Soil groups in moderately inclined mid slope and toe slope positions are Luvisols and Stagnosols developed from (solifluction) mixed carbonate (marlstone) and siliceous (loess) material (Kleber, 1991;Bullmann, 2010) by continued clay relocation to the subsoil (Rau and Unger, 1997).

Infiltration properties of soils and the unsaturated zone
The thickness, clay content and clay distribution in soils influence the infiltration (Smart and Hobbs, 1986).Overall infiltration properties decrease from the upper slope towards the lower slope with increasing thickness of young cover strata like loess loam (Kleber, 1991;Bullmann, 2010).Infiltration is assumed to be inhibited by thick argillaceous soils in alluvial valleys, uvala-like depressions and local toe slope settings, with accumulations of alluvial/colluvial clay (Fig. 2).However, plateau positions (with reduced erosion) at the crest of the Hainich hillslope also show thicker cover strata and thus low infiltration.Fracture and karstification zones like sinkholes or dry valleys (in upper slope positions of the study area) promote the preferential infiltration (Smart and Hobbs, 1986;Suschka, 2007).Rendzic Leptosols of upper slope positions offer better infiltration properties (Hiekel, 2004) than Cambisols, Lithic Udorthents and argillaceous Chromic Cambisols (Brandtner, 1997).Young/initial Luvisols, developed from predominantly silty loess loam show better infiltration properties than mature Luvisols and Stagnosols with layers of subsoil clay.Two layer soil profiles (Cambisols/Luvisols overlying Chromic Cambisols) inhibit infiltration and promote soil water interflow within the topsoil or at the topsoil-subsoil interface.
Water of short and heavy rain events preferentially infiltrates through macropores (desiccation cracks and biopores/earthworm burrows), bypassing the soil matrix pores (Luxmoore, 1991;Edwards and Bohlen, 1996).Intense surface karstification and fracturing is found in valleys, which represent zones of local geostructural weakness (Klimchouk, 2005) as well as the more regional fracturing found in the upper slope (Mempel, 1939;Hoppe and Seidel, 1974;Jordan and Weder, 1995).
The land use type also influences the infiltration properties (Hohnvehlmann, 1995), as it is discussed and classified by Dunne and Leopold (1978)  land use (44.5 to 96.5 % forest).However we detected potentially agriculture related substances (NO 3 -, K + , Cl -) in the wells drilled to all HTU and HTL-aquifer assemblages in middle and lower slope locations (H3/4/5) wells.This points to a significant influence of surface waters entering the aquifers vertically via fractures or sinkholes.

Flow path continuity and aquifer grouping inferred from geology and hydrochemistry
Although we identify two major aquifer assemblages, the chemistry of the groundwaters fall into five distinct clusters.The lithologically defined aquifer-aquitard succession in combination with the increase in karstification towards the base, and the preferential presence of groundwater at the base of limestone packages point to confined aquifers with stratigraphic flow control (Klimchouk and Ford, 2000;Goldscheider and Drew, 2007).Furthermore, hydraulic heads and chemical groundwater compositions of different aquifers at the same sites [H13/14 (cluster 1); H41/42 (cluster 2 and 4); H51/53 (cluster 3 and 5)] are significantly distinct, indicating no vertical connection.Thus we interpret the thin aquifer beds as confined, "sandwich flow type" aquifers (term: White, 1969;Klimchouk, 2005) which are interbedded with confining aquitards.The assumed lateral continuity of aquifers and aquitards results in a "layer-cake" aquifer architecture which has been demonstrated for the Upper Muschelkalk of Central Europe (Aigner, 1982(Aigner, , 1985;;Merz, 1987;Simon, 1997;Borkhataria et al. 2005).
Secondary mineralizations, that indicate groundwater flow (for instance Liesegang banding), occur on limestone fractures.
Fracturing of aquifer rocks increases close to the (NE-SW-striking) fault zones (Hoppe, 1962) and karstification follows the network of faults (Goldscheider and Drew, 2007).Red and brown iron and manganese oxides on upper to middle slope fractures (H1/2/3; cluster 1 and 2) signal temporarily unsaturated conditions, whereas green/gray fracture minerals in the lower slope domains (HTU, H4/H5, cluster 4/5) point to permanently saturated conditions (and likely anoxic conditions, for instance no oxidized Fe/Mn-minerals).Corrosion occurs as intrastratal karstification, which means one or more layers of soluble strata is covered or sandwiched between insoluble beds (Palmer, 1995).Karstification is consistently high in the thick bedded aquifer storey (moTK-1, HTL) and less pronounced in thinly bedded aquifer storeys (moM-1/4/5/6; HTU; Fig. 6) as the bed thickness controls continuity, spacing and width of joints (Goldscheider & Drew, 2007).
Principal component analysis (PCA) using the complete parameter set (a) does not simply reflect aquifer stratigraphy, as each aquifer assemblage contains more than one type of groundwater chemistry (Fig. 6).By contrast, the PCA (b) carried out with the limited parameter set (without redox-related parameters) shows a clear separation of HTL and HTU aquifer assemblages (except cluster 2).The comparison of groundwater samples, that fall in the individual clusters, with the control factors derived from the PCA (using parameter set a), depends on aquifer rock type, slope position, land use and soil group in the recharge area as well as with the proximity to karstification zones results in the following interpretation of the clusters: Cluster 1 (upper slope HTL wells H13/21 and HTU well H14) comprises the moTK-1 aquifer storey in shallow (near surface) positions with recharge areas, exclusively used as forest.Relatively high HCO 3 -and Ca 2+ as well as low Mg 2+ , SO 4 2-, Na + , K + , Si 4+ concentrations (Fig. 7) correspond with the pure calcite limestone (CaCO 3 ) aquifer lithology without any gypsum and very thin marlstone (clay mineral) interlayers.
Cluster 2 (HTL in middle slope wells H31/41 and HTU well H32) includes aquifer storeys in shallow and moderate depth with mixed forest and agricultural catchment.The sites H3/4 are located in the discharge of a prominent karstification zone with sinkholes crossing the line of the research well transect.In addition, round and elongated areas on cropland (recognized on aerial images) with increased soil thickness (inferred from soil mapping), point to former/filled sinkholes.Oxygenated groundwaters in both aquifer assemblages (thus with low Fe/Mn mobility; Hem, 1985) and a high degree in rock fracturing and fracture mineralization with Fe-/Mn-oxide minerals (HTU aquifer storeys moM-9/8/7 and 6), point to a vertical penetration with near surface groundwater via fracture zones through the HTU and in the HTL aquifers.Enhanced concentrations in Na + , K + , NO 3 -, Cl -and TOC (Fig. 7) are likely related to agriculture and fertilizing (Matthess, 1994; Hydrol.Earth Syst.Sci.Discuss., doi:10.5194/hess-2016-374,2016 Manuscript under review for journal Hydrol.Earth Syst.Sci.Published: 5 September 2016 c Author(s) 2016.CC-BY 3.0 License.Kunkel et al, 2004).As nitrate, which is generally derived from agricultural fertilizers (Agrawal et al., 1999;Jeong, 2001), is still present in the deep aquifer waters of site H3, vertical bypassing must be faster than the denitrification process or (alternatively) descending, fault-related waters completely bypass the upper aquifers via large master joints (term: Dreybrodt , 1988;Ford and Williams, 2007), cutting through the complete overlying strata.Vertical fault-related infiltration is likely, as the hydrochemical composition does not match with the forest covered catchment of the HTL (moTK-1) aquifer storey.
Cluster 3 (HTL in footslope well H51) encompasses the moTK-1 aquifer storey in depths of greater than 85 meters, with more distant (4-5 km) forest catchment.Very high concentrations of Ca 2+ , SO 4 2-, Sr 2+ , Cl -(and SO 4 2-/HCO 3 --ratios) that do not reflect the pure limestone (CaCO 3 ) lithology of this aquifer storey, as well as the combination with low K + , Na + , NO 3 concentrations point to a non-surface related source of sulfatic waters.A non-agricultural source of SO 4 2-is supported by significantly lower sulfate concentration in the more surface-near wells of (H52/53) of the same site.Increased Cl - concentrations could be due to dissolution of evaporites (Edmunds and Smedley, 2000).The geostructural position in the discharge of a karstification lineament (fracture zone) and the stratigraphic position (of HTL), directly overlying the Middle Muschelkalk subgroup (containing Ca-(Sr)-sulfate rocks; Jordan and Weder, 1995) point to the import of rising sulfatic Middle Muschelkalk groundwaters (Garleb, 2002;Völker and Völker, 2002).Mixing of Ca-sulfate and Ca-carbonate waters results in an oversaturation and precipitation of calcite and a decrease in alkalinity (Moral et al., 2008), which is supported by a slightly lower pH in well H51.The ascent of confined groundwaters from stratigraphically lower to higher formations, which is well known for artesian settings (Klimchouk, 2005), is assumed to be coupled to a certain threshold value of confinement and hydrostatic pressure, that is probably reached between the well sites H4/5 (cross section in Fig. 8).
Alternatively, the ascent of Middle Muschelkalk groundwaters could be promoted by the lateral permeability decrease from the recharge to the discharge area within the Middle Muschelkalk aquifers of the Hainich region.This decrease in permeability is related to a "subrosion front" (probably between site H4/5; Fig. 8), as the dissolution of sulfate rocks is completed (high permeability) in the recharge area whereas impermeable sulfate rocks are still present in the discharge area (Hoppe and Seidel, 1974;Garleb, 2002).
Cluster 4 (HTU in valley-position wells H42/43/44) includes aquifer storeys in shallow depth with thick (> 5 m) alluvial (argillaceous) cover sediment in valley position, with mixed forest/agriculture/village catchment.The aquifer lithology (limestones) is reflected by high HCO 3 -groundwater concentrations.Low K + , Na + , Si 4+ point to a small influence from marlstones (clay minerals; Bakalowicz, 1994), and thus the alluvial, argillaceous sealing strata at the top is very likely intact.
High Fe (2+ and 3+) , Mn (2+ and 4+) and low SO 4 2-concentrations correspond to a low redox potential and the absence of dissolved oxygen, controlling the mobility of these ions (Hem, 1985;Hsu et al., 2010).Under the given physico-chemical conditions, the NO 3 -concentration is low due to the denitrification by anaerobic bacteria (Agrawal et al., 1999).
Cluster 5 (HTU in footslope wells H52, H53) comprises aquifer storeys covered by very thick (> 40 m), low permeable cover strata (Erfurt formation, Warburg formation) and long (flow) distances (> 3 km) to their recharge area.Based on the aquifer lithology, groundwater discharges through interbedded limestones and marlstones.High K + , Na + and Mg 2+ concentrations point to long residence times with clay minerals (Bakalowicz, 1994).High silica concentrations are proportional to (long) residence times of groundwater in the subsurface (Khan and Umar, 2010).Low NO 3 -concentrations correspond with low redox potentials due to the denitrification of nitrate (Jang and Liu, 2005).
All in all, hierarchical clusters can reflect differences in the length of the flow path (residence time), and therefore differential amounts of rock-water and microbial interaction, the land use type (forest/agriculture) in the recharge area, and external sources of sulfate and magnesium (for instance lower slope well H51; cluster 3).To this end, the effectiveness of the basal and top sealing aquitard layers has a significant influence on the groundwater chemistry.
In this study we distinguish between three different modes of subsurface water flow: (1) Vertical and descending water flow assumed to take place in fracture zones (Worthington, 1999;Goldscheider & Drew, 2007), tracked by lineaments of sinkholes (Mempel, 1939;Hoppe, 1962;Smart and Hobbs, 1986;Jordan and Weder, 1995).Descending flow of surface waters very likely takes place in the middle slope sinkholes with the sites H3/4 downstream (cluster 2, HTL and HTU wells) and ascending groundwater flow is supposed for the lower slope well H51 (cluster 3 HTU, Fig. 8).

Controls for variations in karstification
Karstification and carbonate dissolution is mainly controlled by CO 2 exchange, and thus related to the concentration of dissolved CO 2 in the groundwater (Zötl, 1974;Powell, 1977;Dreybroth, 1981;Hoffer-French and Herman, 1989) as it is exemplified below: Calcite-dissolution: The production of metabolic carbon dioxide within the soils is discussed as a major driving force for the karstification in the near surface carbonate rocks (Krämer, 1988;Ford and Williams, 2007), for instance in epikarst settings (Pan and Cao, 1999).
With increasing distance from soil related CO 2 -sources, the corrosion and the width of fractures and conduits decrease (Ford and Williams, 2007;Williams, 2008).The dissolution of calcite is a relatively fast reaction (Plummer et al., 1979).Kinetic calculations of Dreybroth (1981) infer that CaCO 3 -saturation within karst groundwater in general is reached after a few meters.To this end, the hydrological system in karst groundwaters theoretically grades from wholly karstic to non-karstic (Ford and Williams, 2007) with 50-80 % of carbonate dissolution occurring in the uppermost 10 meters of the subsurface in unconfined settings (Smith and Atkinson, 1976).This contradicts the observed karst features and high CO 2 concentrations in our wells 4 to 5 kilometers away from the groundwater catchment.
One aspect for subsurface carbonate dissolution is the continued release of carbon dioxide by microbes living in the groundwater and rocks far away from the groundwater catchment (Lian et al., 2011).According to Gabrovšek et al. (2000), evenly distributed microorganisms deliver constant production rates of carbon dioxide, as long as organic matter and oxygen is available.In addition, microbes produce organic acids during fermentation under lowered oxygen concentrations that occur in a deprotonated form under pH neutral conditions (for instance acetate or proprionate) or enzymes (carbonic anhydrases) that both enhance limestone weathering (Lian et al., 2011).In addition to organic acids and alcohols released during fermentation, high amounts of carbon dioxide are released.Also other anaerobic processes using Fe(III) or sulfate as terminal electron acceptor produce carbon dioxide which can be applied to the anoxic aquifer storey moM-8 (HTU, cluster 4).Alternatively, karstification may be to a large extent triggered and maintained to the provision of strongly undersaturated (mean ion sums: 123 mg/l for forest, 65 mg/l for grassland) and weakly acidic (pH: 5.9 for forest, 6.0 for grassland), precipitation-fed seepage of CO 2 -containing water via fracture zones and lines of sinkholes.Such deep penetration of surface waters is supported by the presence of agriculture-born substances in the deep aquifer storey moTK-1 (HTL) of the well sites H4/5.As the groundwater compositions of all aquifer storeys are chemically different, mixing corrosion (Bögli, 1964;Wigley and Plummer, 1976) is also possible, for instance through the cross formational mixing of HCO up to depths of 5 meters to reach the host rock.Soil mapping focuses on the influence of geology, relief and land use on infiltration properties.Soil description was carried out according to the German soil survey instruction (Bodenkundliche Kartieranleitung KA5, AG Boden, 2005) and the world reference base (WRB)-scheme (IUSS Working Group WRB, 2006).Soil colors were determined using a Munsell soil color chart.Outcrop mapping was supplemented by the description of rubblestones on cropland, for distinguishing underlying loess loam, marlstone, dolomite and limestone.Strike and dip directions were constructed, based on the boundary between the most Hydrol.Earth Syst.Sci.Discuss., doi:10.5194/hess-2016-374,2016 Manuscript under review for journal Hydrol.Earth Syst.Sci.Published: 5 September 2016 c Author(s) 2016.CC-BY 3.0 License.prominent limestone package (Trochitenkalk formation) and the superposing limestone-marlstone alternations (Meissner formation).

Fig. 1 )
Fig.1) occur at the lower eastern slopes of the Hainich ridge, tracing regional fault/fracture zones.Headwater areas of creeks, small rivers, oblique valleys and even agricultural drainage ditches are mostly dry.During our monitoring period (November 2013 to May 2016) surface water runoff typically occurred from December to March.
succession of the strata outcropping within the study area comprises Middle Triassic sedimentary rocks of the Middle and Upper Muschelkalk subgroup as well as the Lower Keuper subgroup.According to the German stratigraphy, the Middle Muschelkalk is separated into the Karlstadt, Heilbronn, and Diemel formations, the Upper Muschelkalk comprises the Hydrol.Earth Syst.Sci.Discuss., doi:10.5194/hess-2016-374,2016 Manuscript under review for journal Hydrol.Earth Syst.Sci.Published: 5 September 2016 c Author(s) 2016.CC-BY 3.0 License.Trochitenkalk, Meissner, and Warburg formations and the Lower Keuper is synonymously used for the Erfurt formation (Deutsche Stratigraphische Kommission, 2002).The Upper Muschelkalk subgroup, which hosts the target aquifers of the of minor importance.Limestones in the Meissner formation are fracture aquifers and the mineralogical compositions are predominantly calcite and subordinate dolomite and traces of quartz, illite and feldspar.The overlying 16 m thick Warburg formation (moW) with predominantly marlstones (mineralogically: calcite, dolomite, quartz) and the 35 m thick Erfurt formation (kuE) with dolomite rocks (dolomite, calcite for the United States.The soils of the former military (tank) training area very likely inhibit infiltration and favor surface runoff due to compaction.Cropland cultivation modifies soil fabrics, reduces the humic content and increases the soil bulk density.Aquifers in (fertilized) agriculturally used catchment areas (for instance aquifer moM-9), are anoxic, probably due to microbial oxygen depletion resulting from the degradation of organic carbon compounds or oxidation of inorganic electron donors.Catchment areas (defined for all individual aquifers) show a predominance of forest Hydrol.Earth Syst.Sci.Discuss., doi:10.5194/hess-2016-374,2016 Manuscript under review for journal Hydrol.Earth Syst.Sci.Published: 5 September 2016 c Author(s) 2016.CC-BY 3.0 License.
Hydrol.Earth Syst.Sci.Discuss., doi:10.5194/hess-2016-374,2016   Manuscript under review for journal Hydrol.Earth Syst.Sci.Published: 5 September 2016 c Author(s) 2016.CC-BY 3.0 License.(gravitationallydriven) takes place in the recharge areas of the aquifers at the crest and upper slope zones of the Hainich hillslope (including sites H1/2, HTL, cluster 1).(2) Bedding-parallel, confined groundwater flow (stratigraphic flow control, Goldscheider, 2005) occurs in the tectonically undisturbed zones, with intact aquitard interbeds, between, and downstream the two sinkhole lineaments.Confined flow could explain the long water-rock contact times in cluster 5 (HTU) and also the very low concentrations of NO 3 -, NO 2 -and NH 4 + (with potential agricultural origin) of cluster 4 (HTU)-aquifers, which are sealed at the top.Bedding parallel flow is also supposed for fractures with small fault displacements that do not exceed the thickness of the interbedded aquitard (compare to Goldscheider & Drew, 2007).This situation is probably applied to the straight and equidistant oblique valleys, which are probably fault zones.(3) Cross-formational flow through open faults is

Fig. 5 .
Fig. 5. Stratigraphic succession of the Upper Muschelkalk with the aquifer storeys moTK-1 to moM-9 and average chemical 15 compositions of monitored groundwaters.

Fig. 6 .Fig. 7 .
Fig. 6.Correlation of gamma-ray logs, biostratigraphic limestone marker beds and the degree of karstification (red bars).The geological aquifer correlation is cross-checked with the hierarchical clustering of hydrochemical parameters.