High‐Resolution Proxy Records From Two Simultaneously Grown Stalagmites From Zoolithencave (Southeastern Germany) and their Potential for Palaeoclimate Reconstruction

Two small annually laminated stalagmites from Zoolithencave (southeastern Germany) grew between CE 1821 and 1970 (Zoo‐rez‐1) and CE 1835 and 1970 (Zoo‐rez‐2), respectively. Trace element concentrations were determined by Laser Ablation Inductively Coupled Plasma Mass Spectrometry (LA‐ICP‐MS). Samples for δ13C and δ18O analyses were micromilled on annual and subannual resolution. Soil and host rock samples were analyzed by X‐Ray Diffraction (XRD) and their elemental concentrations determined via inductively coupled plasma optical emission spectrometer (ICP‐OES). Trace element concentrations in the stalagmites show two groups in the principal component analyses: one with Mg, Ba, and Sr and another with Y, P, and Al, respectively. The second group reflects the content of detrital material. Increased weathering of soil minerals seems to have a strong influence on the silicate/carbonate weathering ratio controlling the variability of Mg, Ba, and Sr. Meteorological and Global Network of Isotopes in Precipitation (GNIP) station data were used to calculate the δ18O values of the drip water (infiltration‐weighted, mean annual, and the mean of the winter precipitation δ18O values) as well as the corresponding speleothem calcite. The δ18O values calculated by the infiltration‐weighted model show similar patterns and amplitudes as the measured δ18O values of the two stalagmites. This suggests that the δ18O values of speleothem calcite reflect the δ18O values of infiltration‐weighted annual precipitation, which zis related to mean annual temperature, resulting in a significant correlation between mean annual temperature and the measured δ18O values of stalagmite Zoo‐rez‐2. This relationship could potentially be used for quantitative climate reconstruction in the future by extending the time series back in time with further stalagmites from Zoolithencave.


Introduction
Quantifying past temperature and precipitation changes is an important but challenging task to set the recent climate change in relation to variations in the past. To achieve this aim, annually to subannually resolved proxy records from climate archives are needed. The most prominent climate archive providing temperature or precipitation reconstructions with annual resolution are tree rings (e.g., Büntgen et al., 2011;Esper et al., 2014;Wilson et al., 2005). Due to the overlap of the tree-ring chronologies with instrumental data, a calibration of the proxy records with meteorological data is possible, and the quantitative reconstruction of climate parameters back in time using the calibration is a standard technique in dendrochronology (Fritts, 1976;Schweingruber, 1983). Speleothems can provide similar high-resolution proxy records, especially if they show annual laminae. A big advantage of speleothems is that they can be precisely dated by the 230 Th/U method (e.g., Cheng et al., 2013;Richards & Dorale, 2003;Scholz & Hoffmann, 2008), and the derived age-depth models can be further improved by lamina counting, which anchors internal age-models for annual to subannual proxy records (Baker et al., 1993(Baker et al., , 2015Scholz et al., 2012;Smith et al., 2009;Tan et al., 2003;Warken et al., 2018). Speleothem δ 13 C and δ 18 O values can be measured with annual to subannual resolution and have provided information on past temperature and/or precipitation variability (Boch et  van Rampelbergh et al., 2015). Furthermore, trace elements can be measured at very high spatial resolution. Magnesium, Sr, and Ba have provided information about past precipitation as well as P and Y could be used as proxies for variations in soil and vegetation activity (Borsato et al., 2007;Johnson et al., 2006;Kuczumow et al., 2003;Smith et al., 2009;Treble et al., 2003;Warken et al., 2018). Speleothems containing annual laminae and providing multi proxy records, such as δ 13 C and δ 18 O values as well as trace element concentrations, that grew until recent times provide the opportunity to calibrate the annual proxies with meteorological data. This allows to quantitatively reconstruct past climate parameters, such as temperature and/or precipitation. For example, Warken et al. (2018) quantified late Holocene autumn/winter precipitation from annual Mg/Ca ratios of a Romanian stalagmite, and Tan et al. (2003) reconstructed summer temperature anomalies for the last 2,650 years from the lamina thickness of a stalagmite from northern China.
In this study, proxy records (δ 13 C and δ 18 O values and trace element concentrations) from two modern stalagmites from Zoolithencave (southeastern Germany) were tested for their potential to calibrate with meteorological data. This potentially provides the basis for the interpretation of longer Holocene speleothem proxy records from Zoolithencave in the future.

Cave Setting
Zoolithencave is located in the Franconian Jura in southeastern Germany, 30 km southeast of Bamberg ( Figure 1). The host rock of Zoolithencave is Upper Jurassic dolomite (Malm δ, Middle Kimmeridgian), and the cave is famous for its palaeontological inventory (e.g., Esper, 1774;Heller et al., 1972;Rosendahl, 2005). The soil above the cave consists of approximately 15-cm humic A-horizon and approximately 30-40-cm brownish, loamy B-horizon. The average rock overburden of the Entrance Hall is 10-  Groiß, 1988). The location of the Zoolithencave is indicated by the red star (modified after Riechelmann et al., 2019). 12 m (Wurth, 2002). The sampling site of the two Zoo-rez stalagmites is approximately 20 m away from the cave entrance (Riechelmann et al., 2019), and first cave air monitoring results show that the temperature in the Entrance Hall is relatively constant at 8.5 ± 0.4°C (Riechelmann et al., 2014) and relative humidity is 94 ± 4% year-round (pers. Comm., Prof. Dr. D. K. Richter, Ruhr-University Bochum). This is comparable to other caves such as Bunker Cave with 10.9 ± 0.6°C and 93 ± 2% relative humidity (Riechelmann et al., 2011). Furthermore, first results from drip water monitoring at a drip site near the sampling site of the two Zoo-rez stalagmites show drip water δ 18 O values with a mean of −9.88 ± 0.32‰ (pers. Comm., Prof. Dr. D. K. Richter, Ruhr-University Bochum). This drip site shows a seasonality in the drip rate with mean drip rate of 2.68 ml/min (Riechelmann et al., 2014), and a quick response of the drip rate to main precipitation events within hours is observed (Holz et al., 2018).

Stalagmites Zoo-rez-1 and -2
Stalagmites Zoo-rez-1 and -2 were sampled as one block in the Entrance Hall of Zoolithencave and grew simultaneously within a distance of 7 cm. Zoo-rez-1 has a height of 3 cm, and Zoo-rez-2 has a height of 2.7 cm ( Figure 2). Both stalagmites were sampled in August CE 1999 and described by Wurth (2002). The chronologies were constrained by 14 C bomb peak detection at the top of both stalagmites, a 14 C age of a piece of charcoal from the consolidated base part of Zoo-rez-2 (Figure 2b), lamina counting, and cross-dating of the lamina thickness series (Riechelmann et al., 2019). Zoo-rez-1 grew between CE 1821 and 1970, whereas Zoo-rez-2 grew from CE 1835 to 1970. The age uncertainty at the top of the two speleothems due to the uncertainty of the 14 C bomb peak detection is ±5 years; for the older parts, the uncertainty increases to ±19 years for Zoo-rez-1 and ±7 year for Zoo-rez-2. Both chronologies probably still contain missing and false laminae. However, we were able to minimize the number of missing and false laminae by using the cross-dating of replicated measurements of the same track as well as cross-dating of the three parallel measured tracks on an individual stalagmite. Nevertheless, the missing and false laminae could result in a compression and/or stretching of the chronologies in certain periods. Furthermore, the uncertainty of the 14 C bomb peak detection could result in potential differences in the age of the top layers of the two stalagmites (Riechelmann et al., 2019). These uncertainties likely affect correlation coefficients with meteorological data.

LA-ICP-MS Analyses of Trace Element Concentrations
The element concentrations of the two stalagmites were determined with an Element2 ICP-MS (ThermoScientific, Bremen, Germany) equipped with a high-energy Nd:YAG UP213 laser ablation system (wavelength = 213 nm; New Wave, Fremont, USA) at the Max Planck Institute for Chemistry, Mainz. The reference material used for calibration was NIST SRM 612, a synthetic glass with a high trace Figure 2. Pictures of the sampled slices of (a) Zoo-rez-1 and (b) Zoo-rez-2. The laser ablation tracks are indicated by black lines, and the milling tracks for the carbon and oxygen isotope samples are indicated by the red rectangles. The sampling tracks in Zoo-rez-1 are a little to the left from the growth axis (5 mm), due to the milling of the 14 C samples for the bomb peak detection (see Fig. 2a in Riechelmann et al., 2019). However, Zoo-rez-1 has an asymmetric growth, which gives the impression that the growth axis is located much more to the right. Zoo-rez-1 and -2 were sampled as a block from the cave. The two slices were cut from this block and by visual inspection of the cut slices and the other pieces of the block, we could identify the growth axes of the two stalagmites. The asymmetric growth of Zoo-rez-1 is probably due to the uneven basement. element content (Jochum et al., 2011). The MACS3 (carbonate pressed powder) was measured as a second reference material . The measurements were performed in line scan mode near the growth axes of the stalagmites (Figure 2). The spot size of the laser beam was 110 μm, the pulse repetition rate was 10 Hz, and the scan speed was 10 μm/s. The measured isotopes were 25 Mg, 138 Ba, 88 Sr, 27 Al, 31 P, and 89 Y, which were normalized to 43 Ca as an internal standard. The 1σ precision of the related NIST SRM 612 measurements was 5.3% for Mg, 3.6% for Ba, 2.3% for Sr, 8.1% for Al, 5.5% for P, and 3.4% for Y. For a detailed description of the methodology, the reader is referred to Jochum et al. (2012) and Mischel et al. (2017).

Soil and Host Rock Analyses
Four soil samples covering different depths (Zoo B1: 8-23 cm; Zoo B2: 23-34 cm; Zoo B3: 34-41 cm; Zoo B4: 41-43 cm) were collected above Zoolithencave. These samples were analyzed with a Philips Xpert MPD Theta-Theta X-ray Diffractometer (Malvern Panalytical GmbH, Kassel, Germany) using Cu Kα radiation at 45 kV and 40 mA at the Institute for Geology, Mineralogy and Geophysics, Ruhr-University Bochum, to determine the mineralogy of the soil. The scanning range was from 4°to 65°(2θ) with a step size of 0.01°(2θ). The background noise was reduced by using a graphite secondary monochromator.
In order to distinguish between exchangeable and structurally bound ions of clay minerals, a two-step acidleaching procedure was performed following the method of Wimpenny et al. (2014), which is based on the method of Chan and Hein (2007). For the first leaching step, acetic acid was used, which reflects conditions occurring in natural soils, where carbonic acid is produced by microbial and plant activity. Approximately 200 mg of each dry soil sample was used, and in the first leaching step (exchangeable ions), 5 ml of suprapure acetic acid (4 wt.%) was added, whereas 5 ml of suprapure 2M HCl was used for the second leaching step. Approximately 2 ml of a mixture (1/1) of H 2 O 2 (31 wt.%) and HNO 3 (65 wt.%) was added to the dried down leachates to destroy organic matter. Subsequently, the dried down leachates were pestle in a mortar. Additionally, two samples were drilled from the dolomitic host rock.
Both soil and host rock samples were weighed for elemental (Ca, Mg, Sr, and Ba) analyses. The samples were dissolved in HNO 3 (3.5 wt.%) and then diluted with 2-ml MilliQ® H 2 O. The elemental concentrations were measured with an inductively coupled plasma optical emission spectrometer (Thermo Fisher Scientific iCAP 6500 DUO, Bremen, Germany) at the Institute for Geology, Mineralogy and Geophysics, Ruhr-University Bochum. The used reference materials were BCS-CRM512 (dolomite; n = 112) and BCS-CRM513 (limestone; n = 110; Bureau of Analysed Samples Ltd.). The 1σ reproducibility for the two reference materials is 0.19% and 0.32% for Ca, 0.09% and 0.003% for Mg, 21 and 2 μg/g for Sr, and 2 and 22 μg/g for Ba, respectively.

Carbon and Oxygen Isotope Composition
The samples for stable carbon and oxygen isotope composition of the two stalagmites were obtained with a MicroMill (New Wave Research, Portland, USA) using 300-μm carbide metal drill bit (Dettman & Lohmann, 1995) at the Institute for Geosciences, University of Mainz. The sample resolution is 50 μm for Zoo-rez-1, which approximately results in three samples per year, and 130 μm for Zoo-rez-2, which approximately results in an annual resolution. Sample weights ranged between 60 and 120 μg (± 1 μg).  Figure 1). Mean annual temperature is 8.5°C, and the mean annual sum of precipitation is 638 mm for the period CE 1879-2013 ( Figure S1 in the supporting information). Daily potential evapotranspiration (ET pot ) was calculated using the equation of Haude (1954Haude ( , 1955: (1) with P 2pm ¼ 4:54 hPa × 10 7:45×T2pm=235þT2pm ð Þ hPa ½ : (2) P 2pm [hPa] is the saturation vapor pressure, F 2pm [%] is the relative humidity, and T 2pm [°C] is the temperature at 2 pm, respectively. The meteorological station data only provide mean daily relative humidity and not the relative humidity at 2 pm. Therefore, the mean daily relative humidity was used for F 2pm . Furthermore, only daily maximum temperatures are provided. Since these should be reached around noon (i.e., near to 2 pm), we used those for T 2pm . The composition of vegetation is included in the calculation by the vegetation coefficient h (Häckel, 1999). Vegetation coefficients for beeches were used ( Table S1 in the supporting information) as the vegetation above Zoolithencave predominantly consists of this tree species. Dendrochronological analyses of the beeches above the cave in CE 2014 resulted in a maximum tree age of 140 years (pers. Comm., Prof. Dr. A. Bräuning, University Erlangen-Nuremberg). From the daily calculated evapotranspiration, the monthly sum of evapotranspiration was calculated back to CE 1879. The monthly infiltration (Inf) was calculated as  Figure 1). This is the GNIP station with the longest data series  nearby the cave location and has a similar altitude as Zoolithencave (455 m a.s.l.). There is also a GNIP station in Würzburg approximately 100 km northwest of Zoolithencave. The mean annual δ 18 O values of precipitation show a positive correlation between both GNIP stations with r = 0.43 (p < 0.01, n = 36, CE 1978-2013). However, comparing the data of the meteorological station Bamberg (mean annual temperature: 9.1°C; mean annual sum of precipitation: 651 mm), the data are more similar with Regensburg (8.8°C; 658 mm) than with Würzburg (9.6°C; 596 mm). Therefore, the GNIP station Regensburg was used.

Data Analysis
The cross-dated lamina thickness series of both stalagmites provide an assignment of a calendar year to each lamina. However, as described in section 2.2, there is still an age uncertainty for the top lamina of the two stalagmites of ±5 years due to uncertainties in the lag of the 14 C bomb peak increase between the atmosphere and the stalagmite as well as uncertainties in the laminae counting procedure, which could not be exactly quantified. This is due to still missing or false laminae, which could result in a compression and/or stretching of the chronologies in different time spans. The cross-dating results in a growth phase of CE 1821 to 1970 for Zoo-rez-1 and of CE 1835 to 1970 for Zoo-rez-2 (Riechelmann et al., 2019). These lamina-based age-depth models for both stalagmites provide the option to assign an age to the different proxy records, which have subannual resolution. This was done using an R Code (R Core Team, 2018) assigning an age to every proxy data point of the trace element, δ 13 C and δ 18 O time series of both stalagmites using linear interpolation between the counted laminae. For comparison of the proxy data of the two stalagmites and the meteorological data, mean annual values of the trace element, δ 13 C, and δ 18 O time series were calculated. Proxy records and climate parameters were compared using Pearson correlation coefficients (r) and p values (p < 0.05, p < 0.01, p < 0.001).
All elemental data were standardized prior to principal component analysis (PCA; Navarra & Simoncini, 2010;von Storch & Zwiers, 2002), which was performed using the software PAST3 (Hammer et al., 2001). The PCA was used to identify proxies influenced by the same component.
In order to assess potential parameters affecting the speleothem calcite δ 18 O values, the model of Wackerbarth et al. (2010) was applied, which also used the infiltration calculations as presented in section 2.6. The mean annual δ 18 O values of the cave drip water were calculated by the Wackerbarth et al. (2010) approach as the sum of the infiltration-weighted monthly mean δ 18 O values of meteoric precipitation (GNIP station Regensburg): where Gi is the weighting coefficient for each month (Table S2), which depends on temperature, amount of precipitation, relative humidity, and the type of vegetation (see section 2.6 for details). The weighting coefficient is the fraction of the annual amount of water infiltrating into the cave during each month: Then, where T is the cave temperature in K (281.65 K). The cave temperature (8.5 ± 0.4°C) corresponds very well to the mean annual temperature above the cave (8.5 ± 0.8°C, CE 1879-2013, meteorological stations Bamberg and Bamberg Observatory). Changes of cave air temperature occur on multiannual to decadal scales (Moore & Sullivan, 1978). Therefore, we assume a constant cave air temperature for the model period. During the time span of CE 1978-2013, the mean annual outside temperature is 9.1°C (meteorological station Bamberg where T is the mean annual temperature from the meteorological stations Bamberg (CE 1949(CE -1970 and Bamberg Observatory (CE 1879(CE -1948

Element Concentrations
The Mg series of Zoo-rez-1 and -2 shows pronounced cycles ( Figure 4a). However, counting the minima of both series results in 123 minima for Zoo-rez-1 and 93 minima for Zoo-rez-2, which does not correspond to the length of the two time series (150 and 133 years, respectively). Barium and Sr cycles correlate positively with Mg as shown by the results of the PCA (Figure 5), albeit their amplitudes are smaller (Figures 4b and 4c).
The second group of trace elements in the PCA is formed by Y, P, and Al ( Figure 5), which show elevated values at the same time in all three elemental records (Figures 4d-4f). In the PCA of Zoo-rez-2  (Figures 4a-4e) except for Al, where Zoo-rez-1 has a much higher concentration than Zoorez-2 (mean value: 71 compared to 12 μg/g, Figure 4f).

Soil Mineralogy and Elemental Concentrations
The soil above Zoolithencave consists of quartz, feldspar (plagioclase, orthoclase), kaolinite, illite/muscovite, and chlorite (XRD analysis, Figure S2). There are no differences in the mineralogical composition of the soil samples through the soil-depth profile. Calcium, Mg, and Sr concentrations of the soil samples are generally higher for samples treated with acetic acid compared to those treated with HCl (Table 1). Samples from greater depths (34-43 cm) display significantly higher Ca, Mg, and Sr concentrations (Zoo B3 and B4). For Ba, we observed no significant difference between acetic acid and HCl leachates, and the concentrations vary between 148 and 376 μg/g ( Table 1). The dolomitic host rock displays high Ca and Mg concentrations, whereas Sr and especially Ba concentrations are very low (Table 1).

δ 13 C and δ 18 O Values
The

Annual Proxy Data
For comparison of all proxy records, mean annual values were calculated for the element concentrations as well as the δ 13 C and δ 18 O values. The annual lamina thickness series of both stalagmites does not show any highly significant correlations (p < 0.001) with the proxy records ( Table 2). The δ 13 C and δ 18 O series show a positive correlation of 0.50 (p < 0.001, n = 150) for Zoo-rez-1 and 0.69 (p < 0.001, n = 113) for Zoo-rez-2, respectively, which are slightly higher than the correlation coefficients obtained for the measured high-resolution data. Furthermore, the δ 18 O values covary positively with the Mg content (p < 0.001) in both stalagmites and show a negative correlation (p < 0.001) with P in Zoo-rez-2. The δ 13 C values show a positive correlation (p < 0.001) with Y in Zoo-rez-1 and a positive correlation (p < 0.001) with Ba, Mg, and Y in Zoo-rez-2. For the element concentrations, the same pattern is visible for the annually resolved data and the high-resolution data. Mg, Ba, and Sr as well as Y and P are positively correlated (p < 0.001). Yttrium shows a negative correlation (p < 0.001) with Mg and Sr in Zoo-rez-2. Aluminum shows a positive correlation (p < 0.001) with Mg and P in Zoo-rez-1 and with Ba and Sr in Zoo-rez-2 (Table 2).

Trace Elemental Variations of Soil and Host Rock
Leaching the soil by acetic acid, as used in the first step, reflects conditions occurring in natural soils, where carbonic acid is produced by microbial and plant activity. In this first acid-leaching step, mostly exchangeable ions of clay minerals, such as Mg, Sr, and Ca, are detached. Thus, the higher concentrations of these elements for the deeper soil samples leached by acetic acid are probably due to a higher abundance of clay minerals in the deeper parts of the soil. By using HCl, structurally bound ions are leached from the minerals. In this case, Mg is derived from chlorite and possibly also from illite (Okrusch & Matthes, 2005;Wimpenny et al., 2014), which are more abundant in deeper soil sections explaining the higher concentrations for the two deepest samples (Table 1). The mineral feldspar, however, can already be weathered under the presence of carbonic acid and subsequently forms kaolinite (Okrusch & Matthes, 2005). Thus, the high Ca concentrations for both the acetic acid and HCl leachates can be explained by the weathering of Ca-bearing plagioclase. Consequently, Ca and Mg are derived from (i) easily leachable sites (ion exchangeable) and (ii) structural sites. This is also the case for Sr, although no mineral containing Sr in its crystal lattice was observed by XRD. However, Sr most likely substitutes for other cations in the crystal lattice. The relatively constant concentrations of Ba for both acid-leaching steps are best explained by the presence of feldspar containing Ba. Mixed crystals between orthoclase (KAlSi 3 O 8 ) and celsian (BaAl 2 Si 2 O 8 ), the hyalophane, provide Ba due to the high weathering sensitivity of feldspar (Okrusch & Matthes, 2005). The Mg concentration of the host rock is high due to its dolomitic composition; however, Sr concentrations are much lower, and even Ba shows a very low concentration in the dolomite.

Trace Element Variations of Speleothems
Phosphorus and Y show a positive correlation in both stalagmites and form one group in the PCA (Table 2, Figure 5). Both elements are mostly transported into the cave associated with colloids, such as humic and other organic substances from the soil (Fairchild & Treble, 2009), and subsequently incorporated into the stalagmite calcite. The high organic content of both stalagmites is visible in the UV-fluorescence, which is induced by humic and fulvic acids (Baker et al., 1993;Riechelmann et al., 2019;Shopov, 2003). Phosphorus originates most likely from microbial breakdown of organic matter in the soil, and Y is adsorbed to organic matter, which results in a positive correlation of P and Y (Borsato et al., 2007). Therefore, P and Y have been interpreted as proxies for vegetation productivity and microbial activity in the soil in several speleothem trace element studies (Borsato et al., 2007;Fairchild & Treble, 2009;Treble et al., 2003;Wassenburg et al., 2012). Decreasing P and Y concentrations are related to decreasing microbial breakdown of organic matter and lower vegetation productivity, which are affected by decreasing temperatures or precipitation Geochemistry, Geophysics, Geosystems and vice versa (Borsato et al., 2007;Treble et al., 2003;Wassenburg et al., 2012). In many cases, both elements peak after an increase of infiltration in autumn indicating more effective transport of organic colloids from the soil into the cave during this season ("autumnal flush") (Borsato et al., 2007;Frisia et al., 2012;Huang et al., 2001;Treble et al., 2003). Unfortunately, P and Y cannot be used as proxies for vegetation and soil activity in the Zoo-rez stalagmites because the signal is also influenced by detrital contamination related to silicates. Aluminum is a proxy for detrital material in speleothems because it is transported to the cave

10.1029/2019GC008755
Geochemistry, Geophysics, Geosystems as particles, such as clay minerals, which are rich in Al (Fairchild & Treble, 2009;Wassenburg et al., 2012). In particular, Al shows a positive correlation with P in Zoo-rez-1 (Table 2) and forms one group with P and Y in the PCA for both stalagmites ( Figure 5). Several peaks in Al of Zoo-rez-1 are also visible in P and Y of Zoorez-1 (Figures 4d-4f). Although, Zoo-rez-2 contains less Al (mean value 12 μg/g), higher Al values in the top section of the stalagmite are also associated with increased P and Y in the top part of Zoo-rez-2 (Figures 5d-5f). Thus, detrital contribution to P and Y is very likely.
A positive correlation between speleothem Mg, Ba, and Sr ( Figure 5, Table 2) is most commonly interpreted as an indicator of prior calcite precipitation (PCP) (Fairchild et al., 2000(Fairchild et al., , 2006Johnson et al., 2006;Smith et al., 2009;Tooth & Fairchild, 2003;Treble et al., 2003). PCP describes the precipitation of calcite prior to precipitation of carbonate on the top of the stalagmite, which might occur in air filled pockets and cavities in the aquifer above the cave and/or at the cave ceiling as stalactites or other sinter forms. PCP increases the Mg, Sr, and Ba to Ca ratios of the drip water and, hence, in speleothem calcite (Fairchild et al., 2006). Enhanced PCP is most pronounced in the summer season, when infiltration is at a minimum, or under drier climate conditions associated with longer drip intervals (Fairchild et al., 2006;Fairchild & Treble, 2009;Treble et al., 2003;Wassenburg et al., 2012Wassenburg et al., , 2020. A simple test to assess whether PCP is a major controlling factor for variations in the elemental concentrations is plotting ln (Mg/Ca) versus ln (Sr/Ca) and calculating the slope of the regression line (Sinclair et al., 2012). Wassenburg et al. (2020) modeled different PCP slopes compared to Sinclair et al. (2012) and showed that PCP may be associated with slopes between 0.71 and 1.45 depending on whether the Sr partitioning coefficient is related to the Mg/Ca ratio of the speleothem calcite. The slopes calculated for Zoo-rez-1 and -2 are 0.20 and 0.18, respectively ( Figure 7). Therefore, we conclude that the Mg, Ba, and Sr variations in Zoo-rez-1 and -2 are most likely not dominated by PCP and have to be influenced by other factors.
Growth rate could be an influencing factor of the Sr and Ba variations due to their growth rate dependent partitioning coefficients (Borsato et al., 2007;Treble et al., 2005;Warken et al., 2018). However, since the partitioning coefficient of Mg is not growth rate dependent, the positive correlation between Mg, Sr, and Ba can probably not be explained by variations in growth rate. Furthermore, a correlation between the lamina thickness series, which reflects the annual growth rate, and Ba or Sr is not observed in the two stalagmites (Table 2). Therefore, growth rate can be excluded as an influencing factor.
The sources of Ca, Mg, and Sr in the speleothems from Zoolithencave are the soil and host rock (especially for Ca and Mg), whereas Ba mostly results from the soil as discussed in section 3.4 (Table 1). Nevertheless, the Mg of the stalagmites probably originates in large parts from the dolomitic host rock (Table 1) because the amount of Ca originating from the host rock is higher than from the soil. The Sr of the stalagmites originates both from the dolomite and the soil. The exchangeable ions also display an increase of Sr with soil depth, which is not the case for Ba. For comparison to this case, that Ba could be mainly provided from the soil, was also detected in drip water from a cave in southeast Australia, where the host rock did not contain any Ba (Rutlidge et al., 2014). Therefore, the changing amount of leaching from the soil is an important factor causing a positive correlation between Mg, Sr, and Ba. Longer water residence time in the soil, induced by drier climate conditions as well as higher temperatures, increases silicate weathering, which results in changes in the silicate (feldspar and clay minerals) versus carbonate (dolomite) weathering ratio. However, carbonate weathering is always dominant and can thus be considered as more or less constant in the temperate climate zone (Egli et al., 2008;Maher, 2010;Tipper et al., 2006). Therefore, changes in silicate weathering are most likely responsible for variations in the silicate/carbonate weathering ratio. There is a significant negative correlation between temperature and infiltration for the months May to September (CE 1879-2013) ( Figure S1 and Table S2). Warmer and drier summers probably increase the silicate/carbonate weathering ratio, and more Mg, Sr, and Ba are derived from the soil than during colder and wetter climate conditions.
Dry conditions also favor PCP, which enhances the Mg, Sr, and Ba concentrations in the same direction. Significant positive correlations (p < 0.001) of Mg and Ba with the δ 13 C values of stalagmite Zoo-rez-2 indicate that PCP has a minor influence (Fairchild & Treble, 2009;Johnson et al., 2006), due to not very high correlation coefficients (Table 2). For Zoo-rez-1, the correlation between Mg and δ 13 C is weak (p < 0.05), which suggests that the Mg content of this stalagmite is also influenced by another factor. This is probably detrital material as suggested by the weak but significant correlation (p < 0.01) between Al and Mg (Table 2). Due to the smaller partitioning coefficient for Mg in comparison to Sr and Ba, PCP would be most pronounced for Mg. However, it is likely that the lower slopes of the trend lines in Figure 7 are related to changes in the initial concentrations of these elements, which are related to changes in the silicate/carbonate weathering ratio. The two processes, changes in the silicate/carbonate weathering ratio and PCP, cannot be disentangled.
A positive correlation is also observed between δ 18 O values and Mg content for both stalagmites (Table 2). However, these correlations are weak, which shows that PCP and drip rate seem to have a minor effect on these proxies Johnson et al., 2006;Mühlinghaus et al., 2009;Riechelmann et al., 2013). A more likely explanation for the positive correlation between the Mg content and δ 18 O values involves a relation to temperature because δ 18 O values are influenced by mean annual temperature (section 4.3.4) and Mg is related to increased silicate/carbonate weathering ratio during warmer and drier summers, which is recorded at annual resolution in the stalagmite.
Zoo-rez-2 shows significant negative correlations between Mg and Y and Sr and Y, which are not visible for Zoo-rez-1 due to Y and P being strongly affected by detrital material. This negative correlation of Mg and Sr with Y may be explained by warmer and drier conditions inducing an increased silicate/carbonate weathering ratio and also result in a reduction of soil activity (Borsato et al., 2007;Treble et al., 2003;Wassenburg et al., 2012). This would in turn result in lower Y concentrations and explain the negative correlation between Mg and Y ( Table 2).

. Equilibrium Versus Disequilibrium Fractionation During Calcite Precipitation
The significant positive correlation between the δ 13 C and δ 18 O values observed for both stalagmites (Table 2, Figure S3) indicates disequilibrium fractionation during calcite precipitation as suggested by several other studies Fohlmeister et al., 2017;Hendy, 1971;Mattey et al., 2008;Mickler et al., 2004;Mühlinghaus et al., 2009;Riechelmann et al., 2013). The correlation between the δ 13 C and δ 18 O values is higher for Zoo-rez-2 than for Zoo-rez-1, which could indicate stronger disequilibrium fractionation during calcite precipitation for Zoo-rez-2. On the other hand, Zoo-rez-2 shows lower δ 13 C and δ 18 O values than Zoo-rez-1, which suggests stronger disequilibrium fractionation during calcite precipitation for Zoo-rez-1 or more and constant PCP. The signal is transferred relatively fast due to the low rock overburden. This short transfer time is also the reason for the observed annual lamination in the stalagmites (Riechelmann et al., 2019). For speleothem δ 13 C values, the amplitudes are generally higher than for δ 18 O values Scholz et al., 2009), which agrees with the observation for Zoo-rez-1 and -2. Long-term changes in the ventilation of the cave do probably not play a role during the growth of Zoo-rez-1 and -2, because the cave entrance was closed by a door sometime after the 1970s (Lang, 2000). Nowadays, there is a relatively dense beech forest above the cave, which most probably developed during the growth period of the stalagmites. However, we do not see a long-term trend in the δ 13 C values. Therefore, a correlation of these vegetation changes and the δ 13 C values is not clearly visible. However, we could not exclude that vegetation changes influenced the variability in the δ 13 C values of both stalagmites. Furthermore, PCP seems to have a minor influence on the δ 13 C values (section 4.2).  Figure 8). This indicates a pronounced influence of mean annual temperature on the δ 18 O values of the stalagmites. As discussed above, the mean annual temperature signal could still be preserved in the δ 18 O values of the stalagmites in case that the degree of disequilibrium fractionation is constant (Hansen et al., 2019;Mühlinghaus et al., 2009). The other annual proxy records were also tested for correlations with the meteorological data, but no significant correlation coefficients were observed.

δ 18 O Values as Proxy for Mean Annual Temperature
The main source for the δ 18 O signal in the speleothem calcite is the infiltrating precipitation (Lachniet, 2009), which may be influenced by several processes on the way through the soil and karst as discussed above. On the intraannual scale, mean monthly temperature and mean monthly δ 18 O value of precipitation show a high correlation of r = 0.98 (p < 0.001; n = 12; CE 1978-2013; GNIP station Regensburg), which is the expected relation for a midlatitude climate (Rozanski et al., 1993  and mean annual temperature, which is the same value as for mean annual GNIP δ 18 O data correlated with mean annual temperature. For the third regression mean, winter precipitation δ 18 O values were used as δ 18 O values for drip water, and the resulting correlation of the modeled δ 18 O values for calcite with mean annual temperature is r = 0.67 (p < 0.001, n = 36) (detailed description of the calculations, see section 2.7). Comparing the three differently modeled δ 18 O series of calcite, the infiltration-weighted    (Genty et al., 2014;Mischel et al., 2015;Riechelmann et al., 2011Riechelmann et al., , 2017. The model using mean annual δ 18 O values shows a bit higher values (mean value: −6.29 ± 0.62‰) than the model using infiltration-weighted δ 18 O values. Despite a 2σ error for this modeled series of 1.60‰, it indicates that there is a certain loss of precipitation due to evapotranspiration during the summer months ( Figure 9) and that the infiltration-weighted model seems to be most realistic.
Similarly, the slightly increasing trend over time (Figures 8 and 9) is observed in the modeled and the measured δ 18 O values. Furthermore, for Zoo-rez-2, the correlation between the measured and the infiltrationweighted, modeled δ 18 O values is r = 0.43 (p < 0.001, n = 73). After CE 1935, the correlation between mean annual temperature and measured δ 18 O values of Zoo-rez-2 is high, whereas it is low for the older section ( Figure 8b). This may be due to missing or false laminae in this section (Riechelmann et al., 2019). The correlation between mean annual temperature and measured δ 18 O values of Zoo-rez-1 is insignificant (r = 0.08, p > 0.05, n = 92). However, it is obvious from the time series (Figure 8a) that both show a common pattern before CE 1905. This may also be explained by dating uncertainties due to missing or false laminae, which are estimated to be ±7 to 19 years in the lower part of the stalagmites, in addition to the dating uncertainty of the top laminae with ±5 years due to the uncertainty of the detection of the 14 C bomb peak (Riechelmann et al., 2019). In addition to these chronological uncertainties, other factors could have an influence on the δ 18 O values and result in reduced correlations with mean annual temperature as well as the not existent correlation between the δ 18 O values of both stalagmites. These could be changes in the degree of disequilibrium isotope fractionation during precipitation of speleothem calcite, changes in the amount of PCP, or both drip sites could have had different drip rates, as well as differences in the initial drip water chemistry, such as different SI calcite (Riechelmann et al., 2013).
In summary, the overlap with instrumental data provides the opportunity to calibrate measured speleothem δ 18 O values with mean annual temperature and provides a fundament for quantitative temperature reconstruction ( Figure 10). The resulting regression is given by the following equation: T ¼ 0:60 ± 0:15 × δ 18 O calcite þ 12:46 ± 1:02: This describes a potential equation to reconstruct quantitative temperatures from the δ 18 O values of the calcite further back in time. The standard deviation (SD) for this regression is ±0.67, which results in a 2σ error of the reconstructed temperatures of ±1.4°C. This equation can be used for potential speleothem records of the entire Holocene using other stalagmites from Zoolithencave. Nevertheless, these new stalagmites have also to be tested for their response to temperature in the calibration period and by overlapping sections of the stalagmites.

Conclusions
The proxy records of the two stalagmites Zoo-rez-1 and -2 are influenced by several factors, which makes their interpretation in terms of past climate variability challenging. Phosphorus and Y, often interpreted as proxies for vegetation and soil activity, are both also influenced by detrital material. Magnesium, Sr, and Ba are affected by weathering of feldspar and clay minerals in the soil and the resulting changes in the silicate to carbonate weathering ratio and to a lesser extent by PCP, as suggested by the positive correlation of Mg and δ 13 C values. Weathering of silicates in the soil is increased during warmer and drier conditions, which would also increase PCP and shift Mg, Ba, and Sr in the same direction. Therefore, a distinct interpretation of Mg, Ba, and Sr is not possible.
The δ 18 O values of the two speleothems were compared to mean annual temperature, which show a significant correlation of r = 0.43 (p < 0.001, n = 73) for Zoo-rez-2. Geochemistry, Geophysics, Geosystems were modeled using three approaches of the calculation of the potential δ 18 O values of the drip water (infiltration-weighted mean, mean annual precipitation, and mean winter precipitation δ 18 O values) and calculating δ 18 O values of the calcite under equilibrium isotope fractionation conditions. The regressions between the three differently modeled δ 18 O values and the mean annual temperature allow to model the δ 18 O values back in time for the timespan of the two stalagmites. The three differently calculated δ 18 O series show that the infiltration-weighted modeled δ 18 O series correspond best with the measured δ 18 O values of both stalagmites. This indicates that infiltration also takes place during the summer months; however, a pronounced portion of the summer precipitation is lost during evapotranspiration. Differences in the older part of Zoo-rez-2 and in the younger part of Zoo-rez-1 between the measured and infiltration-weighted modeled δ 18 O data as well as with the mean annual temperature are most probably due to dating uncertainties, resulting from missing and/or false laminae. However, also changes in disequilibrium isotope fractionation, PCP, drip rate, and/or drip water chemistry could result in these differences. The δ 18 O values of Zoo-rez-2 show a positive correlation (r = 0.43, p < 0.001, n = 73) with mean annual temperature, which results in a linear regression providing a calibration equation. This would provide the opportunity to calculate quantitative temperatures from stalagmite δ 18 O values. Therefore, the speleothem δ 18 O values can be used as a proxy for mean annual temperature and provides the basis for the interpretation of older stalagmites from Zoolithencave.