Lignin oxidation products as a vegetation proxy in stalagmite and drip water samples from the Herbstlabyrinth , Germany

Here we present the first quantitative record of lignin oxidation products (LOPs) in a Holocene stalagmite from the Herbstlabyrinth Cave in central Germany, as well as LOP results from 16 months of drip water monitoring. Lignin is only produced by vascular plants and is therefore an unambiguous vegetation proxy, which can help to better interpret other vegetation and climate proxies in speleothems. We compared our results with stable isotope and trace element data from the same samples. The drip water monitoring reveals a seasonal pattern of LOPs in a fast drip site with low LOP concentrations in winter 5 and higher LOP concentrations in summer, which is opposite to the behaviour of the drip rate, Mg and PO3− 4 concentrations. In the stalagmite, LOP concentrations are correlated or show a similar behaviour to P, Ba and U concentrations. The LOP ratios C/V and S/V, which are usually used to differentiate between angiosperm and gymnosperm and woody and non-woody lignin sources, are anticorrelated to the LOP concentrations and show a similar behaviour to δC and Mg concentrations. These results highlight the potential of LOPs as a new, highly specific vegetation proxy in speleothems, but also demonstrate current 10 limitations in our understanding of the transport of lignin from the soil into the cave and the speleothems.


Introduction
Speleothems are valuable climate archives, because they can grow continuously for thousands of years and can be dated accurately 500.000years back in time using the 230 Th-U method (Richards and Dorale, 2003;Scholz and Hoffmann, 2008).
Furthermore, the cave provides a preservative environment that protects the recorded chemical proxy signals against outer influences such as light, abrupt changes in temperature and mechanical disturbance.The calcite crystal is able to incorporate and thus preserve not only trace elements and isotopic ratios, but also organic molecules.Until recently, mostly stable isotope ratios and trace elements have been used as climate proxies in speleothems (McDermott, 2004;Fairchild and Treble, 2009).
The methodology is well-established, but the records sometimes cannot be interpreted without doubt, unless other proxies are available for comparison (Lachniet, 2009;Fairchild and Treble, 2009;Mischel et al., 2017;Scholz et al., 2012).Therefore, it Organic matter in speleothems and cave drip water is most often analyzed as total organic carbon (TOC) or by fluorescence spectroscopy (Quiers et al., 2015).Size exclusion chromatography coupled to organic carbon detection (LC-OCD) was applied to divide organic matter in cave drip water in different fractions, such as biopolymers and humic substances (Rutlidge et al., 2015).Only very few molecular organic analytes have been analyzed in speleothems so far.Glycerol dialkyl glycerol tetraethers (GDGTs), which are produced by microorganisms in situ in the cave or the overlaying vadose zone, have been implemented as organic temperature proxies (Blyth et al., 2013(Blyth et al., , 2014)).Lipid biomarkers, such as fatty acids (Bosle, 2014) and especially long chain n-alkanes from plant leave waxes, have been used as vegetation proxies, and there have been approaches to use the chain length distribution of n-alkanes to distinguish between the input of grasses and woody plants (Xie, 2003;Blyth et al., 2007Blyth et al., , 2011)).However, there are uncertainties about the validity of chain length distributions to distinguish between different plant groups (Bush and McInerney, 2013;Blyth et al., 2016).Therefore, a more specific plant biomarker is needed.
Lignin is one of the main constituents of wood and woody plants.It has been widely used as paleo-vegetation proxy in archives such as peat, lake sediments and marine sediment cores (see, for instance, the review by Jex et al., 2014) and as proxy for terrestrial input of plant biomass in natural waters, like rivers and oceans (e.g., Zhang et al., 2013;Standley and Kaplan, 1998;Hernes and Benner, 2002).Blyth and Watson (2009), Blyth et al. (2010) first detected lignin phenols in speleothems, and recently, Heidke et al. (2018) developed a method to quantitatively analyze lignin as a paleo-vegetation proxy in speleothems and cave drip water.The advantage of lignin as a vegetation proxy is that it is only produced by vascular plants and not by microorganisms.Thus, it is very specific.In addition, lignin analysis does not only give information about the abundance, but also about the type of vegetation.The biopolymer lignin consists of three different monomers, coniferyl alcohol, sinapyl alcohol and p-coumaryl alcohol, which are incorporated into the polymer in the form of phenylpropanoids.The proportion of these three monomers varies with the type of vegetation.Lignin from gymnosperm wood consists mainly of coniferyl alcohol, whereas lignin from angiosperm wood contains coniferyl and sinapyl alcohol.Non-woody vegetation, like grasses, leaves and needles, are characterized by a higher proportion of p-coumaryl alcohol.To analyse the lignin composition, the lignin polymer has to be oxidatively degraded into monomeric lignin oxidation products (LOPs).The guaiacyl phenylpropanoid (from coniferyl alcohol) is oxidised to vanillic acid, vanillin and acetovanillone (V-group LOPs), the syringyl phenyl propanoid (from sinapyl alcohol) is oxidised to syringic acid, syringaldehyde and acetosyringone (S-group LOPs), and the p-hydroxyphenyl phenylpropanoid (from p-coumaryl alcohol) is oxidised to p-coumaric acid and ferulic acid (C-group LOPs), but also to p-hydroxybenzoic acid, p-hydroxybenzaldehyde and p-hydroxyacetophenone (H-group LOPs).The latter H-group LOPs can also originate from other sources, like soil microorganisms or the degradation of protein-rich material, and not only from lignin.Therefore, the H-group LOPs are usually not used as vegetation proxies.Usually, the sum parameter Σ8, which is the sum of all eight individual LOPs from the C-, S-and V-group, is used to present the total lignin concentration.In addition, the ratios of the different LOP groups, C/V and S/V, are used to present the type of lignin, where a higher C/V ratio indicates a higher contribution of non-woody vs. woody plant material and a higher S/V ratio indicates a higher contribution of angiosperm vs. gymnosperm plant material (Hedges and Mann, 1979).
Here, we present the first quantitative record of lignin oxidation products from a Holocene stalagmite from the Herbstlabyrinth in central Germany.In addition, we present LOP results from cave drip water from the same cave, sampled monthly

Analytical methods
The methods used for dating of the stalagmite samples, the analysis of trace elements and stable isotopes as well as the calculation of the drip rate for the cave drip water samples and the growth rate of the stalagmite are described in detail in Mischel et al. (2017) and references therein.The method used here for the analysis of lignin oxidation products in speleothems and cave drip water is described in detail in Heidke et al. (2018).In the following paragraph, only a brief description will be given.To compare the LOP results with stalagmite data of stable isotopes and trace elements, which have a much higher resolution of 2 mm per sample, a mean value of the higher-resolution data according to the sample size of the LOP samples was calculated.In the discussion, only these lower-resolution data will be shown and discussed.The original data can be found in Mischel et al. (2017).The standard deviations of the respective mean values were used as error bars for the stalagmite stable isotope and trace element data.For the drip water stable isotope and trace element data, the uncertainty is very small and therefore not shown here, but it is provided in Mischel et al. (2017).For the drip water LOP data, the uncertainty was calculated as described in Heidke et al. (2018), and for the stalagmite LOP data, the calculation of the uncertainty is described in section 3.2.

Analysis of LOPs
The stalagmite samples were cleaned with organic solvents, edged on the outside with diluted HCl, and finally dissolved in 30% HCl.These sample solutions as well as the drip water samples were extracted via solid phase extraction (SPE), and the extracted lignin was degraded via microwave-assisted CuO oxidation.The resulting lignin oxidation products (LOPs) were again extracted via SPE and then analysed via ultrahigh-performance liquid chromatography (UHPLC) coupled to heated electrospray ionisation (HESI) high resolution mass spectrometry (HRMS) using a Dionex Ultimate 3000 UHPLC system and a Q-Exactive Orbitrap mass spectrometer (Thermo Fisher Scientific).Further details of the method can be found in Heidke et al. (2018).

LOPs in drip water samples
In Fig. 1, the LOP results of the fast drip site, D1, the slow drip site, D2, and the cave pool water, PW, are shown.The height of the columns represents the sum of the LOP concentrations, Σ8, and the different colors show the contribution of the C-, S-and V-group LOPs.In general, the V-group LOPs contribute the largest part to the total LOP concentration, followed by the S-group and the C-group LOPs.D1 shows a strong seasonal pattern with high LOP concentrations between 500 and 1800 ng • L −1 from May to August 2014 and from April to August 2015, and low LOP concentrations between 30 and 400 ng • L −1 from September 2014 to February 2015.For July 2014 and March 2015, no data are available.For the slow drip site, D2, the concentrations do not show a seasonal pattern, but are highly variable between 80 and 3000 ng • L −1 .In May, July and August 2014 and June, July and August 2015, there is no data available, mainly because the drip rate was too low to collect a sufficient sample volume.In Fig. 2, the C/V and S/V ratios are shown for D1, D2 and PW.Because concentrations of individual analytes were below the limit of detection, the ratios could not be calculated for every sample.For the different drip sites, the ratios are in a similar range.The ranges of the ratios in D1 are 0.02-0.33 for C/V, 0.14-1.21for S/V, .In D2, the ranges are 0.00-0.20 for C/V, 0.08-0.68 for S/V, .In PW, the ranges are 0.00-0.43 for C/V, 0.18-0.77for S/V.The ratios of vanillic acid to vanillin (Vac/Val) and syringic acid to syringaldehyde (Sac/Sal) are shown in Fig. S1 in the supplementary information (SI).

LOPs in stalagmite samples
One complication of studies with a large number of samples which have to be analyzed over an extended period of time are socalled batch effects, which occur because measurements are affected by laboratory conditions, such as reagent lots, instrumental drift, stability of the reference standards or a varying efficiency of the sample preparation steps (e.g.CuO oxidation step).Since such effects are almost unavoidable, adjustment strategies are generally required (Wehrens et al., 2016;Kirwan et al., 2013;Surowiec et al., 2017), which is especially true when temporal records of the target analytes are the aim of the study.To do so, the stalagmite samples for LOP analysis were analyzed in six batches with nine samples and one blank sample per batch.
To recognize systematic errors leading to batch effects, the samples within each batch were not located side by side in the stalagmite, but evenly distributed over the whole length of the stalagmite.In fact, in the results of the LOP concentrations and the C/V and S/V ratios, we observed a regular pattern, which corresponded to the different batches.To uncover the original signal and overcome the differences in instrumental response between the batches, the individual measurements results were revised based on the following correction procedure.In the oldest part of the stalagmite (11.2-8.6 ka BP), the pattern was most visible and least superimposed by original signals.Therefore, we used only this part to calculate a correction factor using equation ( 1), with x smoothed representing the corrected (smoothed) value of Σ8, C/V or S/V, x representing the original value of Σ8, C/V or S/V, Xbatch k , k = 1, 2, ..., 6, the mean value of all x i in batch k, and X the mean value of all x i of all batches in the oldest part of the stalagmite.The error bars ∆x smoothed for the smoothed values were calculated by equation ( 2), with ∆ Xbatch k and ∆ X representing the standard deviations of the respective mean values and ∆x representing the uncertainty of a single sample analysis as described in Heidke et al. (2018).The smoothed and original data for Σ8 can be compared in Fig. 3, and for C/V and S/V in Fig. 4.
Figure 3 shows the concentrations of the C-, S-and V-group LOPs as well as the Σ8 concentrations, both smoothed and original data for comparison, plotted against the age of the stalagmite.Figure 4 shows the C/V and S/V ratios, both smoothed and original.The ratios of vanillic acid to vanillin (Vac/Val) and syringic acid to syringaldehyde (Sac/Sal) can be seen in Fig.

S2 in the SI (only original data).
The mean Σ8 concentration in the whole stalagmite was 51 ± 15 ng • g −1 .The highest concentrations of up to 92 ng • g −1 occured shortly after the hiatus at the beginning of the middle part at 7.6 ka BP, followed by the lowest concentrations with 19 ng • g −1 at 7.0 ka BP.The mean C/V ratio was 0.32 ± 0.15, with the lowest ratio of 0.12 at 10.9 ka BP and the highest ratios of up to 0.68 at 0.3, 6.5 and 7.2 ka BP.The mean S/V ratio was 2.15 ± 0.71, with the lowest ratio of 1.08 at 11.1 ka BP and the highest ratios of up to 4.18 at 5.1 and 7.1 ka BP.While the sources of lignin are only vascular plants and the source of Mg 2+ is mainly the host rock, PO 3− 4 can originate from both the vegetation and the host rock.Since the seasonal pattern of PO 3− 4 here is similar to that of Mg 2+ and opposed to that of Σ8, we assume that the PO 3− 4 in the drip water also originates from the hostrock.However, it is important to note that this concerns the seasonal timescale only.At longer timescales, as they are recorded in a stalagmite, the release and mobilization of phosphorous from the hostrock can be related to the productivity of the vegetation above the cave, since phosphorous serves as a plant nutrient.For the LOPs, there might be several causes for the observed concentration pattern.One possible reason could be the growth season of the vegetation from spring to autumn and the activity of soil microorganisms during the warmer time of the year, which would lead to a higher abundance of -partially degraded and therefore transportable -lignin in the soil.However, Mischel et al. (2015) assumed that the drip water of the Herbstlabyrinth is fed by a large reservoir in the karst aquifer, where the water has a residence time of approx.10 months and is mixed with the water of 12 months.In addition, the degradation of lignin in the soil takes months to years (Thevenot et al., 2010;Bahri et al., 2006).Therefore, the original seasonality of the input of plant material on the surface will probably be smoothed out by these effects, and it seems more likely that the observed seasonal signal is caused by hydrological effects.A possible reason could be dilution of the -otherwise relatively constantlignin concentration from the reservoir of organic matter in winter, when the recharge of the aquifer and consequently the drip rate are higher.In summer, in contrast, the slow drip rate and lower recharge of the aquifer might cause a higher concentration of the organic matter and therefore the lignin content in the drip water (Fig. 6).A combination of both factors -the acitivity of the vegetation and the soil microorganisms in summer and the dilution effect due to higher recharge of the aquifer in winter -is also possible to explain the observed seasonal concentration pattern of Σ8.
Interestingly, Bosle (2014), who analyzed low-molecular weight saturated fatty acids as biomarkers for microbial activity in drip water samples from the same cave monitoring program, found a similar seasonal trend with high concentrations during the summer months and low concentrations in winter for the longest fatty acid, arachidic acid (C 20 ), but the opposite trend for the shorter chained fatty acids C 12 to C 18 .Their interpretation was that C 20 could be derived from higher plants above the cave, whereas the shorter chained fatty acids could be produced by microorganisms in the cave and the aquifer.This hypothesis seems to be supported by our LOP results.
In Fig. 7, the averaged Σ8 concentrations of the fast drip site, D1, the slow drip site, D2, and the pool water, PW, are compared.On average, the slow drip site D2 has higher Σ8 concentration than the fast drip site D1, and the pool water has the 12  The interpretation of C/V and S/V in respect to seasonal patterns is rather difficult because there are several data points missing during the winter months where the concentration of the V-group LOPs was below the quantification limit and therefore no C/V or S/V ratios could be calculated.Overall, there is no clear seasonal pattern in the C/V and S/V ratios.On the other hand, we did not expect to see a seasonal pattern here, as the overall vegetation mixture above the cave should stay relatively unchanged throughout one year.Of course, the activity of soil microorganisms can change with temperature and moisture conditions, and the productivity of the vegetation varies over the year.However, if we consider the long turnover times of lignin degradation in soil and assume a reservoir of organic matter above the cave, the C/V and S/V ratios should stay relatively constant.In addition, radiocarbon dating studies showed that the organic matter above caves can be a mix of recent material and material that is up to several hundred years old (Trumbore, 2000;Tegen and Dörr, 1996;Fohlmeister et al., 2011).
The only observable pattern is from 04/2015 to 07/2015, where there was a constant decrease in the C/V and S/V ratios.
The same decrease could also be observed in the drip rate, whereas the δ 13 C values show a constant increase in the same time interval (Fig. 8).The increase in δ 13 C values could be related to the decreasing drip rate and consequently a longer residence time of the drops in the cave air resulting in increased degassing (Hansen et al., 2017).An explanation for the decrease of C/V and S/V ratios could possibly be linked to the decreasing drip rate, too.In times of reduced recharge of the aquifer, the transport of the lignin from the upper soil into the cave probably takes longer and involves more phase-changes than in times of higher recharge.According to Hernes et al. (2007) and Hernes et al. (2013), every phase-change that occurs to the lignin on  The C/V and S/V ratios of the drip water samples are lower than the ratios found in the stalagmite.This could be interpreted in different ways: Either there was more input of non-woody vegetation or less input of woody vegetation in former times than now, or the difference is also caused by phase-change fractionation as the incorporation into the calcite of the growing speleothem represents a phase-change, too.To find out more about these effects, more comparative studies of LOPs in speleothems and in cave drip water, and, in addition, in situ experiments in an artificial cave (Hansen et al., 2017;Wiedner et al., 2008;Polag et al., 2010) would be beneficial.

Stalagmite samples
In Fig. 9, the lignin parameters Σ8, C/V and S/V are compared with several trace elements, stable isotopes and the growth rate of the stalagmite.The youngest part of the stalagmite (0-2.5 ka BP) has a lower growth rate and shows a different behaviour of several proxies, e.g.P and δ 13 C, compared to the rest of the stalagmite.Therefore, Mischel et al. (2017)  the stalagmite are shown in Table 1, those for the older part are shown in Table S1 in the SI.In contrast to Mischel et al. (2017), we did not detrend the records before calculating the correlations.
The correlation between Σ8 smoothed and P in the middle part of the stalagmite is R Σ8−P = 0.66 (p < 0.01), and the correlation between Σ8 smoothed and Ba is R Σ8-Ba = 0.60 (p < 0.01).This positive correlation between Σ8 and P in the stalagmite is opposite as in the drip water.This shows that the proxy concentrations in the stalagmite and the drip water are influenced by different mechanisms at different timescales.In the older and younger parts of the stalagmite, there are no significant correlations between P or Ba and Σ8.Uranium shows a similar pattern as P and Ba (Mischel et al., 2017).However, although U is correlated to P and Ba with R U-P = 0.79 and R U-Ba = 0.78, (p < 0.01), there is no significant correlation between U and Σ8.Mischel et al. ( 2017) interpreted P, Ba and U as vegetation proxies, with higher concentrations of these elements indicating a more productive vegetation, coinciding with wetter climate conditions.The correlation of Σ8 as an unambigous vegetation proxy with P, Ba and U confirms this interpretation.
Moreover, there is also a strong positive correlation between Σ8 and the growth rate of the stalagmite, R Σ8-growth rate = 0.79 (p < 0.01).This can be seen especially at the beginning of the middle part (around 8.0-7.5 ka BP, see Fig. 9), where a peak in the growth rate is accompanied by a peak in Σ8 and also in P, Ba, and U concentrations.In the older and younger parts of the stalagmite, there is no such correlation.A higher growth rate can be caused, inter alia, by a higher drip rate or by a higher supersaturation of the drip water with respect to CaCO 3 (Dreybrodt and Scholz, 2011;Fairchild and Baker, 2012).The latter, in turn, can be caused by a higher soil pCO 2 due to a more productive vegetation and higher microbial activity in the soil, which would be in accordance with the observed higher concentrations of all vegetation proxies.On the other hand, this peak in the growth rate and the vegetation proxies occurs shortly after a hiatus.Therefore, it is also possible that a change in the flow path of the drip water feeding the stalagmite was responsible for the faster growth of the speleothem and a higher input of organic matter.
The C/V and S/V records have a distinct positive peak at around 7.2 ka BP, and a second, less distinct, positive peak at 6.5 ka BP. Figure 9 shows that at approximately the same time, the concentrations of Σ8 and P are lower or even show negative peaks, indicating less input of organic material.This is also reflected in the negative correlations of C/V and S/V with  (Fairchild and Baker, 2012).Consequently, these peaks can be interpreted as drier periods with less input of organic material.One possible interpretation of the C/V and S/V records is therefore that at drier times with lower lignin input, there was less woody and more non-woody plant material available as a lignin source and the vegetation possibly consisted more of grasses and shrubs.Vice versa, at times with generally higher lignin input, more woody material was the lignin source and the vegetation probably consisted more of forest.Another possible interpretation, which would also be in accordance with the other proxies, would be that during drier conditions in the aquifer, the lignin was transported into the cave  more slowly and therefore had more time to undergo phase-changes on the way, such as adsorbtion and desorbtion to mineral surfaces.Such phase-changes can also lead to a change in the C/V and S/V ratios (Hernes et al., 2007).
The pollen-based climate reconstruction from the Meerfelder Maar and the Holzmaar in the Eifel region by Litt et al. (2009) shows a relatively stable vegetation development without any abrupt changes, at least not at the times observed in the C/V and S/V record (6.6 and 7.2 ka BP).This could be interpreted in different ways.One interpretation is that the assumed vegetation changes were local to the region of the the Herbstlabyrinth and did not affect the Eifel region.Another interpretation could be that our proxies are more sensitive to vegetation changes than the pollen record.And third, it is possible that the changes in the vegetation proxies and C/V and S/V were rather caused by changes in the flow paths of the drip water than by changes in the overlaying vegetation.However, Litt et al. (2009) state that their reconstruction method rather underestimates than overestimates climate changes.
In Fig. 10, the principal component analysis of all proxies in the middle and older part of the stalagmite is shown with loadings and scores for the first and second principal components.The first principal component (PC1) explains 31.6% of the overall variance and the second component (PC2) explains 26.9%.Together, these two PCs explain 58.5% of the overall variance.The kink in the scree plot (Fig. S3 in the SI) appears after the fourth PC with PC3 contributing 18.8% and PC4 10.21%, summing up to 87.4% of the variance explained by the first four principal components.However, PC3 and PC4 (Fig. S4 in the SI) are not straightforward to interpret and will not be discussed here.PC1 is composed of Sr and Mg with negative loadings and U, P, Ba, C/V and S/V with positive loadings.It can be inferred from the scores of the individual samples that the influence of Sr and Mg is mainly dominant in the older part of the stalagmite.This long-term decrease of Sr and Mg (see Fig. 9) was interpreted by Mischel et al. (2017) as the result of a thin loess cover, deposited during the last Glacial, being progressively leached at the beginning of the Holocene.PC2 is composed of Σ8, Ba and P with positive loadings and C/V and S/V with negative loadings.This can be interpreted as the amount of plant organic matter being flushed into the cave.
Overall, we conclude that samples with low lignin content are accompanied by low P, Ba and U concentrations and high C/V and S/V values and vice versa.This confirms P, Ba and U as vegetation proxies, and, in addition, indicates that at times with lower input of organic matter, the composition of the organic matter was different, too.The concurrence of more positive δ 13 C values and higher Mg concentrations with times of lower vegetational input suggests drier conditions (Mischel et al., 2017), which could have lead to a reduced vegetation cover consisting more of grass than of forest.This highlights the large potential of the lignin proxies for reconstruction of vegetation.However, we cannot say with certainty if the observed changes in the lignin composition result directly from a change in the vegetation cover or from different transport processes in the soil and aquifer.Another example for this ambiguity is the observation that a higher growth rate after the hiatus, at 7.7-7.3ka BP, is accompanied by higher concentrations of Σ8, P, Ba and U.This could either indicate a wetter climate with more recharge and a more productive vegetation at that time, or a different flow path of the drip water that lead to both an increased growth rate of the stalagmite and an increased input of organic matter.

Conclusions and outlook
The total lignin concentration, Σ8, seems to be a reliable proxy for the input of plant material into the stalagmites.The Σ8 records fits well with P as an established vegetation proxy for speleothems from the Herbstlabyrinth and also correlates with Ba and U, which also have been interpreted as vegetation proxies by Mischel et al. (2017).The clear benefit of Σ8 compared to these trace elements is, that the sources of lignin are exclusively higher plants and not, for example, microorganisms or the 5 host rock.Therefore, Σ8 can help to better interpret potential vegetation proxies whose sources are less clear.
The interpretation of the lignin parameters C/V and S/V seems to be more complicated, as these ratios might be influenced not only by the type of the overlaying vegetation, but also by phase-change fractionation due to adsorption and desorption processes on the way from the soil into the cave, which can differ for different flow paths.Greater changes in the vegetation, such as longer periods of deforestation, will probably still cause changes in the C/V and S/V ratios, but more research is needed should be carried out.In addition to stalagmites, flowstones could be valuable sample material because they are often fed by water flows with higher discharge carrying larger amounts of organic material.
Clim.Past Discuss., https://doi.org/10.5194/cp-2019-5Manuscript under review for journal Clim.Past Discussion started: 15 January 2019 c Author(s) 2019.CC BY 4.0 License.In the pool water, PW, the LOP concentrations show a similar seasonal pattern as in the fast drip site, but less pronounced with concentrations between 140 and 650 ng • L −1 from September 2014 to Febrary 2015 and between 440 and 1100 ng • L −1 in the other months, with one exception in April 2015 with more than 2500 ng • L −1 .

Figure 2 .
Figure 2. C/V and S/V ratios in the drip water.

Figure 3 .
Figure 3. Concentrations of the C-, S-and V-group LOPs and Σ8 concentrations plotted against the age of the stalagmite.The uppermost plot shows the smoothed results of Σ8 concentrations.

Figure 4 .
Figure 4. Ratios of C/V (red) and S/V (blue), original data and smoothed data, plotted against the age of the stalagmite.

Figure 6 .
Figure 6.Comparison of Σ8 concentrations and drip rate in summer and winter.

Clim.
Past Discuss., https://doi.org/10.5194/cp-2019-5Manuscript under review for journal Clim.Past Discussion started: 15 January 2019 c Author(s) 2019.CC BY 4.0 License.lowest concentrations.This finding could be explained by dilution effects as discussed above, but also by different reservoirs feeding the different drip sites.

Figure 7 .
Figure 7.Comparison of average Σ8 concentrations of the fast drip site, D1, the slow drip site, D2 and the poolwater, PW.
Clim.Past Discuss., https://doi.org/10.5194/cp-2019-5Manuscript under review for journal Clim.Past Discussion started: 15 January 2019 c Author(s) 2019.CC BY 4.0 License.itsway from the plant litter to the deeper soil (and into the aquifer and the cave), for example from bound in plant material to dissolved in water and from adsorbed onto mineral surfaces to desorbed, can lead to a fractionation and therefore to a change in the ratios of C/V, S/V and Acid/Aldehyde.

Figure 8 .
Figure 8.Comparison of C/V and S/V ratios with the drip rate and δ 13 C values in the fast drip site D1.
discussed this part of the stalagmite separately from the two other parts.As the LOP record of the youngest part only consists of five samples, we decided to focus the discussion on the rest of the stalagmite only and did not consider the first five data points for the calculation of correlation coefficients and the principal component analysis.All correlation coefficients for the middle part of Clim.Past Discuss., https://doi.org/10.5194/cp-2019-5Manuscript under review for journal Clim.Past Discussion started: 15 January 2019 c Author(s) 2019.CC BY 4.0 License.

Σ8
and P in the middle part of the stalagmite, R Σ8-C/V = −0.51(p < 0.05), R Σ8-S/V = −0.54(p < 0.05), R P-C/V = −0.48,(p < 0.05) and R P-S/V = −0.53,(p < 0.05).Furthermore, the records of δ 13 C and Mg show positive peaks at approximately 7.2 and 6.6 ka BP, although there are no significant correlations of C/V or S/V with δ 13 C or Mg.Higher concentrations of Mg coinciding with more positive δ 13 C values in speleothems are often associated with prior calcite precipitation and usually indicate drier conditions Discuss., https://doi.org/10.5194/cp-2019-5Manuscript under review for journal Clim.Past Discussion started: 15 January 2019 c Author(s) 2019.CC BY 4.0 License.

Figure 10 .
Figure 10.Principal component analysis with the variables Σ8, C/V, S/V, trace elements, growth rate and stable isotopes.The red dots are the scores of the individual samples, labeled with their age in ka BP.

10
to understand all possible influences.In future studies, more speleothem samples from different vegetation and climate zones should be analyzed to study the relation of vegetation types above the cave and LOP ratios found in the speleothems.To get more insight into possible fractionation processes occurring on the way from the soil to the cave, comparative studies of LOPs in soil, drip water and speleothems Clim.Past Discuss., https://doi.org/10.5194/cp-2019-5Manuscript under review for journal Clim.Past Discussion started: 15 January 2019 c Author(s) 2019.CC BY 4.0 License.

Table 1 .
Correlation coefficients R for the middle part of the stalagmite.