We are grateful to anonymous referee #1 for this comprehensive and relevant feedback on our preprint. We address all the comments and propose a point-by-point response, starting with the major comments and then the minor ones. We would like to submit a revised version of the paper after receiving feedback from all referees.

Major comments

Outlier statistic test

Referee comment: 

“For the identification of drought-induced anomalies, the authors used a Grubbs test for outlier detection and characterized any outliers during the drought period as a significant drought effect. However, the Grubbs test relies on roughly normally distributed data, while the normality of the data was not tested, and I would doubt that the concentrations are normally distributed. In stream water concentration, assuming a log-normal distribution often gives reasonable results, meaning concentrations need to be log-transformed before applying the Grubbs test. Second, no outliers in the rest of the time series have been shown in the study. Therefore, it remains unclear whether a similar number of outliers occurred during normal conditions, less intense droughts, or wet periods. Especially in combination with the unjustified assumption of normality, it might be that too many outliers have been identified that are not that exceptional after all. It also makes it hard to say for sure if post-drought conditions differ from any other rain event.”

Thank you, you are perfectly right. We propose to change section 2.5 (lines 174-175) by the following method description, to adjust some sentences accordingly to the new statistical analysis, and to add the figure S2 to supplementary material.

After performing a Shapiro-Wilk test (Shapiro test function in the rstatix library) to verify whether or not the data are normally distributed, it appears that the time series of concentrations in soil solutions do not follow a normal distribution (except for Na+, Si, and pH), even after log-transformation. So, Grubbs test cannot be applied. To find the outlier points using the same method for all types of distribution, we propose a simple and robust method based on interquartile range (IQR) with a visualisation by boxplots.

L174-175:

Previous text: 

To highlight outlier concentrations, a Grubbs statistical test is performed on the entire 2015-2023 period for each parameter, using the ‘outliers’ package on R (Komsta, 2022).

Replaced by: 

To find outlier points, we propose a robust method based on the calculation of the first and the third quartiles (Q1 and Q3) and the interquartile range (IQR = Q3 - Q1). It consists of determining an interval with the lower bound (Q1 – 3 x IQR) and the upper bound (Q3 + 3 x IQR), whereby points outside this interval are considered outliers. The value of 3 is chosen to identify only extreme outliers (Schwertman et al., 2004; Grunsky, 2010). A box plot representation allows outliers to be visualized as points outside this interval (Figure S2). This representation, with the points corresponding to those for summer 2022 (June to September 2022) shown with red crosses, allows to visualize the anomaly at 30 cm during and after the 2022 drought.

We also propose to correct Table 1 as follows to show the new statistical method for determining outliers and to mention other points that may be considered as outliers.

Table 1:  Mean and median of pH, Conductivity (Cond. in μS/cm), Dissolved Organic Carbon (DOC in ppm-C), and concentrations (mmol/L) in ions (Na⁺, K⁺, Mg²⁺, Ca²⁺, Al, Si, NH₄⁺ and NO₃^-) for the 2015-2023 period in the soil solutions at 30 cm depth. Q₃ is 75th percentile, IQR the interquartile range and (Q₃ + 3 x IQR) the upper bond above which points are considered outliers. Values of outliers at 30cm depth are given in the fifth line with the corresponding date in the sixth line. ‘X’ means that no outlier is found at 30cm depth during the 2015-2023 period. “<d.l.” means below detection limit.

This new methodology implies different modifications in the Result & Discussion part. All references to the Grubs test have been modified to correspond to this new test.

L225-227: 

Previous text: 

The time series exhibits annual seasonality in the chemical signal, as well as significant anomalies (Grubbs test p-value < 10-2; Table 1) during the exceptionally severe drought of summer 2022 (Fig. 3; Table 1).

Replaced by: 

The time series exhibits annual seasonality in the chemical signal, as well as significant anomalies during the exceptionally severe drought of summer 2022 with numerous outlier concentrations (Fig. 3; Table 1; Fig S2).

L245-247

Previous text: 

These values are well above the inter-annual variability and are the highest ever registered at this depth, considered as outliers by Grubbs tests with p-value about 10-7 and 10-4 respectively (Table 1).

Replaced by: 

These values are well above the inter-annual variability and are the highest ever registered at this depth, considered as outliers because they are above Q3+3*IQR (Table 1; Fig S2).

L251-252

Previous text: 

The peaks are confirmed by the Grubb tests, conducted at 30cm depth over the period 2015-2023, meaning that these maximum concentrations are outliers (Table 1). Focusing on major cations at 30cm depth, the K+, Ca2+, Mg2+ and Al concentrations increase twice successively between 21 June 2022 and 30 August 2022. The highest concentrations ever registered during 2015-2023 period in 30 cm depth soil solution were reached for the second peak (Table 1).

Replaced by: 

Those peaks are confirmed as outlier concentrations for K+, Ca2+, Mg2+, Al, and NO3- on the 30 August 2022 (Table 1). Focusing on major cations at 30cm depth, the K+, Ca2+, Mg2+ and Al concentrations increase twice successively between 21 June 2022 and 30 August 2022. K+, Ca2+, NH4+ also admit outlier concentrations on the 21 June 2022 (Table 1; Fig S2). The highest concentrations ever registered during 2015-2023 period in 30 cm depth soil solution were reached for the second peak on the 30 August 2022 (Fig. 3, Table 1). 

The table 2 must also be changed with the new method of outlier determination.

Table 2: Mean, median of HIX, FI and BIX. Q1 is the 25th percentile, IQR the interquartile range, (Q1 – 3*IQR) the lower bond bellow which points are considered outliers. Values of outliers at 30cm depth are given in the fourth line with the corresponding date in the fifth line. ‘X’ means that no outlier is found at 30cm depth during the 2015-2023 period.

L286-287

Previous text: 

… with an FI of 1.32, considered as an outlier by the Grubbs test (p-value = 4.8e-3), compared with a mean value of 1.74 ± 0.14 at this depth (Fig. 5; Table 2).

Replaced by: 

with an FI of 1.32 on the 21 June 2022, considered as an outlier because the value is bellow (Q1 – 3*IQR = 1.42; Table 2).

L347-348

Previous text: 

In addition, the Grubbs test also confirms that Na and Si are the only elements that do not show abnormal values (outliers) during the drought of 2022 (Table 1). 

Replaced by: 

X (Deletion of the sentence)

Writing 

Referee comment: 

“The writing is generally acceptable, but before publication in BG, additional improvement in style and grammar is needed. This also includes checking the tenses, which sometimes switch from past to present and future for no apparent reason. Additionally, the separation between the Chapters is not always adequate: Some parts of the results already include discussion points, but mostly, there are many new methods and results at the start of the discussion that do not belong there.”

Once the revision process is complete, we will proofread the new manuscript before submitting the revised version. We will improve the writing of the paper by checking the grammar and phrasing and paying particular attention to the use of tenses. The corrected version will fully meet the journal’s language requirements.

Minor comments

L1: 

Referee comment: 

“The title could be misinterpreted as a number of 2022 drought consequences. To clarify, I would rephrase it to something like “Consequences of the 2022 drought …”

Previous text: 

2022 drought consequences on nutrient dynamics in forest soil solutions of a declining spruces plot in the Strengbach catchment (Vosges Mountains, France)

Replaced by: 

Consequences of the 2022 drought on nutrient dynamics in forest soil solutions of a declining spruces plot in the Strengbach catchment (Vosges Mountains, France).

L13: 

Referee comment: 

“The same here and elsewhere. I would avoid numbers at the start of a sentence. One could start with “The year 2022 …”, for example.”

Previous text: 

2022 was the hottest and driest year ever recorded in France,…

Replaced by: 

The year 2022 was the hottest and driest year ever recorded in France, …

L18-19:

Referee comment: 

“When first reading this sentence, I was not sure what this was supposed to tell me. Is that work done by others, or in this study? What is the interdisciplinary part here? It becomes a little clearer later, but I recommend sticking more to the active form to make clear that this was done as part of this manuscript.”

Previous text: 

Thanks to interdisciplinary monitoring of soil solution chemistry (in particular, the combination of analytical tools for inorganic and organic chemistry), the impacts of drought on biogeochemical processes—and more broadly, on forest soil fertility—are now better understood.

Replaced by: 

Through interdisciplinary monitoring of soil solution chemistry, we gain a better understanding of the impacts of drought on biogeochemical processes—and, more broadly, on forest soil fertility.

L53-57: 

Referee comment: 

“That is not restricted to the US, but occurs massively across Europe as well. Hartmann et al. (2022) documented elevated tree mortality globally.”

Previous text: 

In addition, global warming and frequent droughts have accelerated the spread and intensity of insect attacks, as documented in US forests (Weed et al., 2013; Vose et al., 2016; Frank, 2021).

Replaced by: 

In addition, global warming and frequent droughts have accelerated the spread and intensity of insect attacks, as documented in US forests (Weed et al., 2013; Vose et al., 2016; Frank, 2021) and in Europe (Hartmann et al. 2022).

L57-59:

Referee comment: 

“I suggest differentiating that this is especially true when spruce is growing out of its natural distribution range.”; “Species names need to be in italics”

Previous text: 

Spruces are boreal that are particularly vulnerable to drought, leading to an increase in the tree's vulnerability to parasites, such as the bark beetle (Ips typographus) which devastates Vosges spruce forests (Saintonge, 2022; Gomez et al., 2023; Knutzen et al., 2025).

Replaced by: 

Spruces are trees from boreal region. Out of its natural localisation, spruces can be more affected by droughts, leading to an increase in the tree's vulnerability to parasites, such as the bark beetle (Ips typographus), which devastates Vosges spruce forests (Saintonge, 2022; Gomez et al., 2023; Knutzen et al., 2025).

L65: 

Referee comment:

“In which way? Does high nutrient availability make them more vulnerable (as is the case for plants that are over-fertilized), or does a lack of sufficient nutrients weaken the trees?”

Previous text: 

Another factor of tree vulnerability to drought is nutrient availability

Replaced by: 

Another factor of tree vulnerability to drought may be a reduce of nutrient availability 

L68-69: 

Referee comment: 

“I suggest adding Winter et al. (2025) here.”

Previous text: 

Nitrogen is an essential plant nutrient, and the consequences of drought on the nitrogen cycle are documented (Lamersdorf et al., 1998; Muhr et al., 2008; Deng et al., 2021; Krüger et al., 2021; Winter et al., 2023).

Replaced by: 

Nitrogen is an essential plant nutrient, and the consequences of drought on the nitrogen cycle are documented (Lamersdorf et al., 1998; Muhr et al., 2008; Deng et al., 2021; Krüger et al., 2021; Winter et al., 2023, 2025).

L79-82: 

Referee comment:

“If I understood it right, diverse forest types are not covered in this study. Hence, it does not go well with the introduction to this work, but would rather fit into the discussion.” 

Previous text: 

However, studies addressing the impact of natural drought on mineral nutrient dynamics in soil solution remain scarce and require further extension to diverse forest types.

Replaced by: 

However, studies addressing the impact of natural drought on mineral nutrient dynamics in soil solution remain scarce and require multiple observations to be better understood.

L115: 

Referee comment:

“This needs to be specified. How did the rainfall distribution change?”

Previous text: 

But we can already observe some climate change consequences at the local scale in the OHGE, so that the OHGE databank is a key tool to understand environmental change consequences (Pierret et al., 2018; Strohmenger et al., 2022).

Replaced by: 

But we can already observe some climate change consequences at the local scale in the OHGE, with wetter summers and winters, and dryer autumns comparing before and after 2006 (Strohmenger et al., 2022). The OHGE databank is a key tool to understand environmental change consequences (Pierret et al., 2018; Strohmenger et al., 2022).

L163-171: 

Referee comment:

“This needs more explanation of the model. What type of model is that? It also requires some discussion on the model's uncertainty in the discussion section, which is currently missing.”

To address the reviewer’s comment, we provide additional details about the model; these should be sufficient to understand the approach. However, since the model does not allow for a quantitative estimation of water flow and transport of associated nutrients, a discussion of model uncertainties or sensitivity would be beyond the objectives of our study.

New section 2.4:

2.4 Hydrological soil modelling

Drought events can create variations in soil water content. To evaluate changes in water availability in the soil profile induced by rainfall decrease, an in-house daily water balance calculation code is applied to a soil profile under spruces. The BILHYDAY code, whose structure is inspired by the model BILJOU developed by Granier et al. (1999), takes meteorological data as inputs with stand parameters (leaf area index, soil porosity and residual water content). Based on a conceptual approach, the model operates on a daily time step to describe the hydrological processes of the soil–tree–atmosphere continuum. Potential evapotranspiration is estimated following the Penman formulation. Precipitation feeds a canopy reservoir that simulates interception up to a threshold capacity, while transpiration is represented as a function of potential evapotranspiration, soil water storage, and leaf area index. Transpiration is distributed among soil layers according to the root density profile. Soil evaporation is estimated from the incident energy (induced from direct radiation) and an extinction coefficient. Five soil layers are represented as cascading reservoirs that drain sequentially from top to bottom according to their field capacity. The uppermost layer is supplied by non-intercepted precipitation and is also subject to evaporation losses. Root water uptake associated with transpiration is distributed among the different soil layers according to the root density profile. The BILHYDAY model is used qualitatively as complementary information to support the interpretation of geochemical measurements of soil solutions. Model outputs are corroborated by soil moisture measurements, which are not shown here since they are not directly co-located. The relative extractable soil water is calculated similarly to Granier et al. (1999) and can be used to assess water stress. Although the model does not simulate photosynthesis by integrating a dynamic stomatal conductance, transpiration is nevertheless adjusted according to soil water availability, with regulation occurring below a threshold of 0.4, as proposed by Bréda et al. (2006).

L175-178 : 

Referee comment: 

“NMDS needs further explanation.”

Previous text: 

Multivariate statistical analysis helps to highlight atypical observations in long-term monitoring and can reveal the impact of extreme events on soil when many parameters are measured (Knight et al., 2024). A non-metric multidimensional scaling (NMDS) statistical analysis was conducted using the metaMDS function in the ‘vegan’ library in R (Oksanen et al., 2001).

Replaced by: 

Non-metric multidimensional scaling (NMDS) analysis helps to highlight atypical observations in long-term monitoring and can reveal the impact of extreme events on soil when many parameters are measured (Knight et al., 2024) or the consequences of extreme meteorological events on hydrochemistry (Setia et al., 2021). We ran NMDS statistical analysis using the metaMDS function in the ‘vegan’ library in R (Oksanen et al., 2001). NMDS were carried out on the dataset of chemical measurements (pH, conductivity, Na+, K+, Mg2+, Ca2+, Al, Si, NH4+, NO3-, DOC) and during the period 2015-2023, separating points by depth of sampling (5, 10, 30 and 60 cm) to visualise anomalies at each depth. Plotting the coordinates of different soil solution samples highlights atypical points. The scores assigned to the chemical parameters provide information on which parameters distinguish the samples from one another and which ones explain the anomalies observed.

L184:

Referee comment:

“1 September – 31 August is not a typical hydrological year!”; A figure illustration how anomalous the drought was would be a great benefit here or in the SI. For example, one could show precipitation vs. temperature anomalies. That would also better characterize the drought, rather than being restricted to information about lower precipitation.”

We Update values with hydrological year from 1 October to 30 September and we propose a new first paragraph for the section 3.1. as follow: 

On the OHGE site, considering the hydrological years over the recorded period from 1987 to 2023 (year running Y from 1 October to 30 September of year Y+1), 2021/2022 (including drought 2022) was the third driest year on record (after 1990 and 1995) with a cumulated precipitation amount of 970 mm (-20% compared to the average value over the period), the fourth warmest year with an average annual temperature of 7.47°C (+20% compared to the average for the period – just after 2019, 2023 and 2022), and with the 2nd highest number of days with an average daily temperature above 18°C, i.e. 39 days compared to an average of 17 days. Among the years with the least precipitation, 2021/2022 was the warmest year, which in combination makes 2021/2022 the most drought-affected year. 

The longest period of summer meteorological drought observed between 1987 and 2024 was 19 days in 2018, and then 17 days in 2022, 2013 and 1990. Nevertheless, the summer of 2022 was exceptional because this first drought period from 2 July to 19 July was very quickly followed by a second period of 24 days during which daily rainfall did not exceed 0.7 mm (between 21 July and 13 August 2022; Fig 2-A). It is important to notice that a rainfall of 8.9 mm, on 20 July 2022, separates those two meteorological drought events (Fig. 2.A). The summer months of July and August, with only 61 mm of rainfall, were the driest ever recorded at the site (Fig. S1), representing a decrease of nearly 70% relative to normal conditions. The particularly dry July month (10.2 mm cumulated; Fig. S1; Fig 2A) was amplified by the previous dry March and May months with respectively 33 mm and 36 mm cumulated, compared with the 102 mm/month usually observed on the Strengbach watershed. 

We provide several graphs of precipitation, temperature and precipitation versus temperature anomalies in SI (Fig. S1). You’re right that these can help illustrate the significance of the 2022 drought.

L198: 

Referee comment: 

“I suggest not using an abbreviation for water content; it is not an especially complicated word.”

Previous text: 

…the average water content (WC) in each layer is simulated. The WC of the three intermediate layers…

Replaced by: 

…the average water content in each layer is simulated. The water content of the three intermediate layers…

L213:

Referee comment: 

“due to stomatal closure or downregulation”

Previous text: 

…tree transpiration reaches a very low value due to stomatal regulation, and this continues until rain returns.

Replaced by: 

…tree transpiration reaches a very low value due to stomatal regulation or downregulation, and this continues until rain returns.

L222: 

Referee comment:

 “The background does not look orange to me, but rather light red.”

Previous text: 

The orange background indicates the meteorological drought period,

Replaced by: 

The red background indicates the meteorological drought period,

L273: 

Referee comment: 

“Isn’t that discussion already?”

Previous text: 

Throughfall of 16 August 2022 (red cross on Fig. 4) corresponds to the lixiviation of needle depositions after the dry period (Fig. 2-A).

Replaced by: 

Throughfall of 16 August 2022 (red cross on Fig. 4) corresponds to the end of the dry period with the return of rainfall on the 14 August 2022 (Fig. 2-A).

L295-340:

Referee comment:

“A lot of this is what I would classify as methods and results, not discussion.”

Previous text: 

In order to gain an overall view of the data, and visualise anomalies at each depth, multivariate NMDS statistical analyses were carried out on the dataset of chemical measurements (pH, conductivity, Na+, K+, Mg2+, Ca2+, Al, Si, NH4+, NO3-, DOC) and during the time period 2015-2023, separating points by depth of sampling (5, 10, 30 and 60 cm). Soil solution samples from the 2022 drought (from June to November) have a statistically distinct signature only at 30 cm depth (Fig. 6).

Replaced by: 

NMDS plots at each depth (Fig. 6) show that soil solution samples from the 2022 drought (from June to November) have a statistically distinct signature only at 30 cm depth (Fig. 6).

We have developed the section on materials and methods for NMDS analysis in greater detail and have therefore removed this information from section 4.1. However, NMDS analyses are data processing and their description is already part of the interpretation of the data. Thus, we prefer to leave this part to the discussion and not to the results.

L344: 

Referee comment:

“ How is significance defined here?”

Previous text: 

… ,which is within the range of inter-annual variability with a mean concentration during 2015-2023 period of 0.070 ± 0.019 mmol/L and 0.066 ± 0.032 mmol/L for Na and Si respectively (Table 1).

Replaced by: 

… ,which is within the range of inter-annual variability with a mean concentration during 2015-2023 period of 0.066 ± 0.025 (median ± IQR) mmol/L and 0.061 ± 0.046 mmol/L for Na and Si respectively (Table 1).

L365-367 : 

Referee question: 

“At what time scale would weathering be expected to play a role?”

Belyazid et al. (2022), with numerical modelling, describe that weathering rate varies considerably within the year. Weathering is expected to play a role at the seasonal scale. But depending on the element studied, the mineral weathering may be (or not) a dominant flux that will influence significantly the composition of soil solution. For instance, in the soil of the Strengbach catchment, the silicium is almost completely controlled by the mineral weathering, as it is in negligeable concentration in throughfall and litter (Pierret et al., 2018; Oursin et al., 2023). A decrease of weathering would be expected to play an immediate role in silicium concentration in soil solution. In another hand, an element such as calcium, is mainly provided by litter degradation and throughfall (Beaulieu et al. 2020). A decrease in mineral alteration should not have a significant impact on the calcium concentration of the soil solution.

L485:

Referee question: 

“Was there even enough soil water to sample during the drought?”

Yes, even though some samples had low volumes, particularly on 20 July 2022 and 30 August 2022. This is possible thanks to the highly draining soil of the Strengbach watershed.

L525:

Referee question:

“ I assume this is a fast recovery? How does this result align with the discussion on accumulated material flushed with rewetting?”

The flushing of accumulated material results in highly concentrated soil solution (nutrients and OM), with relatively high temperature and sufficient water content. That is an ideal medium for microbial (or fungal) recovery. For instance, the high ammonium concentration, higher pH, relatively high temperature and sufficient water content promotes growth of nitrifiers bacteria after rewetting as observed by Krüger et al. (2021).  

L580 : 

Previous text: 

Figure 6

Replaced by: 

Figure 7

CLL :

Referee comment: 

“Some context on the impact of drought on water quality as an ecosystem service provided by healthy forests would broaden the scope of this conclusion, which is, so far, a little narrow on forest management.”

Previous text: 

Forest management must take this into account to achieve more sustainable forests.

Replaced by: 

Forest management must take this into account to achieve more sustainable forests and preserving the ecosystem services provided by forests, such as carbon sinks, biodiversity preservation, island of coolness that help mitigate heatwaves and protection of water quality in mountain areas.  

L625 

Referee comment: 

“Then the data should be made available elsewhere. “Not yet available” is not an argument to bypass open science.”

Previous text: 

Data concerning soil solutions are not yet available online. However, weather data and data on the chemistry of throughfall are available on the BDOH platform: https://bd-ohge.unistra.fr/OHGE/

Replaced by: 

Weather data and data on the chemistry of throughfall are available on the BDOH platform: https://bd-ohge.unistra.fr/OHGE/. Data concerning soil solution are provided in the supplementary information (Table S1).  

We supply a data table showing time series of concentrations in soil solutions (Table S1) .

References to add: 

Grunsky, E. C.: The interpretation of geochemical survey data, GEEA, 10, 27–74, https://doi.org/10.1144/1467-7873/09-210, 2010.

Hartmann, H., Bastos, A., Das, A. J., Esquivel-Muelbert, A., Hammond, W. M., Martínez-Vilalta, J., McDowell, N. G., Powers, J. S., Pugh, T. A. M., Ruthrof, K. X., and Allen, C. D.: Climate Change Risks to Global Forest Health: Emergence of Unexpected Events of Elevated Tree Mortality Worldwide, Annual Review of Plant Biology, 73, 673–702, https://doi.org/10.1146/annurev-arplant-102820-012804, 2022.

Schwertman, N. C., Owens, M. A., and Adnan, R.: A simple more general boxplot method for identifying outliers, Computational Statistics & Data Analysis, 47, 165–174, https://doi.org/10.1016/j.csda.2003.10.012, 2004.

Setia, R., Lamba, S., Chander, S., Kumar, V., Singh, R., Litoria, P. K., Singh, R. P., and Pateriya, B.: Spatio-temporal variations in water quality, hydrochemistry and its controlling factors in a perennial river in India, Appl Water Sci, 11, 169, https://doi.org/10.1007/s13201-021-01504-3, 2021. 

Winter, C., Müller, S., Kattenborn, T., Stahl, K., Szillat, K., Weiler, M., and Schnabel, F.: Forest Dieback in Drinking Water Protection Areas—A Hidden Threat to Water Quality, Earth’s Future, 13, e2025EF006078, https://doi.org/10.1029/2025EF006078, 2025.

We thank anonymous referee #2 for this relevant review. We propose responding to all comments one by one and then suggesting corresponding changes to the paper.