Comment 1: Firstly there were few replicates. No repetitions for the control soil and organic amended soil, and only duplicates for the lime amended soil (with and without organic amendment). Soil columns were sliced after the end of the experiments, but apparently slices were analysed without dividing into subsamples. Surely there was enough material to do so, thereby allowing repetitions and a better appreciation of variability.

Reply 1: Indeed, we agree that more column replicates per treatment would have enhanced the statistical power of the experiment. Unfortunately, the column set-up did not allow for this in practice because this set-up is a rather unique one for unsaturated flow using the suction plates. The standard column set-up with free drainage is unsuitable, leading to water-saturated conditions at the outlet and local pH increases due to anaerobic conditions. Even though that artefact is straightforward, it is still overly used, and the suction plate method is a much better alternative, with the drawback of limited replication and statistical power. In a revised version, we will reference the artefacts in other studies to justify this set-up. Despite this, we proved to find significant differences between the lime & lime+OA treatments (both in duplicate).  Apart from that, there were indeed no analytical replicates within one slice of a single column. Since we were mainly interested in the depth profile of the soil pH and DOC concentration, we chose to sample one representative composite sample rather than a few point samples.  

Comment 2: Since there were no comparisons of the amount of lime, or organic amendment nor comparison of organic materials, I suggest a better explanation of the choice of amount of lime and amendment should be given.

Reply 2: Indeed, this part of the preliminary work was not included in the draft manuscript, but will be included partly in the Supplementary Information and partly in the Materials & Methods section as detailed below. The selection was done as follows:

• Type of organic amendment: 7 different organic amendments (three types of green compost, a pig manure, a mix of green compost+pig manure, and two types of farmyard manure) were analyzed for pH, DOC concentration, and aromaticity (SUVA method). A titration curve was also constructed for each type of amendment to determine its buffer capacity. A high DOC concentration was desired to maximize leaching of organic anions, while a low aromaticity was desired to minimize sorption to iron and aluminum oxides in the soil; based on this, the “ILVO green compost” was selected. A table with the tested organic amendments and their characteristics will be included in the Supplementary Information of the revised manuscript.
• Dose of lime & organic amendment: typical application rates in field experiments on highly weathered soils are 2 tons/ha lime and 10 tons/ha farmyard manure. The application rates in this leaching experiment were chosen to be somewhat higher as a proof of concept, in an attempt to ensure a leaching effect. If the doses of 5 g/kg lime and 10 g/kg compost are converted to field scale, assuming a plough layer of 10 cm depth and a soil bulk density of 1400 kg/m³, the resulting field application rates are 7 ton/ha lime and 14 ton/ha compost. This information will be included in the Materials & Methods section of the revised manuscript.

Comment 3: Secondly the leachate could be collected, but there is no information on the composition of the leachate.

Reply 3: In this study, we did not collect the leachates. The primary research objective was to understand how alkalinity (either or not in combination with DOC) is retained and distributed within the soil matrix, rather than how much was lost via leaching. As alkalinity leaching is a notoriously slow process, it was expected that leaching losses, if any, would be minimal. The data (Figure 1 & S3, Table S2 & S3) also show that pH and DOC concentration changes are restricted to the upper 5 cm of the soil columns.

Comment 4: Third the leaching experiments were relatively short, three weeks for the comparison of soil with different adsorption affinities for Dissolved organic carbon, and eleven weeks for the comparison of repacked and intact soils for a third soil. Why were these periods chosen? I understand that the experiments may have been limited by equipment availability.

Reply 4: These periods were chosen based on the number of leached pore volumes, rather than the duration needed to achieve this. The number of leached pore volumes is calculated as the total volume of rainwater applied to the column during the experiment divided by the pore volume of the column (which depends on the soil bulk density). In the first experiment, we aimed for at least one leached pore volume. At the highest rainfall intensity that was possible without overflowing the columns, this took three weeks. Since the results of the first experiment indicated a very limited alkalinity leaching, we aimed for two leached pore volumes in the second experiment. However, the rainfall intensity had to be lowered since the undisturbed columns were more prone to overflowing. The duration needed to achieve two leached pore volumes was eleven weeks in this case. We will include a short explanation on the choice of the duration of the experiment in the Materials & Methods section.

Comment 5: I found it difficult to follow the adsorption/release of DOC in organic amended soil.  Kd are given as soil and pH dependent (Table 1), and the adsorption isotherms in Figure S2 are nonlinear and show classical form. However in the main text, the authors explain difficulties with desorption of soil DOC when suspended.

Reply 5: Indeed, the isotherms in Figure S2 are derived by first adding the negative intercept of the y-axis of net DOC sorption to the measured values for the Podzol and Retisol, resulting in the total DOC sorbed. Then, a Freundlich isotherm was fitted to this data (result is shown in Figure S2). This was done because there was a net release of DOC during batch sorption experiment at every concentration of added DOC for these two soils. The K_D in Table 1 was then derived by fitting a linear line through the lower DOC concentration range data points, of which the slope represents the K_D. This is described in line 118-133 of the draft manuscript.

Comment 6: The authors attempt to use optode images to identify heterogeneity with preferential movement of amendment. Why was this method only used for intact soil? It is thus not correct to claim that there was different movement in repacked and intact soil columns. The authors explain that poor contact may introduce artefacts in the pH measurement using these planar optodes. I was certainly perplexed by the images. pH at depth appeared to be very acid, much more acid than the soil. Could a better calibration not have been found?

Reply 6: The method was used for all 12 columns of the second experiment (both intact and sieved), however, only three images representative of the leaching process were selected for the manuscript (Figure 2). For the sieved soils, pH imaging of the lime+OA treatment failed due to insufficient wetting of the sensor foil and soil before sensor foil deployment. The pH sensor foils are designed for use in aqueous solutions; the use in unsubmerged soils is possible if the sensor is sufficiently pre-soaked, but this was not always equally effective. Regarding the calibration of the optode, we fully agree that better calibration should be established in future experiments. The technique is quite new and the foils currently do not cover the full pH range. There are two types of foils available: one for pH range 2.5-4.5 and one for pH range 5.5-7.5. This is the “main response region”, meaning that the sensors still respond outside of this range but with lower sensitivity. Both sensors need to be calibrated separately and preferably at the ionic strength of the soil solution. In our study, we faced following problems that probably caused the shift in calibration:

• Application of the sensor foils to unsubmerged soils (insufficient wetting)
• The soil pH range in the columns superseded that of each sensor foil separately (e.g. pH 3.3-6.5 (0.01 M CaCl₂, 1:5) in undisturbed limed columns). Therefore, the two sensor foils were applied next to each other. In theory, the images of both foils should be combined to achieve the best result, but this was not possible in our case due to the complex pattern of preferential flow pathways. Therefore, we chose to only show the result of the lower pH foil, but since the pH of the upper soil layers falls outside of the main response region, the calibration is shifted.
• The sensor foil calibration buffers should be prepared at the same ionic strength as the sample. However, the ionic strength of the soil solution within one column ranged from 14 mM (control soil, deeper layers) to 41 mM (lime+OA, upper layers). Therefore, an average value of 25 mM was chosen for the ionic strength of the calibration buffers. This is quite low for the planar optode (use in samples with ionic strength of 200 mM was recommended) and can cause a shift in calibration.

Comment 7: Amendments were added to the top of the undisturbed soil column mixed with sieved soil, to a depth of 2 cm. Was sieved soil similarly added to the top of un amended undisturbed soil?

Reply 7: Yes, indeed, the top layer of the undisturbed control column is unamended sieved soil. In Table S3, it can be seen that the soil pH in the upper two cm of the undisturbed control column is about 0.2 units higher than that of the layers below, due to air-drying and rewetting of the sieved soil.

Comment 8: I find the authors too affirmative and positive in their conclusions. Sieved Retisol actually shows less pH increase at depth with addition of organic amendment in addition to lime, the effect is not significant for ferralsol, and small, even if significant for the podzol.

Reply 8: Point taken and the revised text will make less strong claims.

Comment 9: The soils are identified by geographical origin and not by soil type in Table 1 and the Supplementary information.

Reply 9: Thank you, we will adapt this for the revised manuscript.

Comment 10: The difference in pH and DOC with respect to the control soil is given as delta, not D in axis labels.

Reply 10: At the moment, the difference is written as “delta” but we can definitely change it to “Δ” in axis labels.

Comment 11: I suggest colour and symbols should be used to distinguish treatments instead of shades of grey.

Reply 11: Thank you, we will adapt this for the revised manuscript.

Comment 12: Findings should not be described as troublesome, line 259, I suggest disappointing or limits the relevance to …

Reply 12: Thank you, we will adapt this for the revised manuscript.

Comment 13: Did the authors consider attempting to quantify the movement of Ca? It is possible that any change would have been too small to be detected.

Reply 13: Unfortunately, the Ca concentration in the soil slices was not measured in this study.

Comment 14: The sigmoidal fit used is not appropriate for DOC profiles (Figure S3).

Reply 14: Indeed, the DOC concentrations in unamended soils decrease with depth and then, a sigmoidal fit is not appropriate. However, our main objective here was to compare the depth of the inflection point of the curves (so the depth of enrichment in DOC) between different treatments, which we believed could be best achieved with a sigmoidal fit.

Reply 1: Thank you for your comment, you are absolutely right. The statistical program JMP does give results for the Tukey test in this case, it uses a pooled variance estimate from ANOVA and adjusts for unequal group sizes by modifying the standard error in each comparison. However, it still assumes equal variances across group and requires enough degrees of freedom in the residuals to be reliable, and as you noted, this is not the case here. We will adapt the manuscript in the way you propose here.

Reply 2: Thank you for your recommendation. The Freundlich isotherms in Figure S2 are now of the form s + D = K_F * cⁿ but will be modified in the revised manuscript to s = K_F * cⁿ – D with D describing the desorbable DOC concentration (y-axis intercept).

Comment 1: The results, although not truly novel, are important for improving and further developing strategies to mend subsoil acidification occurring in various soils around the world. The probably weakest point of the study is the little number of soils being tested. Especially, the effect of soil structure has been tested only for one soil and therefore warrants more in-detail studies. Given that small number of soils tested, I am reluctant to fully support the immediate publication of the manuscript. I therefore suggest re-working the manuscript for pointing out even more clearly the preliminary nature of the study.

Reply 1: Thank you for your feedback. The goal of our study was to identify the underlying mechanisms of organic matter-mediated alkalinity leaching, not to prove effects on a bigger scale or to make claims about all Ferralsols, Retisols or Podzols. For our mechanistic study, we selected a few soils that were representative examples of soils exhibiting low, average and high DOC sorption, with the intention of creating a gradient in the success of organic matter-mediated leaching. To prove the concept of organic matter-mediated alkalinity leaching in soils all around the world, a higher number of soil types and replicates per soil type would indeed be needed; unfortunately, this was not logistically possible in this study. I refer back to our response to the first comment of Referee #1: Unfortunately, the column set-up did not allow for more replicates in practice because this set-up is a rather unique one for unsaturated flow using the suction plates. The standard column set-up with free drainage is unsuitable, leading to water-saturated conditions at the outlet and local pH increases due to anaerobic conditions. Even though that artefact is straightforward, it is still overly used, and the suction plate method is a much better alternative, with the drawback of limited replication and statistical power. In a revised version, we will reference the artefacts in other studies to justify this set-up.

Comment 2: Also, I am puzzled by description of the statistical evaluation of results since the number of replicates per soil and treatment was hard to guess. The tests used would require three replicates per treatment at minimum, resulting in each 12 columns used for the three homogenised soil. Is that correct? The authors may consider to provide a sketch depicting to entire experiment setup for.

Reply 2: The experimental set-up allowed for twelve columns to be leached at the same time. In Experiment 1, six columns were filled with the Podzol and six columns with the Ferralsol. For each of these soils, there was one control column, one compost-amended column, two lime-amended columns and two lime+compost amended columns. In Experiment 2, six columns were filled with sieved Retisol and six columns with intact Retisol. The treatments were the same as for Experiment 1. This information is given in line 139-141 of the draft manuscript. The statistical analysis comparing pH or DOC concentration values between the different soil treatments within one soil and at each depth was done with an ANOVA followed by a Tukey test, but as Prof. Jan Siemens also noted in the Community Comment below, this does not allow for the derivation of robust conclusions. The statistical program JMP uses the Tukey-Kramer adjustment to allow for unbalanced designs but the validity of the test is compromised. Therefore, we will adjust this analysis for the revised manuscript in the way proposed by Prof. Jan Siemens. However, the statistical comparison of leaching depths in lime versus lime+OA treatments for each soil was possible since these treatments were in duplicate per soil. Duplicates is sufficient for statistical analyses. For each soil, data of all four columns were fitted in one set with an assumed difference in parameter values of the sigmoidal fits for each of the four parameters (∆pH_max, ∆pH_back, slope and b). The statistical difference in parameter values were tested using dummy variables to identify treatment effects on the extent and depth of penetration of the alkalinity. 

Comment 3: The authors assume that the results of the optode imaging are probably hampered by incomplete contact to soil. I therefore suggest to consider transferring the result to supplement section since presumably not reliable.

Reply 3: In our opinion, the optode images still have an added value since they show preferential flow patterns that can not be detected by the conventional soil pH measurements. For the lime+OA treatment in the sieved soil, the imaging failed due to insufficient contact with between the soil and the sensor foil, therefore we only showed images for the intact soil in Figure 2. The sentence  “The absolute values of the soil pH are solely indicative since incomplete contact between the sample and sensor foil may occur in unsaturated samples” was added more out of caution than that we actually think this happened.

Comment 4: The derivation of partition coefficients from the sorption data is hampered by the mostly less than ideal fits of the applied sorption model. The authors may consider using other approaches. Several have suggested and tested by Kothawala et al. (2008, Geoderma 148, 43–50. doi: 10.1016/j.geoderma.2008.09.004).

Reply 4: Thank you, we also fitted a Langmuir isotherm but fits were not markedly better than for the Freundlich isotherm. An Initial Mass isotherm is less suited for highly sorbing soils like the Ferralsol (doi:10.1016/j.geoderma.2007.01.012).

Comment 5: Finally, please refer to the soils by type and not regional origin.

Reply 5: Thank you, we will adapt this in the revised manuscript.