the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 07 May 2025
Coupling of soil carbon and water cycles in two agroforestry systems in Malawi
Abstract. Consequences of climate change are likely to pose severe challenges on agriculture in Southern Africa. Agroforestry systems (AFSs) can potentially alleviate some of the adverse effects and offer adaptation solutions to a sustainable land use. Positive effects of AFSs which have been shown include increasing soil carbon (C) and nitrogen concentrations, sustaining favourable nutrient cycling, protection against erosion and increased carbon sequestration. The influence of the AFS tree component on the water cycling of the crops, however, is still relatively unknown.
In this study we assessed the influence of gliricidia-maize intercropping on carbon cycling and water fluxes compared to maize as a sole crop at two well-established long-term experiments in central and southern Malawi, run by the World Agroforestry (ICRAF). The controlled setup and different durations of the experiments (>10 and >30 years) at the two sites provided information regarding soil-specific impacts of gliricidia on water dynamics. We examined soil C contents and density fractionation as proxy for organic matter stability, soil physical and soil hydrological characteristics. We also monitored soil moisture and matric potential in different depths, determined retention curves on samples in the lab and from field data and analysed soil moisture responses to rainfall events to assess the influence of the AFS on water fluxes.
Our results show a clear increase in C contents and stability as a result of the gliricidia impact compared to the control, especially pronounced at the site with the generally lower baseline C contents. The treatment effect is also visible in soil physical characteristics such as porosity and bulk density, which is mirrored by a greater saturated hydraulic conductivity. These treatment effects were, however, not directly translatable into soil water dynamics as the latter were influenced by several additional factors such as soil texture and interception. The gliricidia plots showed greater soil water storage capacities and retained overall more water, while generally both treatments were not under severe water stress during the observation period. We also noticed a protective effect against soil drying below the topsoil facilitated by the gliricidia. Furthermore, infiltration shifted towards more immediate/macropore infiltration under gliricidia.
We conclude that the AFS treatment of adding gliricidia into maize cultivation has a considerable effect on nutrient and water cycling in the crop, while the effect on water fluxes is not straightforward. While the differences in soil moisture and matric potential never lead to a shortage of water for the crops, a detailed examination of water fluxes require respective measurement in the field as they cannot be deduced from soil physical characteristics directly. AFS can thus not only support carbon accumulation and stabilization, but a sensible combination of trees and crops can also be beneficial for a more sustainable use of the available water.
- Preprint
(6522 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-1719', Anonymous Referee #1, 08 Jun 2025
- AC1: 'Reply on RC1', Svenja Hoffmeister, 30 Jun 2025
-
RC2: 'Comment on egusphere-2025-1719', Anonymous Referee #2, 10 Aug 2025
The manuscript titled "Coupling of Soil Carbon and Water Cycles in Two Agroforestry Systems in Malawi" addresses an important topic relevant to sustainable agriculture and climate adaptation in Southern Africa. It explores the coupling of soil carbon and water cycles within maize-Gliricidia-based agroforestry systems in Malawi, drawing on experimental data from well-established long-term research sites. Conducting research on such long-term experiments provides valuable insights into the sustained impacts of agroforestry management on soil health and water dynamics over time.Given the increasing challenges posed by climate change in Southern Africa, understanding how agroforestry influences both carbon sequestration and water availability is critical for developing sustainable land-use practices. This study’s focus on soil carbon content, physical properties, and hydrological responses under Gliricidia-maize intercropping and maize monocropping provides important information for enhancing agroecosystem resilience.
While the topic is relevant and timely, the manuscript would benefit from clearer explanations of data analysis, consistent use of terminology, and more thorough contextualization of the results. Detailed feedback is provided below.
Title
It is better to modify the title because the two agroforestry systems are not actually different. Both sites use a Gliricidia sepium-based agroforestry system, so there is no need to refer to them as two separate systems.
Here is my suggested title: Coupling of Soil Carbon and Water Cycles in Gliricidia sepium*-Based Agroforestry Systems in Malawi. Still I have question on the term “water cycles”.
Abstract
The abstract provides a clear overview of the study’s objectives, methodology, and key findings related to soil carbon and water dynamics in Maize-Gliricidia-based agroforestry systems and Maize monocropping in Malawi. It effectively highlights the importance of agroforestry in addressing climate challenges and improving soil health.
However, some areas could be improved for clarity and precision:
Consistency in Terminology: The abstract refers to “controlled setup,” which may be misleading since the experiments are conducted under field conditions. Clarifying this will avoid confusion.
Focus on Key Results: While the abstract summarizes results well, some sentences could be more concise. For example, the relationship between soil physical properties and water dynamics could be stated more directly.
Contextualization: The abstract would benefit from briefly mentioning the implications of the findings for sustainable agriculture or land management in Malawi.
No keywords were provided
Overall, the abstract is informative but could be polished to improve readability and impact.
Introduction
I noticed some redundancy in the introduction section of the article. For example, the effects of intercropping Gliricidia trees with maize are repeatedly discussed in relation to soil structure, soil fertility, soil organic carbon (SOC) content, carbon stocks and stability, water dynamics, and maize yield. These points should be streamlined and organized clearly to avoid repetition.
Additionally, I observed a disruption in the flow of ideas within one paragraph. You wrote, “Consequently, in this study we examine the link between carbon input and soil hydrology,” which is immediately followed by, “There is still a lack of knowledge about the extent to which carbon-induced short- and long-term changes in soil structure affect water fluxes in these systems.” This sequence weakens the coherence of the introduction.
I recommend revising the introduction to improve clarity, coherence, and engagement. The background should succinctly summarize existing knowledge, highlight the specific knowledge gap, and effectively convey the importance of your study without redundancy. It would also be helpful to clearly and concisely differentiate your research from previous studies such as Chirwa et al. (2007), Hoffmeister et al. (2025), Kerr (2012), Kirsten et al. (2021), Maier et al. (2023), and Makumba et al. (2006, 2007).
Line 72: What do you mean by "farmers might be concerned"? Is there any evidence or findings that show how farmers perceive the impact of agroforestry systems (AFS) on soil evaporation?
Line 64: You should clearly define what the tree component is?
Line 72. Some of the findings are not fully reviewed or presented in the introduction section. For example, the relationship between increased SOC and soil evaporation is still debated in the literature and may depend on contextual factors such as soil texture and climate, as reported by Feifel et al. (2024).
Line 76. The findings were not fully presented while making the argument. However, they are still influenced by seasonal variation. For instance, Chirwa et al. (2007) reported that the available soil water content at the start of the cropping season was generally lower in tree-based systems, indicating that trees likely continued to extract soil moisture throughout the dry season.
Line 83. Lack of information is provided concerning the distance between the two sites. Sites that are 316 km apart were generalized as being in the same climatic zone. Does this make sense? If so, please show us on a map, using a DEM or a table with long-term climate data from meteorological stations.
Line 103. Sometimes you use the wrong words. For example, calling it a "controlled agroforestry experiment" is odd for an experiment conducted under field conditions..
It is better to include the hypothesis! It would strengthen the manuscript to explicitly state the study’s hypothesis (or hypotheses) in the introduction. Presenting the hypothesis provides readers with a clear understanding of the research’s expected outcomes and the rationale behind the study design. This also helps in framing the results and discussion, allowing the audience to assess whether the findings support or refute the initial assumptions.
Method
Line 102: The map is not sufficiently explanatory, as it lacks essential elements and does not clearly represent the study area’s landscape and elevation. Could you please provide a DEM map of the study area that includes all necessary features, such as coordinates, a legend, and other relevant details? This DEM map will help to clearly highlight the climate zones within your study area.
Line 103. Misleading information is included, such as referring to the study as a controlled agroforestry experiment. A controlled experiment has a very different meaning, whereas the agroforestry experiment in question was conducted under field conditions.
Line 105: Information that should be presented in the introduction, which could help to highlight the research gap, is misplaced here in the Methods section. For example, see line 105.
I feel that it is better to provide topics for each section of the methodology, such as: Study Area Description, Experimental Design, Description of Agroforestry System / Cropping System, Sampling Procedures, Measurement Techniques, Data Collection on Environmental Variables, Laboratory Analysis, Data Analysis Methods, Quality Control and Assurance, and Ethical Considerations or Permissions.
The management practices in the control (maize monocropping) were not explained at all. If it was conducted without any fertilizer, it does not reflect the reality of maize production in Malawi or other African countries. It would make more sense to compare maize production under local fertilizer rates and typical management practices. The experiment would have been improved by including three treatments: control (no fertilization), T1 (fertilizer at the recommended local rate), and Maize-Gliricidia based agroforestry.
Line 115: There is a lack of consistency in reporting temperature: for the Chitedze site, the mean annual temperature is reported, whereas for the Makoka site, the mean daily temperature is given.
Line 116: Lack of consistency in soil group classification by the IUSS Working Group: For the Chitedze site, you cited IUSS Working Group (2014) for the classification of Chromic Luvisols, while for the Makoka site, you used IUSS Working Group (2022) for Ferric Lixisols. I suggest using the latest version (e.g., IUSS Working Group, 2022) for both sites. Typically, such data are established at the country level, and you may be able to find updated classifications for both locations in the newer version.
Figure 1. Figure 1 requires changes based on the above comments.
Line 125: Would the ridges in the maize plot intercropped with gliricidia trees affect soil sampling for bulk density and other parameters that require undisturbed soil samples for analysis?
In line 134. Previous data were included in this article, but details about the data and the source link (local or international repository) were not provided. I personally recommend presenting this information in the appendix section of the article. Additionally, you should mention it in the introduction for example, by stating that previous work has addressed certain aspects, but further analysis is still needed due to a lack of [specific information or data] or inconsistencies in the results (short- or long-term).Furthermore, It is a fact that both pedological and hydrological characteristics vary with time and season, so you should explain why you chose to include data from 2019 and 2021.
Line 135: Some of the topics seem overly broad and include irrelevant information. It would be better to provide concise main topics and split the rest into subtopics. For example, having 'Soil sampling with an auger' as a main topic in the article seems quite unusual.
Line 138: There is a lack of standardized names for sampling instruments. Use ‘Kopecky rings’ instead of ‘small and large cylinders,’ as Kopecky rings are standardized tools for collecting undisturbed soil samples in the field. Since there are ridges inside the plot, how did you manage to collect undisturbed soil samples from these areas? To what extent can the samples collected from this site be considered truly undisturbed?
Line 142: There is a lack of information regarding the soil sampling system and sample collection. Since there are ridges inside the plots, how did you manage to take samples from both disturbed and undisturbed soil depths (e.g., see line 170) in both treatments at the two sites? Furthermore, some studies report that in tree-based agroforestry systems and annual crop production areas, the 31-90 cm soil layer typically represents the depth where leached nutrients accumulate and where root growth contributes organic matter and nutrients. Do you think sampling only the 0-10 cm and 10-20 cm layers is sufficient to represent the actual soil conditions at the site? I also have doubts about the depth ranges (e.g., 0-10 and 10-20 cm); I personally suggest using 0-10 cm and 11-20 cm ranges.
Lines 141-147, line 170, and line 199 show inconsistency in sample collection for physicochemical and hydrological data collection. Sometimes you use depth intervals of 0-10 cm, 11-20 cm, other times 0-20 cm, or specific depths like 5 cm, 7 cm, 10 cm, 15 cm, 25 cm, and 60 cm, which makes it very difficult to follow the overall research flow. There is consistent inconsistency in sampling and sample depth selection, making it hard to interpret the results. This variability in sampling could itself introduce additional variation beyond the treatment effects (control and Gliricidia agroforestry plots).
Line147-150: The sampling times vary, with some campaigns conducted in 2021 and others in 2022, which is methodologically challenging for soils in tropical areas. Both soil and hydrological characteristics change over time, and using data from different periods could complicate the results and overall findings of this research.
Line 154: How reliable is the distilled water method for measuring water-dispersible clay (WDC), considering that some clay particles are strongly bound and will not disperse without the use of chemical dispersants such as sodium hexametaphosphate (Na₆P₆O₁₈)? Furthermore, the results can be influenced by soil type.
Line 154: You analyzed CEC but you don’t analyzed Na⁺? Please also correct the notation for Ca²⁺, Mg²⁺, and K⁺.
Line 164: What are the similarities and differences between this study and the one by Hoffmeister et al. (2025)? Is there any connection to this study (e.g., in terms of data, methods, etc.)? What is the purpose of mentioning it here?
Line 173: The majority of the measurements, such as hydraulic conductivity (Ksat), were conducted in the laboratory, despite the availability of accepted field measurement methods. Why choose to measure Ksat using laboratory instruments in Germany when field measurements using a double-ring infiltrometer or Guelph permeameter are easier and more cost-effective than transporting soil samples to a distant location?
Line 194-195: The falling-head method is generally recommended for fine-textured soils due to its sensitivity to low permeability. Could you clarify why this method was selected, especially if the soil in your study includes coarser textures? For example Guelph permeameter and double-ring infiltrometer.
Line 204: Despite the actual rainy season being from November to April, why were you interested in collecting some hydrological data (e.g., water flux data such as soil moisture and matric potential) from March to May 2022? Did you also consider calculating antecedent soil moisture content?
Line 210: The soil water storage change was calculated based on the difference between two successive soil water content measurements. At what time intervals did you conduct the soil moisture and matric potential measurements, and why did you select those intervals?
In general, there is no information on data analysis, such as the type of software used to analyze the soil and hydrological characteristics, or how the mean values of these characteristics were compared between treatments at the two sites. It is also difficult to conduct ANOVA with only two treatments. I suggest performing a two-factor factorial analysis for each site, introducing treatment and depth as factors. Additionally, if interested, you could consider treatment and site as two factors to analyze their interaction effect.
Result
Why do you want to compare the different soil and hydrological characteristics between sites? It is quite clear that the soil classes at the two sites are different (i.e., Chromic Luvisols vs. Ferric Lixisols). Differences are expected due to soil formation processes, parent materials, and stages of weathering.
You are only presenting the results in the Results section, but it is important to include explanations of your own reasoning, justifications, and the implications of the high and low values of the soil and hydrological characteristics between treatments.
Table 1. No information about the sand, silt and clay faction for Makoka site 0-20 cm depth ranges.
The entire Results section lacks statistical t-test analyses with significant p-values and letters indicating differences between treatments (e.g., a, b, c in mean values in the tables or figure bars). In general, the soil physicochemical and hydrological characteristics are presented simply as high or low without any statistical comparison or significance values.
Table 1. The table captions sometimes include too much information, such as abbreviation details. They should follow the standard format of the journal.
As described in the Methods section, the presented result data show a lot of inconsistency in depth ranges for soil physicochemical and hydrological characteristics, which makes it quite confusing and hard to understand the overall flow of the results.
There is inconsistency in the number of samples used for analysis in the results presented in the tables. For example, results analyzed with 10 samples at one site are compared with results from only 5 samples at the other site. For example in Table 1, the soil characteristics from 2019 (Chitedze) with a larger sample number were compared with those from the other site (Makoka) in 2021, which had only 5 samples. This is methodologically incorrect, if you don’t you proper analysis tools for such type of data. Additionally, there is no information on the number of replications for each treatment at the two sites.
Figure 2. It is very hard to understand and visualize the differences in Figure 2a, which seems to be based on subjective judgment. In Figure 2b, at the Makoka site, the experiment is designed as a randomized complete block design. It is important to show how the other replications look so that it is easier to visualize the differences. In general, since these figures are not the major aim of this research, it would be better to include them in the supplementary section of the article. Furthermore, no standard sampling procedures was reported for this data.
Table 2-: It would be better to move this table to the supplementary section, as it is not directly relevant to the Results section because it presents results from previous research. I am unsure of the intention behind including secondary data in the Results section, especially since soil properties (such as soil carbon) change over time. You could discuss these changes in the Discussion section while referring to the data in the supplementary materials.
There is also mixing of sections, with literature frequently cited in the Results section, reflecting an organizational problem. The manuscript should be structured according to the journal’s guidelines. Furthermore, some information that belongs in the Methodology section is instead presented in the Results section; see, for example, lines 265 and 304-305.
Figure 3. A beautiful figure was presented but they statistically don’t show anything and seem like static, meaningless figure.
Line 251. You are comparing the C content between sites. In my opinion, these two locations are incomparable due to differences in sampling and methodology. It would be better to focus on the treatment effects at each site separately.
Figure 4: The relationship between WDC and Colf was negative, but the R² value (R² = 0.65) is positive. Additionally, the legend for AFS is not clearly identified in Figure 4.
Line 265: sometimes you compare soil characteristics between sites, but if interested, you could instead perform a two-factor ANOVA to examine the interaction effect of site and treatment. As I understand it, you want to evaluate the effect of Maize-Gliricidia based agroforestry compared to maize monocropping; therefore, it would be better to compare these results separately for each site.
Line 274: The treatment differences (per depth) for the small cylinders were tested using the Mann-Whitney U test, as the sample size was sufficient and results showed no significant difference. However, this is not the correct place to explain the analysis method; such details should be included in the Methodology section.
Line 278: Bulk density (BD), porosity, and Ksat values did not show any statistical evidence of differences between treatments at the two sites. For example, although Ksat was higher in the Gliricidia treatment than in the control, this difference was not statistically significant (see Figure 5).
Why don’t you show the correlation between organic carbon and both bulk density and Ksat for the two treatments?
Line 292: Sometimes, measurement values from the laboratory and the field are presented in the Results section for example, retention curves (tension and soil water content) but there is no explanation or statistical comparison between the two sets of measurements. Since the methods used are standardized, it is unclear why repeated measurements were made, especially when the study’s focus is not on evaluating these methodologies. This presentation is confusing (see Figure 6 results).
Figure 6. The figures contain many values for different hydrological characteristics but lack labels for each panel. Please add labels such as Soil Water Content (a) Chitedze, (b) Makoka, and PAW (c) Chitedze, (d) Makoka at the top of Figure 6, and explain them clearly in the figure caption. This will enable you to refer to them easily after presenting the results.
Sometimes the result for one treatment is high at one site but low at the other. For example, in line 298, in Makoka, PAW in the control showed “slightly higher values” compared to the AFS treatment; however, in Chitedze, higher PAW values were recorded in the AFS treatment. There is no reasoning or explanation provided for these contrasting results throughout the Results section.
Line 305: Did you check the antecedent moisture content and its relationship with overall water content and matric potential?
Line 312. What is the reason for the strong drying after rainfall, even in the gliricidia treatment, based on your observations or experience? Could the difference in soil water content between the sites be due to variations in geology or parent material?
Figure 7 lacks labeling for P and WC in both the control and gliricidia treatments.
Line 335: The statement ‘the water slowly percolated downwards in the gliricidia plot as demonstrated by the sequential storage increases with increasing depth’, does this truly reflect real field conditions? I have doubts because real-world soil water dynamics are often more complex due to factors such as soil heterogeneity, lateral flows, plant activity, and varying environmental conditions. Therefore, relying solely on this observation may not fully capture the actual behavior of water movement. Additionally, the implications of this for shallow- and deep-rooted plants should be elaborated. Furthermore, the relevance of these findings to the specific climatic zone and potential supplementary irrigation strategies should also be discussed.
Discussion
Figures are misplaced for example, Figure 10 is placed inside the discussion section instead of the appropriate section.
Sometimes unreasonable generalizations are made, even though the facts are only presented in the Results section. For example, in line 356, it states that the soil at the Chitedze site already has a high carbon saturation level. Did you calculate the Carbon Saturation Index (CSI), which is the ratio of observed SOC to the estimated maximum SOC storage capacity, or use other metrics such as the Mineral-Associated Organic Carbon to Clay + Silt Ratio (MAOC/CS)? In tropical climates, reaching carbon saturation is more challenging than in temperate regions due to higher microbial activity.
Line 357: Some supporting literature is presented in the discussion without relevant information. These references are off-topic, not related to agroforestry systems or maize, and are from non-tropical climatic zones. In research focused on the tropics, your citations come from Japan and Germany. Couldn’t you find any literature related to tropical regions?
Line 366: Have you identified any literature on the relationship between organic matter and soil aggregation in tropical regions, particularly in Africa? You cite Bronick and Lal (2005), but the argument you present is not consistent with the WRB/FAO classification (i.e., Alfisols are distinguished from Luvisols and Lixisols).
Line 376. Generalizing the results as ‘high’ by what index did you classify ‘high’ and ‘very high’? I am also surprised that there was no statistical difference in pedogenic oxide values between the two treatments, yet it was reported that these contribute to the potential structural benefits typically attributed to increased SOM in AFS treatments.
Line 390. The fact is that there is no statistical difference in bulk density and Ksat between the two treatments under different management practices. How do you support your argument in light of this?
Line 397: We did not see any statistical evidence showing a variation in BD and Ksat between the treatments (e.g., Figure 5)
4413: “The treatment differences appeared smaller than differences in depth” this means that depth is another important factor, or that depth mattered more than the treatment in explaining the differences. I suggest conducting a two-factor analysis to assess the interaction effect between treatments and depth.
445-450: In the gliricidia treatment, water movement down the soil profile is context-dependent, particularly influenced by soil texture. What are the implications of this?
Line 490: This conclusion is difficult to accept because it appears to be based on simple observation without a proper sampling procedure for plant growth and health attributes. Please review how samples are taken with replicates in experimental plots in other published studies.
Line 498: Does this mean that the effect of Gliricidia residue input on CEC is site-context dependent?
Conclusion
Line 511: “In maize and gliricidia intercropping, we saw a clear treatment effect on soil nutrient and C contents as well as on soil structure.” However, regarding the treatment effect, no statistical differences were observed between treatments at either site. You simply label the values as ‘high’ and ‘low’ based on the numbers, which is not scientifically valid. Moreover, the treatment effects were not consistent across both sites for all parameters. Please review your tables and figures.
Line 518: “In the context of climate change, we saw that some of the envisioned challenges for agriculture in Southern Africa can be alleviated with the adoption of AFSs.” However, your research does not consistently support this conclusion. How did you arrive at this statement?
-
AC2: 'Reply on RC2', Svenja Hoffmeister, 12 Sep 2025
We would like to thank the referee for their thorough and detailed feedback.
We appreciate the large effort put into reading the manuscript, asking valuable questions and making helpful suggestions.
In the attached document we address all points made by the referee in detail.
-
AC2: 'Reply on RC2', Svenja Hoffmeister, 12 Sep 2025
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 675 | 72 | 15 | 762 | 18 | 30 |
- HTML: 675
- PDF: 72
- XML: 15
- Total: 762
- BibTeX: 18
- EndNote: 30
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The authors present a concise study in which they investigate effects of Gliricidia-Maize intercropping on soil carbon and water cycling compared to sole maize cultivation. With that they exemplarily address the question of whether agroforestry systems can be adaptation solutions for sustainable land use. In particular, they analyze the relationship between the observed increased and more stable SOC in the agroforestry area and the associated changes in soil hydraulic properties and soil water dynamics. The effects of the altered soil hydraulic properties on infiltration and water holding capacity of the soils after selected rain events are examined and discussed in detail. The data basis for the study comes from two similar long-term agroforestry trials in Malawi. Existing data from these trials were complemented by data collection from various measurement campaigns and soil moisture and matrix potential monitoring in 2022. The results show clear correlations between effects on the carbon cycle, soil physical parameters and partly on the soil water regime. Although treatment effect on soil physical parameters were not directly translatable to soil water fluxes, the results of the study confirm several presumptions about the positive effects of agroforestry systems on soil carbon and soil hydraulic properties, which have rarely been demonstrated experimentally on a process-specific basis.
Thus, the paper address a relevant scientific question which fits well within the scope of Biogeosciences journal. The data set offers interesting insights, which support the results and justify the conclusions drawn, even if a longer monitoring of soil moisture and an extension of the analysis over more than one growing season and further precipitation events would be desirable, to confirm the findings. The applied scientific methods and assumptions are valid and clearly outlined. The experiment and data analyses are well described, except (in my opinion) the calculation of the changes in soil water content. Presenting a formula would be appropriate and helpful here. The authors always give proper credit to related work by others and clearly indicate their own contribution. The title is well chosen and clearly reflects the content of the paper. The abstract provides a concise and complete summary of the study. The manuscript is well structured, each section is clear and coherent. This, together with the precise and fluent language, makes the paper easy to read and to understand. However, a few carelessnesses crept in during the preparation of the manuscript that need to be corrected (see below, specific comments). The number and quality of references as well as the amount and quality of supplementary material is appropriate.
Despite the very positive overall impression the paper makes to me, the reader is left with a certain amount of uncertainty. This concerns Figure 6, which shows retention curves for the topsoil and subsoil at the two sites determined by means of hyprop. For site Chitedze, the curves for the topsoil (control, but also gliricidia) show unrealistically high saturation water contents/porosities (θsat > 0.9 m3/m3!). This is not adequately discussed. How do these high values come about? What are the values of the van Genuchten parameters derived from the hyprop measurements? Is the scaling of the x-axis correct? As a significant part of the data interpretation, discussion and conclusions relate to the shape of these curves, these questions definitely need to be clarified. Possibly some passages in the results section and in the discussion relating to the flattening of these curves near saturation need to be revised. The calculation of plant-available water may also be affected.
Specific comments:
Line 188: Last sentence of section 2.3. repeats sentence in lines 173-175.
Line 211: From where does the 0.18 m come ("sensor depth increment")? It is not align with the depths indicated in line 199. Can you provide a formula for the calculation of change in water storage?
Line 228: The ratio of Fe_d/Al_d was lower(!) in Chitedze
Table 1: no texture data is given for Makoka. Why? You refer to it in lines 216-218.
Line 236: Here you mention a C/N ratio for the intercrop of 15.4. However, in table 1 a value of 11.3 is given. Which value is correct?
Table 2: Difference C content at Chitedze: 1.5 (=30.2-28.7) (instead of 1.8)
Line 249: 4.4 gC kg-1 or 4.3 gC kg-1 (table 2)?
Line 252: unit g C kg-1 ff.
Lines 267-8: lower bulk density in gliricidia sites only at 5 cm, but reverse at 15 cm!
LineS 277-278: In my opinion, this statement is misleading as it contradicts the data from the large cylinders. These show a decrease in Ksat! Indeed, data from the small cylinders show an increase, but that goes along with a decrease in porosity. Here it is certainly better to rely more on the results from the large cylinders, which are probably more reliable.
Figure 6 gives me some puzzles. I am irritated by the presented retention curves for Chitedze measured in the laboratory, as they show saturated water contents of 0.75-0.95 m3/m3 (assuming that x-axis indicates volumteric soil water content as indicated in the label). How can this be? Values of this magnitude are not realistic. They also contradict the results from soil cylinders. In which pF-range were the retention curves actually recorded using the Hyprop? What are the values of the van Genuchten parameters (eq. 1) derived from Hyprop measurements? When fitting the retention curve to Hyprop data, it would probably have been more expedient to fix theta_sat to the value measured by soil cylinders. Since the PAW values are directly derived from these retention curves, I also cannot fully trust them.
Figure 8, figure caption: E5-E8 in Fig. 7 instead of Fig. 6
A few abbrevations are not defined before first use ("BD", "OM")
Table A1: re-arrange rows and indicate differences in sampling depth for small cylinders (5, 15) and larger cylinders (5, 25)
Table A3: When comparing figs. 7 & 9 with tabele A3, I think event E4 corresponds to C5 with 17.8 mm (instead of C4 with 3 mm only)