InVEST Soil Carbon Stock Modelling of Agricultural Landscapes as an Ecosystem Service Indicator
, Ana Flávia Boeni 2,3, Viola Judit Prohászka 4, Alfréd Szilágyi 1, Eszter Tormáné Kovács 5, László Pásztor 6 and Csaba Centeri 5
Round 1
Reviewer 1 Report
To authors:
Comments and suggestions need to be addressed:-
Line 29: This national is referring to Hungary specifically. If so, please, rephrase, as ... National inventories of the InVEST in Hungary is unknown.
Line 154: Both site were explained. If slope of the site (%) were included would have been better.
Line 147 – 148 and 159 – 160: The vegetation cover at each site is different, therefore the source of variation at each site was expected because of spatial and vegetation variation?
Line 287: what the x-axis indicating (soil depth (cm))? What are these? Soil depth (cm)?
Make it clear. It looks also carbon stock ranges. Give the x-axis a title name? – Figure 6?
Line 297 – 298: The (F2.73 = 2.413, ns) and (F1.74 = 5.25, p < 0.05) statistics expressing p- value is more accurate than the F-value. The p-value expression is more specific and accurate than the F-test.
Line 383: Check the units on the y-axis the units of each graph (Mg or Mg ha-1). Figure 10?
Line 403: “. . . study determines and reports a distinct. . .” these verbs need to be in past tense form. Check the text accordingly.
General comments:
· Avoid very long sentences as they distract or misguide readers.
· Verbs need to be checked throughout the manuscript.
Most of the comments are highlighted and text were also included as Insert Text At Cursor in the pdf format.
Comments for author File: 
Comments.pdf
Author Response
Response to Reviewer 1 Comments
Point 1: Line 29: This national is referring to Hungary specifically. If so, please, rephrase, as ... National inventories of the InVEST in Hungary is unknown.
Response 1: Agreed. Line 28-29 was updated to “However, the accuracy of National carbon inventories of Hungary is unknown”. As carbon inventories are used by many carbon modelling software, and their accuracy is undetermined, “InVEST” was left out of the sentence.
Point 2: Line 154: Both site were explained. If slope of the site (%) were included would have been better.
Response 2: Thank you for this suggestion. The study areas are both very large (> 20000 ha) and include various slope levels. As described in Line 155, the southern study site is a hilly landscape. Text on the elevation ranges was included for both sites, in Line 141-142 and Line 157-158 to better describe the areas as suggested.
Point 3: Line 147 – 148 and 159 – 160: The vegetation cover at each site is different, therefore the source of variation at each site was expected because of spatial and vegetation variation?
Response 3: Indeed, there are several sources that cause SOC variances in the soil properties, we explore these in the Discussion’s 4th paragraph (Lines 424-431). However, one of the goals of the study was also to understand how the land-use land-cover can influence the soil carbon stock, and therefore to create a more accurate mapping of CS stocks. Because of that samples of different land uses were taken and we unpack these CS differences in the Discussion (Lines 418-449).
Point 4: Line 287: what the x-axis indicating (soil depth (cm))? What are these? Soil depth (cm)? Make it clear. It looks also carbon stock ranges. Give the x-axis a title name? – Figure 6?
Response 4: Agreed. Titles have been added to the bottom two graphs to name the x-axis, see Line 298.
Point 5: Line 297 – 298: The (F2.73 = 2.413, ns) and (F1.74 = 5.25, p < 0.05) statistics expressing p- value is more accurate than the F-value. The p-value expression is more specific and accurate than the F-test. The F value that corresponds with your alpha value = 0.05, 0.001 or 0.0001, then the difference among groups is deemed statistically significant. Instead F value, please use the alpha or p-value which is more specific.
Response 5: In the first version, we performed the One-Way Anova, that by default uses the F statistics and in the text, we have reported all the values (F and p-values). However, we accepted the suggestion and indicated only the p-values (if higher or lower than the alpha) in Lines 301, 302, 304.
Point 6: Line 383: Check the units on the y-axis the units of each graph (Mg or Mg ha-1). Figure 10?
Response 6: Y-axis titles’ units were checked and we confirm they are correct. In Figure 10, Line 385, the left graph shows the Total potential aggregated soil carbon stock (Mg) stored between 0 and 30 cm soil depth in the study area landscapes. This value is sourced from the results of the InVEST spatial models, as described in Line 268-269. To improve understanding, Lines 386-387 were updated to, “Total potential aggregated soil carbon stock (Mg) stored between 0 and 30 cm soil depth in the Vác-Pest-Danube Valley and South-Zselic study area landscapes, Hungary...” and “..., from 0 to 30 cm soil depth...“ was added to Line 269 in the Methods to improve the description. As the study areas are of different size, the right graph was included to compare soil carbon values between the study areas as carbon per ha, and therefore the y-axis units are Mg.ha-1.
Point 7: Line 403: “. . . study determines and reports a distinct. . .” these verbs need to be in past tense form. Check the text accordingly.
Response 7: Agreed. Lines 406-409 now reads as, “The methodology presented in this study determines determined and reports reported a distinct potential range of landscape-level soil carbon stock, a new approach to evaluating and reporting soil carbon storage on this scale.”
Point 8: General comments:
- Avoid very long sentences as they distract or misguide readers.
- Verbs need to be checked throughout the manuscript.
Response 8: Sentences were checked for their length and edited and shortened were necessary. Verbs were checked throughout the text as well, and edited accordingly.
In addition to the above comments, all spelling and grammatical errors pointed out by the reviewers have been corrected. A new relevant reference was added at Line 104. We look forward to hearing from you in due time regarding our submission and to responding to any further questions and comments you or the reviewers may have.
Sincerely,
Lyndre Nel
Reviewer 2 Report
This is the revision of the manuscript number sustainability-1816084 Title: “InVEST soil carbon stock modelling of agricultural landscapes as an ecosystem service indicator”, proposed by Ms. Samira Jiao and colleagues for consideration for publication in Sustainability
The manuscript raises an interesting issue regarding the assessment of productive soils, one of the essential elements of the vegetable cycle is C, so it is interesting to have storage of C in the soil, thus helping to mitigate the increases in atmospheric CO2 between other questions, even so I have many doubts about the study, such as the criteria that were taken to decide the number of soil samples for each zone for the analysis of the SOC
Material and methods:
Line 75 I recommend that the units that are to be related to another unit are separated by a space and not by a period, such as Mg.ha-1, this should be indicated as Mg ha-1
Line 244 I believe that the data should be treated with statistical studies that indicate whether or not the differences are significant, for which the Tukey or Duncan tests can be applied.
Line 231 What should be done with the crop residues in the areas of farmland
Results and discussion:
Line 306 In figure 7 there is a large amplitude in the error bar of the areas of the South-Zselic, what can they be due to?.
I think it is appropriate that the COS values should be accompanied by other parameters to know what state that COS is in, such as labile carbon, water-soluble carbon, fulvic acids and humic acids
General comments:
The author refers to the comparison of the models as a solution for the decision-making of government agencies in terms of soil valuation and management, but, as referred to in the discussion of this article, carbon stocks in the soil are difficult to model accurately since it can show exceptionally large variations (line 451-453) due to the great heterogeneity in small areas and even more so in large areas, which makes it difficult for this study to show conclusive values of SOC in areas with large expanses.
Author Response
Response to Reviewer 2 Comments
Thank you for giving us the opportunity to submit a revised draft of my manuscript titled ‘InVEST soil carbon stock modelling of agricultural landscapes as an ecosystem service indicator’ to Sustainability. We are grateful to reviewer 2 for their insightful comments and suggestions. Changes have been incorporated to reflect most of the suggestions provided. We have highlighted the changes within the manuscript.
Point 1: The manuscript raises an interesting issue regarding the assessment of productive soils, one of the essential elements of the vegetable cycle is C, so it is interesting to have storage of C in the soil, thus helping to mitigate the increases in atmospheric CO2 between other questions, even so I have many doubts about the study, such as the criteria that were taken to decide the number of soil samples for each zone for the analysis of the SOC
Response 1: The basic decision on sampling was to sample “typical” and representative areas/land-use land-cover of the sampling sites, i.e., farmland, forested areas and grasslands. The other major sampling principle was to make a scientifically sound comparison of the major land use forms. This is why we have 5 replicates, i.e., the general minimum number per habitat/land cover type required for statistical analysis for soil science. The Northern site is 20 704 ha while the Southern site is 51 100 ha – this is the reason for more sample collection from the southern site.
Point 2: Material and methods: Line 75: I recommend that the units that are to be related to another unit are separated by a space and not by a period, such as Mg.ha-1, this should be indicated as Mg ha-1
Response 2: Thank you for bringing this to our attention. We have seen units reported differently in other Sustainability articles. We will check with the editors on their publishing requirements and edit them accordingly.
Point 3: Line 244: I believe that the data should be treated with statistical studies that indicate whether or not the differences are significant, for which the Tukey or Duncan tests can be applied.
Response 3: A Shapiro-Wilk Test and a Levene's Test for the Homogeneity of Variance of the residuals verified the normality of the SOC data from the soil samples. One-way Analysis of Variance (ANOVA) tests were done, comparing SOC between the different LULC and the study areas. Statistical differences among means were checked through a post hoc Tukey-test (α = 0.05). Explained in Lines 250-254.
Statistical analyses between all the LULC classes in general showed no significant difference between farmland (n = 356), forested areas (n = 20) and grassland (n = 20) LULC (F2.73 = 2.413, ns: p > 0.05). Analyses showed a statistical significance in the carbon stock measurements between the two study areas (F1.74 = 5.25, p < 0.05), regardless of LULC. An ANOVA on the soil carbon stock means of the LULC, for each study area, yielded significant variation (F2.12 = 4.02, p < 0.05), where a post hoc Tukey test showed that a significant difference was only found between carbon stock from forested area and grassland for the northern Vác-Pest-Danube Valley microregion study area (p < 0.05), see Figure 7. Explained in Lines 302-310.
Point 4: Line 231: What should be done with the crop residues in the areas of farmland
Response 4: We do not understand the question, can you please rephrase it? Lines 230-233 read, “Soil carbon stock was classified in ranges (10-20, 20-40, 40-60, 60-80, 80-100, 100-120, 120-140, and 140-160 Mg) for specified LULC (farmland, forest, and grassland) per area hectare within each mesoregion in Hungary (data received from the MTA Institute of Soil Science).”
Point 5: Results and discussion: Line 306 In figure 7 there is a large amplitude in the error bar of the areas of the South-Zselic, what can they be due to?.
Response 5: Indeed, the carbon stock of all land-use land-covers of South-Zselic presented large variation, while in the North, this variation was especially noticeable between the land-uses. In a study from Europe forest soils, Baritz et al. (2010) also found higher variations in soil carbon stocks from forests of different classifications. As in the case of our study, the authors also found a large variation in carbon stocks in Luvic soils rather than in Fluvisoils.
However, our hypothesis to explain the variations in our study is related to the complexity of the soil changes regarding different positions on the relief and to the different soil management practices (Duarte-Guardia et al., 2020). The South-Zselic region covers a broader area, with higher heterogeneity in the relief, as observed in Figure 3. Nevertheless, in this area, the soil is basically formed with a loam texture (LVh - Haplic Luvisol), which has a higher potential to keep the organic matter in its aggregates. In this area, the soil samples were closer but concentrated in three different spots. On the other hand, Vás-Pest-Danube Valley is smaller and it has basically the same relief, although there are three types of soils, especially with sandy texture (FLe - Eutric Fluvisols).
So these areas presented different factors that affect the soil carbon stock. It can show that some soils are more vulnerable to the effect of land-use management, the socio-economic factor that can be controlled, particularly ones that have sandy textures. It points out that these soils should receive more attention from policy measures in order to avoid carbon losses, while others are characteristic for presenting such variations which are probably connected to other biophysical factors and that which cannot be managed.
Point 6: I think it is appropriate that the COS values should be accompanied by other parameters to know what state that COS is in, such as labile carbon, water-soluble carbon, fulvic acids and humic acids
Response 6: Thank you for this suggestion. It would have been interesting to explore a complete analysis of the soil characteristics. However, in the case of our study, it seems slightly out of scope because (1) the goal was to present a more viable, resource-efficient option for researchers and land-use managers that do not have the resources required for full soil assessments, and (2) InVEST soil carbon stock mapping does not typically report such additional parameters (see Reference 1, Reference 2, Reference 3). Of course, we agree that when it comes to soil, the more you measure and analyse, the better it would be for environmental research. However, extensive sampling and lab analysis costs money and time that many land-use managers and researchers do not have.
For our future research, we are looking at including additional soil characteristics/parameters to accompany the InVEST soil carbon models, but this is a work in progress.
Point 7: General comments: The author refers to the comparison of the models as a solution for the decision-making of government agencies in terms of soil valuation and management, but, as referred to in the discussion of this article, carbon stocks in the soil are difficult to model accurately since it can show exceptionally large variations (line 451-453) due to the great heterogeneity in small areas and even more so in large areas, which makes it difficult for this study to show conclusive values of SOC in areas with large expanses.
Response 7: Thank you for your comments. We fully agree with the difficulties. The purpose of the sampling was to show the importance of sampling versus using the pre-set values/parameters from the model or national inventories. Exact values, measured in an accredited laboratory from the samples collected at the study area are assumed to be more accurate, and better reflect “real” soil carbon values compared to the over-generalised values used in national soil carbon inventories.
This study is geared toward providing a resource-efficient tool to land-use managers and researchers investigating sustainable soil management on the landscape scale. As pointed out by the FAO’s “Soil Organic Carbon Mapping Cookbook 2nd Ed.”, “Understanding the status of a given soil in different land uses, including its properties and functions, and relating this information to the various ecosystem services provided by them, becomes mandatory for sustainable soil management decisions. As the availability of soil data and information is fundamental to underpin these decisions, partners of the GSP decided to establish a Global Soil Information System (GLOSIS) based on a distributed model whereby the system is fed by national soil information systems.” So as more soil data becomes available, e.g., soil parameters of specific landscapes included in national inventories/datasets, soil carbon stock modelling will become more accurate. However, until such time, we present a methodology of reporting more contextualised and accurate data of soil carbon stock of landscape areas. As pointed out in Lines 508-513, “Given the limitations to this modelling method, it should be considered as a complementary tool for decision-making and a guide for understanding general CS trends on the landscape-scale. It should not be taken as an exact representation of soil carbon in these landscapes (as should be the case for the majority of soil carbon models). The InVEST soil carbon models can be used for regional land-use and change management recommendations and future spatial planning”.
In addition to the above comments, all spelling and grammatical errors pointed out by the reviewers have been corrected. A new relevant reference was added at Line 104. We look forward to hearing from you in due time regarding our submission and to responding to any further questions and comments you or the reviewers may have.
Sincerely,
Lyndre Nel
Reviewer 3 Report
There are few items that must be clarified and improved before this article can be considered for publication.
1) Lines 84-90: It was mentioned that the data used in soil carbon models are not sufficiently accurate. How are the data used in this study improved compared to the previous ones? What are the quality assurance and quality control (QA/QC) used for the data in this study?
2) Lines 238-241: How could soil organic carbon (SOC) be the product of soil weight and bulk density? The equation is NOT dimensionally consistent; hence, it cannot be correct. You will have to re-exam this equation and all the contents/data relevant its calculations. I suspect a lot of changes need to be made consequently.
3) If the main data processing approach was incorrect, the results and discussion wouldn't be right.
Author Response
Response to Reviewer 3 Comments
Point 1: Lines 84-90: It was mentioned that the data used in soil carbon models are not sufficiently accurate. How are the data used in this study improved compared to the previous ones? What are the quality assurance and quality control (QA/QC) used for the data in this study?
Response 1: You have raised an important point here, one that we focus on in this study. Generally, soil carbon inventories used for soil carbon modelling do not have any quality assurance and quality control (QA/QC) apart from a few statistical tests done on the data to check for normality in distribution (we did not do these in this study as we used limited data compared to large-scale studies on soil carbon, and this type of testing needs advanced computer software capacities) (Reference 1, Reference 2) .
We state in Lines 92-95 that it would be ideal to do vast soil survey sampling across large areas, as this is the only way to improve the quality of soil carbon datasets – but this is not feasible due to associated high resource costs. Thus, this study presents a very practical and resource-efficient method for landscape-level researchers and analysts to determine and report soil carbon models that improve the quality of soil carbon determination, i.e. reporting soil carbon ranges from minimum, mean to maximum based on selective soil sampling in combination with the National soil carbon inventory of Hungary.
Point 2: Lines 238-241: How could soil organic carbon (SOC) be the product of soil weight and bulk density? The equation is NOT dimensionally consistent; hence, it cannot be correct. You will have to re-exam this equation and all the contents/data relevant its calculations. I suspect a lot of changes need to be made consequently.
Response 2: Thank you for pointing this out and we agree. Between manuscript drafts, the last part of the formula seems to have been removed somehow and this mistake was overlooked in the final check before submission. We apologize for this. The FAO’s 2018 report Soil Organic Carbon Mapping Cookbook, page 15 [Reference 7], shows their formula to determine SOC stock for mineral soils as
SOC = d * BD * (Ctot-Cmin) * CFst
where:
SOC = soil organic carbon [kg/m2]; d = depth of horizon/depth class [m]; BD = bulk density [kg/m3]; Ctot and Cmin = total and mineral (or inorganic) carbon [g g-1]; CFst= correction factor for stoniness and gravel content.
Our formula used
SOC = d * BD * (Corg)
SOC = soil organic carbon [kg/m2]; d = depth of horizon/sample [m]; BD = bulk density [kg/m3]; and Corg = organic carbon [g g-1].
Total carbon, stoniness and gravel were not used as this was not calculated. Lines 240-243 were updated to include Corg as it was directly analysed in the lab from soil samples, and we replaced “soil weight” with depth as this was incorrectly stated in the manuscript.
Point 3: If the main data processing approach was incorrect, the results and discussion wouldn't be right.
Response 3: Please see Response 2.
In addition to the above comments, all spelling and grammatical errors pointed out by the reviewers have been corrected. A new relevant reference was added at Line 104. We look forward to hearing from you in due time regarding our submission and to responding to any further questions and comments you or the reviewers may have.
Sincerely,
Lyndre Nel
Round 2
Reviewer 1 Report
Dear authors,
All the comments have been addressed as per the previous suggestion.
With best,
The reviewer
Comments for author File: Comments.pdf
Reviewer 2 Report
I accept the small modifications made by the author.
Reviewer 3 Report
The previous comments have been properly addressed.
