Assessing Biodiversity Conditions in Cocoa Agroforests with a Rapid Assessment Method: Outcomes from a Large-Scale Application in Ghana

Round 1

Reviewer 1 Report

The manuscript presents the method of assessing biodiversity conditions in cocoa agroforests, developed by the authors of the manuscript. Personally, I would consider it as preliminary method than the final method. The described method is relatively simple and cheap to perform. It also does not require extensive specialist knowledge. This is a plus. However, it should then be verified on a more specialized and detailed level.

The text of the manuscript itself is a bit too convoluted. I suggest using shorter and more communicative sentences.

Fig. 3 is very hard to read. Diagrams will be very useful in this manuscript. However, they must be legible for the recipients. It is worth considering adding more illustrations of the presented method. More detailed and more readable.

Statistical analyzes are definitely lacking in the manuscript. They are necessary to be able to evaluate the performance of the presented method and draw appropriate conclusions. There is no indication of dependencies, variability, etc.

The discussion could be more elaborate. At the moment it is at the minimum level.

Author Response

The manuscript presents the method of assessing biodiversity conditions in cocoa agroforests, developed by the authors of the manuscript. Personally, I would consider it as preliminary method than the final method. The described method is relatively simple and cheap to perform. It also does not require extensive specialist knowledge. This is a plus. However, it should then be verified on a more specialized and detailed level.

R1: We appreciate all the comments and suggestions provided. We understand that information on how the method was developed and how its robustness was verified is required in this manuscript, despite being detailed elsewhere (Raneri et al. 2021, [28]). It is not a preliminary method, although there is always room for improvement and adaptations according to the conditions of its application. The choice of the method and its robustness and representativeness were tested and confirmed by a related study, described in detail in Raneri et al., 2021. We understand that further info on the method and how it was developed is required in this manuscript, as such, we have explained the method further, please see lines 117-159.

The text of the manuscript itself is a bit too convoluted. I suggest using shorter and more communicative sentences.

R2: We appreciate the suggestion, and we have thoroughly revised the text of the manuscript.

Fig. 3 is very hard to read. Diagrams will be very useful in this manuscript. However, they must be legible for the recipients. It is worth considering adding more illustrations of the presented method. More detailed and more readable.

R3: We have revised Figure 3, adding some points to clarify the steps of the procedure.

Statistical analyzes are definitely lacking in the manuscript. They are necessary to be able to evaluate the performance of the presented method and draw appropriate conclusions. There is no indication of dependencies, variability, etc.

R4: The statistics regarding the method accuracy have been presented in the article dedicated to the method testing and selection (please see Raneri et al., 2021, 28). The sub-section added provides further information on how the method was evaluated (Lines 117-159). After the evaluation process done to the method beforehand, the purpose of this manuscript is, thus, to demonstrate how it can be applied to obtain data at large scale and how these data can be used to show field performance and to compare indicators status across different spatial units.

The discussion could be more elaborate. At the moment it is at the minimum level.

R: We have revised the discussion, elaborating further on topics regarding the method potential and limitations, future pathways of research, the involvement of the farmers in the process. Please see the revised discussion section (L407-464).

Author Response File: Author Response.docx

Reviewer 2 Report

The study presents the results of a rapid assessment method applied in Ghana, developed to measure non-cocoa tree characteristics based on easily observed parameters using sample surveys and mapping tools.

1.      It is a well-written and interesting work, although it can be improved mainly in the theoretical and practical support of the thresholds of the indicators proposed for the evaluation, one of the most relevant assumptions of the work. For this, a table could be included that supports how the limits of the thresholds and the categories were established from the theoretical-practical review since the numbers are continuous (for example, what is the real difference in that there are 18 trees per hectare for a high threshold concerning 17. I think it must be placed in a broad context and, above all, understand why these numbers are occurring in each context and assess the relevance of its extrapolation to other contexts.

2.      It is established that the selection of the study units was related to their participation in the Traceability and Mapping System promoted by the cocoa industry. How representative would these units be of the diversity of other forms of agroforest management in the region?

3.      What are the limitations of the sampling method? For example, for larger units size or for distribution of agroforestry practices (trees in other vertical and horizontal arrangements).

 4.      Another element that I suggest is that the relevance, limitations, and risks of the proposal be discussed, for example, concerning indicators such as the presence of native tree species, one of the most relevant elements for the maintenance of biodiversity and the most beyond the richness of species, in agroforestry systems.

5.      The study can be favored using multi-criteria analysis of the indicators to identify how they operate together in the different systems and how they can be improved.

6.      The study could be greatly enriched if the authors included the reasons of the agroforesters for the decisions regarding the presence of trees, their identity, and characteristics that would allow us to see to what extent the categories proposed in the study are achievable in the different socioecological contexts where the events occur cocoa agroforests.

Author Response

The study presents the results of a rapid assessment method applied in Ghana, developed to measure non-cocoa tree characteristics based on easily observed parameters using sample surveys and mapping tools.

1. It is a well-written and interesting work, although it can be improved mainly in the theoretical and practical support of the thresholds of the indicators proposed for the evaluation, one of the most relevant assumptions of the work. For this, a table could be included that supports how the limits of the thresholds and the categories were established from the theoretical-practical review since the numbers are continuous (for example, what is the real difference in that there are 18 trees per hectare for a high threshold concerning 17. I think it must be placed in a broad context and, above all, understand why these numbers are occurring in each context and assess the relevance of its extrapolation to other contexts.

R1: We thank the reviewer for the feedback and suggestions.

We agree that the thresholds defined must be justified, therefore we have provided background information as supplementary material. The thresholds were defined based on existing studies, specifically regarding certification schemes, and by a process of consultation with different stakeholders, described in the reference provided (28). The 18 trees per hectare, for example, results from the thresholds defined by Rainforest Alliance and UTZ certified for certification. Regarding the example provided, it is a fact that close values may be placed in different classes without representing significant changes in biodiversity levels, however thresholds are required to define a common baseline, to compare at different scales and to monitor status over time.

2. It is established that the selection of the study units was related to their participation in the Traceability and Mapping System promoted by the cocoa industry. How representative would these units be of the diversity of other forms of agroforest management in the region?

R2: Indeed, this study was implemented in combination with a traceability and mapping system promoted by the cocoa industry. We have added some more info on this process in the text (L117-158) to clarify the process. Despite the selection of farms being limited to the mapping system coverage, the sample size obtained is rather large, it contains nearly 9000 cocoa fields spread by 6 regions and 19 districts, and different types of agroforests (size, location, configuration) are included. The purpose of the study was to demonstrate the use of a rapid assessment method at large scale and to present a straightforward classification at field level and across regions, enabling harmonized baseline and comparisons. In any case, the data obtained can be used and visualized in different ways, and this point was added in the discussion, please see L442 and following.

3. What are the limitations of the sampling method? For example, for larger units size or for distribution of agroforestry practices (trees in other vertical and horizontal arrangements).

R3: As any other sampling method, the one used has some limitations, some have been previously discussed in an article dedicated to the design and testing of the method (Ref 28). The limitations of the method and potential adaptations and improvements are further discussed now, please see L397-411, and L435-446.

4. Another element that I suggest is that the relevance, limitations, and risks of the proposal be discussed, for example, concerning indicators such as the presence of native tree species, one of the most relevant elements for the maintenance of biodiversity and the most beyond the richness of species, in agroforestry systems.

R4: We agree that there are other relevant indicators that were not included, specifically tree abundance and the native tree species, because the collection of such data requires specialized knowledge and additional resources. To ensure the replicability of the method at large scale, the need for specialized knowledge that would require additional resources had to be reduced. However, together with the sampling procedure, it is possible to collect information on existing tree species from the farmers that accompany the mapping of the farm borders and the sampling, as now highlighted in the discussion (please see L442-451). Despite the usefulness of such data, which could allow the calculation of tree abundance and the proportion of native species, for example, there are some issues regarding the need to present the species by their local names to be identified by farmers, and the need to validate the information provided by the farmers with in-depth assessments.

5. The study can be favored using multi-criteria analysis of the indicators to identify how they operate together in the different systems and how they can be improved.

 R5: We appreciate and agree with the suggestion, although we think it is more suitable to a future study, as this one is focused on demonstrating the application at large scale of the method and how straightforward information could be obtained to check the current status of fields and compare among districts and regions. We added some points for potential research pathways in the discussion.

6. The study could be greatly enriched if the authors included the reasons of the agroforesters for the decisions regarding the presence of trees, their identity, and characteristics that would allow us to see to what extent the categories proposed in the study are achievable in the different socioecological contexts where the events occur cocoa agroforests.

R6: We appreciate the suggestion. This type of questions, on the reasons for the choice of specific trees, was not included in the method, which was focused on a quick sampling procedure. We think this would be a potential approach for a future use of the method, with the adjustment regarding the addition of a farmer’s questionnaire, as mentioned now in the discussion, see L442-445.

Author Response File: Author Response.docx

Reviewer 3 Report

The manuscript “Assessing biodiversity conditions in cocoa agroforests with a rapid assessment method. Outcomes from a large-scale application in Ghana” reports an original study. The results represent a very extensive field survey, and the data are likely of interest to cocoa researchers and those interested in shade tree management and accounting for biodiversity metrics. The ‘rapid assessment’ method tested here at large scale across multiple cocoa growing regions in Ghana would be of interest to both researchers and practitioners interested in scaling traceability programs for cocoa agroforestry. The authors describe the advantages and limitations of the rapid assessment method, which is relevant for future work in this area. However, there are some important gaps in describing the process and potential application, which I elaborate in ‘major comments’ below.

Generally, the manuscript is written clearly, but there are several places where sentences are very long and difficult to follow. I made a few notes in ‘minor comments’ on where these are, but a larger editing effort should be done to streamline sentences to be more succinct and clearer.

Major comments:

11)  More is needed on the selection of this specific rapid assessment method. While a reference to previous work evaluating this process is listed [ref 28], it would be informative to know the key findings from that work in terms of accuracy.

2 2) More detail is required to improve the reproducibility of data analysis. The statement in the methods “The main procedure for data analysis was developed in R software [39], implementing automated routines based on open-source tools, to facilitate adjustments, up-dates and the replication to other areas (Figure 3)” is unclear. What tools? What kind of “updates” and what “areas”? Again, in the discussion, “open-access tools and automated routines” (line 382), what are these?

33) The manuscript lacks a discussion on the use a generalized classification scheme for all regions. The authors point out the data could be adapted to new threshold levels (lines 399-401), this could be a point to elaborate on other ways to improve the classification system to be more ecologically relevant to a region; notably distinct climatic conditions that are experienced between different cocoa regions of Ghana that likely have different interactions with shade tree densities and types etc.

44)  Given the large data set, I think more empirical testing is appropriate or worth suggesting for future work.  Notably, how would these indicators be interpreted together? Especially given the ones measured are not independent (many with number of trees and dbh in the calculation).

Minor comments:

I suspect the journal will want decimals (.) instead of commas (,) in numbers with decimal places. E.g., Table 3.

lines 22-23: This is hard to follow. What is meant by succession potential? It becomes clearer in the manuscript. Perhaps re-word here just to clarify, at least something like potential for tree succession. And  "required strong improvements" is also unclear.

line 40: I don't think you are talking specifically about tree-based intercropping, a term typically in reference to intercrops of trees and annuals (alley cropping).

lines 56-59: Can you break this very long sentence up into two? It is hard to follow and understand what "their" is referring to specifically.

lines 72: I think "biodiversity indicators" is more descriptive. I think explanation is needed on how this assessment can bridge analysis at “different spatial scales”. And it is unclear how it can be analyzed at different temporal scales.

Lines 74-75: Statement on recommended benchmark values is too vague.

Lines 75-77: This statement is unclear; too vague.

lines105-114: This lacks justification. There is a reference to [28] but I do think some explanation as to what this method does is needed.

Lines 129-130: This is too vague. What environmental conditions?

line 148: I do not understand "extracting the number of recommended and other classes". Please revise.

line 149: what is meant by "overall status" is this defined yet?

lines 136-151: Very long and hard to follow sentences. Please revise to make more succinct and clearer.

line 158: "Further examination" of what?

lines 166-167: "implementing automated routines based on open-source tools, to facilitate adjustments, up-dates and the replication to other areas " is vague. All these steps would need some more detail to understand exactly what was done to improve the reproducibility of this study,

lines 178-179: Can you contextualize the mean farm size surveyed in this study with typical farm size distributions in the country?

line 179: superscript 2 in 200m2

Table 2 needs more detail on the units for each indicator.

line 210: I suggest changing "not desirable" to "not optimal" for cocoa production. This statement should also have references.

line 245: "and succession may be undermined" should be reworded to something like "indicating a lower potential for tree succession over time" or something like that.

line 261: Change "didn't" to "did not".

Table 4: This is difficult to interpret - is "species" species richness? "trees" tree density? Make sure these are more accurately labelled.

line 347: "exotic species" do you mean "non-native tree species"?

lines 374-375: This idea is unclear - why "particularly when combined with organic cocoa production"?

line 375: What does "This supports the maintenance of other trees in cocoa fields" mean? What is "this" in reference to?

lines 381-382: What are "the use of open-access tools and automated routines".

lines 384-392: Needless to say farmers are very busy. How would this scheme role out in practice? Is there compensation? A statement about the labour demands (not just in the assessments but in training and additional time demands) should be discussed.

lines 395-399: This is a very long sentence.

Author Response

The manuscript “Assessing biodiversity conditions in cocoa agroforests with a rapid assessment method. Outcomes from a large-scale application in Ghana” reports an original study. The results represent a very extensive field survey, and the data are likely of interest to cocoa researchers and those interested in shade tree management and accounting for biodiversity metrics. The ‘rapid assessment’ method tested here at large scale across multiple cocoa growing regions in Ghana would be of interest to both researchers and practitioners interested in scaling traceability programs for cocoa agroforestry. The authors describe the advantages and limitations of the rapid assessment method, which is relevant for future work in this area. However, there are some important gaps in describing the process and potential application, which I elaborate in ‘major comments’ below.

Generally, the manuscript is written clearly, but there are several places where sentences are very long and difficult to follow. I made a few notes in ‘minor comments’ on where these are, but a larger editing effort should be done to streamline sentences to be more succinct and clearer.

R: Thank you for the positive feedback and for the comments provided. We appreciate the thorough review done and the detailed suggestions. We try to respond below to each point raised, and the suggested changes were applied whenever possible, to improve the clarity and quality of the manuscript.

Major comments:

11)  More is needed on the selection of this specific rapid assessment method. While a reference to previous work evaluating this process is listed [ref 28], it would be informative to know the key findings from that work in terms of accuracy.

R: Although the method is detailed elsewhere, we understand that information is lacking in the manuscript. We added further information on the rapid assessment method applied, explaining briefly how the method was selected and how its robustness was verified. Please see L105-159 and the added sub-section on the rapid assessment method applied.

2 2) More detail is required to improve the reproducibility of data analysis. The statement in the methods “The main procedure for data analysis was developed in R software [39], implementing automated routines based on open-source tools, to facilitate adjustments, up-dates and the replication to other areas (Figure 3)” is unclear. What tools? What kind of “updates” and what “areas”? Again, in the discussion, “open-access tools and automated routines” (line 382), what are these?

R: We explain in more detail the use of these tools, adding a sub-section dedicated to the statistical and spatial analysis tools (see L213-232). We are also able to provide the script (automated routine) developed to perform the analysis, as indicated in the supplementary material. The method was tested and applied in Ghana but it is prepared to be used elsewhere, in other countries and regions, and for other commodities. The use of open-source tools mentioned were r software and qgis, which are free and widely available, and do not require additional financial efforts to be used (besides initial training).

33) The manuscript lacks a discussion on the use a generalized classification scheme for all regions. The authors point out the data could be adapted to new threshold levels (lines 399-401), this could be a point to elaborate on other ways to improve the classification system to be more ecologically relevant to a region; notably distinct climatic conditions that are experienced between different cocoa regions of Ghana that likely have different interactions with shade tree densities and types etc.

R: We agree that the classification scheme can be adjusted to regional contexts, but in this case, the purpose was to demonstrate the potential use of this method for a large scale application that could enable a wider assessment and comparisons between regions. We think this is a pertinent point and added it in the discussion, please see L456-463.

44)  Given the large data set, I think more empirical testing is appropriate or worth suggesting for future work.  Notably, how would these indicators be interpreted together? Especially given the ones measured are not independent (many with number of trees and dbh in the calculation).

R: We thank the reviewer for raising another relevant point, and we added further discussion on potential pathways for future research that can result from the application of such method.

Minor comments:

I suspect the journal will want decimals (.) instead of commas (,) in numbers with decimal places. E.g., Table 3.

R: Change as suggested throughout the manuscript.

lines 22-23: This is hard to follow. What is meant by succession potential? It becomes clearer in the manuscript. Perhaps re-word here just to clarify, at least something like potential for tree succession. And  "required strong improvements" is also unclear.

R: Following the suggestion, the phrase was revised.

line 40: I don't think you are talking specifically about tree-based intercropping, a term typically in reference to intercrops of trees and annuals (alley cropping).

R: In cocoa agroforestry, intercropping is used as well, since banana and plantain are planted together with the trees in a cocoa field, for example.

lines 56-59: Can you break this very long sentence up into two? It is hard to follow and understand what "their" is referring to specifically.

R: We have changed the text according to the suggestion, thank you.

lines 72: I think "biodiversity indicators" is more descriptive. I think explanation is needed on how this assessment can bridge analysis at “different spatial scales”. And it is unclear how it can be analyzed at different temporal scales.

R: We have provided further explanation on how different spatial scales can be analysed, please see lines 77-86. Analysis of the collected data can be done at field level, or aggregating at district and regional levels, for example. For temporal scales, the same data procedure applied in different periods allows comparisons over time of the indicators’ levels. In this study, we only applied it once, therefore it was not possible to compare the performance of the fields over time, but the methodology enables this step.

Lines 74-75: Statement on recommended benchmark values is too vague.

Lines 75-77: This statement is unclear; too vague.

R: This sub-section was revised accordingly and clarification was added in the text (lines 73-86)

lines105-114: This lacks justification. There is a reference to [28] but I do think some explanation as to what this method does is needed.

R: We understand that further explanation on the rapid assessment method is required in this manuscript, although it has been presented in detail elsewhere. As such, we have added information on the method and briefly describe the calibration and validation procedures applied. Please see L105-159.

Lines 129-130: This is too vague. What environmental conditions?

R: The environmental conditions assessed are presented in Table 1 and Figure 3; we have added also in the text to clarify the indicators calculated.

line 148: I do not understand "extracting the number of recommended and other classes". Please revise.

R: We have clarified the meaning of the sentence. We counted the number of recommended levels each field attained, out of the 6 indicators evaluated, to compare their overall environmental performance.

line 149: what is meant by "overall status" is this defined yet?

R: The text was revised to clarify the analysis carried out, please see L185-195.

lines 136-151: Very long and hard to follow sentences. Please revise to make more succinct and clearer.

R: The text was revised to clarify the analysis carried out, please see L185-195.

line 158: "Further examination" of what?

R: Thank you for pointing out the missing text. This section of the text was revised to clarify the steps done, see L201 and following.

lines 166-167: "implementing automated routines based on open-source tools, to facilitate adjustments, up-dates and the replication to other areas " is vague. All these steps would need some more detail to understand exactly what was done to improve the reproducibility of this study.

R: In this part, the text followed the common description used in spatial analysis topics. We understand that further detail may be needed to clarify the use of these spatial analysis tools in the context of biodiversity/environmental conditions in agroforestry. As such, we have added information on how the analysis tools and procedures were implemented and their advantages for replication and possible updates. Please see the added sub-section in the manuscript (“2.3.3 Statistical and spatial analysis tools”), L212-231.

lines 178-179: Can you contextualize the mean farm size surveyed in this study with typical farm size distributions in the country?

R: This information was added in the new sub-section dedicated to the method, as it was designed based on the mean size of cocoa fields in the country, and mapped within the TMS initiative. According to Ghana COCOBOD, cocoa farms size range between 0.4 to 4 hectares.

line 179: superscript 2 in 200m2

R: Corrected as suggested

Table 2 needs more detail on the units for each indicator.

R: Corresponding units were added in Table 2.

line 210: I suggest changing "not desirable" to "not optimal" for cocoa production. This statement should also have references.

R: Text was changed as suggested. We added the indication of the supplementary material, where supporting references are presented for each indicator and threshold defined.

line 245: "and succession may be undermined" should be reworded to something like "indicating a lower potential for tree succession over time" or something like that.

R: Text was changed as suggested

line 261: Change "didn't" to "did not".

R: Changed as suggested.

Table 4: This is difficult to interpret - is "species" species richness? "trees" tree density? Make sure these are more accurately labelled.

R: We have added the units of measurement to each parameter

line 347: "exotic species" do you mean "non-native tree species"?

R: Yes. Exotic species are those that occur in areas outside of their natural geographic range and are introduced in other areas. They can be referred by other names, such as non-native or non-indigenous. We added clarification in the text.

lines 374-375: This idea is unclear - why "particularly when combined with organic cocoa production"?

R: Text was revised; indeed, this sentence was misplaced.

line 375: What does "This supports the maintenance of other trees in cocoa fields" mean? What is "this" in reference to?

lines 381-382: What are "the use of open-access tools and automated routines".

R: Following prior suggestions, the use of such tools has been explained in the additional sub-section in the methods (L212-229).

lines 384-392: Needless to say farmers are very busy. How would this scheme role out in practice? Is there compensation? A statement about the labour demands (not just in the assessments but in training and additional time demands) should be discussed.

R: Farmers were voluntarily participating in the surveys, as the mapping of the border requires their presence and permission is needed to measure the trees in their fields. But they were not the surveyors, who were hired instead by the cocoa industry in the framework of the traceability and mapping system. A direct result for them was the mapping of their own field, and they could have access to the results, as the purpose was to help them implement biodiversity-friendly measures and ensure both sufficient yield and sustainability of the cocoa production. We understand this was not clear enough and we added some points in the discussion (L442-454).

lines 395-399: This is a very long sentence.

R: Sentence was revised, as it was the discussion section.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The authors acted on the reviewers' comments and significantly improved the manuscript.

Reviewer 2 Report

No additional comments