Impact of Cropland Management on Invertebrate Richness and Abundance in Agroforestry Systems in Bali, Indonesia

Round 1

Reviewer 1 Report

The study evaluates the impact of crop management (i.e., polyculture vs rustic) on the taxonomic richness and abundance of invertebrates in West Bali, Indonesia. A total of 53 plots were sampled using pitfall and pan traps for invertebrates. Shade cover was estimated at each plot. Whenever possible, taxa were determined to species level. If not possible, specimens were classified by family and assigned a morphospecies for analysis. Generalized linear models were used to evaluate the relationship between invertebrate taxonomic richness and abundance explained by crop management type, canopy cover, tree species richness, crop species richness and yields. A total of 2,603 specimens were collected, belonging to 133 morphospecies. Main findings include a positive relationship between canopy cover and the abundance of non-pest dipterans; and with the abundance and species richness of Formicidae. A negative relationship between canopy cover and the abundance of Orthoptera. Tree richness positively related with the abundance and species richness of biological control Hymenoptera and Orthoptera. Crop species richness positively related with Zygentoma abundance.

            The manuscript has a good sample size and it is well written. In the introduction, more emphasis should be made on the habitat-heterogeneity hypothesis since it is the source of the study’s predictions. In various parts of the text there are misleading statements on the habitat-heterogeneity hypothesis which need to be corrected (see below). My main concern is the sampling design. There are 31 plots for rustic and only 22 for polyculture systems. This is a wide difference in sampling units that is very likely masking the actual invertebrate species richness and abundance. Sampling effort needs to be controlled for and analyses run again before the manuscript can be considered for publication. Another important concern is the location of five polyculture plots which are in the production forest and not in the agricultural land as is the case for the rest of polyculture systems (see below for specific recommendations). 

Comments for author File: Comments.pdf

Author Response

The study evaluates the impact of crop management (i.e., polyculture vs rustic) on the taxonomic richness and abundance of invertebrates in West Bali, Indonesia. A total of 53 plots were sampled using pitfall and pan traps for invertebrates. Shade cover was estimated at each plot. Whenever possible, taxa were determined to species level. If not possible, specimens were classified by family and assigned a morphospecies for analysis. Generalized linear models were used to evaluate the relationship between invertebrate taxonomic richness and abundance explained by crop management type, canopy cover, tree species richness, crop species richness and yields. A total of 2,603 specimens were collected, belonging to 133 morphospecies. Main findings include a positive relationship between canopy cover and the abundance of non-pest dipterans; and with the abundance and species richness of Formicidae. A negative relationship between canopy cover and the abundance of Orthoptera. Tree richness positively related with the abundance and species richness of biological control Hymenoptera and Orthoptera. Crop species richness positively related with Zygentoma abundance.

The manuscript has a good sample size and it is well written. In the introduction, more emphasis should be made on the habitat-heterogeneity hypothesis since it is the source of the study’s predictions. In various parts of the text there are misleading statements on the habitat-heterogeneity hypothesis which need to be corrected (see below). My main concern is the sampling design. There are 31 plots for rustic and only 22 for polyculture systems. This is a wide difference in sampling units that is very likely masking the actual invertebrate species richness and abundance. Sampling effort needs to be controlled for and analyses run again before the manuscript can be considered for publication. Another important concern is the location of five polyculture plots which are in the production forest and not in the agricultural land as is the case for the rest of polyculture systems (see below for specific recommendations).

We thank the reviewer for recognising the importance of the paper, and we accept the important revisions suggested. For some of them we thought it required more clarification and we have also edited the manuscript accordingly. We specify the edits for each point below.

L31-32: Please correct. “The habitat-heterogeneity hypothesis, developed initially by MacArthur and MacArthur (1961), proposes that an increase in number of different habitats can lead to an increase in species diversity” (Cramer & Willig 2002, J Mammal, 2002).

Data on birds and plants the led to this hypothesis, not on invertebrates. Further, the habitat heterogeneity hypothesis does not propose a higher abundance. Please carefully check statements on this matter in all the manuscript.

We have now added a section in the introduction about this. There is a lot of evidence of applicability of this theory on inverts, and also linking with a abundance not just richness.

L68: Please add reference.

Added

L67-68: In “it is the presence of species, such as invertebrates, that creates productive landscapes”, what do the authors mean by“productive”? Is it agricultural or ecological productivity? Please explain.

Edited

L72: presence invasion of invasive species

Changed

L94-97: Is there any research on the abundance and richness of other taxonomic groups in the production forests? If there is, please briefly include. If not, please mention that.

We are the first ones to work in the area (at least legally with permits) as far as we know, that is confirmed by the local forest department and university. We have now added a sentence.

L98: Since the habitat-heterogeneity hypothesis is the foundation of your predictions, it should be mentioned, explained and exemplified earlier in the introduction.

This is a very good suggestion, we have now added a section earlier in the introduction

L110: Why sampling occurred in the drier season? I think it would be good to include this in the text.

Good point, we did so to limit the impact of weather on the survey method, we have now added a line.

Figure 1: Is the temperature scale correct? According to the graphic, from January-May and Mid September-December temperatures are above 50 Celsius.

We realised that the figure could have been confusing, we have now highlighted the temperature line in red.

L133: Why sample size is not the same for both systems? How can you be sure that abundance and richness is not a consequence of sample size instead of production system?

We expand on this point below, but to summarise we did not intend to provide total estimates of diversity and abundance (see section 4.2), we are aware that there would be more species (in both habitats). We aimed to understand how the factors linked with land management influenced richness and abundance of invertebrates. In that sense we used plots as statistical unit, so it is not a major issue if we do not have exactly the same number of plots in each system (in an ideal word all treatments should be the same or evenly distributed in case of continuous variables, including for example shade cover, crop richness, etc., but it is often not the case as we need to work in tropical context and with local communities) providing that we have enough to infer mean estimates from a statistical model. 

L135: Fifty meters seem a short distance between traps to be considered as independent samples. Many insects can move hundreds of meters in short periods of time (i.e., one night).

We agree that this might be a short distance for some taxa but we based our choice on several papers who used that distance in forested areas (we have now added the references)

Figure 3: There are 6 plots in the production forest area instead of 5, as mentioned in the text and figure legend. Please verify and correct. I am concerned about those five sites. They are embedded in a forest/rustic production matrix. Therefore, one would expect the presence of more forest species than in the rest of the polyculture systems. My suggestion is to check if the abundance and species composition in these five plots statistically differ from the other polyculture plots before including them in this class. If they in fact are similar, then they, in theory could be considered as polyculture plots.

We thank the reviewer for this comment, that was also highlighted by other reviewers. We have now added a section in the methods. Mainly, production forests are a political land zonation that does not necessarily reflect into practice. In theory, production forests should be used sustainably, but in practice many production forests in Indonesia are converted into agricultural lands, including monocultures. In the study area the production forest is well preserved, but there were some areas that were cleared in the past and does not host native forest trees, that is why those plots were classed as polycultures.

In the process of checking the data we also realised one plot with 1 native forest tree species was incorrectly assigned to polyculture so we redid the analyses

L159: tree richness or tree species richness?

It is tree species richness but we shorten to tree richness, we have now mentioned that the first time.

L172: Please add manufacturer.

The manufacturer is Olympus the model is SZ61

L193-195: As read, it seems that profit from yields was estimated for each farm. However, I assume that the total area per farm is different in each farm. Below in Figure 4, total yields are calculated as USD/ha. Please explain in the text if this is how it was calculated for the statistical analysis.

We have now added more text in the description.

Results: Please include your findings by system for: shade cover, crop species richness, tree species richness, and productivity.

Also based on the request from another reviewer, we have now added this information in the results (Table 1 and linked text)

Table 1. Here the problem of sampling sites per system becomes evident. Richness and abundance is considerable higher in rustic system, which coincides with the original prediction. But sampling sites in polyculture systems is nearly 30% lower than in rustic ones.

Of course, if we added nine polyculture plots, abundance and richness would increase. As a practical example, if we consider that abundance and richness increase steadily by plot, in 31 polyculture plots we would have an abundance of 827, which of course remains way below 2016 but it’s quite different from the original 587. In the case of richness, one could expect 100 taxa. In this case, richness from polyculture would be close from richness in rustic systems of 105. Of course, the relationship between sampling effort and richness is not linear but do you see my point?

It is imperative that you find a way to control for sampling effort prior to run statistical analyses again.

We understand the point from the reviewer, but we would like to note that the scope of the paper was not to estimate the total richness and abundance of invertebrates per system, but rather show how richness and abundance of invertebrates are influenced by system, shade, tree richness, crop richness, and yields. We did include this reflection in the section 4.2. The data were analysed by plots, so it is not a major issue if the sample size is unbalanced between the two systems, provided that there is enough sample size to estimate model averages. We understand that the total number of invertebrates is not comparable, that is why we did not discuss that, but we still think it is important to show the inventory of species found.     

L236: Please rephrase. Yields from a crop do not have any impact on the abundance of any taxa.

Sorry we cannot reconnect this to the original text, maybe the line numbering is not the same. Could you please indicate the new line number?

Figure 4: Maybe lowering the y scale to 5000 would be good. In legend, as it is written it seems that all rustic and polyculture systems in the Yeh Embang Kauh village are included in the graphic, which I guess is not the case. Please change accordingly.

We have now edited the graph accordingly

L251: Same as in L236.

Again sorry we were not able to reconnect this to the original text

L273-276: What about your findings?

We did not threat canopy cover as categorical variable so we are just referring to what known in literature (also based on our previous research).

- Please make sure to specify to what richness you are referring to (i.e., species richness) throughout the text.

We have now added this the first time we mention crop and tree species richness

Reviewer 2 Report

This is an interesting research showing the impact of farm management on invertebrate richness and abundance in agroforestry systems in Bali, Indonesia. The paper is well written according to academic standards, clearly describing research gap, objectives, applied methodology and obtained results as well as conclusions. The only suggestion I have is to describe some basic characteristics of the polyculture system and rustic system.

Author Response

This is an interesting research showing the impact of farm management on invertebrate richness and abundance in agroforestry systems in Bali, Indonesia. The paper is well written according to academic standards, clearly describing research gap, objectives, applied methodology and obtained results as well as conclusions. The only suggestion I have is to describe some basic characteristics of the polyculture system and rustic system.

We thank the reviewer for the kind comments. We have now added more description of the two systems so that it is easier to understand the methods and design. 

Reviewer 3 Report

The authors present an original paper titled ‘Impact of farm management on invertebrate richness and abundance in agroforestry systems in Bali, Indonesia’, which I can recommend for publication in Land after minor corrections.

These are pretty simple studies on invertebrate biodiversity, which, in general, indicate that increased habitat complexity promotes higher invertebrate species richness and abundance. However, the studies may be significant locally due to the rising awareness of how the management of different areas can lead to increases in biodiversity. The authors are aware of their study limitations and highlighted them in the manuscript. I am satisfied that they plan to improve their research workshop and expand their study to areas with a different management system.

In my opinion, the reviewed manuscript has a clear structure and is written in a very reader-friendly way. The title is clear and adequately reflects the content. The abstract and the introduction are suitable and sufficient. The methods are appropriately chosen, and the material is well collected. The results are clearly presented, and the discussion is interesting and comprehensive. The references are adequately selected and well-balanced.

 My only objection is to the conclusions, which, in my opinion, are not conclusions but rather a further discussion or summary. I would ask you to draw synthetic conclusions from your study.

In general, the manuscript, as the preliminary study, is interesting and can be published after minor corrections.

Author Response

The authors present an original paper titled ‘Impact of farm management on invertebrate richness and abundance in agroforestry systems in Bali, Indonesia’, which I can recommend for publication in Land after minor corrections.

These are pretty simple studies on invertebrate biodiversity, which, in general, indicate that increased habitat complexity promotes higher invertebrate species richness and abundance. However, the studies may be significant locally due to the rising awareness of how the management of different areas can lead to increases in biodiversity. The authors are aware of their study limitations and highlighted them in the manuscript. I am satisfied that they plan to improve their research workshop and expand their study to areas with a different management system.

In my opinion, the reviewed manuscript has a clear structure and is written in a very reader-friendly way. The title is clear and adequately reflects the content. The abstract and the introduction are suitable and sufficient. The methods are appropriately chosen, and the material is well collected. The results are clearly presented, and the discussion is interesting and comprehensive. The references are adequately selected and well-balanced.

My only objection is to the conclusions, which, in my opinion, are not conclusions but rather a further discussion or summary. I would ask you to draw synthetic conclusions from your study.

We thank the reviewer for the kind comments. We accept the edits suggested and reworded the conclusion so that it does not sound like we have an additional discussion. We moved part of the conclusion to the discussion.

In general, the manuscript, as the preliminary study, is interesting and can be published after minor corrections.

Reviewer 4 Report

The paper investgated the impacts of farm management on invertebrate richness and abundance in agroforestry systems in Bali. Plots from two  systems (rustic system and polyculture system) were set up and used for comparision. The study is very interesting. I have some comments about the experiment and result.

1.  How many times did the author retrieve invertebrate traps? You descriped that collected data from May to August 2023, so you retrieve invertebrate traps for one time, or every day from May to August 2023?

2. The data in Table A1 is very important ,and I suggested to show these data in the main text. Table A1 is too long in this version, simple it (just give Z value and * label, and may also delete some taxa with low abundance).

3. what about the effects of system type (rustic system and polyculture system) on canopy cover, crop richness, tree richness, and yield?  I suggested also show thest results, and can be used to indicate which management is better for biodiversity.

Author Response

The paper investgated the impacts of farm management on invertebrate richness and abundance in agroforestry systems in Bali. Plots from two  systems (rustic system and polyculture system) were set up and used for comparision. The study is very interesting. I have some comments about the experiment and result.

We thank the reviewer for the kind comments and useful suggestions.

1. How many times did the author retrieve invertebrate traps? You descriped that collected data from May to August 2023, so you retrieve invertebrate traps for one time, or every day from May to August 2023?

We have now added “once in each plot” at line 162

2. The data in Table A1 is very important ,and I suggested to show these data in the main text. Table A1 is too long in this version, simple it (just give Z value and * label, and may also delete some taxa with low abundance).

In the original version, we have added the results shown in the table in text. But we understand that a simplified version of that table can be useful. We accept the suggestion to include a short version of Table A1 in text but we decided to show all significant results only and then leave A1 for extra information.

3. what about the effects of system type (rustic system and polyculture system) on canopy cover, crop richness, tree richness, and yield? I suggested also show thest results, and can be used to indicate which management is better for biodiversity.

We have now added a table with a summary of the vegetation parameters for the two systems, with linked statistics. That will also help the reader to understand the difference between polyculture and rustic systems.

Reviewer 5 Report

The article deals with an important topic. However, there are significant shortcomings in the methodology, that negatively affect its value. 

The Authors could not explain the experimental design. They studied the invertebrate populations in very diverse agroecosystems, diverse ecosystems that were tried to be classified as rustic and polycultures. In my opinion, the biggest difference between these ecosystems is their use as copland and forest. Secondly, certain crops will undoubtedly affect the presence of particular insects, such as pests, to a greater extent than the canopy cover suggested by the authors.

The authors should first rethink the experimental design and data analysis. 

Please find below some comments that may help improve the article:

Introduction: 

98-99. Please explain what is the heterogeneity hypothesis

Lines 100–101: “Based on the habitat heterogeneity hypothesis [26], we predicted a decrease in abundance and richness of invertebrates as a consequence of a reduction in habitat quality of the farm (measured by the richness of crops and non-crop trees), a reduction in shade cover, and an increase in profits obtained.”.

Try to be more specific. What are “crop” and “non-crop trees”?

 Line 101” “increase in profits obtained” - what does this mean?

2. Materials and Methods

Please very carefully describe the polyculture systems and rustic systems. It is not clear to me what the specific features of those ecosystems are.

Figure 3. After analyzing this map, I understand even less about the pattern of the experiment. So the Authors sampled from two ecosystem types: agricultural land and production forest. Within those ecosystems, the other two types were distinguished: rustic and polyculture.

Please try to very specifically determine the factors in your experiment. 

Data analysis: 

When you run the GLM model, you need to specify the factors and factor levels. It is not possible, as you did not describe the experimental scheme. There is one major factor: the agriculture field and a forest. Then there are the types of crops. Further perhaps the land-use management, ect.

Results: 

Figure 4. There is no explanation of these results in the text.

Author Response

The article deals with an important topic. However, there are significant shortcomings in the methodology, that negatively affect its value.

We thank the reviewer for recognising the importance of the manuscript and we revised the manuscript based on the comments

The Authors could not explain the experimental design. They studied the invertebrate populations in very diverse agroecosystems, diverse ecosystems that were tried to be classified as rustic and polycultures. In my opinion, the biggest difference between these ecosystems is their use as copland and forest. Secondly, certain crops will undoubtedly affect the presence of particular insects, such as pests, to a greater extent than the canopy cover suggested by the authors.

We understand that species-specific relationships can be beneficial, but showing them was not the purpose of our paper. We wanted to give an overview of inverts biodiversity in relation to farming practices. Rustic vs polyculture is an important factor to consider as, by definition, rustic means that there is a cover of natural forest. We have tried to make this distinction clearer also based on the comments from other reviewers.  

The authors should first rethink the experimental design and data analysis.

We hope with the clarifications and additions now the paper is clearer

Please find below some comments that may help improve the article:

Introduction:

98-99. Please explain what is the heterogeneity hypothesis

We have now added a section in the introduction

Lines 100–101: “Based on the habitat heterogeneity hypothesis [26], we predicted a decrease in abundance and richness of invertebrates as a consequence of a reduction in habitat quality of the farm (measured by the richness of crops and non-crop trees), a reduction in shade cover, and an increase in profits obtained.”.

Try to be more specific. What are “crop” and “non-crop trees”?

Changed “non-crop trees” with “native forest trees”

 Line 101” “increase in profits obtained” - what does this mean?

Added “from crops”.

2. Materials and Methods

Please very carefully describe the polyculture systems and rustic systems. It is not clear to me what the specific features of those ecosystems are.

We have now added a clear definition of the two systems, also showing the differences in the results. Mainly rustic systems have cover of native forest trees while polyculture systems does not. In the process of checking the data we also realised one plot with 1 native forest tree species was incorrectly assigned to polyculture so we redid the analyses.

Figure 3. After analyzing this map, I understand even less about the pattern of the experiment. So the Authors sampled from two ecosystem types: agricultural land and production forest. Within those ecosystems, the other two types were distinguished: rustic and polyculture.

Please try to very specifically determine the factors in your experiment.

We understand the confusion and we thank the reviewer for suggesting to better explain the differentiation. Mainly, production forest is a political land zonation that does not necessarily reflect into practice. In theory, production forests should be used sustainably, but in practice many production forests in Indonesia are converted into agricultural lands, including monocultures. In the study area the production forest is well preserved, but there were some areas that were cleared in the past and does not host native forest trees, that is why those plots were classed as polycultures.

Data analysis:

When you run the GLM model, you need to specify the factors and factor levels. It is not possible, as you did not describe the experimental scheme. There is one major factor: the agriculture field and a forest. Then there are the types of crops. Further perhaps the land-use management, ect.

We hope that the above clarifications highlighted that we did have a clear experimental scheme and that the factors included are described in details. We do accept suggestions on how to further improve the description.

Results:

Figure 4. There is no explanation of these results in the text.

We have now added a section on profits and vegetation characteristics between rustic and polyculture systems.

Round 2

Reviewer 1 Report

Suggestions were implemented or a justification in the response letter was provided. 

Author Response

The reviewer has no further comments

Reviewer 5 Report

Dear Authors, 

I appreciate your effort in improving this manuscript. 

However, please remark on some shortcomings in this version of this manuscript: 

1. Introduction: 

"That said, we expect some taxa of invertebrates, especially the ones considered as pests, to increase with an increased level of agricultural intensification given the relationship between pests and reduced habitat complexity" 

In the results, the Authors indicate some taxa related to biological control, but not as pests. There is only a short explanation in the discussion. As the Authors want to focus also on pests, additional analysis is needed.

Fig. 3. "Figure 3. After analyzing this map, I understand even less about the pattern of the experiment. So the Authors sampled from two ecosystem types: agricultural land and production forest. Within those ecosystems, the other two types were distinguished: rustic and polyculture.

Please try to very specifically determine the factors in your experiment.

We understand the confusion and we thank the reviewer for suggesting to better explain the differentiation. Mainly, production forest is a political land zonation that does not necessarily reflect into practice. In theory, production forests should be used sustainably, but in practice many production forests in Indonesia are converted into agricultural lands, including monocultures. In the study area the production forest is well preserved, but there were some areas that were cleared in the past and does not host native forest trees, that is why those plots were classed as polycultures. "

I understand the explanation. The "political land zonation" presented on the map makes the experiment difficult to understand. I propose to improve the map and show the factual land use management. 

Data analysis: 

Please add more detail about the statistical model, including the random and fixed variables.

Results: 

Figure 4 - Please interpret the results shown in the figure. 

Table 2. The abundance of certain taxa is very low (4, 5, 7, 1). Please consider, if a proper sampling method has been used for all the invertebrates. Is it necessary to show the abundance of all taxa? Please consider if showing the abundances of the ecological groups or only the most abundant taxa will make the data more representative.

Author Response

Dear Authors,

I appreciate your effort in improving this manuscript.

However, please remark on some shortcomings in this version of this manuscript:

We thank the reviewer for finding time to check this new version and provide more useful comments.

1. Introduction:

"That said, we expect some taxa of invertebrates, especially the ones considered as pests, to increase with an increased level of agricultural intensification given the relationship between pests and reduced habitat complexity"

In the results, the Authors indicate some taxa related to biological control, but not as pests. There is only a short explanation in the discussion. As the Authors want to focus also on pests, additional analysis is needed.

We agree with the reviewer that we could have expanded more the discussion on pests. We want to highlight the fact that we were only able to split taxa into different roles only when the sample size was enough in taxa that are not homogeneous in their role (e.g. we have a category Diptera pest). We have now made more examples of other pest taxa in the discussion, in the section where we talk about pests and biological control (but we mention pests in different sections of the discussion).  

Fig. 3. "Figure 3. After analyzing this map, I understand even less about the pattern of the experiment. So the Authors sampled from two ecosystem types: agricultural land and production forest. Within those ecosystems, the other two types were distinguished: rustic and polyculture.

Please try to very specifically determine the factors in your experiment.

We understand the confusion and we thank the reviewer for suggesting to better explain the differentiation. Mainly, production forest is a political land zonation that does not necessarily reflect into practice. In theory, production forests should be used sustainably, but in practice many production forests in Indonesia are converted into agricultural lands, including monocultures. In the study area the production forest is well preserved, but there were some areas that were cleared in the past and does not host native forest trees, that is why those plots were classed as polycultures. "

I understand the explanation. The "political land zonation" presented on the map makes the experiment difficult to understand. I propose to improve the map and show the factual land use management.

We understand the concern of the reviewer but we still think it is fundamental to show that in Indonesia there is this political zonation, and we are repeatedly discussing this in our paper. The fact that most of the times the production forest is cleared just indicates unsustainable use of it. But in our study area we had very nice rustic systems to monitor. We have changed the colours of the map as we think the green made a bit of confusion, but we think it is important to show the political zonation (readers from Indonesia/experts of Indonesia would like to see that). We still show the land use management by separating the plots into rustic vs polyculture.

Data analysis:

Please add more detail about the statistical model, including the random and fixed variables.

We have now specified the fixed effects and response variables. We do not have random effects.

Results:

Figure 4 - Please interpret the results shown in the figure.

We have now added more explanation.

Table 2. The abundance of certain taxa is very low (4, 5, 7, 1). Please consider, if a proper sampling method has been used for all the invertebrates. Is it necessary to show the abundance of all taxa? Please consider if showing the abundances of the ecological groups or only the most abundant taxa will make the data more representative.

We refer the reviewer to our section 4.2. We think it is important to show all the taxa that we found, even if for some we could not proceed with the data analysis (we describe this in our data analysis section). In section 4.2 we provide an explanation for our study design as well as explaining limitations. We think it is important to show what we do not get much if we just use pitfall traps and pan traps. We cannot change our study design but we can still provide some suggestions to other researchers, especially if they focus on the taxa that were under sampled.