The Importance of Scale and the MAUP for Robust Ecosystem Service Evaluations and Landscape Decisions

Round 1

Reviewer 1 Report

In my opinion, the work "The importance of scale and the MAUP for robust Ecosystem Service evaluations and landscape decisions" is interesting and concerns important issues, but requires some parts to be clarified, detailed and explained. Because it is illegible in its present form. Perhaps it is understandable to people dealing with this issue. It is incomprehensible for a reader who is not an expert in this topic. I have the impression that we are looking at the same but we are seeing something different.

Detailed comments:

1. Title: Including the abbreviation MAUP in the title is not a good idea as it was only explained in the abstract. Not all readers need to know what the abbreviation refers to and what it means.
2. The language should be improved to be more scientific. Phrases should be removed from slang or colloquial language, e.g. "scale blindness", "coarser" Very long sentences making it difficult to understand the arguments
3. Chapter background is too short. The problem can be described in more detail in the works of other authors.
4. No description: What is on the axes of graph 1. If there is a top geographical map, there should be informed where the north is located, as well as the scale.
5. What does the CRS 27700 abbreviation mean in the title of chart 1.
6. I do not understand why the problem is to take the smallest scale, i.e. 50m. This should be clarified in the introduction, not later (lines 138-139)
7. Row 129 Table ??
8. How is the sentence in lines 129-131 to be understood? What is "... to some sort of regulating ES"?
9. Please describe the results of Table 2 in more detail. I didn't understand what optimization was about (line 137). Would it be appropriate to describe more precisely what is maximized?
10. A description of the method of operation of the genetic algorithm is redundant (lines 144-160).
11. Line 183 - Figure ??
12. This is incomprehensible lines 223-227
13. What do the dashed curves in Figure 5 mean?
14. Is the value of 2.44 worse than 2.69? (lines 198-206) How should this be understood?
15. Line 254 - 262 Could the authors explain why the results indicate that the differences are due to aggregation?
16. Figure 8, lines 261-262 I disagree with this claim because they are not perfectly normal distributions. You can see it occurring here skewness.
17. No explanation as to why this is happening according to the Authors (lines 244-253)
18. What do the numbers in figure 10 mean?
19. What does "the best" mean? What are the criteria for this? (line 304)
20. Where and how is it shown that "The effect of the spatial support was shown to be non-linear, non-nested and non-hierarchical ...?" (line 315)
21. Figure 11 is completely incomprehensible to me. I don't know how it can be used? If this is an area for further research, it probably is not needed here.
22. The discussion of the results is very extensive. There were no confirmations in the statistics that the authors refer to. 

Author Response

Reviewer 1

 

In my opinion, the work "The importance of scale and the MAUP for robust Ecosystem Service evaluations and landscape decisions" is interesting and concerns important issues, but requires some parts to be clarified, detailed and explained. Because it is illegible in its present form. Perhaps it is understandable to people dealing with this issue. It is incomprehensible for a reader who is not an expert in this topic. I have the impression that we are looking at the same but we are seeing something different. 

 

Response: we thank the reviewer for their comments. We fundamentally disagree as do Reviewers 2,3 and 4. Therefore, we have selectively addressed some of the points below but not all. The work is aimed at people and experts working in the field of land use, ES and NC.

 

Detailed comments: 

1. Title: Including the abbreviation MAUP in the title is not a good idea as it was only explained in the abstract. Not all readers need to know what the abbreviation refers to and what it means. 

Response: This has not been done. We like MAUP in the title. It is a well recognised term.

1. The language should be improved to be more scientific. Phrases should be removed from slang or colloquial language, e.g. "scale blindness", "coarser" Very long sentences making it difficult to understand the arguments 

Response: we have tried to do this. We have reduced length of some sentences. But very few run over 3 lines in the manuscript.  

1. Chapter background is too short. The problem can be described in more detail in the works of other authors. 

Response: this has not been done.

1. No description: What is on the axes of graph 1. If there is a top geographical map, there should be informed where the north is located, as well as the scale. 

Response: This has not been done. Figure 1 is a map. Scale and position are indicated by the axes, the geographic reference is given by the figure caption.

1. What does the CRS 27700 abbreviation mean in the title of chart 1. 

Response: This has not been done. As above… Figure 1 is a map. Scale and position are indicated by the axes, the geographic reference is given by the figure caption.

1. I do not understand why the problem is to take the smallest scale, i.e. 50m. This should be clarified in the introduction, not later (lines 138-139) 

Response: this has not been done. We do not understand the point being made here

1. Row 129 Table ?? 

Response: this has been corrected.

1. How is the sentence in lines 129-131 to be understood? What is "... to some sort of regulating ES"? 

Response: This has not been done. This is perfect English and explains how the scores in Table 2 were obtained.

1. Please describe the results of Table 2 in more detail. I didn't understand what optimization was about (line 137). Would it be appropriate to describe more precisely what is maximized? 

Response: This has not been done. This is explained in lines 127-131 and it is commonly how ES values are derived from land use.

1. A description of the method of operation of the genetic algorithm is redundant (lines 144-160). 

Response: We disagree. Optimisation of any kind requires heuristic or linear searches through the decision space. We have provided a cursory description of the approach we used with links to references providing greater detail. We have left the description in the manuscript with some edits.

1. Line 183 - Figure ?? 

Response: this has been corrected.

1. This is incomprehensible lines 223-227 

Response: We disagree. This has not been changed.

1. What do the dashed curves in Figure 5 mean?

Response: this has been done. This has been clarified in the modified Figure caption as follows “The proportions of different land use classes in the study area (solid line) and the proportion of the total ES score associated with each land use (dashed line) when optimised using an ES gradient aggregated to different resolutions.”

1. Is the value of 2.44 worse than 2.69? (lines 198-206) How should this be understood? 

Response: this has been clarified.

1. Line 254 - 262 Could the authors explain why the results indicate that the differences are due to aggregation?

Response: This has been clarified.

1. Figure 8, lines 261-262 I disagree with this claim because they are not perfectly normal distributions. You can see it occurring here skewness. 

Response: This has been clarified. The point is that scores the peak of the distributions – the most frequent not the spread around the mean. We have edited the text accordingly.

1. No explanation as to why this is happening according to the Authors (lines 244-253) 

Response: This is explained in the text in lines 254 – 262 (original submission) and lines 255-266 in the resubmission).

1. What do the numbers in figure 10 mean? 

Response: this has been clarified in the Figure legend text. These are the land use grid cell ES scores that are summed to generate the overall ES score.

1. What does "the best" mean? What are the criteria for this? (line 304) 

Response: this has been clarified as follows “For this case study, the highest ES scores results are generated under ES gradients aggregated to 100 m” 

1. Where and how is it shown that "The effect of the spatial support was shown to be non-linear, non-nested and non-hierarchical ...?" (line 315) 

Response: this has been clarified  

1. Figure 11 is completely incomprehensible to me. I don't know how it can be used? If this is an area for further research, it probably is not needed here. 

Response: this has not been done. We have kept the figure in. In the text we explicitly indicate this as an area of further research when we say “However this is very much an area for further research.”

1. The discussion of the results is very extensive. There were no confirmations in the statistics that the authors refer to. 

Response: we thank the reviewer for this observation. The aim of the whole paper was to generate items for discussion in order to make the overall case in the paper.

Reviewer 2 Report

Dear Authors,

This paper is really important and originality. I am really impressed of it, but the figures/legends should be improved. Please check the size of fonts etc.

Author Response

Reviewer 2

 

Dear Authors,

This paper is really important and originality. I am really impressed of it, but the figures/legends should be improved. Please check the size of fonts etc.

Response: the figures legends and captions have been reviewed and some of them have been revised.

Reviewer 3 Report

The article examines the issue of geographical scale in Ecosystem Service evaluations. It is a well-written and clearly described analysis. I would recommend it for publication in the journal. However, there are two issues mentioned in the article that need additional tests. I strongly recommend testing the following two observations before the paper will be published:

Lines 198-199: The total ES scores decline because parcels with higher ES scores are small and disappear as a result of aggregation. In a different spatial setting, where parcels with higher ES scores have larger areas, the overall ES score might increase. It should be tested.

Lines 277-278: Non-hierarchical behaviour can pop up due to arbitrary grid configuration when the coarser grids’ position is less optimal than the position of the finer level. That is, it is possible that at the same coarser scale there would be an optimal grid configuration, which is shifted north-south, east-west or both, or in an intermediate direction compared to the grid resulting by default. I suppose that this effect can be tested by, for instance, comparing the results of shifting the grids by half or by 1/3 of their side with the non-shifted situation. I think it can be illustrated based on the original data, without optimization.

 

Typos and minor recommendations:

Lines 29 and 30: names the authors of the cited sources

Line 137: should be “the task is to identify”

There are repetitions in lines 149, 176, 195, 237

The discussion sections should be divided into discussions and conclusions.

Author Response

Reviewer 3

 

The article examines the issue of geographical scale in Ecosystem Service evaluations. It is a well-written and clearly described analysis. I would recommend it for publication in the journal. However, there are two issues mentioned in the article that need additional tests. I strongly recommend testing the following two observations before the paper will be published:

 

Lines 198-199: The total ES scores decline because parcels with higher ES scores are small and disappear as a result of aggregation. In a different spatial setting, where parcels with higher ES scores have larger areas, the overall ES score might increase. It should be tested.

Response: this has not been done although it has been addressed. In this case, although potentially a valid point for the results in Section 4.2 using the gridded land use data, this is incorrect for the field (vector) land use data. Here the land use parcels are not being aggregated to grids. The decline is being driven by the ES grids. The same trend is evident when the land use data are aggregated to grids in Section 4.2. However, we would ask the reviewer to note that we explicitly make this point – the trend are driven by the ES aggregation – in our discussion of the results in the original version of the paper (lines 295-300 and 306-312) which are retained in the revision.

 

Lines 277-278: Non-hierarchical behaviour can pop up due to arbitrary grid configuration when the coarser grids’ position is less optimal than the position of the finer level. That is, it is possible that at the same coarser scale there would be an optimal grid configuration, which is shifted north-south, east-west or both, or in an intermediate direction compared to the grid resulting by default. I suppose that this effect can be tested by, for instance, comparing the results of shifting the grids by half or by 1/3 of their side with the non-shifted situation. I think it can be illustrated based on the original data, without optimization.

Response: this has not been done although it has been addressed. In the Discussion section (Line 317)we say “There is the possibility that this non-hierarchical behaviour may be due to due to arbitrary positioning of the ES grids, although the same trends were observed in the analyses of both field (vector) and gridded land use data, suggesting that the agriculture-related processes in this study are best captured at that resolution.”

 

Typos and minor recommendations:

Lines 29 and 30: names the authors of the cited sources

Response: this has been done

Line 137: should be “the task is to identify”

Response: this has been corrected

There are repetitions in lines 149, 176, 195, 237

Response: these have been corrected

The discussion sections should be divided into discussions and conclusions.

Response: this has been done

Reviewer 4 Report

This study investigates the influence of the aggregation of ES (ecosystem services) units on land use decisions using the GIS methodology and a genetic algorithm. The topic is of interest to the readers of this journal. The figures are illustrative and have significant implications. The data and codes are publicly available, and this practice is highly recommendable. I have the following questions about this study; please answer them.

L. 29–30: The use of [reference number] as a noun looks strange.

Table 1: Neutral grassland. Natural grassland?

L. 129: Table ?? Please correct this. “[A] local ES study” should be cited as a reference.

L. 174: Why do you need to have 1.2x the number of observations as the initial populations? I think the number of observations would be sufficient.

L. 179–181: Why does reallocating broadleaf forests not make sense?

L. 183: Figure ??

L. 219–222: Interesting Interestingly ?

L. 237: The word “analyses” appears twice.

L. 240: The word “format” appears twice.

L. 262: 10,000m (1ha) -> 10,000m^2 (1ha).

Figure 8: I do not see the point of this graph in relation with the analyses in this study. This is just a description. Do you want to say 1ha or 500m has some meaning? Please explain this in the main text.

L. 271–278: I could not follow the explanation here. Please revise this part.

Figure 10: I believe the caption for Figure 10 is wrong.

In Figure 10, the values of the most north-west corner cell range from 1 to 16. Given the scores from 1 to 5 in Table 2, I do not understand how the values can change so much. This means that the values can change as much as five times; there could be errors.

L. 308–311: “A different case study, containing field parcels with different typical spatial properties would show the same overall trends (of declining allocation effectiveness with increasing ES gradient aggregation), but with trend discontinuities at aggregation scales specific to that case study data.”
<- I think this is not substantiated because a different case study is not conducted in this study.

L. 330–: In general, the recommendations given here are general textbook guidelines, and I hope to see more ES and natural capital-oriented recommendations.

L. 336–337: “The scale of spatial data aggregations should be matched to the granularity of the processes being evaluated.”
<- I cannot find what parts of this study support this statement. I intuitively suppose this would be true; however, I cannot be convinced by the analyses in this paper.

L. 338–339: “…considered to be stationary (stable) with respect to their variances covariances and other moments”
<- What parts of this study support this statement? I have not seen an analysis involving these points.

L. 379–383: Please explain briefly how local stability is linked to the notion of process non-stationarity and what is the “duality.”

 

Author Response

Reviewer 4

 

This study investigates the influence of the aggregation of ES (ecosystem services) units on land use decisions using the GIS methodology and a genetic algorithm. The topic is of interest to the readers of this journal. The figures are illustrative and have significant implications. The data and codes are publicly available, and this practice is highly recommendable. I have the following questions about this study; please answer them.

 

1. 29–30: The use of [reference number] as a noun looks strange.

Response: this has been changed

Table 1: Neutral grassland. Natural grassland?

Response: No this is correct – the UK has Acid, Calcareous and Neutral Grassland classes

1. 129: Table ?? Please correct this. “[A] local ES study” should be cited as a reference.

Response: this has been done

1. 174: Why do you need to have 1.2x the number of observations as the initial populations? I think the number of observations would be sufficient.

Response: this has been clarified

1. 179–181: Why does reallocating broadleaf forests not make sense?

Response: this has been clarified

1. 183: Figure ??

Response: this has been corrected

1. 219–222: Interestingà Interestingly ?

Response: this has been corrected

1. 237: The word “analyses” appears twice.

Response: this has been corrected

1. 240: The word “format” appears twice.

Response: this has been corrected

1. 262: 10,000m (1ha) -> 10,000m^2 (1ha).

Response: this has been corrected

Figure 8: I do not see the point of this graph in relation with the analyses in this study. This is just a description. Do you want to say 1ha or 500m has some meaning? Please explain this in the main text.

Response: this has been clarified in the text

1. 271–278: I could not follow the explanation here. Please revise this part.

Response: we have tried to clarify this in the text

Figure 10: I believe the caption for Figure 10 is wrong.

Response: this has been corrected

In Figure 10, the values of the most north-west corner cell range from 1 to 16. Given the scores from 1 to 5 in Table 2, I do not understand how the values can change so much. This means that the values can change as much as five times; there could be errors.

Response: we have tried to clarify this in the text after Figure 10 is first introduced, reflecting back on how the scores are calculated.

1. 308–311: “A different case study, containing field parcels with different typical spatial properties would show the same overall trends (of declining allocation effectiveness with increasing ES gradient aggregation), but with trend discontinuities at aggregation scales specific to that case study data.”
   <- I think this is not substantiated because a different case study is not conducted in this study.

Response: this has been re-phrased to accommodate this point.

1. 330–: In general, the recommendations given here are general textbook guidelines, and I hope to see more ES and natural capital-oriented recommendations.

Response: this has not been done. We would like to do that with a case study more strongly grounded in the real world of ES and NC – multiple objectives, trade-offs, synergies / bundles etc – but the aim in this paper was to emphasise the importance of scale which we do not this has been or is adequately addressed in the ES literature.

1. 336–337: “The scale of spatial data aggregations should be matched to the granularity of the processes being evaluated.”
   <- I cannot find what parts of this study support this statement. I intuitively suppose this would be true; however, I cannot be convinced by the analyses in this paper.

Response: we believe this point *is* indicated by the results of this paper – the results show the inflections at 100m and 500m ES aggregations scales indicating the mismatches at other scales.

1. 338–339: “…considered to be stationary (stable) with respect to their variances covariances and other moments”
   <- What parts of this study support this statement? I have not seen an analysis involving these points.

Response: this has been addressed. This is our expertise – we have indicated how this can be done

1. 379–383: Please explain briefly how local stability is linked to the notion of process non-stationarity and what is the “duality.”

Response: we believe local stability has been explained – and in the response to the point above Line 338 above. The mention of duality has been removed – it is confusing here.

Round 2

Reviewer 3 Report

My comments have been addressed fairly enough. I would recommend the paper for publication, however, minor proofreading should be done in advance, as there might be some repetitions in the newly added text: at least, I found one in line 331.

Author Response

Response: we thank the reviewer for their patience. We have proof read the re-submission and tried to correct all of the repetitions and errors.

Reviewer 4 Report

I am not convinced by the authors' explanation on Figure 10 (why 16 times difference?) and their recommendation of matching of spatial aggregations with the granularity of the processes being evaluated (it seems "the finer, the better" from Figure 7). This may be due to my unfamiliarity with this field and other works by the authors. I would like to leave the decision to the academic editor despite my following recommendation.

Author Response

Response: we have tried to clarify this in a revised description of the figures in the Results section (p12).