Roads in the Southwestern Amazon, State of Acre, between 2007 and 2019

Round 1

Reviewer 1 Report

It is an important study to monitor road extension in the southwestern Amazon. Technically it would be better to highlight that the extraction of roads by machine-learning techniques is difficult at the Landsat scale level in the Amazon region for some specific reasons. Otherwise, manual road extraction sounds like an OpenStreetMap campaign and far from academic activities. Furthermore, the accuracy assessment is mandatory.

To improve the quality of this paper, I would like the authors to give the following comments, which would be helpful for publication.

• L37: There are no descriptions for BR-163, 230, 319. It is hard for international readers to understand the background on road development in this area.
• L43: Remove “on the other hand”
• L67: The authors do not seem to discuss the highest number of roads, but the total length of roads instead.
• L68: “What is the correlation between roads and deforestation?“ deforestation of what? The degree, severity, or total areas? Also, several types of deforestation can be considered including natural-oriented forest fire or tree disease, but it supposes that the authors only consider deforestation by anthropogenic activities. The definition of “deforestation” in this article is essentially needed.
• L70-71: Coordinations are strange.
• Figure1. Closer colored classes make us confuse. Design the map with a distinct color palette.
• The title in Figure 1: please rephrase it. does it mean “Map of Indigenous lands, conservation units, and deforestation areas in the State of Acre in 2019.”?
• Figure 1&2. In a geographic coordinate system, the scale bar and the north arrow do not function precisely. Remove them.
• L124: what CAR represents?
• Figure 2: do 4 images show the same area? Are these images from Landsat 8?
• How do the authors digitize roads in detail? Full descriptions are necessary.
• What do you think about the accuracy of the digitized roads? How can I trust the results?
• L148-149: “The sample unit of the model was the annual rate of roads up to 2019 per municipality.” I don’t understand what it means.
• Figure3. For the same reason as the comment for Fig 1 and 2, the north arrow should be removed.
• Figure 3: Use color for the class: “Not identified”.
• Fig5: what the mean “Ate 2006”?
• L204-205: “The regression analysis showed a predisposition to decrease deforestation concerning the increase in distance from roads” it is not clear why the regression analysis was applied here and why it is needed.
• Fig6: what is the meaning of “95% of deforestation”? In terms of areas, Severity, or something else? And why is the threshold of 95% selected?
• L212: the municipality limits?
• Fig7: Does each dot represents a municipality?
• Fig8. What do the authors indicate from Fig8? I don’t think it makes sense. Instead, how do you think about a correlation plot between road length and total deforestation areas in each land class?
• L246: “in the Amazon context” → “in the Amazon.”
• L257: “the highest growth rates in the Brazilian Amazon” of what?
• L261-266: An example of fig is appreciated in what these sentences represent.
• L308-309: “The increase in deforestation in 2019 was accompanied by a higher annual rate of growth of roads.” The correlation analysis shows only the relation between two variables, and never tells the causes. It is a leap discussion about the cause of deforestation by road extension.

Author Response

It is an important study to monitor road extension in the southwestern Amazon. Technically it would be better to highlight that the extraction of roads by machine-learning techniques is difficult at the Landsat scale level in the Amazon region for some specific reasons. Otherwise, manual road extraction sounds like an OpenStreetMap campaign and far from academic activities. Furthermore, the accuracy assessment is mandatory.

We understand the reviewer's concern with the quality of the scientific information in this study. We expanded the literature review and included text between lines 59-65, bringing the perspective that automatic road mapping methods exist and have been applied to high resolution images, which have high cost and little time coverage. Studies in the Brazilian Amazon and Africa show that the manual road mapping technique

 

To improve the quality of this paper, I would like the authors to give the following comments, which would be helpful for publication.

L37: There are no descriptions for BR-163, 230, 319. It is hard for international readers to understand the background on road development in this area.

We include contextual information about the state level location. Line 36-39.

 

L43: Remove “on the other hand”

Removed

 

L67: The authors do not seem to discuss the highest number of roads, but the total length of roads instead.

Yes, our analyzes were focused on the total length of length, we corrected the text. Line 77

 

L68: “What is the correlation between roads and deforestation?“ deforestation of what? The degree, severity, or total areas? Also, several types of deforestation can be considered including natural-oriented forest fire or tree disease, but it supposes that the authors only consider deforestation by anthropogenic activities. The definition of “deforestation” in this article is essentially needed.

We have included a description of the type of deforestation we used in this study. Line 77-78 and 136-137

 

L70-71: Coordinations are strange.

We rewrote the sentence. Line 80-81

 

Figure1. Closer colored classes make us confuse. Design the map with a distinct color palette.

We edit the colors.

 

The title in Figure 1: please rephrase it. does it mean “Map of Indigenous lands, conservation units, and deforestation areas in the State of Acre in 2019.”?

Reescrevemos o título da Figura 1. Line 96-97

 

Figure 1&2. In a geographic coordinate system, the scale bar and the north arrow do not function precisely. Remove them

We excluded the north arrow but kept the graphic scale with color modification for better reading.

 

L124: what CAR represents?

CAR is Rural Environmental Registry as mentioned in line 108. We kept the acronym in Portuguese because it is widely known and accepted in the scientific literature (see Costa et al 2018. https://doi.org/10.1525/elementa.260; L ' Roe et al. 2016. https://doi.org/10.1016/j.landusepol.2016.05.029; Viana et al. 2016. https://doi.org/10.1002/eet.1720)

We specify that it is a Portuguese acronym. Line 108

 

Figure 2: do 4 images show the same area? Are these images from Landsat 8?

Figures 2a 2b and 2c 2d are the same areas for different years.

The images are from Landsat8 and red band (band 4). We include this information Line 127

 

How do the authors digitize roads in detail? Full descriptions are necessary.

We write a better explanation of this information in the section 2.2. Lines 108-112

 

What do you think about the accuracy of the digitized roads? How can I trust the results?

Kappa coefficient applied to test mapping accuracy as described in the lines 113-116

 

L148-149: “The sample unit of the model was the annual rate of roads up to 2019 per municipality.” I don’t understand what it means.

We rewrote part of this section to make our analysis clearer. Line 141-142

 

Figure3. For the same reason as the comment for Fig 1 and 2, the north arrow should be removed.

Revised.

 

Figure 3: Use color for the class: “Not identified”.

Revised.

 

Fig5: what the mean “Ate 2006”?

We write a better explanation of this information in the section 2.2. Lines 108-112

 

L204-205: “The regression analysis showed a predisposition to decrease deforestation concerning the increase in distance from roads” it is not clear why the regression analysis was applied here and why it is needed.

We rewrote part of this section to make our analysis clearer. Line 115-116

 

Fig6: what is the meaning of “95% of deforestation”? In terms of areas, Severity, or something else? And why is the threshold of 95% selected?

We include more explanation in section 4. Lines 141-147

 

L212: the municipality limits?

Revised. Line 216

 

Fig7: Does each dot represents a municipality?

Yes, we include this information in the figure caption.

 

Fig8. What do the authors indicate from Fig8? I don’t think it makes sense. Instead, how do you think about a correlation plot between road length and total deforestation areas in each land class?

The title of Fig8 has been corrected as suggested by the reviewer.

 

L246: “in the Amazon context” → “in the Amazon.”

Revised. Line 240

 

L257: “the highest growth rates in the Brazilian Amazon” of what?

Revised. Line 250

 

L308-309: “The increase in deforestation in 2019 was accompanied by a higher annual rate of growth of roads.” The correlation analysis shows only the relation between two variables, and never tells the causes. It is a leap discussion about the cause of deforestation by road extension.

We understand that our analysis and data do not reveal all the factors that influenced the increase in deforestation and roads. We decided to remove this sentence.

 

Author Response File: Author Response.pdf

Reviewer 2 Report

General comments:

The authors of the study, use satellite imagery (specifically Landsat) to determine the amount of new roads in the state of Acre and their relation to deforestation. Although this research has a large potential in determining policy towards the reduction of deforestation in the Amazon and is extremely needed, it still needs some significant improvements.

The English language and phrasing of the text must be entirely reviewed, as it looks very unpolished. Also some figures and tables do not use English words in legends or headings, and some maps need visual improvement.

In the methods section, you need to explain more how did you identified the roads, and how did you quantified deforestation.

In the discussion section you need to give a few possible explanations for the found results (why such a large increase in 2019, why such an increase in the CU etc).

Title: I suggest you reword it, so it sounds better (maybe something like  ...roads in the southwestern Amazon state of Acre …?

Specific comments:

L17: no comma after year

L28: 92% increase…by 240%

L36: but this was…

L51: two type of roads

L61: expressive increase? What does it mean in English?

L63: which limits

L75: no need to say country of Europe

Figure 1: Need to redo map legend (no need for South America, not clear which is the state of Acre is, better colors)

Figure 3: needs scale bar, no need to say international limit, better choice of background color

L178: ano-1 ??

Table 1: Table headings to English

Figure 5: English legend

L202: Out of

L205: regression analysis does not use decision tree com split

Figure 8: what do the colors mean?

L241: so that only wide roads…

L249: in the State .. of roads are built

L249: At the western end of the Arc of Deforestation

L298: the monitoring must be continuously monitored?

L320: Move the relevant information to funding from the Acknowledgements section

Table S2 and S3: Table caption does not make sense in English and the regions are not ranked. Please rephrase

Table S3: What does TI mean ?

Author Response

The authors of the study, use satellite imagery (specifically Landsat) to determine the amount of new roads in the state of Acre and their relation to deforestation. Although this research has a large potential in determining policy towards the reduction of deforestation in the Amazon and is extremely needed, it still needs some significant improvements.

the manuscript went through the review of English by a native researcher

 

In the methods section, you need to explain more how did you identified the roads, and how did you quantified deforestation.

We write a better explanation of this information in the section 2.2. Lines 108-112

 

In the discussion section you need to give a few possible explanations for the found results (why such a large increase in 2019, why such an increase in the CU etc).

We expanded the discussion on the suggested topics. 289-297

 

Title: I suggest you reword it, so it sounds better (maybe something like  ...roads in the southwestern Amazon state of Acre …?

We accept the suggestion of the review, keeping the date of analysis. We believe that the date information can better guide the reader. Line 2-3

 

L17: no comma after year

Revised, Line 17

 

L28: 92% increase…by 240%

We have revised this sentence. Line 27-28

 

L36: but this was…

We revised the sentence. Line 35

 

L51: two type of roads

We revised the sentence. Line 53

 

L61: expressive increase? What does it mean in English?

They are redundancies of the Portuguese line. We reviewed the term. Line 70

 

L63: which limits

We revised the sentence. Line 70-72

 

L75: no need to say country of Europe

Excluded term. Line 84

 

Figure 1: Need to redo map legend (no need for South America, not clear which is the state of Acre is, better colors)

Revised

 

Figure 3: needs scale bar, no need to say international limit, better choice of background color

Revised

 

L178: ano-1 ??

Revised. Line 179-180

 

Table 1: Table headings to English

Revised.

 

Figure 5: English legend

Revised.

 

L205: regression analysis does not use decision tree com split

we review the statistical analyzes. Line 146-147

 

Figure 8: what do the colors mean?

Revised.

 

L241: so that only wide roads…

Using medium spatial resolution satellite images, road identification needs to be close to deforested areas.

 

L249: in the State .. of roads are built

Revised. Line 242-243

 

L249: At the western end of the Arc of Deforestation

Revised. Line 424

 

L298: the monitoring must be continuously monitored?

Revised. Line 299

 

L320: Move the relevant information to funding from the Acknowledgements section

Revised

 

Table S2 and S3: Table caption does not make sense in English and the regions are not ranked. Please rephrase

Revised

 

Table S3: What does TI mean ?Tabela S3: O que significa TI?

Revised.

 

Author Response File: Author Response.pdf

Reviewer 3 Report

The manuscript deals with the very important topic of roads in the Amazon. The overall results are sound and the paper seems to reflect appropriate statistical analysis and interpretation of trends.

However, the text tends to be over-descriptive with limited discussion of the forces behind road expansion, the politico-institutional responsibilities and socio-ecological consequences. I suggest that the authors further discuss those issues and more directly connect roads with development and environmental management in the State of Acre. 

For the development history in the first paragraph:

"I would include a full stop after '60s and 80s' and then start a new sentence,
because the programmes mentioned in the text (i.e. Brasil em Acao, Avanca Brasil, PPAs & PAC) were introduced in the 1990s and 2000s, not before.

the word 'currently' is slightly misplaced, given that PAC has been significantly reduced and paralyzed by the new Brazilian administration (in office since 2019). I would remove the word 'currently' "

In addition, the analysis is very aggregate and it would be important to discuss, to some extent, differences between the Rio Branco (eastern) region and the Cruzeiro do Sul (western) region of Acre. 

Finally, some information on the graphs are still in Portuguese.

 

 

Author Response

REVIEWER 3

1 - The manuscript deals with the very important topic of roads in the Amazon. The overall results are sound and the paper seems to reflect appropriate statistical analysis and interpretation of trends.

Answer. We are glad that the Reviewer liked the narrative of the paper, its message, and the methodology used.

 

2 - However, the text tends to be over-descriptive with limited discussion of the forces behind road expansion, the politico-institutional responsibilities and socio-ecological consequences. I suggest that the authors further discuss those issues and more directly connect roads with development and environmental management in the State of Acre.

Answer. Thanks for the comment. The reviewer is mentioning that the discussion is -descriptive. This writing was partly intensive to provide the reader with the characteristics of the first study on the spatio-temporal distribution of roads in Acre. But believing that it is important to expand the discussion, we rewrote some points.

 

3 - For the development history in the first paragraph:

"I would include a full stop after '60s and 80s' and then start a new sentence,

because the programmes mentioned in the text (i.e. Brasil em Acao, Avanca Brasil, PPAs & PAC) were introduced in the 1990s and 2000s, not before.

Answer. Thanks for the suggestion. We changed the sentence as suggested.

 

4 - the word 'currently' is slightly misplaced, given that PAC has been significantly reduced and paralyzed by the new Brazilian administration (in office since 2019). I would remove the word 'currently' "

Answer. Thanks for the suggestion. We changed the sentence as suggested.

 

5 - In addition, the analysis is very aggregate and it would be important to discuss, to some extent, differences between the Rio Branco (eastern) region and the Cruzeiro do Sul (western) region of Acre.

Answer. Thanks for the comment. We have included a paragraph in the discussion section, showing how the history of the occupation process was important for differences between the region of Rio Branco (east) and Cruzeiro do Sul (west) of Acre.

6 - Finally, some information on the graphs are still in Portuguese.

Answer. Thanks for the observation, we reviewed all the figures and tables.

Round 2

Reviewer 1 Report

All improvements are acceptable.

Author Response

We thank the Reviewer for his contributions.

Reviewer 2 Report

I believe the authors have made significant improvements to the article, and have made some good points specifically the use of deep learning to detect roads.

Specific comments:

Figure 4: Please specify what is the dashed line for C and D (if its the 1:1 line)

Author Response

REVIEWER 2

I believe the authors have made significant improvements to the article, and have made some good points specifically the use of deep learning to detect roads.

Answer. We are happy that the Reviewer appreciated the revised version of our manuscript.

 

Specific comments:

Figure 4: Please specify what is the dashed line for C and D (if its the 1:1 line)

Answer. Thanks for the observation, we reviewed the figure legend.