Relationship Between Fire and Forest Cover Loss in Riau Province, Indonesia Between 2001 and 2012
Round 1
Reviewer 1 Report
This study presents some interesting and important statistics on forest loss and fire occurrence in the Riau province in Indonesia, but I would like to see some changes to the manuscript.
What the manuscript is mostly lacking in is explanation of the patterns that are observed. The results section is very elaborate, but the discussion section pales in comparison. For example: a striking result is the large difference in fire occurrence between areas with peat and non-peat soils; the manuscript could improve considerably if more mechanistic explanations of the observations are provided, rather than comparing the statistics with those found by other studies.
Related to this regards the relation between forest loss and fire occurrence. The authors report that fire occurrence is higher in areas with forest loss. However, there is a chicken-and-egg problem here that is not discussed: possibly cleared lands are set on fire, but fire itself is also a mechanism of forest loss. In other words, forest loss and fire may feed back to each other, and either of the two causal relations between fire and forest loss could explain the results. It is reported that forest loss and fire are often in the same year or one year apart. There is no discussion of which tends to precede which, while this is important, because it has a large effect on the implications of this study. Based on figure 9 and lines 309-311, it seems that, indeed, the relation between fire and forest loss is complex, as fire occurs both before and after forest loss. For data on, and implications of, the fire-forest feedback in the tropics, see Staal et al. (2018, Global Change Biology 24:5096-5109). Mentioned study, as well as others (e.g. Staver et al. 2011 Science 334:230-232), also report that fire occurrence is higher in low forest cover areas; this contradicts the statement in lines 328-329. Another recent paper on “How contemporary bioclimatic and human controls change global fire regimes” (Kelley et al. 2019 Nature Climate Change) might also be of interest.
The study area is the Riau province, but the introduction focuses on Indonesia as a whole. This leaves the reader wondering to which extent the context provided applies to the study area and why this specific study area was chosen. This needs some improvement. Also, lines 95-97 report a lack of a quantitative relationship between fire and land cover change, which motivates this study, whereas in the preceding paragraph a range of such relations are taken from the literature. Thus, I found the knowledge gap not clearly identified and strongly recommend that to be more specifically addressed.
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
General comment:
The authors investigated quantitative relationship between forest loss and fire in Indonesia using land cover map and several satellite products. Provided results suggested the importance of conservation of forest to reduce forest fire especially in peatland. The quantitative result may be useful and helpful for policy making. However, presentation of the results is not straightforward and the methodology has several technical defects, which must be largely improved.
Specific comments:
Introduction
Authors provided sufficient number of reviews regarding land cover and land use change in Indonesia. However, those regarding satellite monitoring of forest loss and fire were not provided. Given the analysis is largely depending on satellite products, literature review of satellite-based forest monitoring seems important.
Line 95: “there is still a lack of information…”
Is it true? Several researches have already discussed the relation of fire and land-cover change in the forest. For instance:
Langner, A., Miettinen, J., Siegert, F. (2007): Land cover change 2002-2005 in Borneo and the role of fire derived from MODIS imagery. Global Change Biology. 13(11). 2329-2340.
Van der Werf, G. R. et al. (2008): Climate regulation of fire emissions and deforestation in equatorial Asia. PNAS. 105(51). 20350-20355.
You should define your research novelty in comparison to these existing researches.
Line 113-120:
This part should be moved to the caption of Figure 1.
Line 135:
The spatial resolution of the MODIS product is 1 km (1000 m). How can it detect such small-scale fire (50-1000 m2)?
Line 142: Why did you select GFC as forest loss data? How is the accuracy of it in Indonesia? The GFC is based on image classification of surface reflectance in several optical bands of Landsat. My concern is whether it can classify the conversion of the natural forest to palm trees as “forest loss”.
Line 149: Relating to the previous concern, you should clearly define the meaning of “forest loss” here. Does it include homogeneous forest thinning (i.e., forest degradation), conversion to the other commercial plants and fire-induced damages? The last aspect is particularly important because you discussed relationship between fire and “forest loss”. I guess your point is the effect of man-made clearing of the forest on the occurrence of fire. However, if it includes fire-induced forest damage, the correlation between fire and “forest loss” is quite obvious and no surprise.
Result:
Poor presentation of the results degrades readability. Do not give all the tables/figures/statistics from your ad-hoc analysis, but focus on necessary and sufficient results that support your discussion point.
Line 175: “weakly correlated”
But it is not statistically significant (p = 0.2). The regression line (Fig 2a) seems meaningless. In addition, please rewrite the variables of regression lines (also for Figures 2b and 6b).
Line 216: “Further analysis is required…”
The uncertainty of the satellite-based analysis should be discussed. It is better to provide validation of satellite data using in-situ observation and/or other data sources.
Line 225: Is the largest category “other” meaningful?
Discussion:
Line 322: Again, how did you separate the effect of fire-induced forest damage from the “forest loss”?
Line 330-339: The two paragraphs looks duplicative. In addition, do not just repeat the result (e.g., “areas with high forest cover and low forest loss experience low hot spot frequency”) but also provide the concise interpretation. These paragraphs can be summarized like “less fire occurred over the preserved forest, especially in peatland”.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
My specific remarks have been sufficiently addressed in this revised manuscript. However, I found the revisions with regard to my more aspecific, but nonetheless important, remarks on the introduction and discussion rather weak and minor. Reviewer 2 had some more detailed comments on the introduction, results and discussion with which I fully agree. Therefore, I find it important that these are carefully addressed. I have some doubts whether that has been achieved in this revision, but leave the judgment in this regard to reviewer 2.
Author Response
Please see the attachment
Author Response File: Author Response.docx
