Remotely Sensed Changes in Vegetation Cover Distribution and Groundwater along the Lower Gila River
Round 1
Reviewer 1 Report
Introduction
L29 Genus and species?
L34 Do you want to mention the controversy with regard to willow flycatcher and tamarisk beetles as more reason for studying this plant?
L35-42 redundant use of “southwestern United States”
L36-37 suggest “… introduced, salt cedar spreads…”
L43 suggest “However, due to…”
L52 are you going to use “tamarisk” or “salt cedar”?
Materials and Methods
L97 suggest “…electricity, and drought; creating a…”
L119 need space between “3.75” and “min”?
L137 is “Bare Earth” a named dataset? Should it be capitalized?
L155 “…as rasters with 50 cm cells…”
L168-69 probably true but do you need to state this in the paper?
L176 “…stitched together ~!000…”
L177 “1.7 cm”
L244 and elsewhere, are you using litter and detritus interchangeably?
L255 “agricultural land”
Discussion
L369 you have defined “GRACE” earlier
L393 “require”
Overall, well done. No major concerns, just a few editorial comments. Interesting work concerning an important topic in the Southwestern US.
Author Response
Thank you for all your comments. You can find all the edits made based on your suggestions in the attached document.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This manuscript used multiple remote sensing data and C50 tree model to identify species in the study area, carried out a long time span of cover change analysis to detect the invasive processes and hazards of salt cedar. I appreciate the use of different types of remote sensing data and the progressive experiments for an interesting topic problem. On the other hand, the aspects of the analysis that I was able to understand raised serious questions about the validity of the work.
Detail comments below:
1 It is not clear the significance of the buffer analysis. From the description of this section, it is only for verify “most gains in vegetation are evident along stream channel”.
2 C50 algorithm was used many times in this manuscript, but there is no description of this method. Besides, if the conclusion does not need the interpretable advantage of a tree structure, why not use other algorithms?
3 Lack of description of accuracy verification method. From the information in Table 2, I would think the three species verification samples are inhomogeneous.
4 Extensive results described by these experiments will be more valuable if some management strategies are added in the conclusion part.
5 The abbreviation “UVA” needs to be completed.
6 The numerical information in Figure 5 needs consistent with the description of the text.
In my opinion, this manuscript does not rise to the standards of the Journal in the current form. I would be happy to review a revised version of the manuscript if there is a significant improvement.
Author Response
We would like to thank you for your comments on our manuscript. Your insights have helped us clarify our methodologies. In the attached document you can find responses to each of your detailed comments.
Thank you again.
Author Response File: Author Response.pdf
Reviewer 3 Report
This is an interesting paper that presents a classification method of riparian woody plant species as well as its application to report vegetation change over +2 decades in a desert river system of SW North America. The topic is relevant to the international literature as there is a need to refine the techniques for vegetation mapping from remote sensing imagery. The case study chosen is useful to illustrate the idea for the presence and high occurrence of an invasive tree (saltcedar) which may be highly benefit from global change. The article is well written, the methods look sound (even though it is hard for me to assess the methods entirely as I am a riparian plant ecologist but not a remote sensing expert). The results are overall clearly presented and the discussion answers the research questions. My main suggestion for improvement is a better integration of the water-vegetation relations (and in general the ecological interpretation of the maps) in the paper. As presented now, the paper raises too many expectations about these relations and deviates a little from the main strength of the paper that is the classification method presented. There are a couple of ideas for reorganization that you will find in my comments by section. The modifications that I suggest are not really major. I also recommend below some small improvements in the figures and in the writing style, but again, I hope they shouldn’t take too much effort from the authors. I enjoyed reading this manuscript.
Abstract
Line 13. In my opinion, saltcedar mainly occupies the space vacated by native species because of regulation and changes in land use, but yes, there is some competition too.
Line 23. Not sure if the paper really assesses biodiversity, but “only” the cover of three of the most abundant woody species.
Introduction
General comment:
Very well written interesting context. I agree in that a better capacity to map riparian vegetation is needed and the potential of new technologies to do this hasn’t been fully exploited yet.
Comments by line:
Lines 29-34. I would tune it down here as some studies have shown that the relationship between Tamarix and biodiversity is not so linear. Bird species for example do better in mixed stands of Tamarix and Salicaceae (see Charles Van Riper studies). Where native species are no longer viable, Tamarix represents a viable alternative for hosting wildlife (see long debate about southwestern willow flycatcher and other key bird species). Also see my comment in abstract about saltcedar replacing rather than displacing native species (see Stromberg et al., 2007 Freshwater Biology). This revolves around the debate of Tamarix as being a passenger or driver of ecosystem change (see for example Sher and Quigley 2013 book).
Line 45. Please use SI metric in the paper.
Line 67. A vegetation map of the main woody species is not really an assessment of biodiversity (plant composition is strictly not reported).
Line 74. Following your logic, is not biodiversity one (maybe the main) component of ecosystem health? There is some subjectivity in how to define ecosystem health that is maybe not necessary for this paper. Identifying how vegetation changes and how it relates to water use and groundwater dynamics is more than enough as a goal.
Methods
Disclaimer: remote sensing is not my field of expertise; I hope the other reviewers can give a fair assessment of the techniques used.
General comment: The explanation about the image classification seems well organized and very detailed. However, I was surprised not to find anything about the comparison of vegetation maps and hydrological and climatic data (last paragraph of the intro).
Comments by line:
Line 90. Please use SI.
Lines 96-98. With so many alterations, not really sure we can really refer to it as a “natural” riparian system. I guess it is fine to leave it like this.
Lines 103-107. Give botanical authorities.
Lines 168-169. Kind of poetic, probably unnecessary (I kind of like it as I empathize)
Results
General comment: From the intro and methods, I expected more formal analyses of the relations between vegetation, hydrological and climatic variables.
Comments by line:
Lines 283-284. And maybe not such a good job with the creosote and mesquite. Most of the accuracy value is due to saltcedar. The accuracy to identify creosote and mesquite was more moderate but in general this has been acknowledged.
Discussion
General comment: Despite being mainly a methodological paper about vegetation mapping, the discussion revolves around the explanation of vegetation decline. I am ok with keeping this approach but I the discussion about the new methodology of classification presented, which is the main contribution to the paper, should go first (section 4.2). If the goal is to explain vegetation decline, then the water-vegetation relations need to be a central part of the results section (and more info in the methods too) and do more formal (statistical) tests of water-vege relations (don’t know if that is even possible with the spatial resolution of the water data).
Another general comment is that in section 4.1 there are too few references to other works. Is the situation you are describing similar in other rivers of arid and semi-arid regions, or in rivers of the region? What is different? Where is room/need for research?
Comments by line:
Line 333. Nearest to the main channel of the Gila River (the entire riparian corridor may be seen as the Gila River).
Line 340. Remove the comma.
Line 355. Say which USGS stream gauge is that one.
Line 357. Just for curiosity, how far is the Painted Rock Dam from the study area? If you improve the study area figures, consider adding it to the map.
Lines 364-366. This is results not discussion (or even methods, description of the study area).
Lines 367-377. This fits better in results.
Conclusions
Line 397. Dominant woody plant species to be more precise.
Line 398. The amount of invasive salt cedar (would be good to insist here in the fact that salt cedar is invasive as that makes it of particular interest).
Line 399. I would say the diversity of habitats as biodiversity can be misleading (this is not a study of plant composition that is generally more informative of biodiversity).
Line 417. I fully agree but avoid poetic writing.
Line 421. And other parts of the world, maybe cite the situation of the Tarim and/or Heihe rivers in NW China (dryland, desert ecosystem). I think dust storms are an issue over there too. There are multiple papers that talk about the riparian vegetation collapse in those rivers for overexploitation of water resources, looks like a similar situation.
Figures and Tables
Fig. 1. It would have been nice to have the towns of Sentinel and Dateland depicted in the map or in the photo, but it is not a critical comment.
Fig. 2. Nice figure.
Fig. 3. Caption: Avoid adjectives that may contain opinion such as excellent or overwhelming. Try to remain more neutral. Also, the acronym CART is not explained (does not appear anywhere else in the paper). Clarify in methods which steps are part of the CART. Probably a brief technical explanation of what regression trees are would be nice for readers not familiar with this technique.
Fig. 5. I would remove the horizontal lines.
Fig. 6. No legend for the red and black circles. Maybe could be solved by not having a blue color in the legend (circles depicted in white for example) so we know that the legend refers only to circle size. Indicate the code of the two USGS stream gauges in the map. No legend for the black polygon (I understand it is the study area that corresponds to Fig. 1 but needs to be said). It is hard to read the map in the background of the figure. Consider creating your own map and labels.
Fig. 7. Remove the grid in the background.
Table 1. Indicate the meaning of the OHM acronym in the caption.
Author Response
We would like to thank you for your comments on our manuscript. Your insights have helped improve all portions of the paper. In the attached document you can find responses to each of your detailed comments.
Thank you again.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
Base on the authors' effort and reviewers' suggestions, the manuscript has significantly improved in the scientific context, method, and any more details. I agree to publish this manuscript.
