Can terrestrial laser scanners (TLSs) and hemispherical photographs predict tropical dry forest succession with liana abundance?

Sánchez-Azofeifa, Gerardo Arturo; Guzmán-Quesada, J. Antonio; Vega-Araya, Mauricio; Campos-Vargas, Carlos; Durán, Sandra Milena; D'Souza, Nikhil; Gianoli, Thomas; Portillo-Quintero, Carlos; Sharp, Iain

doi:https://doi.org/10.5194/bg-14-977-2017

Articles | Volume 14, issue 4

https://doi.org/10.5194/bg-14-977-2017

© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 3.0 License.

https://doi.org/10.5194/bg-14-977-2017

© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 3.0 License.

Articles | Volume 14, issue 4

Research article

|

02 Mar 2017

Research article |

| 02 Mar 2017

Can terrestrial laser scanners (TLSs) and hemispherical photographs predict tropical dry forest succession with liana abundance?

Gerardo Arturo Sánchez-Azofeifa, J. Antonio Guzmán-Quesada, Mauricio Vega-Araya, Carlos Campos-Vargas, Sandra Milena Durán, Nikhil D'Souza, Thomas Gianoli, Carlos Portillo-Quintero, and Iain Sharp

Download

Final revised paper (published on 02 Mar 2017)
Preprint (discussion started on 26 Oct 2015)

Interactive discussion

Status: closed

AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment

- Printer-friendly version

- Supplement

RC C7871: 'referee comment', Anonymous Referee #1, 23 Nov 2015
- AC C9128: 'Responses to reviewer No. 1', A. Sanchez-Azofeifa, 15 Jan 2016
RC C7901: 'Anonymous review #2', Anonymous Referee #2, 24 Nov 2015
- AC C9136: 'Response to reviewer No. 2', A. Sanchez-Azofeifa, 15 Jan 2016
AC C9113: 'Response to external reviewers', A. Sanchez-Azofeifa, 15 Jan 2016

Peer-review completion

AR: Author's response | RR: Referee report | ED: Editor decision

ED: Reconsider after major revisions (20 Jan 2016) by Anja Rammig

AR by Arturo Sanchez-Azofeifa on behalf of the Authors (01 Mar 2016)

ED: Referee Nomination & Report Request started (15 Mar 2016) by Anja Rammig

RR by Anonymous Referee #1 (04 Apr 2016)

Suggestions for revision or reasons for rejection

General
In their revised version the authors did some efforts to respond to the comments of both reviewers, which I appreciate. Additional information was provided on the methods and analysis. However, with this additional information included, it became even more clear for me that the analysis is too weak and the assumptions are too shaky to support the conclusions (see my major comments). I can therefore not recommend this manuscript for publication in Biogeosciences.
In my opinion the authors have to make a choice: or (1) they make an assessment of VEGNET to detect differences in tropical forest structure, (2) or they use VEGNET to study the impact of liana’s on forest structure, or (3) they combine both.
In their response letter the authors state that the actual purpose of the paper is to assess whether TLS (but you are actually testing VEGNET, which is due to the fixed zenith angle quite different from most other TLS approaches) can detect differences in aboveground structure across plots. This is actually option (1), but this is not what is presented in the paper.
For each option more data is needed:
(1) To assess the potential of VEGNET to study difference in structure of tropical forests, you need another quantitative method to observe forest structure, to compare VEGNET to. Now the VEGNET is just applied on rather arbitrarily chosen plots, for which you assume they have a different structure, but no quantitative comparison is made with other data. You could compare the vegnet information with inventory data, hemispherical photo’s, tree height measurements, other type of TLS… etc… some of this information is apparently available (page 8 line 168-176 in the manuscript) but not included in a quantitative analysis.
(2) To study the impact of lianas on forest structure you need more information on the plots. (and the plots probably need to be larger to do a decent study). You need background information on the plots to be sure that you are looking to effects of lianas and not to effects of other factors (species composition, soil, hydrology, topography, disturbance). Some of this info might be in the other studies where the authors refer to, but in that case that information should be integrated in this manuscript. Apparently the authors do have access to some information on the number of lianas per hectare (see response to reviewer 2), this information should really be integrated in the paper. In addition you really need young plots with lianas to study the impact of lianas along a successional gradient.
(3) To study both in one paper, you need all the info of 1 and 2.
I hope the authors can use my comments to improve their work in the future, by adding new data and doing a different analysis.

Major comments:
1. It became clear for me that the VEGNET approach is different from other TLS approaches in the sense that it scans at a fixed zenith angle. This really limits the info that you can extract from the scans compared to other TLS approaches that can produce full 3D point clouds. This needs to be mentioned in the discussion.

2. The authors added some information to show that the plots are randomly selected. However, I still doubt that the setup is sufficient to study the selected research questions. The plots are probably independent replicate (yes I meant replicates) samples of ‘the forest’, but are they replicate samples of the categories that you define? (in terms of succession and in terms of liana presence?). I doubt. In addition, the plots are very small in size. Only 0.1 ha, means that the plots are very prone to edge effects. You might have 0 rooting liana stems in such a small plots, but the crowns might be full of lianas due to lateral ingrowth.

3. An additional weak point is for me the fact that only 1 scan per plot is performed. In combination with the fact that VEGNET is scanning at 1 fixed azimuth angle, I suspect that the observed structure is really depending on the scanning location. Did you do any tests by scanning at different locations within the plot? Moreover, no information is give on the actual location of the scanning position within the plot.

4. The authors state that they want to test the impact of lianas on canopy structure, but due to the lack of information about the plots and the rather weak set-up, the resulting patterns seen in the VEGENET data might not be caused by lianas at all.

5. It is now more clear for me what the RG parameter actually means. The authors use this index to explore if successional trajectories could be altered by lianas, by assuming that if they don’t see a change in vertical structural complexity, this would mean that lianas are altering the successional pathway. I really don’t understand this assumption. Why would an altered succession be reflected in a none-change in vertical structure? In addition the authors assume that if they don’t find an RG trend along the successional gradient for the plots with lianas, that this would indicate an arrested succession by lianas… But on the other hand the late successional plots are selected based on their structure. I really think this a weak setup. Also on page 13 (line 278) the authors write that “a change or no significant increase in PAI as a function of RG would suggest that lianas are altering the successional trajectory of vertical structure”. This sentence is very confusing. “a change” is contradicting “no significant increase”. I agree that VEGNET might be used to make observations of canopy structure of tropical forest, but even this probably needs to be tested. But to test if lianas are altering successional trajectories, you really need to present more information on the plots, come up with a setup that really allows you to study lianas of successional gradients, make scans in plots that are large enough, make multiple scans per plot, …

6. The answer of the authors to my comment 8, is not sufficient. The authors provide indeed some more information on the trends they found in fig 3. However this not answer to the key point of my comment: the fact that you find a significant positive relation for plots without lianas is mainly driven by the fact that you have observation in early successional plots. You don’t have observations in early plots with lianas, leading to a non-significant trend for the plots with lianas. Try to fit a curve through the plots without lianas, but only considering the intermediate and late successional plots, the trend will probably be lost…

Specific comments
1. Depending on which option the authors choose, the title of a future study should be adapted: if they chose option (1), the title should be adapted: e.g. ‘the potential of VEGNET to observe canopy structure in tropical forests’
2. Line 25: vegetation ‘components’, I guess you mean ‘indices’?
3. Line 206: PAVD is estimated based on the derivative of LAI, but how is LAI estimated?
4. Line 245: ‘RG is a measured that has..’ something is wrong with this sentence.
5. Line 286: ‘liana treatments’ here and elsewhere this is confusing terminology, because a lot of readers will interpret it as if you experimentally added or removed lianas.
6. Table 1: units of variables are missing: Cx, cy, RG, PAI
7. Table 1: the plots are 20x50 m. I assume that the centroid coorindates Cx and Cy are in meters. Does this mean that the Cy coordinates which are close to 10 are located really at the edge of the plot? What about edge effects?
8. Fig 3, y axis, units are missing. Are these boxplots based on only 3 observations per boxplot? Probably not the best way to represent the data.

Hide

RR by Blaise Tymen (06 Apr 2016)

ED: Reconsider after major revisions (18 Apr 2016) by Anja Rammig

AR by Arturo Sanchez-Azofeifa on behalf of the Authors (02 Nov 2016) Author's response Manuscript

ED: Referee Nomination & Report Request started (09 Nov 2016) by Anja Rammig

RR by Blaise Tymen (21 Nov 2016)

ED: Reconsider after major revisions (25 Nov 2016) by Anja Rammig

AR by Arturo Sanchez-Azofeifa on behalf of the Authors (30 Dec 2016) Author's response Manuscript

ED: Referee Nomination & Report Request started (03 Jan 2017) by Anja Rammig

RR by Blaise Tymen (12 Jan 2017)

ED: Publish subject to minor revisions (Editor review) (16 Jan 2017) by Anja Rammig

AR by Arturo Sanchez-Azofeifa on behalf of the Authors (24 Jan 2017) Author's response Manuscript

ED: Publish as is (30 Jan 2017) by Anja Rammig

AR by Arturo Sanchez-Azofeifa on behalf of the Authors (08 Feb 2017)

Short summary

Lianas (woody vines) tend to respond positively to disturbance and show high densities in secondary forests. Lianas are a key component of tropical forests because they may reduce carbon potential. Identifying tools for liana detection is therefore essential for monitoring changes in tropical forests. In this study, we describe how terrestrial laser scanning can be used to detect the presence of lianas in forest stands of different ages in secondary tropical dry forests at a regional scale.