Vegetation Dynamic in a Large Floodplain Wetland: The Effects of Hydroclimatic Regime

Round 1

Reviewer 1 Report (Previous Reviewer 1)

This manscript has been well improved according to my previous comments. No more other comments.

Author Response

Thanks for your positive comments.

Reviewer 2 Report (Previous Reviewer 2)

The authors largely addressed my earlier comments in this revision. I still have some concerns about this study:

(1) In Introduction, the authors claimed that the purpose of study is "to identify the key flow regime parameters that....." and in Conclusions, they stated that two flow regime indices are the main driving forces for the temporal changes of wetland types. This conclusion is based on the results from EDT where less degree of human disturbance exists. Why these results could lead to the statement in Abstract: "Our findings suggest that flow regime acts in concert with other factors, such as climate change and sand mining in river channels, driving wetland habitat transition in floodplain landscape."? Also, why do these results mean flow regime is not the sole player? Somewhere in Introduction, Discussion, and/or Conclusions, the authors should explain this logic.

(2) I still do not quite understand Figures 6 and 7: what does 'Effect on area' mean? I understand the figures were based on the modeling results. But in figure captions, the meaning of vertical axis needs to be explained.

(3) Figure 1c: the map was distorted. It seems that the authors used a wrong project to produce this map.

In general, English is easy to follow. 

Author Response

The authors largely addressed my earlier comments in this revision. I still have some concerns about this study:

(1) In Introduction, the authors claimed that the purpose of study is "to identify the key flow regime parameters that....." and in Conclusions, they stated that two flow regime indices are the main driving forces for the temporal changes of wetland types. This conclusion is based on the results from EDT where less degree of human disturbance exists. Why these results could lead to the statement in Abstract: "Our findings suggest that flow regime acts in concert with other factors, such as climate change and sand mining in river channels, driving wetland habitat transition in floodplain landscape."? Also, why do these results mean flow regime is not the sole player? Somewhere in Introduction, Discussion, and/or Conclusions, the authors should explain this logic.

Response: Thanks for your comment. We added the following in the Discussion on the reed bed expansion:

“Other broader scale factors, such as the increase of winter rainfall associated with cli-mate change [107]and the elevated atmospheric CO₂ level (108), might be responsible for the continuous spread this cosmopolitan species”.

(2) I still do not quite understand Figures 6 and 7: what does 'Effect on area' mean? I understand the figures were based on the modeling results. But in figure captions, the meaning of vertical axis needs to be explained.

Response: Thanks for you comments. We added “The Y-axis is the modelled partial response of habitat area to Year” (Fig 6) and “The Y-axis is the modelled partial response of habitat area to Date of water level withdrawal” (Fig 7) in the caption for clarity.

(3) Figure 1c: the map was distorted. It seems that the authors used a wrong project to produce this map.

Response: No sure what you meant by “distorted”. We used the WGS 84 map projection.  

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

I appreciate this work and the topic of this manuscript fits the scope of our journal well. Overall, the selection of the study area has some new idea, and the optimate accuracy of the results confirmed the good performance of the RF model in wetland classifications. Moreover, the authors investigated the response of wetland habitats to flow regime variations by using GAMM including DOYD and DOYW, respectively, which improved the reliability of the results, because the hydrological process of large floodplain is highly complex, and varied significantly between intra and inter-annual. Nevertheless, the following problems still exist for improvement. 

 

Detailed comments,

Line 201, What does “for computing convenience” mean? Why you need to shift the time series back by three months?

Line 222-223, What are the values of the two parameters and how to acquired them?

Line 243-247, The nine environmental variables selected in this study is very important, rationality and full representativeness should be explained?

 

Minor comments,

Line 191, Abbreviation of the NDWI

Line 341, table2?

In table 4 ‘DOYW’ or ‘DOYD’?

Line 426-428, list the references

Line 471, GAM models?

Reviewer 2 Report

This study examined the temporal changes of four main habitats in east dongting lake (EDL) of China using machine learning (ML) techniques and statistical modeling. The authors showed that the area of reedbeds has expanded, while that of mudflats decreased over the 34 years, and subsequently concluded that the duration of dry season (DRS) and the date when water level starts to withdraw (DWLW) were the key hydrological factors causing these habitat changes.

 

In my opinion, the main results have less scientific significance because they have been discovered in previous studies as the authors stated. Perhaps the main value of the study lies in the use of the ML technique (RF) and the GAMM model. Unfortunately, the descriptions of the two methods are very vague and the outcomes are not convincible. The RF method requires to split the samples into training and validating classes for each year. The number of samples in each year is essentially based on the number of Landsat images available in each year. By splitting these images into two groups (75% and 25%), I suspect that some sample sizes are very small because the total number of samples is 772 for the entire 34 years. The GAMM modeling results (Table 4: DOYW should be DOYD) are very hard for me to understand. Does Ref.df mean that there are 9 independent variables in the model? How should we explain the result with p = 0.69? The results are not significant? Without reasonable explanation of these results, I am not convinced by these results. Also, a very important issue ignored by the authors when using the GAMM model is the potential impact of spatial correlation on the results. There is definitely certain degree of spatial autocorrelation (SP) for each independent variable. Without assessing the possible impact of SP on the GAMM model, I cannot accept the results reported in Table 4. What further intrigues me is that in conclusion, the authors specifically claimed that the two hydrological factors (DRL and DWLW) are the main cause for the changes of the habitats in the study area, but in Abstract, the authors stated that the driving factors are the combination of hydrological factors and human disturbances. Keep in mind that in the second paragraph of section 4.2, the authors claimed that they did not include tha areas affected by human activities by stating “we excluded areas with reed plantation and winter cropping”.

 

Overall, I think the data obtained in this study are potentially valuable. However, the analyses the authors did fall short. Why not plot the areas of each habitat type again time to examine their temporal trends?

 

Minor points:

 

The Introduction needs to be significantly revised. Froe example, L109-113 and L118-123 should be moved to the earlier sections; L57-58 is hard to understand; and L105 ‘driving’ should be drives;

 

L125: efforts should be effects;

 

L151: calling should be calling for;

 

L167: at should be in;

 

L209-213: this sentence is awkward. It needs to be revised;

 

L257: manually?

 

L338: the number of the title should be 3.3;

 

L411-412: this statement needs to be verified;

 

L426: this sentence needs citations to support;

 

L449: this is not quite right. It oversimplified the relationship shown in Fig. 7;

 

Fig. 1 needs to be significantly revised: (c) and (d) should (a) and (b). They are NOT ‘inset’. Also, (c) has a problem of projection. The shape of mainland China is distorted. For (d), the main channel of Yangtze River should be highlighted using a thicker curve

Reviewer 3 Report

General comments

The paper entitled “Habitat transition in a large floodplain wetland: flow regime is not the sole player” treats about a topic of the highest interest in watershed modelling scope. Therefore, the topic clearly falls under the journal scope. The declared objective of the manuscript is mapping the winter distribution of four wetland habitats by means of the random forest model applied to different areas and datasets. Moreover, Authors pushed their analysis forward by investigating the response of wetland habitats to flow regime variations using generalized additive mixed models (GAMM). Finally, Authors accomplished a change detection analysis by comparing images of wetland captured at different years. This last study has been conceived to support the restoration strategies to be applied where the habitats degradation was sensible.

1.       What is the main question addressed by the research: The declared objective of Authors is to provide an effective source of information for support environmental planner and managers with the knowledge to recover degraded wetland habitats.

2.       Do you consider the topic original or relevant in the field, and if so, why: This aim is pursued through the application of Random Forest machine learning method. Considered the critical objective the work can be considered relevant.

3.       What does it add to the subject area compared with other published material: The present work enhance the knowledge about the processes affecting the wetland habitats.

4.       What specific improvements could the authors consider regarding the methodology: a comparison with other machine learning methodologies at the state of the art.

5.       Are the conclusions consistent with the evidence and arguments presented and do they address the main question posed: yes.

6.       Are the references appropriate: The References are good with recent citations.

7.       Please include any additional comments on the tables and figures: Graphic representations and Tables are fine.

Concerning the manuscript, the typographical outline is satisfactory. The used language is very fluent and effective in explaining the processes of habitats degradation. Abstract and Introduction introduce the reader into the treated topic very well, clarifying the aims of the manuscript. Keywords are pertinent to paper content and appropriate. The highlights are missing. Materials and Methods are of good quality. The References section is rich with recent citations. Graphic representations are generally fine.

Entering in the very merit of the paper is opinion of this reviewer that the manuscript is well written and the experimental part is well conducted. The evidences show that some dependencies between habitats and flow are straightforward, whereas others are more complex and need to be deepened. This last is a very interesting result.

In conclusion, this reviewer recommends considering the paper in the present form.