Unravelling Complex Interaction among Coastal Management and Marine Biodiversity: A Case Study in Southern Spain

Round 1

Reviewer 1 Report

Review of Unravelling complex interaction among coastal management and marine biodiversity: a case study in Southern Spain

This excellent paper seeks to unravel the processes associated with coastal management with particular reference to Spain but also a method termed PLS-PM.  The paper is well written and provides a detailed explanation of the topic.

My comments are minor but important.

1.     The methods section lines 186+ provide a description of the PLS-PM method. However this method is relatively unknown and seldom used. For this paper to work the authors need to create a ‘toy’ model to explain how the steps work. Even as I was very familiar with SEM and BN I struggled to step through the extensive list of specific terms such as Inner and Outer models etc and I was left wondering how it all works. See section 2.2.1 page 24 Sanchez 2013 for ideas although noting that the R package is not available presently.

2.     The methods are detailed but little attention is given to potential alternative approaches. For example Bayesian Networks could have addressed the modelling problem in a similar way with potentially a larger audience. Perhaps present the reasons for the method selection.

3.     Line 318 extra )

4.     Line 168 extreme decline. Line 171 increased.

5.     Line 239. Like all models there are limitations. Areas that need addressing such as; data distributions, sample size limitations, unsatisfactory fitness index, misunderstanding between confirmatory and exploratory research, and poor factor loadings (Afthanorhan et al. 2020 doi:10.7903/cmr.20247)  would help the paper more. The explanation of the GoF was very good within this context. Include a broader examination of the methods than line 522 including applications to other regions.

Author Response

1. The methods section lines 186+ provide a description of the PLS-PM method. However this method is relatively unknown and seldom used. For this paper to work the authors need to create a ‘toy’ model to explain how the steps work. Even as I was very familiar with SEM and BN I struggled to step through the extensive list of specific terms such as Inner and Outer models etc and I was left wondering how it all works. See section 2.2.1 page 24 Sanchez 2013 for ideas although noting that the R package is not available presently.

Answer: We have reviewed with attention the works performed by Sánchez 2013 (ref. 48), Fu et al 2015 (ref. 50) and Selim et al 2016 (ref. 49) and follow their way of thinking in order to build our “toy model”. This analysis is also new for us for this reason we were cautious and scrupulous in describing the model trying to describe carefully and clearly the different concepts required to build the model. The design of our conceptual path model is presented in Figure 2 and we hypothesize that the biodiversity loss is due to the influence of anthropogenic pressures while these ones decrease the biodiversity capacity of ecosystem services delivery. Meanwhile this biodiversity loss triggers the launching conservation and management responses.

2. The methods are detailed but little attention is given to potential alternative approaches. For example Bayesian Networks could have addressed the modelling problem in a similar way with potentially a larger audience. Perhaps present the reasons for the method selection.

Answer: In a first version of the manuscript we used DPSIR to explore the complex and multiple relationships among biodiversity, marine ecosystem services delivery, pressures, conservation responses and management strategies which is a more widespread multicriteria analysis but we read the papers of Fu et al 2015 and Selim et al 2016 developed to explore the relationships among fisheries exploitation, environmental variables and ecosystem services delivery, and decided that the PLS-PM would fit well to our objective because it is a very straightforward method, easy to explain and understand. These authors proved the PLS-PM to be useful to quantify the strength of the links between ecosystem processes and services. So, we decided to use PLS-PM instead of DPSIR or Bayesian networks.

3. Line 318 extra )

Corrected

4. Line 168 extreme decline. Line 171 increased.

Corrected

5. Line 239. Like all models there are limitations. Areas that need addressing such as; data distributions, sample size limitations, unsatisfactory fitness index, misunderstanding between confirmatory and exploratory research, and poor factor loadings (Afthanorhan et al. 2020 doi:10.7903/cmr.20247) would help the paper more. The explanation of the GoF was very good within this context.

We follow the produces explained by Esposito-Vinzi et al. (2010) ref 47 and Sánchez (2013) ref. 48 to assess the quality of the measurement, the structural and the overall models:

1) the measurement model, using the Cronbach's alpha that evaluates how well a block of indicators or manifest variables (MVs) measures their corresponding latent variables (LV), whilst the Dillon- Goldstein's rho focuses on the variance of the sum of variables in the block of interest. As a rule of thumb, for both indices a value alpha greater than 0.7 is considered acceptable.

2) the structural model, evaluated by examining three indices or quality metrics: the predictive power of the model R² where R²>0.6 is considered high; the Redundancy index and the Goodness-of-Fit (GoF) where acceptable values are those >0.7. In addition, the loadings i.e. the correlations between a LV and its MVs, when are greater than 0.7 are acceptable.

3) the quality of the overall model pathways and results, where checked through bootstrapping analysis and using the 95 % bootstrap confidence interval to evaluate whether the parameters are significantly different from zero.

In this context, our results are quite good. We have read carefully the paper of Afthanorhan et al. 2020 and reinforce our idea the use of PLS-PM as a powerful tool to quantify the strength among the complex links between ecosystem processes and services.  

Include a broader examination of the methods than line 522 including applications to other regions.

Included.

Reviewer 2 Report

In general, this paper is well written and researched with new input into the current literature. I do, however, have some comments to improve the overall quality of the paper.

1.      All Tables and Figures presented in the paper should provide their sources.

2.      P2., at the end of 1. Introduction, you need to provide the overall structure of this paper.

3.      P.4, “summarising”; “summarized” you mixed up the British and American spellings. This should be fixed.

4.      P.4, line 169, “…and the decline in the Alborán Sea over the period”. Decline of what?

5.      4. Discussion. It is noted that you are comparing two maritime areas in Spain, namely, “The Alborán Sea” and “The Gulf of Cádiz”. It would be much better if you could tell the reader what are the similarities between these two areas? Differences? What is the reason behind these observations? At the moment, you have all required material in this section but presented as a big whole piece without comparing these two areas in a logical way.

6.      As a research paper, you need to add an extra part as 5. Conclusion.  

Author Response

1. All Tables and Figures presented in the paper should provide their sources.

All the Tables and Figures presented in the paper are made by ourselves. In addition, the Table S1 of “Supplementary Tables” file, includes all the consulted data base sources to gather the information necessary to construct the indicators.

Figure 2 is modified from the EMODnet Bathymetry Viewing and Download portal service available at https://portal.emodnet-bathymetry.eu. on as it is mentioned in its Figure caption.

2. , at the end of 1. Introduction, you need to provide the overall structure of this paper.

We added a final paragraph at the end of the Introduction section, summarising our posed hypothesis.

3. 4, “summarising”; “summarized” you mixed up the British and American spellings. This should be fixed.

Done

4. 4, line 169, “…and the decline in the Alborán Sea over the period”. Decline of what?

Corrected. The phrase is referred to the decline of the state of biodiversity.

5. 4. Discussion. It is noted that you are comparing two maritime areas in Spain, namely, “The Alborán Sea” and “The Gulf of Cádiz”. It would be much better if you could tell the reader what are the similarities between these two areas? Differences? What is the reason behind these observations? At the moment, you have all required material in this section but presented as a big whole piece without comparing these two areas in a logical way.

We added a paragraph at the beginning of the “2.1 Study area” justifying why we choose this areas, that area located in the same administrative region, the Autonomous Community of Andalusia, but that belong to two different marine Demarcations and present different physiographic attributes. Thus, they have their indicators described by area (Gulf of Cádiz and Alborán Sea) except the indicators related with the conservation and management normative delivered at Regional scale. All the data bases consulted for the construction of the indicators is summarised in Table S1 of the “Supplementary Tables” file.

We discuss our results comparing always both marine areas, which are different since the physiographical perspective as explained in the “2.1 Study area” subsection, specially the ones related with the ecosystem service delivery.

6. As a research paper, you need to add an extra part as 5. Conclusion.

The detailed instructions presented in the section of “Research Manuscript Sections” of the Sustainability journal, stated that the “Conclusions” are not mandatory.

Reviewer 3 Report

The manuscript is adequate for publication. Bibliographic references are missing in the introduction and methods. See the PDF for comments. 

Comments for author File: Comments.pdf

Author Response

Specific Comments. Attached pdf file.

Line 36: No need

Line 2: There is no need for this "estimation": for the last tens of thousand years.

We agree, we deleted it.

Line 41: change pelagic to “marine ecosystems”

Done

81-82: I think the introduction contextualizes the background quite well, but the objectives are rather generic and it is unclear what is the specifics. Thus, I am uncertain regarding how the objectives truly connect with this background information, specially with the last paragraph, which is a little dislocated from the others. It seems that only fisheries are being evaluated, so I am uncertain as to whether the biodiversity encompasses all components of coastal and marine ecosystems.

Thus, I suggest to specify the exact questions being addressed, as well as to mention that a specific area is being used as a case study.

We added a final paragraph at the end of the Introduction section, summarising our posed hypothesis which stated the rational of the study. We do not mention in the Introduction the specific area being use as a case study because the former subsection of the Material and Methods section is focused in the presentation and description of the studied marine areas.

112: It is unclear what this means. Please, rephrase.

Done

116: sediments

Corrected

119-122: Both for the Alboran Sea and the Gulf of Cadiz, I suggest to revise the description to focus on the most important information of characterization for the purposes of the study. The characterization is interesting, but is excessive for a study that, at least to this point, do not focus on physical oceanographic characteristics.

We try to summarise at maximum the different characteristics of both areas which are very complex and heterogenous. We think the presentation of this complexity is crucial to understand why they are considered hotspots of biodiversity and why is important to analyse the interactions among the ecosystem service delivery, fisheries and conservation and management responses in these two contrasting marine areas.

129: As the objectives are too generic, I am unsure as to what are the purposes of the indicators. This could be better explored in the objectives, especifying what kind of indicators are being targeted (biological, physical, social, and so on.).

All the data bases consulted for the construction of the indicators is summarised in Table S1 of the “Supplementary Tables” file. In the text we say ”… their assignment to the different blocks of the path analysis dimensions, and the data source used for obtaining the database are summarized in Table S1”. We added that the information was gathered in a yearly basis.

131-132: I suggest to change to: "Indicators were evaluated separately for each of the two study areas, that are characterized by distinct environmental features, species composition and fishing pressure.

Done

136: Please, inform somewhere the periodicity (e.g., yearly) that the indicators were evaluated and if all were measured within the same periodicity.

Done

Looking at Table S1, it seems some of the indices do not have periodic measurements (such as the MTI, or changes in conservation categories). Does this affect the analysis?

MTI is calculated for every year using the specific composition of the catches and their trophic level, the exact method to compute it can be found in the cited literature (Pauly et al., 1998). Changes in the conservation categories have also been calculated for every year checking the UICN red list and national and regional lists of endangered species. And it does not affect the analysis.

162: By when possible, do the authors mean "when available"? Does this limitation affects their comparison with other indicators?

“When possible” has a broader meaning than “when available” since those indicators were only included when we were sure that the amount of data was large enough to be comparable with the other indicators.

168-169: Please, rephrase. I think it needs a language revision

Done

170-799: Here, it seems the results are already being shown. Isn't the discussion of the temporal trends of the results one of the goals?

I think I still miss the true goal of the manuscript, which again, should be better stated in the objectives.

We think that with the inclusion of the last paragraph in the Introduction section presenting our hypothesis will help to understand better the rest of the text.

209: I like this figure. It does not allow to see what MVs truly represent (it would be very hard to summarize all this information anyways), but it does a good job of summarizing the model!

Thank you.

213-214: Here is the kind of information that I believe should be in the objectives. Then, the methods start making more sense.

We agree with you and decides to include our hypothesis at the end of the Introduction section.

218: I miss some information on how the variables were treated. I see that many variables have temporal "replicates" due to periodic measures, but others do not seem to have this pattern. How was this treated? How variables were inserted in the model (as means, or the whole trend was inserted)?

All the variables have yearly measurements

234-236: I enjoy the explanation of the models. I am not very familiar with them, but I was able to understand it very well.

Thank you.

268: R²

Corrected.

325: Are these weights standardized to a range? In other words, how one evaluates what is a high or low weight for the MVs?

We explained the different ways to evaluate the quality of the models in the subsection “2.5. Evaluation of the path regression model”. The formative MVs weights measure the contribution of each MV to the LV, the ones that are causing a LV. The elimination of some MV indicator must be done based on multicollinearity, if high multicollinearity exists, it must be deleted.

350-351: I get the description is important, but I suggest to also focus on the actual meaning of this statistical feature

In the subsection of Material and Methods “2.5. Evaluation of the path regression model” we explained why is important to check the cross loadings for the evaluation of the outer model:

“   Then it is necessary to check the cross-loadings which are the loadings of an indicator with the rest of LVs, to verify that the shared variance between LV and its MVs is larger than the shared with other LVs by looking at the diagonal of the cross-loading matrix block by block.”

So, if most part of the reflective MVs showed that the shared variance between LV and its MVs were larger than the shared with other LV looking at the diagonal of the cross-loading matrix block by block in both outer models, it means that the outer moder is well design. The interpretation of these results is given in the “Discussion” section.

365: Remove the point to unite these sentences

Done.

367-369: But this does not mean a weak relationship, right? I feel maybe these relationships between LV could be further explored, if any interesting patterns can be discussed

The interpretation of these results is given in the “Discussion” section.

373: If this is going to be discussed as a finding, then the temporal trends are better suited for the results rather than be discussed in the methods.

We believe the right section for the temporal trends is the “Material and methods” section since the complex process of selecting the indicators that can be included in the model is an essential part of the PLS-PM methodology. We think that if we take this explanation to the results section the building of the conceptual model would be difficult to follow.

Maybe even the temporal trend of biodiversity (it does not need to be for all MVs) could be used a main figure to illustrate this result.

Figures showing the trends of all indicators are included in the supplementary materials (Figures S1 and S2). We think that the trends of the indicators and the selection process can be better understood when are observed together.

But on a sidenote, it is important to highlight that the MVs are a proxy of potential biodiversity loss and not actual measures of biodiversity (at least not for most components of marine biodiversity, as I understand from the methods); Thus, this result also need to be discussed with some reservations.

We are fully conscious that we are using proxies and not real estimates of biodiversity in the area. Table S1 in the supplementary materials explain the nature and source of every indicator. More classic approaches to biodiversity measurements are impossible to compute for such a long time series considering that the actual biomass and/or relative abundance of most taxa is unknown.

404-405: Please, correct me if I am wrong, but it is not that decrease in pressure did not reduced biodiversity loss rates, it seemed to have a contrasting effect. I understand from the negative path coefficient between these two variables that a decrease in pressure would result in an increase in the biodiversity loss (and vice-versa).

Is it the right interpretation?

If so, what could have brought such result. I miss some explanation for this, especially considering it is stated as a main hypothesis of the work.

This result also surprise us and after many discussions and revisions of the results we think that this strange behavior of the interaction between anthropogenic pressures, biodiversity loss and responses should lie in the lack of efficiency of the given responses and /or the lack of time to probe their effects in an improvement of Biodiversity state and ES delivery. We added a final sentence to this paragraph in the manuscript.

412: As someone who works directly with the management of protected areas, I would add that policies are often inefficient because knowledge is very limited to support them and are often lost in political interests (resulting in the so-called "paper parks"). This is an unfortunate situation where PAs were created to meet global targets, not for actual protection (but I do not know if such is the case for the study area).

Added.

454: great

491-493:  This sentence is a little confusing. Please, rephrase. “although PPR (primary production required to sustain the fishery [94]) was selected as an important  MV in both areas, MTI (Marine Trophic Index [95]) was only considered in the Alborán  Sea (Figures 3 and 4).”

Both, PPR and MTI are indicators related to regulating ecosystem services. PPR was informative enough in both areas, and it was included in both models, but MTI had different trends in each one of them (as can be seen in figures S1 and S2). The model in the Gulf of Cádiz was significantly worse when that MV was included and we decided to left it out following the criteria explained in the methods’  section.

526-527: It is a little unclear to me how this conclusion relates to the result. I suggest to elaborate a little further.

The explanation as to why the implemented responses are not enough are already discussed in detail in the previous paragraphs (Lines 420 to 466 of the corrected version of the manuscript)

Reviewer 4 Report

General Comments. The manuscript aims to evaluate, through a path modelling approach, the complex interactions between biodiversity, ecosystem services, management practices and anthropic activities. It is a really interesting manuscript, that is generally well-written, and tackles an often assessed topic in a original way. The manuscript does a good job of contextualizing the background and connecting the information among the sections. It is also very competent in explaining the modelling approach, which is not trivial, but is well laid out with quality presentation of the results. The discussion also addresses most relevant points without being speculative. Although these qualities leave little doubt regarding the potential contribution of the manuscript, I believe some issues and suggestions can be addressed before acceptance. Those are mainly related to the need of better explaining the objectives, to clarify some points in the methods, and to further elaborate the discussion of some results. I have attached a pdf file with those suggestions and questions.

Specific Comments. Attached pdf file.

Comments for author File: Comments.pdf

Author Response

We don't have a fourth reviewer

Round 2

Reviewer 2 Report

All my comments have been addressed. Accept!

Reviewer 4 Report

I have completed the review of the new version of the manuscript. Initially, I could not find the response to my comments (with only a comment saying "We don't have a fourth reviewer"), but I was able to find them addressed to another reviewer (#3). Having said that, the authors carried out a good review job, addressing most of the issues I have raised. Given that my previous report was already positive to the content, structure and results provided in the manuscript, and that the review process carried out by the authors was satisfactory, I recommend the acceptance of the manuscript.