The Role of Landscape Configuration, Season, and Distance from Contaminant Sources on the Degradation of Stream Water Quality in Urban Catchments

Round 1

Reviewer 1 Report

Paper has been revised adequately, hence it's now in an acceptable stage.

Author Response

Response to revision letter

Title: The role of landscape configuration, season and distance from contaminant sources in the degradation of stream water quality in urban catchments

 

 

The paper "The role of landscape configuration, season and distance from contaminant sources in the degradation of stream water quality in urban catchments" addresses  in a complex way the subject of the degradation of stream water quality in urban catchments. Very good documentation and adequate description of the research method, results clearly presented.

 

Many thanks for the revision. Along the paper have been applied minor English changes.

 

Author Response File: Author Response.docx

Reviewer 2 Report

The paper "The role of landscape configuration, season and distance from contaminant sources in the degradation of stream water quality in urban catchments" addresses  in a complex way the subject of the degradation of stream water quality in urban catchments. Very good documentation and adequate description of the research method, results clearly presented.

Author Response

Response to revision letter

Title: The role of landscape configuration, season and distance from contaminant sources in the degradation of stream water quality in urban catchments

 

 

The paper "The role of landscape configuration, season and distance from contaminant sources in the degradation of stream water quality in urban catchments" addresses  in a complex way the subject of the degradation of stream water quality in urban catchments. Very good documentation and adequate description of the research method, results clearly presented.

 

Many thanks for the revision. Along the paper have been applied minor English changes.

 

Author Response File: Author Response.docx

Reviewer 3 Report

Comments to the authors

 

This paper examines the effects of landscape configuration and season on stream water quality using PMS-PM. Specifically, they applied stratified analyses by distance form contaminant sources. The framework of this paper is quite reliable. However, this paper has lots of points to be improved.

 

The most critical shortcoming of this paper is sample size of each subsample model. As the authors suggested, the sample size is only twelve. It is very much smaller than frequently recommended minimum sample size, 200 to 400. Because, the models include seven measured variables, current sample size is extremely small. There are too much typos and grammatical errors. The authors should carefully check the full manuscript before submitting it.
- lines: 33, 125, 237, 348, and so on.. In general path model, the coefficient should be -1 to 1. And, the researcher have to doubt whether it is ‘Heywood case’ when the models shows smaller than -1 or larger than 1. In figure 4, three coefficient are larger/smaller than 1/-1. The authors should check Heywood case. And I think small sample size can be cause of this error. Conclusion section is not qualified. There are only three sentences and each of them belongs to single paragraph. And two of them do not have dot.

Author Response

Response to revision letter

Title: The role of landscape configuration, season and distance from contaminant sources in the degradation of stream water quality in urban catchments

 

This paper examines the effects of landscape configuration and season on stream water quality using PMS-PM. Specifically, they applied stratified analyses by distance form contaminant sources. The framework of this paper is quite reliable. However, this paper has lots of points to be improved.

The authors are very pleased to hear that the framework is quite reliable. We believe that this study design is quite innovative, not only to understand the role of the scale effect in the interplay among ecological integrity, pollution sources and land use metrics, but also because it is used PLS-PM in this thematic.

The most critical shortcoming of this paper is sample size of each subsample model. As the authors suggested, the sample size is only twelve. It is very much smaller than frequently recommended minimum sample size, 200 to 400. Because, the models include seven measured variables, current sample size is extremely small.

We fully agree that the sample size is small, and this is one limitations of this research. The present work was made in the scope of a research project where there were economic constraints, and that is the reason why it could not be measured IPTIN in more Ave River basin sites. The measurement of this bioindicator is quite expensive, since requires that big teams go to the field to do an extensive sampling and also to do further laboratory procedures to identify and count the macroinvertebrates.

When CB-SEM is used, the models require high sample sizes, some authors recommend at least 200 samples. The presented study resorted to PLS-PM (also called SEM-PLS) formative models, that do not require such high sample sizes, and normally, lower numbers of samples can be used if the models are simple, which is the case. Still, high sample sizes are always preferable, but unfortunately, economic restrains did not allowed us to have them. Anyhow, it was possible to achieve high R-squared values for a biggest part of the models (Figure 6), VIFs < 5 (can be checked in supplementary material) and statistical significance for the latent variable Land Use (figure 7). This shows that even with a small sample size it was possible to have reliable statistical models. In this study the models were used to explore cause-effect relationships. If it was intended to use the models for prediction purposes that would require a higher sample size. This because in prediction models it is important to have enough data to have statistical significance for all variables, and to use sub-datasets for calibration.

 

There are too much typos and grammatical errors. The authors should carefully check the full manuscript before submitting it.
- lines: 33, 125, 237, 348, and so on..

 

Many thanks for this review. The authors checked lines 33, 125, 237, 348 and other parts of the document and corrected the English. Changes were highlighted in yellow.

 

In general path model, the coefficient should be -1 to 1. And, the researcher have to doubt whether it is ‘Heywood case’ when the models shows smaller than -1 or larger than 1. In figure 4, three coefficient are larger/smaller than 1/-1. The authors should check Heywood case. And I think small sample size can be cause of this error.

 

Many thanks for this comment. It is rather pertinent and constructive.

The analysis of Heywood cases is a mandatory for CB-SEM models and other statistical methods such as Confirmatory Factor Analysis. When it comes to PLS-PM models it has the advantage to reach convergence in few iterations, and Heywood cases are not a problem of PLS-PM, while in CB-SEM Heywood cases can be a barrier for solution convergence. CB-SEM it is covariance based and uses maximum likelihood estimation whereas standard PLS-SEM uses ordinary least squares.  Anyhow, it was checked if any manuscript or technical document reflected on the   Heywood case impact for PLS-PM formative models, and it was found none.

 The presented study resorted to PLS-PM formative models. In such methods VIF are the most critical aspect that should be analyzed by researchers. In the present study the highest VIF value was 4.541, and normally it is suggested that VIF should be below 5, to do not promote an exceeded variance inflation of the estimates and in all the models this aspect was carefully analyzed by the authors.

We fully agree that normally weights and/or path coefficients varies from -1 to 1.  Still this it is possible that path coefficients and weights can be out of this range, but not very far, if such values are too high it is due to multicollinearity issues (high VIF values).  To analyze all the models and compare them, it was multiplied the weight of each measured variable for the correspondent path coefficient (as it is exampled in equation 6 for the example model). In such analysis the models equation take a linear regression form, and since all the data is normalized, beta coefficients (which for this case are the product between the weight and path coefficient) should be inside a range from -1 to 1. As figure 5 portrays, all the values are inside the desirable range, from -1 to 1.

 

 

 

Conclusion section is not qualified. There are only three sentences and each of them belongs to single paragraph. And two of them do not have dot.

 

The manuscript conclusion has been changed. Instead of being written in topic form, it has been changed to continuous text.

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report

I just checked the paper. But I already rejected this paper during my first review. Because I still think it is not qulified, I do not need to review it again. Of course, I fully read authors` responds, and I still think that the sample size issues cannot be adressed.

Author Response

Dear reviewer,

We agree that the sample size is a limitation, but that it does not hamper the achievement of our main goals that were to explore the influence of scale and season on the cause-effect relationships between anthropogenic pressures and water quality.

 

In this second round review we added, in the abstract and at the end of the discussion, a couple of texts recognizing the limitations of the present study, namely those related to the sample size. We also justified the importance of publishing this study despite the aforementioned limitations. Here we reproduce the fundamental text, incorporated at the end of the Discussion section.

 

"The most limiting factor in this study was probably the small number of sampling points used to assess the IPtI_N, just 12. However, these few samples allowed to provisionally expose the significant role (p < 0.05) of land use metrics for a satisfactory (R² » 0.85) explanation of water quality (IPtI_N) in the long range (> 4km from the contaminant sources). This result is noteworthy. It can (and was) argued that more samples would render the possibility to reveal the influence of other anthropogenic pressures, eventually hidden in this study by the sample’s coarse resolution. Nevertheless, it would not be a surprise if the results obtained in this study were replicated with a finer resolution, because contaminant emissions are subject to larger inter annual variations than are land uses and hence could eventually be inefficient in the studied period. A larger sample would probably capture fine-resolution effects, for example related to point-source contaminant emissions, but is not certain that would change the general outcomes and conclusions taken from this study. The main goals were achieved, which were to explore the influence of anthropogenic pressures on water quality as function of scale and season using a novel statistical method. The 26 PLS-PM models implemented in a predefined sequence were capable to identify the most important variables and distances from contaminant sources that controlled water quality in the Ave River Basin in the studied period. The model results may not be directly used in management initiatives without prior verification using a larger sample, but suggested how scale and season can affect the conclusions about cause-effect relationships involving anthropogenic pressures and water quality. In that context, the outcomes from this study provided interesting clues for managers of water quality at catchment scale, which are inherently an important scientific result."