Verification of an Environmental Impact Assessment Using a Multivariate Statistical Model

Round 1

Reviewer 1 Report

Methods for nutrients determination are now well explained, statistical analysis results are better presented. I have no further comments/suggestions.

Author Response

Methods for nutrients determination are now well explained, statistical analysis results are better presented. I have no further comments/suggestions.

Ans: Thank you very much for the suggestions in the earlier version of the manuscript.

Reviewer 2 Report

The Authors did agree with remarks concerning separate points in the manuscript, and introduced suggested changes, however they did not look more generally upon their work, and seem not to have recognised the need for shifting stresses and rebuilding the discussion.

As it is now, the paper tries to describe the influence of river input upon the plankton community, simultaneously showing almost negligible impact of the outflow from the NMMBA. This, however, may not be the case, especially taking into consideration what the Authors wrote in their reply to my previous review. It appears that there is much variation in the volume of both the water input from the river(s) and the outflow from the NMMBA. If the ratio of water volume from the two sources is not close to constant in time, one can assume that at periods of low water level in the river(s), the wastewater from the NMMBA could exert significant impact. Presumably, data regarding river flow and the amount of wastewater are to be found, so the Authors could incorporate them, and discuss, into their paper.

In their statistical analyses the Authors adopted an unorthodox approach using the factor loadings instead of actual values of environmental factors. Such move requires an explanation or a reference to a source containing such explanation.

The final structural equation model (Fig. 4.) seems to have lost some important links present in the conceptual model from the Fig. 3. To name one example: according to Fig. 4, river input influences phytoplankton only via suspended matter, and zooplankton only via phytoplankton. At the same time river input affects salinity, and salinity has impact on both phyto- and zooplankton diversity and density.

Author Response

The Authors did agree with remarks concerning separate points in the manuscript, and introduced suggested changes, however they did not look more generally upon their work, and seem not to have recognised the need for shifting stresses and rebuilding the discussion.

As it is now, the paper tries to describe the influence of river input upon the plankton community, simultaneously showing almost negligible impact of the outflow from the NMMBA. This, however, may not be the case, especially taking into consideration what the Authors wrote in their reply to my previous review. It appears that there is much variation in the volume of both the water input from the river(s) and the outflow from the NMMBA. If the ratio of water volume from the two sources is not close to constant in time, one can assume that at periods of low water level in the river(s), the wastewater from the NMMBA could exert significant impact. Presumably, data regarding river flow and the amount of wastewater are to be found, so the Authors could incorporate them, and discuss, into their paper.

A: The average flow from Sichong River is 341 million cubic metre, 92% (314 mcm) during rainy season vs 27 mcm in dry season (https://www.wra07.gov.tw/en/). No flow data for the smaller Baoli River, where 100% flow volume occur during rainy season. If NMMBA exert significant impact, we would see significant difference at S10 during dry season (Q3 & Q4, especially Q4), which did not happen.

 

In their statistical analyses the Authors adopted an unorthodox approach using the factor loadings instead of actual values of environmental factors. Such move requires an explanation or a reference to a source containing such explanation.

Ans: This is not a method commonly used but can still be found in some manuscripts, eg. Conant (1988, BioScience 38(4): 254-257), Wakeling & Rozitis (2004; J Exp Biol 207 (14): 2519–2528), Sarnelle (2005, J Plank Res 27(12): 1229-1238) are a few manuscripts that used a similar analysis. Source of explanation could be found in statistiXL (https://www.statistixl.com/features/principal-components/):

Principal Component Analysis is a technique that attempts to reduce complex data sets consisting of many different variables to a smaller set of new variables that still manage to describe much of the variation in the original data. These new variables, called Principal Components, are chosen to be independent (i.e. the new variables are not correlated whereas the original, untransformed variables may have been correlated) and to maximise the variance found in the original data set. The more significant PCAs are selected based on their eigenvalues, and hopefully far fewer PCA variables (e.g. one or two) are required than there were original variables. These fewer Principal Components can then be further analysed by Regression Analysis or ANOVA/MANOVA. Thus, the role of Principal Component Analysis has been to reduce a large number of variables into fewer, simpler ones.

 

The final structural equation model (Fig. 4.) seems to have lost some important links present in the conceptual model from the Fig. 3. To name one example: according to Fig. 4, river input influences phytoplankton only via suspended matter, and zooplankton only via phytoplankton. At the same time river input affects salinity, and salinity has impact on both phyto- and zooplankton diversity and density.

Ans: Fig. 3. is a conceptual model where we assume all the possible abiotic factors will affect the phytoplankton and zooplankton. However, after we carry out SEM analysis by putting all the abiotic factors, as well as phytoplankton and zooplankton data, the final results tells us that only these factors are significantly related (Fig. 4). That’s why some links were not in Fig. 4.

Author Response File: Author Response.pdf