Distribution Characteristics of Dissolved Oxygen in Spring in the Northern Coastal Beibu Gulf, a Typical Subtropical Bay

Round 1

Reviewer 1 Report

This manuscript describes dissolved oxygen and other sensor data processing study in which the authors attempt to describe how environmental conditions dictate observations. The data are represented spatially within a bay system impacted by natural and anthropogenic processes. The various sub-environments are described (particularly within the discussion section) and observations are explained in light of specific sub-environment characteristics. The authors use “fuzzy clustering” to “bin” observations into 4 clades and attempt to explain how these clades fit within the sub-environments.

The manuscript could offer useful information to tie environmental observations to environmental conditions – particularly if the “binning” was done with rigor. The likely valid argument is that sensor data could provide a snapshot into the environmental condition.

While the conclusions are likely valid, it’s difficult to “know” they are because the methodology is somewhat non-routine and in some critical ways is not sufficiently described to ensure rigor. From the materials and methods, it appears the authors took relatively high-resolution temporal data and averaged it by month, then attempted to correlate the measurements. This was done in Matlab. Matlab is made for working with large datasets. What I cannot understand is why the authors averaged data, then attempted correlation? Wouldn’t any correlation be more valid (and more robust in terms of the output statistics) with the entire data set? I can’t fathom the need to average? Next, the authors used “fuzzy cluster analysis” to ordinate and cluster data. Are we to presume this is part of the fuzzy logic toolbox? The methods section should be very explicit how this was done. I have used clustering in Matlab (statistics toolbox) and there are many options which can be confusing. If another investigator were to try and reproduce the results obtained in this MS, he/she would need this information. It’s unclear what clustering algorithm(s) was/were used. It appears the data were centered (z-score standardization) – but this should be explained better. It’s unclear if the standardization was a centering (z-score?) or by using ordination distances (Euclidean distance). Z-scores may be useful in understanding a measurement within context of the mean (where it falls in a distribution), but if the data were averaged, where do individual z-scores fall? Was the averaging a way to generate z-scores for each time point? I suggest the authors explain exactly how calculations were made and be very specific with the reasoning for averaging a rich dataset before exploring statistical or ordination relationships.

Aside from the methodological issues, it is difficult to judge this MS in publishable form. It was very difficult to read. Based on the authors’ affiliations, I assume they are not native English speakers. I always feel hypocritical – as there is no way I could write a MS in a foreign language. However, “Water” states that articles are published in English. I am including a pencil-marked first page with the abstract. It is very difficult to read and has grammatical errors. The first sentence is lines long and the subject-verb-object is lost throughout. I suggest the authors go through each sentence, identify the subject, then match the verb, object and any modifiers. Make the sentences clear, short and concise, e.g. “Oxygen was measured over the course of one month.” No confusion. Not only is clear English the journal’s requirement, but wouldn’t any author want their paper to be easy to read? Everyone is time-limited. Why would anyone bother if a paper is tedious to read and the reader has to go back and forth to figure out what’s trying to be conveyed? I was generally able to figure out what the authors were trying to say, so a careful proofing (there are services for non-English speakers) shouldn’t be too difficult to accomplish.

I started to work on line-by-line, but as stated above, this paper needs some work before being fully evaluated. I am recommended to reject as is – but encourage a re-submission with the issues fixed. I suspect the results are compelling – probably not novel – as we know certain water quality parameters tend to co-correlate, but it’s just not possible to be sure based on the current description.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The manuscript devotes the actual environmental problem. Dissolved oxygen in coastal waters decreased globally in the last few decades, mainly because of excessive human-induced input of nutrients and organic matter from river runoff.

Authors showed that DO level tended to increase at offshore sites in the study area. In the inner bay and estuaries DO is lower mainly due to decomposition of land-derived organic materials. DO concentrations in the study area were also influenced by phytoplankton primary production, coastal mangrove ecosystems and by aquaculture and offshore industries.

 

There are shortcomings that should be corrected before publication.

 

General comments

Text is not properly written that makes the manuscript hard to follow. English correction is recommended.

 

I am also interested to know more about correlations between all indexes. I suggest that matrice of correlation coefficients would be useful here. Next, I suggest authors to think about multiple regression analyses with DO as dependent factor if your data would meet the multiple regression criteria.

 

Specific comments:

 

L. 184-187 and Fig. 3b. DO saturation concentrations are not findings of authors. I suggest that this text with Fig. 3b can be deleted

 

L. 192-194. It was already mentioned in Section 2.3. analytical method.

 

Figure 4. I suggest a map, where these four groups of stations would be denoted. Currently, it is difficult to follow.

 

Table 1. Table is not mentioned in the text of manuscript.

I suggest to add line with combined data from all sites. I also suggest to show significance levels in Table 1 according to general rules as * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001.

 

L. 226. “surface” It is first mention (!) in manuscript. It should be clearly indicated in Methods that you deal with surface oxygen level

 

L. 229. “Class I seawater (DO>6mg/L)” Reference should be here

 

L. 284-285. Oversaturation by oxygen can also be result of active hydrodynamic

 

L. 310. Suggest ‘likely, probably’ etc. instead of ‘mainly’.

L. 366. ‘mainly’ or ‘likely’?

 

L. 397. “eutrophication E value” What is it? Please, explain.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Re- Review of “Distribution characteristics of dissolved oxygen….” By Zheng, Zhao, Shi and Duan

I received this revised manuscript for comments. The authors have provided responses to all my comments. Some of the grammatical errors and awkward sentence construction have been addressed. Two long and difficult sentences are used in the abstract to describe DO saturation and DO concentration. Is it really necessary to describe both (saturation encapsulates the physical water properties). Anyway, the authors have attempted to make the writing better. Perhaps the editor(s) will fix what remains. I find it still very difficult and tedious to read.

One of my main issues with the data processing was using averaged sensor data (monthly average). As I originally stated, it seems unproductive to take a high resolution dataset and remove all the short-term variability before analyzing, correlating and ordinating. The authors provided several references showing other investigators doing similar time-averaging. I guess it’s valid. I still don’t understand why they would want to take an interesting dataset and make it uninteresting….. DO saturation correlates with chl-A. We’ve known that for over 100 years. I understand that high-resolution sensor data may have flaws (bubble over an optical sensor). There are ways to remove outliers (>< 2 StDev from mean). Diurnal variation between chl-A and DO alone would be interesting. It doesn’t make sense to me, but if there are published papers using the same techniques, so be it. I leave it up to the editor(s). I know there is tremendous pressure to publish and generate numbers… If I were the authors, I wouldn’t want a paper that had few citations just to up my numbers.

The algorithms were described in a supplemental document and better described in the text. Seems OK.

As you can tell, I am tepid as best on this manuscript. I see it as probably valid and shows mostly what we’ve known for many years (perhaps for a “new” location). I think if the English were improved and there was something “interesting” pulled out of the high-resolution data it would be far improved and be a more compelling addition to the literature. I marked it largely as average. I will leave it up to the editor to decide on the ultimate fate. 

Author Response

请参阅附件。

Author Response File: Author Response.docx