The Key Impact on Water Quality of Coral Reefs in Kenting National Park

Round 1

Reviewer 1 Report

I enjoyed reading this manuscript and appreciated its clear and concise writing. My biggest concern with the work is that the underlying data are too limited for the statistical tests that were applied. The combination of a Principal Components Analysis (PCA), Cluster Analysis (CA), and Discriminant Analysis (DA) is well-supported for water quality research. However without sufficient samples, the value of the results are limited.  

Based on my understanding, this study relied on seasonal sampling over one year to conduct the PCA analysis. From Figure 2 and the Supplementary Material we see that five samples (one at each site) were collected each season for one year, for n = 20. A second dataset was used for CA and DA.  From Figure 3 it's clear that one sample was collected at each of 14 sites, for n = 14. 

The suitability of the data for a PCA needs to be tested. The Kaiser-Meyer-Olkin and Bartlett's test would be required, especially yin this case of such a low sample number. The results of those text need to be reported and addressed in the analytical approach. The absence of these, or any other adequacy tests, suggests that the dataset is inadequate for the analyses that were conducted. 

The manuscript needs to be revised so that the limitations of the datasets are made clear and the analytical approach adequately addresses these limitations. If the number of observations is insufficient for PCA, then an alternate approach is required. 

Author Response

The result of the Kaiser-Meyer-Olkin and Bartlett's test is as below, which indicating the analysis was acceptable.

 

Descriptive Statistics

	Mean	Std. Deviation	Analysis N
Sal	32.8401	2.56994	869
pH	8.2502	0.16725	869
DO	7.3657	1.29688	869
O2 Sat.	112.5098	19.88516	869
BOD5	1.2115	0.63111	869
NO3	0.0485	0.06704	869
NO2	0.0028	0.00878	869
PO4	0.006	0.0161	869
SiO2	0.2496	0.3593	869
NH3	0.0552	0.19141	869
Turb.	4.7371	13.30448	869
Chl.a	0.2499	0.89781	869
SS	14.5801	28.99512	869

 

 

KMO and Bartlett's Test

Kaiser-Meyer-Olkin Measure of Sampling Adequacy.		0.648
Bartlett's Test of Sphericity	Approx. Chi-Square	6588.574
	df	78
	Sig.	0

 

 

Reviewer 2 Report

An interesting short paper incorporating a range of pertinent statistical analyses and parameter correlations consistent with similar studies of seasonal nutrient loading to nearshore marine waters and of rainfall induced runoff. Specific comments/inquiries are as follows.

Lines 23-24:  This sentence seems to indicate the drop in salinity, rather than the freshwater inputs, was the main impact to coral reef ecosystems. Is this what you meant?

Lines 82-83: Did you conduct any analysis of the bacteria to assess their probable source (e.g., human wastes, soil microbes, animal wastes)?

Lines 88-92: What time of day were these samples collected? Whereas algae’s photosynthesis will consume CO2 and raise the pH during the day, respiration during the night should have the opposite effect. It appears that BOD generally correlated well with other indicators of reducing redox conditions, such as the presence of ammonia, except at the Kenting station. Is there something other than photosynthesis that could result in DO supersaturation there?

Lines 99-100: Are you suggesting that the negative relationship between silica and salinity was a result of rainfall directly on the seawater or of runoff from the land induced by rainfall? Silica is normally attributed to terrestrial sources rather than to the precipitation itself.

Lines 125-126: While you refer to high nutrient loads near discharge areas, did you actually calculate the mass introduced to the various sample locations or are you referring to the concentration of nutrients in those locations?

Lines 190-191: How did you surmise that the nutrients were mainly from household wastes and not from other potential sources?

Author Response

 

Com: Lines 23-24:  This sentence seems to indicate the drop in salinity, rather than the freshwater inputs, was the main impact to coral reef ecosystems. Is this what you meant?

Res: We think we didn’t make the sentence clear. So, we change the sentence to “we found that increasing in the nutrient levels during the summer rainy season, together with the drops in salinity led by freshwater inputs (land- & rainfall-derived), was the main impact to coral reef ecosystem of Kenting.

Com: Lines 82-83: Did you conduct any analysis of the bacteria to assess their probable source (e.g., human wastes, soil microbes, animal wastes)?

Res: Yes, we did conduct total coliform bacteria counts on the water samples in several years. However, long term data were not collected. So we changed the sentence into the followings. “In Table 1, total counts of coliform bacterial (TCB) showed no correlation with any of water parameters measured in this survey, which seemed to discount the impact of sewage. However, most of the seasons displayed high counts of coliform bacteria (median values = 4.88×10³ colony-forming units (cfu)/100 mL, n = 20) in the water samples collected from the streams and outlet station. Due to lack of long term data on TCB, more information is required to explain this phenomenon.”

Com: Lines 88-92: What time of day were these samples collected? Whereas algae’s photosynthesis will consume CO2 and raise the pH during the day, respiration during the night should have the opposite effect. It appears that BOD generally correlated well with other indicators of reducing redox conditions, such as the presence of ammonia, except at the Kenting station. Is there something other than photosynthesis that could result in DO super-saturation there?

Res: The water sampling was conducted during day time referring to the results of real time monitoring (Tew et al. 2014). To explain super-saturation of DO, the sentence was changed to “this could be evidence for high concentrations of phytoplankton and/or dense of coral population (which would provide oxygen at high concentrations via photosynthesis).”

Com: Lines 99-100: Are you suggesting that the negative relationship between silica and salinity was a result of rainfall directly on the seawater or of runoff from the land induced by rainfall? Silica is normally attributed to terrestrial sources rather than to the precipitation itself.

Res: Yes, silica is normally derived from land sources. The highly negative correlation between silica and salinity indicated that the significant amount of water from rainfall flushed the silica in the unprotected land into the sea and the water further reduced the salinity.

Com: Lines 125-126: While you refer to high nutrient loads near discharge areas, did you actually calculate the mass introduced to the various sample locations or are you referring to the concentration of nutrients in those locations?

Res: The data were indicating the concentration of these locations. Detail nutrient load was measured in reference [3] (Meng et al. (2008)).

Com: Lines 190-191: How did you surmise that the nutrients were mainly from household wastes and not from other potential sources?  

Res: Actually, there were two nutrient sources in Kending reef area, upwelling and terrestrial waste water. If the nutrient source came from upwelling, there would be a negative correlation between temperature and nutrient. If the source came terrestrial, the increase in nutrient concentration would be negatively correlated with salinity. Our finding suggested a high negative correlation between nutrient concentration and salinity, so we suggested the impact from terrestrial would be the major one.

 

Round 2

Reviewer 1 Report

The size of the dataset that these analyses are based on is confusing. Figure 2 and the Supplemental Material suggest that for PCA, n = 20. It is reassuring to see that the dataset passes the Kaiser-Meyer-Olkin and Bartlett’s test, but it’s surprising that N for KMO is 869. This conflicts with those sources. Are those other sources subsets of a larger dataset? Was KMO based on a bootstrap analysis? Either way, the manuscript is not clear. It needs to be.

For comparison, the studies that this manuscript was inspired by, as listed at line 51, relied on a lot of observations. Olsen et al. (2012) relied on a minimum of 340 observations of a single parameter. Zeilhofer et al. (2006), Shrestha and Kazama (2007), and Muangthong and Shrestha (2015), all relied on a minimum of closer to 1000 observations of a single parameter. 

Currently, this manuscript suggests that this study relied on only 20. The manuscript must address this shortcoming. This can be done in the Introduction, by including other studies with small sample size; in the Methods, by providing the number of samples used, the KMO results, and supporting techniques; and in the Results and Discussion.  

Author Response

Thank you for the comments. I think there were some unclear description in our manuscript. The data for Fig. 2 were derived from the example data set with N=20 as shown in the supplement. The data for PCA analysis were derived from a long term data set with N=869, which is uploaded in this reversion. So, the sample numbers for PCA analysis were resonable. 

Author Response File: Author Response.pdf