Investigation into the Cause of Iron-Related Clogging of Groundwater Bores Used for Viticulture in the Limestone Coast, South Australia

Reviewer comments

Authors response

Reviewer 1

1) I find it quite problematic that the authors do not provide the full dataset of the article to the review. It is quite a small dataset, that is easy to provide under an excel file. While I understand the possible fear of seeing these data stolen, it prevents the reviews to be done in a complete way. In particular since all discussed information are not presented in the results. I’ll show this below in the detailed comments.

The data presented in this ms was obtained from bores located on properties of wine growers. As these were commercial enterprises, the data itself was deemed confidential. However, summary statistics as well as detailed graphing of the data was possible and has been used within the ms to show the data and its interpretation in detail.

2) A number of references are missing (22 to 26) in the reference list. In several places in the article, more references should also be provided to support the authors statements.

We agree with the reviewer and have provided added references to support statements and have added any missing references.

3) The lack of multivariate analysis is detrimental to the study. In general, no statistical work is provided, which is quite disappointing for such a dataset. Plotting all chemical data into a PCA would help the readers (and authors) understand which parameters are varying together, and which one are independent from another. ANOSIM/ANOVA tests are also recommended to answer questions statistically, such as : is there an existing correlation between X and the clogging status of the bore. Unless it is provided, some statements will lack scientific support.

Authors agree that statistical analysis would be valuable; however, given the small sample size of this pilot study we have avoided statistical interpretation of the data to prevent potential over-interpretation. Our qualitative interpretation of the data identified a number of potential correlations between strongly redox controlled aqueous species such as Fe, Mn, As, DO, NO₃ as well as saturation indices of strongly redox controlled mineral phases such as iron-oxides. This provide a solid basis for future research. This has been identified as a limitation of the study.

Text now reads:

A larger sample size is also needed to explore the statistical significance of observed correlations and relationships.

4) In relation with the previous comment, plotting data against Fe or Mn concentrations is sometimes helpful. Here, it is done quite regularly and affects the interpretation in several instances since Fe and Mn show a rather clear separation between clogged and unclogged bores. It drives the reader into a subjective approach, and in many times does not make sense scientifically. Additionally, it also induces the author in potentially wrong interpretations (i.e. correlation between DPOC and clogging).

We thank the reviewer for the comment. Several aspects are mentioned here, such as plotting data against Mn and Fe and correlations with DPOC. Detailed responses to these queries are provided in response nr 17, 18,19, 20, 21, 22, 23, and 37 below.

5) A more precise statement on the conflicts of interest is needed in my point of view.

Authors can confirm the conflict of interest statement is correct.

6) I am convinced this study is of major interest and is done in a sound way, indeed providing sufficient information to provide the conclusions presented by the authors. I hope the reviews provided here will help enhance the manuscript. I remain available to the authors if needed for clarification

The authors thank the reviewer for their considered comments

7) Line 58 : can the authors insert a reference about their presence in groundwater please?

Change accepted, reference added

8) Line 63: iron sulphides

Change made.

9) Line 64 : also consider reference involving iron bacteria or pipes clogged by EPS (e.g Fleming & Rowe 1999 and 2004)

Change accepted, references added.

10) Line 80-82: can you provide a reference to these saying? I know it is quite hard when it’s mostly heard through the grapevine… but local newspapers or interviews on that topic may be available?

Change accepted, reference to irrigator survey report added.

11) Line 87-91: jumping from correlation to causation is risky, and needs to be at least supported by statistical tests.

Change accepted, now reads:

Correlations between the severity of clogging and the age and depth of bores, the micro-biology and the inorganic chemistry including redox conditions in these groundwaters was able to deduce the most likely cause of the clogging problem. Hypotheses were able to be drawn as to why the problem appears to be becoming more widespread.

12) Line 98-102: references 22 to 27 are missing in the reference list.

Change accepted, references added to the reference list.

13) Figure 1 : what is drawn in blue? Lakes? I am not sure since they are sometimes crossed by rivers. Additionally, since accurate locations cannot be disclosed right here, could it be possible to add a line or two somewhere (at the end of the material and methods for example) stating the support of the authors for sharing data and locations for the sake of reproducibility of the study? Can a “location identifier” be added to the data set. For example zone A for such and such sample, zone B for those ones. In this way, it would be possible to assess local effects even without an absolute location.

We agree with the reviewer that location identifiers would allow “local effects” to be evaluated. This in fact, formed part of the work, i.e. to identify correlations with e.g. bore location, depth of bore, age of bore etc. No correlation could be found based on bore location, while bore depth and age of bore showed some correlation with the clogging status of the bore. The latter is discussed in the manuscript. Besides the lack of a clear correlation between the location of a bore and its clogging status, the participating wine growers asked for locations and bore identifiers to be kept confidential. We therefore are not disclosing identifiers within the ms, but we are confident, that the correlations between clogging and e.g. redox chemistry; bore depth; bore age provide a solid interpretation of the data, and do not suffer from the constraint of having to keep locations confidential. We therefore have not added identifiers within this manuscript.

14) Line 149 : the method described here does not correspond to “whole genome sequencing” which aims at reconstructing the complete set of genes in an organism. Here, the authors are carrying genomic analysis : one single gene (16S rRNA, targeting exclusively Bacteria) is sequenced, and no genome reconstruction is carried out. Please also refrain from using the term “metagenomics” which generally applies to the study of all genes found in a given environment. Although Thermofischer names its material “metagenomics Kit”, they only detect one type of gene : 16S rRNA. The use of this term is therefore misleading to readers used to molecular microbiology techniques.

Change accepted, this has been removed from the manuscript.

15) Line 157-161 : can you provide more information in this paragraph : type of SEM (normal or environmental), how was the sample dehydrated (if it was, as this will affect the EPS structure and texture under SEM), what was the energy and type of coating used? Also, please provide the sample/location name for the observed sample, so that it can be linked to chemical measurements too.

Change accepted, now reads:

A single sample of the clogging material was obtained from an affected screen retrieved from a bore located on one of the properties included in the study. Samples were fixed in 4% glutaraldehyde and dehydrated by immersion in sequential ethanol solutions for 10 min per rinse starting at 1 × 70% v/v, followed by 1 × 90% v/v, 1 × 95% v/v and 2 × 100% v/v, with a final rinse in hexamethyl-disilazane (HMDS) for 30 min. Samples were then air dried and sputter coated with 3 nm of platinum prior to analysis on Inspect FEI F50 Scanning Electron Microscope.

16) Line 167-169: Several comments here: 1. How were the SI measured? please provide the equations and conditions used (pH, temperature).

2. While I understand why the SI of calcite is calculated, why dolomite, gypsum and quartz? Do the authors expect these minerals to precipitate in the system? Were they identified somewhere? 3. Saturation indexes are a function of the concentration of ions involved in the formation of the mineral. What are the authors trying to show by plotting SI vs Si, or sulfate for quartz and gypsum? I personally would be interested in observing Si and sulfate concentrations if they are relevant to the study. But I’m not sure they are in this context. Furthermore, there is no use of these data in the discussion.

Many thanks for this comment and we have now added the method of obtaining the SI data within the text. The SI data is calculated on the basis of the activity of species, pH, water temperature and is calculated using the program PHREEQC (Parkhurst and Appelo, 1999). PHREEQC is the most widely used thermodynamic geochemical simulator for quantifying chemical reactions and transport processes in aquatic systems. The program is based on equilibrium chemistry of aqueous solutions interacting with minerals, gases, solid solutions, exchangers, and sorption surfaces, but also includes the capability to model kinetic reactions with user defined rate equations. PHREEQE calculates concentrations of elements, molalities and activities of aqueous species, pH, pe, saturation indices, and mole transfers of phases to achieve equilibrium as a function of specified reversible and irreversible geochemical reactions. PHREEQC calculates the mineral saturation states based on Ω≡?/?_SP, where K_SP is the equilibrium constant of the mineral dissolution reaction, and Q is the corresponding reaction quotient of this reaction, calculated from the activities of all species involved in the dissolution reaction. At equilibrium, ?=?_SP so that Ω=1. In PHREEQC, the saturation index (SI), defined as SI=log10Ω is used, so that at equilibrium SI=0. We have added the following clarification in chapter 2.2.:

 “Aqueous speciation and saturation state of groundwaters in respect to common mineral phases were computed using the PHREEQC code, utilizing the PHREEQC-2 thermodynamic database (Parkhurst and Appelo 1999). “

We have added the saturation indices of four most common aquifer phases to show and characterise the overall water chemistry of the sample set in relation to the aquifer geology. Figure 2 shows that water samples are generally oversaturated in respect to Calcite, Dolomite and Quartz and reflect the major mineral phases of the Limestone aquifer, while gypsum is absent with all waters being undersaturated in respect to gypsum. It now reads in the Results section:

“All groundwaters were well buffered in a pH range of 7.4 to 8.0 with the dissolution of carbonate minerals providing an effective pH buffering mechanism, due to fast dissolution kinetics (e.g. Plummer et. Al.; 1979; Vandenbohede et al. 2013). The groundwater was further oversaturated in respect to quartz, a further major mineral phase within the target aquifer, but was undersaturated with respect to gypsum, despite some groundwater samples showing elevated S0₄^2- concentrations (Figure 2). “

17) Line 184: can the authors provide a figure supporting the “clear correlation between affected bores and redox potential? A multivariate analysis showing all redox and non-redox sensitive elements and color-coded bores would be perfect. An additional ANOSIM test would likely provide statistical support to this statement. I agree with the overall trend but as it is the main finding of the article, it should be statistically supported in my opinion.

Authors agree that statistical analysis would be valuable; however, given the small sample size of this pilot study we have avoided statistical interpretation of the data to prevent potential over-interpretation. Our qualitative interpretation of the data identified a number of potential correlations between strongly redox controlled aqueous species such as Fe, Mn, As, DO, NO₃ as well as saturation indices of strongly redox controlled mineral phases such as iron-oxides. This provide a solid basis for future research. This has been identified as a limitation of the study.

Text now reads:

A larger sample size is also needed to explore the statistical significance of observed correlations and relationships.

18) Line 194: I disagree with this statement. Four affected samples are < 1 mg/L. Furthermore, the readers cannot observe anywhere in the manuscript NPOC vs severity of clogging (only clogging vs non-clogging).

Change accepted, text removed.

19) I also think the x-axis is misleading in this situation as it artificially discriminates affected vs non-affected sites. Same comment for the NO3 vs Fe plot.

The plots shown in Figure 3 serve two purposes: they demonstrate that i) groundwaters with elevated iron concentrations are reducing waters (no DO, no NO₃, high Mn), and additionally they show that ii) reducing waters are predominantly bores affected by clogging. We therefore would prefer to keep this figure within the text.

20) Finally, please define NPOC somewhere in the manuscript.

Change accepted, NPOC defined

21) Fig. 4 : the As vs Fe plot is not needed in my opinion. Please correct Do to DO in the figure. Same comment regarding the use of Fe in the x-axis.

Many thanks for pointing out the spelling of DO in Figure 4. This has now been changed in the revised manuscript.

Correlations between As and Fe are common due to iron hydroxides commonly providing surface binding sites for arsenic (e.g. Wallis et al. (2011); Wallis and Pichler 2018; Wallis et al, 2020). In reducing waters, iron hydroxides are unstable and their dissolution commonly leads to elevated Fe concentrations, but also often to elevated trace metal concentrations such as As, and this is shown in Figure 4 and postulated in the ms as a possible mechanisms for the elevated arsenic concentrations observed in reducing waters in the study area. We therefore prefer to keep this Figure in the ms but we fully agree with the reviewer that the interpretation of the graph may not have been articulated strongly enough in the ms. As such we have added the following under chapter 3.1:

“Elevated arsenic concentrations were found in reducing groundwaters, with some bores exhibiting concentrations above the drinking water guidelines (e.g. As > 10 µg/L). Under reducing conditions, iron hydroxides are unstable. Their dissolution is commonly associated with a release of trace metals such as arsenic due to the corresponding loss of surface binding sites (e.g. Smedley and Kinniburgh, 2002; Wallis and Pichler, 2018) which could explain the correlation of elevated As and Fe shown in Figure 4. “

22) Fig 5 : Using both Fe and Mn in the x-axis is redundant and does not provide additional information. A plot of SI FeoH vs SI siderite may provide a stronger message here. Please explain the parameters used for SI calculations.

Change accepted and figures updated.

We have now added a method description of how the SI is calculated under chapter 2.2. It now reads: “Aqueous speciation and saturation state of groundwaters in respect to common mineral phases were computed using the PHREEQC code, utilizing the PHREEQC-2 thermodynamic database (Parkhurst and Appelo, 1999).”

We have also added this information to the caption of Figure 5.

Mn was plotted in order to show whether groundwaters may also be oversaturated in regards to manganese oxides. Iron and manganese oxide mineralization is the most common cause of clogging encountered in irrigation systems fed by groundwaters (e.g. Burte et al. 2019). We therefore also determine the potential for Mn minerals to add to the clogging problem in the study area by establishing whether groundwaters were oversaturated in respect to common Mn minerals. Based on the reviewer comment, we have added the following text in the ms to clarify this:  

“Besides iron mineralization, oxidation of manganese is often implicated in clogging of bores and irrigation systems (e.g. Burte et al. 2019). However, despite elevated Mn concentrations, reducing waters remained below saturation of manganese carbonates (MnCO₃) as well as manganese oxides (Figure 5).”

23) Fig 6: As mentioned above, plotting both depth vs Fe and Mn is not very informative. You may used Age vs depth instead Fe or Mn, to show that mostly deep new bores have high Fe.

Change accepted. We agree with the reviewer and have removed the depth versus Mn plot.

24) Line 233 : as discussed earlier, this is not whole genome sequence data. Please change to “16S rRNA gene sequence analysis” here and everywhere else. I have not been able to find these data, can they be provided somewhere. They cannot be discussed as such. Please also deposit the sequences into an open repository (such as NCBI).

Change accepted. As this analysis was outsourced the raw sequence data was not provided. As such this has been removed from the manuscript.

25) Line 235-241: no information is provided in the manuscript to allow discussion on this topic. Can you provide such data (samples vs environmental parameters + assay response + IRB occurrence)

Additional supplementary data has now been included. However, we cannot provide all the raw data as we want to ensure that we maintain confidentiality and that individual sites cannot be identified.

26) Line 245-248: First, I am not sure the coliform and e. coli information is useful to the story. If so, the authors should try to integrate it better in the manuscript. Second, if bacterial samples were taken before well purging, this needs to be described in the material and method. As such, we are under the impression that all samples were taken once the well was purged.

Change accepted. Text updated and now reads:

The presence of coliforms and E. coli indicate potential faecal contamination and are the most commonly used indicator organisms for microbial water quality [35].

27) Line 255: very diverse : please be more specific. Diverse in structure, minerals, chemistry organic content?

Change accepted. Now reads:

The precipitate was found to be very diverse with both crystalline structures, suggesting amorphous iron oxides, and biological structures, suggesting bacterial growth. This suggests the co-occurrence of inorganic as well as microbiological clogging.

28) Line 256: iron oxides : how can you tell it is an iron oxide? Have you performed EDX measurements? If so, can you provide the data?

Change accepted. Now reads:

The precipitate was found to be very diverse with both crystalline structures, suggesting amorphous iron oxides, and biological structures, suggesting bacterial growth. This suggests the co-occurrence of inorganic as well as microbiological clogging.

29) Line 256 : bacterial growth : how can you interpret bacterial growth from one picture? It looks like iron sheaths but quite hard to tell based on this picture. Do you have a better quality one? Can you label it maybe? Understanding how the sample was prepared would facilitate the readers understanding.

Change accepted. Caption now reads

Figure 8: SEM of clogging material: a: crystalline structure suggesting amorpheous iron oxides; b: blue circles show the evidence of biological structure suggesting microbial growth; c: incrustations of a clogged screen, which was sampled for SEM analysis.

Change accepted, text now reads:

A single sample of the clogging material was obtained from an affected screen retrieved from a bore located on one of the properties included in the study. Samples were fixed in 4% glutaraldehyde and dehydrated by immersion in sequential ethanol solutions for 10 min per rinse starting at 1 × 70% v/v, followed by 1 × 90% v/v, 1 × 95% v/v and 2 × 100% v/v, with a final rinse in hexamethyl-disilazane (HMDS) for 30 min. Samples were then air dried and sputter coated with 3 nm of platinum prior to analysis on Inspect FEI F50 Scanning Electron Microscope.

30) Line 266 : I am not sure I understand what the authors refer to with “contrary to the above”.

Change accepted, text removed.

31) Line 267-269 : please provide data showing IRB vs environmental parameters to support this statement.

Additional supplementary data has now been included. However, we cannot provide all the raw data as we want to ensure that we maintain confidentiality and that individual sites cannot be identified.

32) Line 271-272 : As mentioned earlier, the opposite is truer: samples without clogging have a tendency to have elevated organic carbon. I think this statement is erroneous based on the available data.

Change accepted, text removed.

33) Line 275: please specify which nutrients you are referring to.

Change accepted, text removed.

34) Line 276: I would not generalize the use of SEM findings, as only one sample seems to have been investigated.

Change accepted, now reads:

The aqueous chemistry and microbiology (which is supported by the SEM analysis of one sample), suggest that the widespread occurrence of reducing groundwaters with elevated iron concentrations is the primary cause of iron clogging, however, biological clogging may contribute to the problem in the area.

35) Line 278: Same as above, with only one sample observed, you cannot really say that biological clogging contributes to the problem in the area.

Change accepted, now reads:

The aqueous chemistry and microbiology (which is supported by the SEM analysis of one sample), suggest that the widespread occurrence of reducing groundwaters with elevated iron concentrations is the primary cause of iron clogging, however, biological clogging may contribute to the problem in the area.

36) Fig 9 and line 289 : can you provide more information as to where these data were collected from?

Change accepted.

Figure 9: Details of data source added to figure.

Line 289: Reference added for ‘increased viticulture and irrigation demand’.

37) Line 292 : a figure DO vs depth would help supporting this statement.

We agree with the reviewer and we have plotted Depth versus Fe (Figure 6) and DO versus Fe (Figure 3), to show that tendentially declines in DO and stronger reducing waters become more frequent with depth.

38) Line 296: Just out of curiosity: do people know where the clogging occurs? Emitters? Screens? Everywhere?

Clogging occurs in bore screens, pumps and dripper systems. This is described in paragraph 5 of the Introduction and has been added as a clarification to abstract.

39) Line 329-330 : a more precise conflict of interest statement should be provided. In particular, the readers should be able to know if the authors affiliations are involved with drilling activities or authorizations, water and groundwater treatments or decontamination, or any other kind of potential intervention on the farming environment.

Authors can confirm the conflict of interest statement is correct.

This manuscript examines the iron related bacteria and inorganic water chemistry of bore water in order to evaluate the cause of clogging in two different regions within the Limestone Coast region in the south east of South Australia. Certainly, the research topic is very interesting because the importance of microbial as opposed to chemical iron oxidation is not fully established. The authors provide a thorough experimental investigation and data analysis. The figures are clear and captions descriptive. The references are thorough and appropriate. The manuscript is written very well. Certainly, the results are important, and deserve to be published. Consequently, I recommend that the manuscript is accepted for publication in Waterin its present form.

Authors thank reviewers for their considered comments.

What is the sequencing evidence or further proof for the microbiome detected here?

Change accepted. As this analysis was outsourced the raw sequence data was not provided. As such this has been removed from the manuscript.

Can authors present more electron microscope images of studied samples that are affected and show the changes that occurred (i.e the occurrence of the biofilm).

Change accepted, now reads:

A single sample of the clogging material was obtained from an affected screen retrieved from a bore located on one of the properties included in the study. Samples were fixed in 4% glutaraldehyde and dehydrated by immersion in sequential ethanol solutions for 10 min per rinse starting at 1 × 70% v/v, followed by 1 × 90% v/v, 1 × 95% v/v and 2 × 100% v/v, with a final rinse in hexamethyl-disilazane (HMDS) for 30 min. Samples were then air dried and sputter coated with 3 nm of platinum prior to analysis on Inspect FEI F50 Scanning Electron Microscope.

In such samples, preparing electron microscope samples is not easy. How did you do that and what are the imaging details?

Change accepted, now reads:

A single sample of the clogging material was obtained from an affected screen retrieved from a bore located on one of the properties included in the study. Samples were fixed in 4% glutaraldehyde and dehydrated by immersion in sequential ethanol solutions for 10 min per rinse starting at 1 × 70% v/v, followed by 1 × 90% v/v, 1 × 95% v/v and 2 × 100% v/v, with a final rinse in hexamethyl-disilazane (HMDS) for 30 min. Samples were then air dried and sputter coated with 3 nm of platinum prior to analysis on Inspect FEI F50 Scanning Electron Microscope.

Can authors better explain how bacteria related clogging and other reasons for clogging can be separated and what is the overall impact of each on the study area. 

Change accepted now reads:

Analysis of the groundwater microbiology and inorganic chemistry found no apparent correlation between the presence of IRB and the clogging status of wells. In fact, IRB proved to be widespread throughout the limestone aquifer. However, a clear correlation could be found between clogging affected bores and the redox potential of the groundwater with the most severely affected bores strongly oversaturated in respect to iron oxide minerals

Can authors propose some mitigative methods and elaborate on the importance of this study?

Due to the small sample size of this study authors are hesitant to suggest mitigative strategies and risk over extrapolating the data. Rather the findings form this pilot study have identified the direction for future research.