Hydrological tracers for assessing transport and dissipation processes of pesticides in a model constructed wetland system

Fernández-Pascual, Elena; Bork, Marcus; Hensen, Birte; Lange, Jens

doi:https://doi.org/10.5194/hess-24-41-2020

Articles | Volume 24, issue 1

https://doi.org/10.5194/hess-24-41-2020

© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/hess-24-41-2020

© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 24, issue 1

Research article

|

08 Jan 2020

Research article |

| 08 Jan 2020

Hydrological tracers for assessing transport and dissipation processes of pesticides in a model constructed wetland system

Elena Fernández-Pascual, Marcus Bork, Birte Hensen, and Jens Lange

Download

Final revised paper (published on 08 Jan 2020)
Supplement to the final revised paper
Preprint (discussion started on 06 Feb 2019)
Supplement to the preprint

Interactive discussion

Status: closed

AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment

- Printer-friendly version

- Supplement

RC1: 'Review hess-2019-19', Anonymous Referee #1, 20 Feb 2019
- AC1: 'Answers to referee #1', Elena Fernández-Pascual, 01 May 2019
RC2: 'Review Fernandez-Pascual et al', Anonymous Referee #2, 08 Mar 2019
- AC2: 'Answers to referee #2', Elena Fernández-Pascual, 01 May 2019

Peer-review completion

AR: Author's response | RR: Referee report | ED: Editor decision

ED: Reconsider after major revisions (further review by editor and referees) (17 May 2019) by Erwin Zehe

Dear Fernández-Pascual

After reading your study again, I am, in line with both reviewers, impressed by the high quality of the underlying experiments. Nevertheless, both reviewers came up with a long list of critical and constructive comments which need to be thoroughly addressed in a round of major revisions.

Reviewer 1 recommended rejection – her/his main critique is that it is not clear how the findings of your experiments can be generalized and the lack of replication. In your reply you better explained that the overall objective of your study is to introduce the approach and more specifically to show the potential of fluorescent tracers to explore the interplay of transport and transformation of reactive compounds. Seems like a good idea, however, this implies that the story line of the manuscript needs to be strongly revised. Moreover, I think that it is very much worth to consider her/his suggestion of “Putting the characteristics of the artificial wetlands (texture, organic carbon content, water residence time, redox conditions etc.) into the context of real-world wetlands, to reflect – based on scientific theory – what follows for pesticide retention in such wetland and to demonstrate respective insights that go beyond prior knowledge.”

I see that the lack of replicates is a weakness and that the abundance/absence of a single macropore will at this scale drastically change outcome of the experiment (you operate below the REV!). A proper design would need at least three replicates. But I do not think that it is a killing argument, if the story line will be revised as you proposed. An alternative is the combination with modelling to overcome this limitation. Solute transport has a theory, means that we are not thrown to a pure statistical learning paradigm.

While reviewer 2 recommended minor revision, her/his main critique points deserve much attention as well. A mass balance closure of 30% leaves us with much uncertainty about the entire concert of transport and transformation. In a field study this would be a nice achievement, but for a controlled environment it is poor and this puts the value of these experiments into question (also with respect to the issue of being representative and working below the REV). This needs to be discussed.

Given coarse texture of the soil, a rapid reaction does not necessarily speak for preferential flow! This is surely related to fuzzy definition of preferential flow. Is transport in the “near field” preferential flow (transport times < than lateral mixing times) or is it a non-Gaussian residence time distribution? I think the manuscript will benefit from being precise in this respect, maybe a straight forward model exercise with could help to quantify the degree of non-normality of the breakthrough curves. Last but not least I think it is worth to discuss the role of retardation (which introduces a time lag) when doing a correlation analysis of breakthrough curves. One could thus also infer on the retardation using lag correlations?

Look forward to receive the revised manuscript.

Erwin Zehe

Hide

AR by Elena Fernández-Pascual on behalf of the Authors (16 Aug 2019) Author's response Manuscript

ED: Referee Nomination & Report Request started (05 Sep 2019) by Erwin Zehe

RR by Anonymous Referee #3 (13 Sep 2019)

RR by Anonymous Referee #2 (26 Sep 2019)

Suggestions for revision or reasons for rejection

Second revision

The authors have made some efforts to rewrite and clarify the questions raised in the first reviews. Some issues, however, still need to be resolved. The authors should again check their line of argument in critical appraisal of their data and the peculiarities of their setup, and the manuscript could be streamlined and condensed. The resulting paper would surely deserve publication as a valuable source for further studies.

Perhaps the authors might want to revisit the objectives stated in the introduction, also in light of the comments made by the other referee. Maybe objectives i) and ii) are enough, and can also be formulated as three bullet points (investigation of transport, investigation of dissipation, comparison of tracers and pesticides). The design of the experiment does include vegetation and alternating saturation, but was perhaps not adequate to investigate the influence of vegetation and alternating saturation in great details, due to the lack of replica and controls. Of course, these things can still be discussed, and answering the remaining questions is still challenging enough.

The authors refrain from including modeling in the paper, as they point out in their responses to the reviews. Without a quantitative analysis, however, large parts of the conclusions on transport, retardation, sorption and transformation remain rather speculative. A comprehensive numerical simulation would indeed be out of scope of the paper, but perhaps a first assessment of the results using solute transport theory (transfer functions) would have provided a more adequate means to analyze the transport behavior. Instead, the analysis of the data relies heavily on correlations between the breakthrough curves, but correlation ignores any shifts due to retardation. The analysis of lag correlation (which neglects any dispersion) was newly introduced in the revised version, but it is limited to the bromide BTC. Comparing the different tracers and assessing possible retardation and transformation is difficult, if not impossible, with this approach. Looking at lag correlations between the different BTCs (and reporting the significant findings) could possibly improve this weakness a bit, if further analyses should really be out of scope.

Any quantitative assessment requires getting a good description of the flow behavior. The presented perceptual model of the flow field is that tracers enter at the inlet, percolate downwards to the bottom gravel layer, from where they are transported to the upper layers by upwelling. The observations show that the middle layers are not participating in this transport, which is attributed to preferential flow. That water should have passed by all observation points in this layer is unlikely, but if all other explanations can be ruled out, this surprising finding would have to be considered in the interpretation. In my opinion, it is much more likely that lateral transport also occurred at or near the surface, perhaps on top of the saturated, “unconnected” middle layers. This would much better explain the higher concentrations near the surface compared to the bottom. Preferential flow could still be present in the second run, when solutes also appear in the middle layers. I encourage the authors to revisit this point.

The discussion and interpretation of the results should be checked critically, streamlined and condensed. The structure with 5 sub-sections is fine, but perhaps it can be rearranged to separate more clearly the interpretation of the experimental results from the inferences on wetlands and the referencing to the literature. For example, last paragraph of 3.2 and last half of 3.3 could go to the end; 3.3 and 3.4 could be combined, etc.

Specific Comments

p 2, l 29: Strictly speaking, it is not a control, because the difference is not only that on part is being planted, but also is located at the inlet.
p 4, ll 12 to 15: The tracer injection is not fully clear. It is stated that solutes were contained in 40 L of water. This solution was pumped into the system until saturation. According to Fig. 2, saturation was reached in 1-2 days. On page 3, l 41, it is stated that saturation with target substances was one week. In Tab. 2 (B) the pumping rate is given with 21.6 L h-1. Was the tracer solution injected in two hours or two days? Was other water used to further saturate the system after the 40 L were injected?
p 6, l 11: Due to its conservative and non-sorbing character, Br– can hardly serve for investigating retardation, which is commonly caused by reversible adsorption/desorption
p 6, l 18: Correlating BTC between the vegetated and non-vegetated zones could also be interesting, especially for comparison of the zones
p 6, l 24: As stated in the next paragraph, an overall mass balance for pesticides was not possible.
p 7, ll 8 to 18: Why can these observations not be due to flow along the surface? Higher values in the upper layers than in the bottom are hardly possible if solute flow was first to the bottom and then up - or you missed important parts of the breakthrough with your sampling frequency
p 7, ll 19 to 24: What are the time units of the lag correlation? If it is days, why are only lags up to 7 days tested/displayed? From the BTC it appears that lag could be in the range of weeks – months. Why were only Br-BTC analysed?
p7, ll 24/25: That is difficult to understand. Do you argue that transport was from the bottom to the top, but delayed due to low connectivity? Why was there a transport then at all? How fast would a transport have been that was not delayed? How do you rule out transport along the surface? Isn't that much more likely in view of the higher concentrations at the top compared to the bottom?
p 7, l 29: The contrast between "strong" correlation and "not any" correlation sounds greater than it actually is (0.77 vs. 0.55). More important, that does not tell much about the transport characteristics other than that they are different. How about lag correlation here?
p 7, ll 30 to 33: Why not mentioning that Br also has experienced plant uptake? For example, S Xu et al., Environ. Sci. Technol. 2004, 38 (21), 5642-5648
p 8, ll 7/8: These parts are more or less the beginning and the end of the transport regime through the tank, right?
p 8 l 17: “confirmed creation of preferential flow paths” - Or maybe it was only high conductivity and low sorption in the gravel layer, and overland/near-surface flow?
p 9, l 40: “evidencing their great mobility and persistence” – But the total recovery of metazachlor and TPs was low, < 20 %. This is a contradiction.
p 10, ll 9 to 14: These sentences are directly contradicting each other: Either DT50 values let expect higher recoveries, or DT50 values lead to lower recoveries
p 10, l 24 to p 11, l 2: This part not only related to recovery, perhaps put it towards the end (discussion of meaning of results for pesticide removal in wetlands in general)
p 11, l 35 to p 12, l 6: No plant uptake, no transformation, no recovery, no mineralization, no volatilization - where did it go? Is it really plausible that the bulk of the substances was sorbed on sand and gravel particles? What would that imply for future applications of these substances – would sorbing capacities be depleted?
p 12, ll 7/8: Is this difference between vegetated and non-vegetated for Br and UR really significant? Can it only be attributed to the vegetation? How does the distance to the inlet influence this finding?
p 12, ll 10/11: How do you rule other possibilities? Couldn’t preferential flow paths also be due to the flushing in between runs that contributed to a change in pore space? Which effect would a stronger hydraulic gradient invoked by root water uptake in the vegetated zone have?

p12, ll 26/27: From Fig 8 and 9 (A1 and B1) I would rather argue that SRB and the two pesticides are not behaving that similar.
p12, ll 33/34: I do not see where you have shown unequivocally that UR was transformed bio-chemically – you assume that, because there is a large part missing in the mass balance (cf. p 10, l19/20).
p12, l 38: How realistic / relevant is that long time period? Elsewhere, you have reported mean residence times of 6 days for typical wetland systems.
p13, l 12 to 14: Absence in middle layers because of the flow regime (local effect of this particular setup)? (related to comments above )
p 19, Table 2(A): Mean initial organic carbon content - how is zero error achieved?
p 19, Table 2(A): Conductivity: Is Br breakthrough detectable in the conductivity measurements with the 5 TE sensors?
p 21, Table 6: is this really “removal “- in light of the low probability of degradation stated in your text? Or just "trapping" in the sediments with the potential of being slowly released?
In case of Met it is obviously wrong – it was not removed by 92.6 %, but at least another 6 % were transformed.
p 24, Fig. 3-1): Why is there no flow on surface? This was changed from the earlier version.
p 25, Fig 4: The soil moisture sensors probably were not calibrated for the specific material? Gravel is reported to have 45% porosity, but does not exceed 33% moisture during saturated conditions?
Even though soil moisture is not calibrated - could simple mass balance shed light on the proportion of stagnant zones in the system? 40 Liters in a given volume with given porosity require a certain antecedent soil moisture to effect saturation as shown here.
p 27, Fig 6: Time units? Why only +-7?
p 28, Fig 8: In my opinion, this figure shows that all substances were transported with surface flow and vertically downwards from the inlet. Only after that first period of transport, behavior seems to differ.
p 29, Fig 9: Check caption.

Hide

ED: Publish subject to minor revisions (review by editor) (01 Oct 2019) by Erwin Zehe

AR by Elena Fernández-Pascual on behalf of the Authors (24 Oct 2019) Author's response Manuscript

ED: Publish as is (28 Oct 2019) by Erwin Zehe

AR by Elena Fernández-Pascual on behalf of the Authors (25 Nov 2019)

Short summary

In this study we explore the use of hydrological tracers coupled with high vertical resolution sampling and monitoring to evaluate temporal and spatial mechanisms that dominate transport and dissipation of pesticides in a laboratory-scale constructed wetland. Our results reveal different transport vectors and dissipation pathways of solutes over time and space that are influenced by the constructional design, the presence of plants and the alternation of different hydrological conditions.