the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 02 Sep 2024
Water hyacinths as riverine plastic pollution carriers
Abstract. Plastic pollution is an emerging entity threatening freshwater and marine ecosystems. Rivers play an important role in the transport and retention of plastic from land to sea. Tropical rivers are among the most polluted globally, and are assumed to emit substantial amounts of plastic into the marine environment. Concurrently, tropical rivers are invaded by water hyacinths, a free-floating vegetation species native to the Amazon. Water hyacinths grow rapidly, forming dense mats of plants and other material including plastic pollution. With only limited anecdotal and scientific evidence of plastic-water hyacinth trapping, its full spatial extent along river systems remain unknown. Here, we demonstrate the consistent role of water hyacinths as carriers of plastic pollution along a river. Over 69 k plastic items and 57 k water hyacinth patches were identified along the 42 km most downstream section of the Saigon river, Vietnam. More than 73 % of all floating plastics were carried by water hyacinths, ranging between 58–82 % per specific location. The highest trapping ratio was found at the most upstream locations. Although water hyacinths only covered 1.3 % of the total river surface, the plastic concentration in water hyacinths was 197 times higher than in open water. Most downstream, the lowest water hyacinth coverage (0.2 %) corresponded to the largest difference between surface plastic concentration in water hyacinths and at the open water (factor 781). Previous work demonstrated the effective trapping of plastic pollution by water hyacinths at individual sites. Here, we show that plastic-water hyacinth aggregates consistently occur at the river scale. We quantified plastic and water hyacinths at five locations along the Saigon river, Vietnam, using drones and fixed cameras, in combination with a custom-trained YOLOv8 deep learning model. With our paper, we support the theory that water hyacinths effectively concentrate and carry plastic pollution along rivers. Further work on plastic-water hyacinth interactions is needed to better understand the transport, fate and impact of plastic in the world's most polluted rivers. Our results also support the idea of plastic monitoring from space using well-detectable floating vegetation as a proxy. Finally, our work suggests that current removal practices of water hyacinths may be optimized to also recover plastic pollution from rivers.
- Preprint
(3070 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-2270', Anonymous Referee #1, 10 Oct 2024
The article is well-written; no major issues were found during my review. Below is a list of a few minor issues and technical corrections the authors should address before publication.
Page 4, caption of Fig. 1: there is a mismatch between what is said in the caption and what is reported in Panel A, i.e.: “The triangles indicate the camera locations, and the crosses the UAV locations”, on the map instead I think that circles indicate “camera locations” while triangles indicate “UAV locations”, right? Please correct.
Line 162: the word “being” is repeated twice. Please remove.
Line 204: this should be “Kruskal–Wallis” right?
Line 259: …or that water hyacinths could get stuck and trapped on riverbanks upstream and stop flowing downstream? Is this another option that would explain this observation right? In other words, a large fraction is trapped before and this would explain why fewer plants are detected downstream right? However in this case, also the plastic would be trapped in river banks right? Can the authors briefly discuss and elaborate on these points?
Line 306: “pattern”Citation: https://doi.org/10.5194/egusphere-2024-2270-RC1 -
RC2: 'Comment on egusphere-2024-2270', Anonymous Referee #2, 28 Oct 2024
Please check the annotated manuscript for revisions that are mainly about forms and light clarifications.
The manuscript is well written and clear.
Additional coments:
- Does the paper address relevant scientific questions within the scope of BG?
Yes, interactions between plastic pollution and an invasive plant (water hyacinth)
- Does the paper present novel concepts, ideas, tools, or data?
Not a novel concept, but new data in space and time reinforcing a concept proposed recently (by the same authors)
- Are substantial conclusions reached?
Yes, key role of water hyacinth for plastic transport along the entire length of tropical rivers.
- Are the scientific methods and assumptions valid and clearly outlined?
For what I can judge, yes. But I am not an expert in machine learning. However, it is well presented and convincing.
- Are the results sufficient to support the interpretations and conclusions?
Yes
- Is the description of experiments and calculations sufficiently complete and precise to allow their reproduction by fellow scientists (traceability of results)?
Yes
- Do the authors give proper credit to related work and clearly indicate their own new/original contribution?
Yes
- Does the title clearly reflect the contents of the paper?
Yes
- Does the abstract provide a concise and complete summary?
Yes
- Is the overall presentation well structured and clear?
Absolutely clear
- Is the language fluent and precise?
Yes
- Are mathematical formulae, symbols, abbreviations, and units correctly defined and used?
As far I know, yes
- Should any parts of the paper (text, formulae, figures, tables) be clarified, reduced, combined, or eliminated?
Minor corrections are requested in the annotated manuscript.
- Are the number and quality of references appropriate?
Yes
- Is the amount and quality of supplementary material appropriate?
Yes
- RC3: 'Comment by RC3 on egusphere-2024-2270', Anonymous Referee #3, 04 Nov 2024
-
CC1: 'Comment on egusphere-2024-2270', Massimiliano Scalici, 05 Nov 2024
General Comments
Here Authors focused on the interaction of macroplastics with the water hyacinth in a part of a tropical river. They investigated whether water hyacinth could act as a carrier of macroplastics. Beyond the fact that the manuscript seems to be lacking innovation because there are many elements recalling previous publications by some co-authors of this manuscript, the manuscript presents a great number of major issues that should be fixed although the paper should be rejected. Firstly, the not-so-large number of images used to calibrate the model is a problem and they should be implemented. Secondly, self-citations occur a lot within the paper and important and recent papers are not quoted: this is a big mistake that can prejudicate the paper from being published. Furthermore, “Not all locations were covered equally during data collection”, the Authors compared findings from different locations and ways of collection. I should say that, for this and other reasons, this paper should be rejected.
Specific Comments
- Abstract: L4, all tropical rivers are invaded by water hyacinth? L5: what plants? L11: open waters? What’s that?
- L10: do you have also a downstream location? You should make it clear. From your paper, I understood that your locations are both in the last part of the river, therefore it is not proper to name them “upstream and downstream”.
- L14-15: The abstract needs to be restructured. This part is methodological, it should be moved upper in the abstract.
- L16: it is already known, as you stated, so what is the novelty of the paper?
- L18-20: “Our results also support the idea of plastic monitoring from space using well-detectable floating vegetation as a proxy. Finally, our work suggests that current removal practices of water hyacinths may be optimized to also recover plastic pollution from rivers”. These lines are out of the blue.
- L23: reword and add that you are assuming this based on models
- L28: “invasive water hyacinths.”: are there native water hyacinths? Please reword
- L54: “Building on previous assessments, we further explore the role of water hyacinths as plastic carriers at the river scale.” This has been already investigated. Where is the novelty of your paper?
- L59: what is open water? Why to use it?
- L62-66: “Our findings emphasize that plastic-water hyacinth interactions are an integral component of the plastic transport and retention dynamics in tropical rivers.”. This part should be deleted as here it is a spoiler and do not add information to the hypotheses of your paper. Furthermore, I am wondering about the hypotheses of your aims.
- Introduction: some studies published in the literature are not here quoted, this is the major issue that makes the ms not ready to be published. Address all those papers about macroplastics in rivers.
- Introduction and discussion: the self-citation rate is a great issue in the paper. Even though there is no written rule about it, anyway software reports about 50% rate of self-citations, which is not appropriate for a high-quality international journal. Self-citations should be deleted and other literature papers on the same topic should be solved.
- Methods, Line 165: 272 labelled images are not so representative to calibrate a model. This should be addressed in the limits of your approach in the discussion. Moreover, the methods proposed used this calibration in different ecosystems (i.e., small rivers or canals), therefore this could not work in other systems.
- L189: “However, the Modeltiles may count many hyacinth patches in image tiles as individual hyacinths, leading to an overestimation of their abundance.” If the model is overestimating is not a good model. You should revise this part.
- Results, 3.1 subheading should be revised as “river plastic density” because the use of “upstream and downstream” is not properly used here.
- L213: “Our results support the hypothesis that water hyacinths act as important carriers..” this is not true, reword. Moreover, avoid using terms like “important, effective, consistent, etc” throughout the ms (e.g. “important role”, Line 230 effectively, subheading L260 consistent, etc).
- L221: “open water..”: what about open water? How has it been calculated? Does it occupy the same area as plants? This should be clearly stated in Methods.
- L226: “constant Cwh may yield a reasonable first estimate, but may result in large uncertainties for individual locations. Understanding the spatial and temporal variation in Cwh are therefore crucial for plastic detection methods that use water hyacinths as a proxy.” The constant is different, so you cannot compare these findings and the constant should be revised in the equation.
- L234-238: “Downstream, a similar amount of plastic is trapped by a smaller water hyacinth area compared to upstream. As these are also the least accessible areas for conventional plastic monitoring, this emphasizes the potential of using water hyacinths as a proxy for plastic pollution. Furthermore, it suggests that the water hyacinths that travel all the way down to the confluence have accumulated the most plastics. Removing water hyacinths in the downstream regions may therefore also lead to capturing the largest amount of plastics” This paragraph does not stick to your findings and should be revised.
- Line 277: “Not all locations were covered equally during data collection” This is the big issue: you compared findings from different locations and ways of collection. I should say that, for this and other reasons, this paper should be rejected.
- L289: “Lit erature recommends a DSG of 0.5 to 1.25 cm/pixel for macroplastic detection (Andriolo et al., 2023), and this range cannot be reached when the image resolution has to be reduced with a factor 32 (camera) to 56 (UAV). Our initial YOLOv8 configuration could not deal with this, and therefore we trained two separate models (i.e., Modelresize and Modeltiles) using different image processing methods: (1) resizing images, and (2) cutting images into tiles” this study has been carried out in other ecosystems, therefore this findings should not be compared to yours.
- Discussion: Some studies conducted in the literature are not here quoted, this is the major issue that makes the ms not ready to be published. Address all those papers about macroplastics in rivers.
- Discussion: you should consider recent publications and the current literature in the Introduction and Discussion sections.
- There are some issues and limitations that should be added to the discussion
- L311, section 3.5 Outlook reports a list of not-proven results where the Authors made several speculations – therefore it is strongly suggested that it should be deleted.
- Conclusions should be revised according to the results obtained and possible future perspectives.
- Please check the English throughout the manuscript
Citation: https://doi.org/10.5194/egusphere-2024-2270-CC1 -
RC4: 'Comment on egusphere-2024-2270', Anonymous Referee #4, 10 Nov 2024
The article titled "Water Hyacinths as Riverine Plastic Pollution Carriers" addresses an urgent issue relevant to tropical river systems, which are among the most plastic-polluted globally. The study’s methodological approach and its potential implications are both pertinent; however, certain methodological aspects concerning the application of deep learning techniques warrant scrutiny (outlined below).
First, while the study’s premise is valid, the lack of references to other studies examining the role of riparian and aquatic vegetation in plastic trapping is concerning. Including comparative analyses, such as Cesarini and Scalici's (2022) work, 'Riparian Vegetation as a Trap for Plastic Litter', which explores eight rivers across the same climatic zone with varied vegetation types, could enhance the depth of the discussion. Additionally, citing Anantachaisilp et al. (2021) 'An Eco-Friendly Bioplastic Film Obtained from Water Hyacinth' would expand on the ecological context and underscore the multi-faceted environmental impact of water hyacinths in these systems.
In the Introduction, it would be beneficial to delineate explicitly how this study diverges from the findings of Schreyers et al. (2024) 'Water Hyacinths Retain River Plastics'. While there is overlap, particularly in the reported trapping efficiency, a more precise articulation of the novelty of this study’s contribution would be advantageous.
Further clarification is requested on whether the authors accounted for plastic particle shape during deep learning-based detection. Particle shape analysis could guide future research directions and indicate whether other tropical plant species could potentially have higher plastic trapping capacities than water hyacinths.
The UAV imaging during the dry season, as outlined in section 2.2, raises questions regarding seasonal variability. Could the authors speculate on how results might differ in the wet season? Additionally, were there significant lighting variations due to cloud cover during data acquisition that could have affected model performance? This aspect merits a detailed discussion in section 2.2.
A primary concern is the highly variable volume of images used for analysis and the lack of clear calibration between camera and UAV images. The calibration discrepancy, especially where both sources were not consistently applied across locations, could critically affect the interpretative accuracy of results. Data presented in Table 2 require clarification. Specifically, why were only 0.3–1.9% of images labeled (depending on location)? This sample size appears insufficient to support robust deep learning modeling. For the study to be published, this methodology must be refined, as the outcomes could be significantly over- or underestimated. Furthermore, this limitation re-emerges in lines 150-151 of the methods section, subsequently impacting algorithm performance results shown in Table C1. Metrics such as Recall and (m)AP50 suggest moderate to low accuracy, which does not entirely support the conclusions presented.
In summary, substantial methodological revisions are required for this work to advance to further rounds of review.
Citation: https://doi.org/10.5194/egusphere-2024-2270-RC4
Model code and software
Deep learning plastic model YOLOv8 Tianlong Jia https://github.com/TianlongJia/deep_plastic_YoloV8
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 370 | 134 | 34 | 538 | 4 | 13 |
- HTML: 370
- PDF: 134
- XML: 34
- Total: 538
- BibTeX: 4
- EndNote: 13
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1