Widespread Microplastic Pollution in Central Appalachian Streams: Implications for Freshwater Ecosystem Sustainability

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The authors quantify microplastic (MP) contamination in hogsucker fish in central Appalachia. MPs were counted visually after digestion in 10% KOH solution. Observations of MPs were correlated with agricultural land use and waste water discharges. My major comments are related to the statistical approaches used and the laboratory methods. Count data, especially when zero-inflated or highly skewed, should be handled with the appropriate distribution on model error. Broadly, the best is to select a statistical approach that fits the structure of the data rather than transforming the data to fit the assumptions of a statistical approach, more details are in the minor comments. Visual identification is a qualitative measure of MPs, in my experience each sample is counted by two individuals and the results are often different (high observational error). The state of the science now is to include more quantitative measurement techniques, e.g., Raman, infrared spectroscopy, or FTIR. For this manuscript, the use of only visual methods is enough to test the authors’ hypotheses if some measure of observer error could be included in the text and a more robust discussion of the results of laboratory control samples.

Minor comments below.

Introduction

The introduction uses adequate sources and presents a clear narrative for the manuscript. An addition that should be considered is a direct text reference to other studies which used fish GI tracks as indicators of pollutant sources such as heavy metals. The authors could briefly discuss the connection between MPs and other emerging contaminants such as PFAS. Additionally, the use of fish as bioindicators of aqueous pollution sources which if analyzed directly from water samples would have concentrations too low for adequate coverage.

Methods

Lines 133-134 are written referring to the results of this study, meaning they should be in the results or discussion section. Descriptions of the dispute over total length and contaminant concentration should be in the introduction.

Stats – details below, in general, the data type should be considered carefully in these analyses. MPs can be counts or concentrations, both are left censored at zero and require non-gaussian distributions. Even MP length is continuous both left censored at zero, these details should be considered carefully in selecting what statistical approaches will be used.

1) Linear model; MP ~ fish length

# Are the microplastic data in this regression represented as a count or concentration?

2) ANCOVA; log transform MP count and MP length ~ watershed + fish length

# This is not an optimal approach to handling these data, log transformed data can introduce bias in the analysis and model residuals may still fail to meet assumptions. A generalized linear model using a Poisson or negative binomial error distribution will allow for zeros, non-negative nature of count data, and perform better.

3) Pearson correlation analysis of environmental variables and MPs

# What are the environmental variables you examined? I missed these if they were listed previously. How are the MP data represented in this analysis; counts or concentrations?

4) Mixed effects model; MP ~ land cover types, E. coli, water conductivity, watershed area, fish length + (1|sampling location)

# Again, with MPs as counts a transformation is not the optimal approach, using a distribution that describes the inherent properties of count data should be assessed.

5) Likelihood Ratio Test

# Add a sentence that you did a step-wise iterative model approach (either iteratively removing or adding covariates), otherwise readers may assume that your testing the models in 1, 2, and 4 of this section against each other.

Was there a method description for how the lab controls were used?

Results

Minor point; should the site map come before the initial results figure? Readers might expect to see a site map in the introduction or methods sections.

Lines 209 to 227, presenting the best and second best model results is a little confusing here, could just the best model be presented?

Line 221, for hogsuckers, would it be assumed that larger rivers (increased watershed area) would have larger fish? So, would fish length be more appropriate as a random effect?

Were the results of the lab controls presented?

Discussion

Line 239, if all your fish were juvenile it should be mentioned somewhere in the methods. Additionally, some quick life history data would help readers; do they always live in these rivers? Do they feed on the benthos or pelagically? Are the omnivores/predators/detritivores? (If I missed this somewhere, please ignore).

Author Response

Dear Mr. Tian, 

We thank you and the three reviewers, all of the reviewers were constructive and will increase the clarity of our paper.  For your convenience, each reviewer suggestion was cut and pasted below, then our response is immediately below each reviewer suggestion.  Additionally, we have attached a copy of the responses in pdf form if that is easier to look over. We appreciate you extending the deadline a week, that has given us time to thoroughly and meaningfully address each comment.  We look forward to your decision.  

 

Thank you,

 

Isabella Tuzzio, on behalf of the author team

Reviewer 1

1. The authors quantify microplastic (MP) contamination in hogsucker fish in central Appalachia. MPs were counted visually after digestion in 10% KOH solution. Observations of MPs were correlated with agricultural land use and waste water discharges. My major comments are related to the statistical approaches used and the laboratory methods. Count data, especially when zero-inflated or highly skewed, should be handled with the appropriate distribution on model error. Broadly, the best is to select a statistical approach that fits the structure of the data rather than transforming the data to fit the assumptions of a statistical approach, more details are in the minor comments. Visual identification is a qualitative measure of MPs, in my experience each sample is counted by two individuals and the results are often different (high observational error). The state of the science now is to include more quantitative measurement techniques, e.g., Raman, infrared spectroscopy, or FTIR. For this manuscript, the use of only visual methods is enough to test the authors’ hypotheses if some measure of observer error could be included in the text and a more robust discussion of the results of laboratory control samples.
1. We address statistical methodology concerns in more detail in the following responses. Regarding microplastic identification methodology,  we agree that visual identification is subject to error, and in order to mitigate observational error due to visual inspection we filtered the stomach contents through a sieve tower with five mesh sizes to reduce the number of plastics on any one filter and improve count accuracy. We also used a single observer (lead author) to count every sample so that interobserver bias is not present in addition to count bias. We feel that this methodology combined with the control data and hot needle test ensure relatively accurate ratios to determine microplastic levels per site.
2. The introduction uses adequate sources and presents a clear narrative for the manuscript. An addition that should be considered is a direct text reference to other studies which used fish GI tracks as indicators of pollutant sources such as heavy metals. The authors could briefly discuss the connection between MPs and other emerging contaminants such as PFAS. Additionally, the use of fish as bioindicators of aqueous pollution sources which if analyzed directly from water samples would have concentrations too low for adequate coverage.
1. Thank you for the recommendations, they have been very helpful in diversifying the introduction!  We incorporated a study utilizing benthic fish GI tracts to evaluate bioaccumulation of heavy metals (line 64) as well as discussing microplastics as vectors for PFAS/chemical transport throughout ecosystems (line 33). We also incorporated explicit mentions of bioaccumulation in fishes to more clearly illustrate the methodological choice to use fish as a bioindicator instead of taking direct water samples. 
3. Methods: Lines 133-134 are written referring to the results of this study, meaning they should be in the results or discussion section. Descriptions of the dispute over total length and contaminant concentration should be in the introduction.
1. Thank you for catching that; we moved lines 133-134 into the Appendix where they will fit more clearly with the discussion of controls (line 382-383). Regarding the total length and contaminant concentration dispute we feel that it is important to discuss that relationship within section 2.5 because it provides the rationale behind our data analysis methodology/process, though we acknowledge that it is not traditional to highlight diverging hypotheses in a methods section. If this does not aid in the flow of the paper or the reader’s understanding as we had hoped, please follow up, we have no problems reworking the structure for increased clarity and greatly appreciate your thorough review!
4. Stats – details below, in general, the data type should be considered carefully in these analyses. MPs can be counts or concentrations, both are left censored at zero and require non-gaussian distributions. Even MP length is continuous both left censored at zero, these details should be considered carefully in selecting what statistical approaches will be used.
1. Our understanding is that log-transforming data then running Gaussian statistics is equivalent to using a lognormal distribution, however, we also acknowledge that using an appropriate distribution is preferred / the better approach.  Therefore, we reran all statistics using a negative binomial distribution. Surprisingly, we did not have zero inflated data, but it is 0-truncated and over-dispersed, so, in alignment with the reviewers suggestion we opted for the negative binomial distribution.
5. 1) Linear model; MP ~ fish length: # Are the microplastic data in this regression represented as a count or concentration?
1. The microplastic data are represented as a count, we have included the data in the supplementary files.
6. 2) ANCOVA; log transform MP count and MP length ~ watershed + fish length: This is not an optimal approach to handling these data, log transformed data can introduce bias in the analysis and model residuals may still fail to meet assumptions. A generalized linear model using a Poisson or negative binomial error distribution will allow for zeros, non-negative nature of count data, and perform better.
1. We reran all statistics using a negative binomial distribution. Surprisingly, we did not have zero inflated data, but it is 0-truncated and over-dispersed, so, in alignment with the reviewers suggestion we opted for the negative binomial distribution. There were no qualitative changes, however, we did update the statistics to reflect the revisions using the negative binomial distributions.
7. 3) Pearson correlation analysis of environmental variables and MPs: # What are the environmental variables you examined? I missed these if they were listed previously. How are the MP data represented in this analysis; counts or concentrations?
1. We examined specific conductivity, watershed area, land cover types, and E. Coli concentrations at each site (Section 2.1 Environmental Variables) we have included an environmental data table (lines 110-111) to clarify. The microplastic data is in counts. 
8. 4) Mixed effects model; MP ~ land cover types, E. coli, water conductivity, watershed area, fish length + (1|sampling location) : # Again, with MPs as counts a transformation is not the optimal approach, using a distribution that describes the inherent properties of count data should be assessed.

1. Thanks for the observation. We re-ran the models using a negative binomial distribution. We included the results of the most parsimonious models in the results section and assessments of the residuals in the appendix. The results were consistent with previous analyses, except that in the new model selection process, E. coli was selected in the second best model, and watershed area was not among variables in the two best models. 

 We note, however, that while the models converged in the new analyses using a negative binomial distribution and DHARMa's residual vs. predicted analyses and plots demonstrate "no significant problem was detected” (please, see residuals Appendix A3 and A4), comparison of residuals distributions among sets of models indicate to us that the first set of models (first version of the MS) based on transformed data (lognormal distribution) fitted better.

1. 5) Likelihood Ratio Test: # Add a sentence that you did a step-wise iterative model approach (either iteratively removing or adding covariates), otherwise readers may assume that your testing the models in 1, 2, and 4 of this section against each other.

1. We included the explanation. Thanks.

1. Was there a method description for how the lab controls were used?
1. Yes, section 2.4 Contamination Protocol and Appendix A contain control details. A brief overview is as follows: Each sample was paired with a corresponding procedural blank in order to better quantify background contamination levels. Blanks were petri dishes opened during dissection of specimens, filled with equal amounts of 10% KOH, sealed, filtered, and enumerated in the same manner as their corresponding specimen simultaneously. Appendix A details the type, count, mean length, and total length of control microplastic particles for comparison with the sample particles. 
2. Results: Minor point; should the site map come before the initial results figure? Readers might expect to see a site map in the introduction or methods sections.
1. We agree, we moved the site map into the methods section (line 99)
3. Lines 209 to 227, presenting the best and second best model results is a little confusing here, could just the best model be presented?
1. We updated the wording, the results of the best and second best model are separate paragraphs, which we feel makes things less confusing - we hope reviewers agree!  We wish to retain both models, our goal is not predictive (which we agree we should focus on the best model), but instead, we are hypothesis testing and the second model indicates a role sewage pollution (i.e. E. coli), which we feel is a valuable piece of the story, which would be missing without inclusion of the second best model.  Plus the difference in AIC between the top two models was relatively small.  We feel inclusion is important as it helps complete the story, the full suite of drivers, that will ultimately encourage further research.  The connection between microplastic content in the hog sucker stomachs and environmental drivers, we believe is what makes our paper a stronger contribution and separates it from other microplastic fish papers that largely just document occurrence and variation (e.g. temporal or size-related).
4. Line 221, for hogsuckers, would it be assumed that larger rivers (increased watershed area) would have larger fish? So, would fish length be more appropriate as a random effect?
1. The streams we sampled represent typical mid-order Appalachian streams, which hold all life stages of northern hog suckers, though our study included only juveniles (~5-12cm, which may not have been clear in the earlier draft, but we strove to make clear in the revised draft).  This species has a common adult length of ~30cm and an asymptotic length of ~61cm (https://www.fishbase.se/summary/SpeciesSummary.php?ID=2990&AT=northern+hog+sucker).  Northern hog suckers are adapted for cooler waters and are generally confined to low- and mid-order streams, not in large rivers. We do not anticipate any systematic relationship between fish size and stream or watershed size for this species in the streams studied

We opted to reduce complexity in our models and therefore used a random intercept model with sites treated as random effects, while length as fixed factors. This assumes length has the same effect across all watersheds but still accounts for baseline differences between them. This is particularly necessary in our analyses where the dataset is relatively small and there is not enough variation in lengths across watersheds (although there is some variation as shown by the ANCOVA), all fishes are juveniles. By using a simpler random effect structure  i.e. sites as a random effect, we were still able to account for variation in responses among different sites/watersheds. 

We ran a set of models to explore the reviewer’s suggestion. We ran models with fish length as a random effect and compared these models with the two best models presented in the MS (both with fish length as fixed effect). The models with length included as a fixed term had a significantly better performance. Importantly, all of the models indicated that proportion of agricultural lands was the dominant driver of microplastic contamination, or qualitatively all the models produced the same results. As a note, we expect the model with random slope (length | watershed) could indeed be appropriated in our case to account for effects of length variations across watersheds. However, the random slope structure model  (where the effect of a length could vary across watersheds) did not converge possible due to the nature of the data set (unbalanced data across watersheds, small sampling numbers/watershed numbers). Thus, simplifying the random effects structure while still accounting for length seemed a reasonable solution. 

1. Were the results of the lab controls presented?
1. Yes, they can be found in Appendix A (line 380).
2. Discussion: Line 239, if all your fish were juvenile it should be mentioned somewhere in the methods. Additionally, some quick life history data would help readers; do they always live in these rivers? Do they feed on the benthos or pelagically? Are the omnivores/predators/detritivores? (If I missed this somewhere, please ignore).
1. We agree, we added information about feeding ecology, juvenile status, and fish length range in the introduction and methods (lines 55-64, 93-94)

Reviewer 2

1. The problem of microplastics in the modern world is very acute, and it is getting worse. Non-biodegradable materials are present everywhere, enter organisms, accumulate in tissues, but it is unclear what happens to them later in organs, tissues and cells. This is a problem of problems. Due to insufficient epidemiological and etiological studies, there is no complete understanding of the possible symptoms or diseases caused by microplastics. Therefore, the fight against excess microplastics requires the attention and efforts of every citizen, chemical enterprises, scientific institutions and governments. Of course, it is necessary to monitor the spread of microplastics in all biotopes – in soil, air, in inland freshwater bodies, as well as in seas and oceans. In this regard, this article is quite relevant. I found the authors' conclusion convincing that the local intake of microplastics in streams in north central West Virginia is the dominant source of microplastics compared to global factors such as atmospheric deposition.
1. Thank you for your feedback and support! We have incorporated the above and below information from other reviewers’ suggestions to improve clarity and contextualization, please let us know if you have any additional questions/suggestions.

 

Reviewer 3

1. The paper reports an interesting study. The following changes may benefit the manuscript: a.) a more detailed discussion of the research results - here, it would be important to review the results in context of table 1 (rather than comparing the MP contamination found in the Northern hogsucker with data coming from other parts of the world) - meaning a comparison with published results stemming from the Eastern part of the US. Also, other reports of MP presence from Appalachia and surrounding area may be helpful to give a background to the study. For instance, in the discussion of the possibility of atmospheric entry of MPs, the following paper could be used as a reference: Atmospheric Deposition of Microplastics in South Central Appalachia in the United States (Adam Elnahas, Austin Gray, Jennie Lee, Noora AlAmiri, Nishan Pokhrel, Steve Allen, and Hosein Foroutan, ACS ES&T Air 2025 2 (1), 64-72. DOI: 10.1021/acsestair.4c00189). Also, a more detailed description of the individual sites and their surrounding area would be beneficial for the reader. 
1. Thank you for your feedback! We agree that a review of eastern US microplastic pollution within fishes would be ideal but as we point out there is a research gap in the investigation of microplastics in freshwater fish so we were unable to find many comparable studies in similar environments and conducted a more holistic review globally. We included the paper you suggest in the discussion (line 322) which greatly improves experimental context, thank you for the tip! We have also included an environmental variable table (lines 110-111) to give more context to the individual sites and their surrounding areas.
2. Figures 1 and 2 should be interchanged in the text.
1. Thank you for catching that! We moved figure 1 to the methods section for increased clarity.
3. Is there any more information on the MPs - such as color distribution? chemical make-up?
1. For this study the only information we recorded about the microplastics themselves were type, count, and length, though we hope to expand data collection in future iterations!
4. The make of the dissecting microscope should be given (in line 112).
1. Added: Amscope 3.5X-180X Trinocular Stereo Zoom Microscope
5. Looking at Figure 3B, it seems that the data is mostly affected by the variation of MP length ingested by fish of 10-12 cm in length. Is the breakdown of fish size equal over all sites or do fish of a specific length come from mostly one site?
1. All fish, from all sites were juveniles varying is total length between 5.6-12.8mm.  There were no apparent patterns in fish length among sites. All of the streams we sampled contained adult fish (common length per FishBase.org https://www.fishbase.se/summary/SpeciesSummary.php?ID=2990&AT=northern+hog+sucker, is 30cm), while it would have been interesting to have sampled these larger fishes, they were not part of the collection of fish we had access to.  Plus, by focusing on early life fish (estimated 1.5-2.5 years old) we assess recent feeding events, less accumulation, which we feel is a separate question.
6. This sentence from appendix A is not readily understood:  A total of 498 microplastic particles were identified in control groups, ... Are these the blank samples? If so, the number is very high.
1. Correct, these are the blank samples and although 500 environmental microplastics sounds like a lot, that is only about 9 particles per control on average compared to about 40 per sample. Unfortunately, a minimal level of contamination was difficult to mitigate in the absence of a clean room but it is accounted for by the controls. 
7. The authors write in line 270: excepting those with extremely stringent protocol [38]. Does this mean to say that globally and in general the reported MP counts are too high because of external MP contamination?  
1. We meant that the studies performed with extremely stringent protocol, ie intensive digestion/chemical analysis processes often find low levels of microplastic contamination as the cited study states. However, this is not consistent with the majority of global research we reviewed and could potentially be due to degradation of plastics because of such aggressive/selective filtration and enumeration techniques. We added an explicit statement about digestion and enumeration (line 294) to clarify, thank you for pointing that out!

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The problem of microplastics in the modern world is very acute, and it is getting worse. Non-biodegradable materials are present everywhere, enter organisms, accumulate in tissues, but it is unclear what happens to them later in organs, tissues and cells. This is a problem of problems. Due to insufficient epidemiological and etiological studies, there is no complete understanding of the possible symptoms or diseases caused by microplastics. Therefore, the fight against excess microplastics requires the attention and efforts of every citizen, chemical enterprises, scientific institutions and governments. Of course, it is necessary to monitor the spread of microplastics in all biotopes – in soil, air, in inland freshwater bodies, as well as in seas and oceans. In this regard, this article is quite relevant. I found the authors' conclusion convincing that the local intake of microplastics in streams in north central West Virginia is the dominant source of microplastics compared to global factors such as atmospheric deposition.

Author Response

We thank you and the three reviewers, all of the reviewers were constructive and will increase the clarity of our paper.  For your convenience, each reviewer suggestion was cut and pasted below, then our response is immediately below each reviewer suggestion.  Additionally, we have attached a copy of the responses in pdf form if that is easier to look over. We appreciate you extending the deadline a week, that has given us time to thoroughly and meaningfully address each comment.  We look forward to your decision.  

 

Thank you,

 

Isabella Tuzzio, on behalf of the author team

Reviewer 1

1. The authors quantify microplastic (MP) contamination in hogsucker fish in central Appalachia. MPs were counted visually after digestion in 10% KOH solution. Observations of MPs were correlated with agricultural land use and waste water discharges. My major comments are related to the statistical approaches used and the laboratory methods. Count data, especially when zero-inflated or highly skewed, should be handled with the appropriate distribution on model error. Broadly, the best is to select a statistical approach that fits the structure of the data rather than transforming the data to fit the assumptions of a statistical approach, more details are in the minor comments. Visual identification is a qualitative measure of MPs, in my experience each sample is counted by two individuals and the results are often different (high observational error). The state of the science now is to include more quantitative measurement techniques, e.g., Raman, infrared spectroscopy, or FTIR. For this manuscript, the use of only visual methods is enough to test the authors’ hypotheses if some measure of observer error could be included in the text and a more robust discussion of the results of laboratory control samples.
1. We address statistical methodology concerns in more detail in the following responses. Regarding microplastic identification methodology,  we agree that visual identification is subject to error, and in order to mitigate observational error due to visual inspection we filtered the stomach contents through a sieve tower with five mesh sizes to reduce the number of plastics on any one filter and improve count accuracy. We also used a single observer (lead author) to count every sample so that interobserver bias is not present in addition to count bias. We feel that this methodology combined with the control data and hot needle test ensure relatively accurate ratios to determine microplastic levels per site.
2. The introduction uses adequate sources and presents a clear narrative for the manuscript. An addition that should be considered is a direct text reference to other studies which used fish GI tracks as indicators of pollutant sources such as heavy metals. The authors could briefly discuss the connection between MPs and other emerging contaminants such as PFAS. Additionally, the use of fish as bioindicators of aqueous pollution sources which if analyzed directly from water samples would have concentrations too low for adequate coverage.
1. Thank you for the recommendations, they have been very helpful in diversifying the introduction!  We incorporated a study utilizing benthic fish GI tracts to evaluate bioaccumulation of heavy metals (line 64) as well as discussing microplastics as vectors for PFAS/chemical transport throughout ecosystems (line 33). We also incorporated explicit mentions of bioaccumulation in fishes to more clearly illustrate the methodological choice to use fish as a bioindicator instead of taking direct water samples. 
3. Methods: Lines 133-134 are written referring to the results of this study, meaning they should be in the results or discussion section. Descriptions of the dispute over total length and contaminant concentration should be in the introduction.
1. Thank you for catching that; we moved lines 133-134 into the Appendix where they will fit more clearly with the discussion of controls (line 382-383). Regarding the total length and contaminant concentration dispute we feel that it is important to discuss that relationship within section 2.5 because it provides the rationale behind our data analysis methodology/process, though we acknowledge that it is not traditional to highlight diverging hypotheses in a methods section. If this does not aid in the flow of the paper or the reader’s understanding as we had hoped, please follow up, we have no problems reworking the structure for increased clarity and greatly appreciate your thorough review!
4. Stats – details below, in general, the data type should be considered carefully in these analyses. MPs can be counts or concentrations, both are left censored at zero and require non-gaussian distributions. Even MP length is continuous both left censored at zero, these details should be considered carefully in selecting what statistical approaches will be used.
1. Our understanding is that log-transforming data then running Gaussian statistics is equivalent to using a lognormal distribution, however, we also acknowledge that using an appropriate distribution is preferred / the better approach.  Therefore, we reran all statistics using a negative binomial distribution. Surprisingly, we did not have zero inflated data, but it is 0-truncated and over-dispersed, so, in alignment with the reviewers suggestion we opted for the negative binomial distribution.
5. 1) Linear model; MP ~ fish length: # Are the microplastic data in this regression represented as a count or concentration?
1. The microplastic data are represented as a count, we have included the data in the supplementary files.
6. 2) ANCOVA; log transform MP count and MP length ~ watershed + fish length: This is not an optimal approach to handling these data, log transformed data can introduce bias in the analysis and model residuals may still fail to meet assumptions. A generalized linear model using a Poisson or negative binomial error distribution will allow for zeros, non-negative nature of count data, and perform better.
1. We reran all statistics using a negative binomial distribution. Surprisingly, we did not have zero inflated data, but it is 0-truncated and over-dispersed, so, in alignment with the reviewers suggestion we opted for the negative binomial distribution. There were no qualitative changes, however, we did update the statistics to reflect the revisions using the negative binomial distributions.
7. 3) Pearson correlation analysis of environmental variables and MPs: # What are the environmental variables you examined? I missed these if they were listed previously. How are the MP data represented in this analysis; counts or concentrations?
1. We examined specific conductivity, watershed area, land cover types, and E. Coli concentrations at each site (Section 2.1 Environmental Variables) we have included an environmental data table (lines 110-111) to clarify. The microplastic data is in counts. 
8. 4) Mixed effects model; MP ~ land cover types, E. coli, water conductivity, watershed area, fish length + (1|sampling location) : # Again, with MPs as counts a transformation is not the optimal approach, using a distribution that describes the inherent properties of count data should be assessed.

1. Thanks for the observation. We re-ran the models using a negative binomial distribution. We included the results of the most parsimonious models in the results section and assessments of the residuals in the appendix. The results were consistent with previous analyses, except that in the new model selection process, E. coli was selected in the second best model, and watershed area was not among variables in the two best models. 

 We note, however, that while the models converged in the new analyses using a negative binomial distribution and DHARMa's residual vs. predicted analyses and plots demonstrate "no significant problem was detected” (please, see residuals Appendix A3 and A4), comparison of residuals distributions among sets of models indicate to us that the first set of models (first version of the MS) based on transformed data (lognormal distribution) fitted better.

1. 5) Likelihood Ratio Test: # Add a sentence that you did a step-wise iterative model approach (either iteratively removing or adding covariates), otherwise readers may assume that your testing the models in 1, 2, and 4 of this section against each other.

1. We included the explanation. Thanks.

1. Was there a method description for how the lab controls were used?
1. Yes, section 2.4 Contamination Protocol and Appendix A contain control details. A brief overview is as follows: Each sample was paired with a corresponding procedural blank in order to better quantify background contamination levels. Blanks were petri dishes opened during dissection of specimens, filled with equal amounts of 10% KOH, sealed, filtered, and enumerated in the same manner as their corresponding specimen simultaneously. Appendix A details the type, count, mean length, and total length of control microplastic particles for comparison with the sample particles. 
2. Results: Minor point; should the site map come before the initial results figure? Readers might expect to see a site map in the introduction or methods sections.
1. We agree, we moved the site map into the methods section (line 99)
3. Lines 209 to 227, presenting the best and second best model results is a little confusing here, could just the best model be presented?
1. We updated the wording, the results of the best and second best model are separate paragraphs, which we feel makes things less confusing - we hope reviewers agree!  We wish to retain both models, our goal is not predictive (which we agree we should focus on the best model), but instead, we are hypothesis testing and the second model indicates a role sewage pollution (i.e. E. coli), which we feel is a valuable piece of the story, which would be missing without inclusion of the second best model.  Plus the difference in AIC between the top two models was relatively small.  We feel inclusion is important as it helps complete the story, the full suite of drivers, that will ultimately encourage further research.  The connection between microplastic content in the hog sucker stomachs and environmental drivers, we believe is what makes our paper a stronger contribution and separates it from other microplastic fish papers that largely just document occurrence and variation (e.g. temporal or size-related).
4. Line 221, for hogsuckers, would it be assumed that larger rivers (increased watershed area) would have larger fish? So, would fish length be more appropriate as a random effect?
1. The streams we sampled represent typical mid-order Appalachian streams, which hold all life stages of northern hog suckers, though our study included only juveniles (~5-12cm, which may not have been clear in the earlier draft, but we strove to make clear in the revised draft).  This species has a common adult length of ~30cm and an asymptotic length of ~61cm (https://www.fishbase.se/summary/SpeciesSummary.php?ID=2990&AT=northern+hog+sucker).  Northern hog suckers are adapted for cooler waters and are generally confined to low- and mid-order streams, not in large rivers. We do not anticipate any systematic relationship between fish size and stream or watershed size for this species in the streams studied

We opted to reduce complexity in our models and therefore used a random intercept model with sites treated as random effects, while length as fixed factors. This assumes length has the same effect across all watersheds but still accounts for baseline differences between them. This is particularly necessary in our analyses where the dataset is relatively small and there is not enough variation in lengths across watersheds (although there is some variation as shown by the ANCOVA), all fishes are juveniles. By using a simpler random effect structure  i.e. sites as a random effect, we were still able to account for variation in responses among different sites/watersheds. 

We ran a set of models to explore the reviewer’s suggestion. We ran models with fish length as a random effect and compared these models with the two best models presented in the MS (both with fish length as fixed effect). The models with length included as a fixed term had a significantly better performance. Importantly, all of the models indicated that proportion of agricultural lands was the dominant driver of microplastic contamination, or qualitatively all the models produced the same results. As a note, we expect the model with random slope (length | watershed) could indeed be appropriated in our case to account for effects of length variations across watersheds. However, the random slope structure model  (where the effect of a length could vary across watersheds) did not converge possible due to the nature of the data set (unbalanced data across watersheds, small sampling numbers/watershed numbers). Thus, simplifying the random effects structure while still accounting for length seemed a reasonable solution. 

1. Were the results of the lab controls presented?
1. Yes, they can be found in Appendix A (line 380).
2. Discussion: Line 239, if all your fish were juvenile it should be mentioned somewhere in the methods. Additionally, some quick life history data would help readers; do they always live in these rivers? Do they feed on the benthos or pelagically? Are the omnivores/predators/detritivores? (If I missed this somewhere, please ignore).
1. We agree, we added information about feeding ecology, juvenile status, and fish length range in the introduction and methods (lines 55-64, 93-94)

Reviewer 2

1. The problem of microplastics in the modern world is very acute, and it is getting worse. Non-biodegradable materials are present everywhere, enter organisms, accumulate in tissues, but it is unclear what happens to them later in organs, tissues and cells. This is a problem of problems. Due to insufficient epidemiological and etiological studies, there is no complete understanding of the possible symptoms or diseases caused by microplastics. Therefore, the fight against excess microplastics requires the attention and efforts of every citizen, chemical enterprises, scientific institutions and governments. Of course, it is necessary to monitor the spread of microplastics in all biotopes – in soil, air, in inland freshwater bodies, as well as in seas and oceans. In this regard, this article is quite relevant. I found the authors' conclusion convincing that the local intake of microplastics in streams in north central West Virginia is the dominant source of microplastics compared to global factors such as atmospheric deposition.
1. Thank you for your feedback and support! We have incorporated the above and below information from other reviewers’ suggestions to improve clarity and contextualization, please let us know if you have any additional questions/suggestions.

 

Reviewer 3

1. The paper reports an interesting study. The following changes may benefit the manuscript: a.) a more detailed discussion of the research results - here, it would be important to review the results in context of table 1 (rather than comparing the MP contamination found in the Northern hogsucker with data coming from other parts of the world) - meaning a comparison with published results stemming from the Eastern part of the US. Also, other reports of MP presence from Appalachia and surrounding area may be helpful to give a background to the study. For instance, in the discussion of the possibility of atmospheric entry of MPs, the following paper could be used as a reference: Atmospheric Deposition of Microplastics in South Central Appalachia in the United States (Adam Elnahas, Austin Gray, Jennie Lee, Noora AlAmiri, Nishan Pokhrel, Steve Allen, and Hosein Foroutan, ACS ES&T Air 2025 2 (1), 64-72. DOI: 10.1021/acsestair.4c00189). Also, a more detailed description of the individual sites and their surrounding area would be beneficial for the reader. 
1. Thank you for your feedback! We agree that a review of eastern US microplastic pollution within fishes would be ideal but as we point out there is a research gap in the investigation of microplastics in freshwater fish so we were unable to find many comparable studies in similar environments and conducted a more holistic review globally. We included the paper you suggest in the discussion (line 322) which greatly improves experimental context, thank you for the tip! We have also included an environmental variable table (lines 110-111) to give more context to the individual sites and their surrounding areas.
2. Figures 1 and 2 should be interchanged in the text.
1. Thank you for catching that! We moved figure 1 to the methods section for increased clarity.
3. Is there any more information on the MPs - such as color distribution? chemical make-up?
1. For this study the only information we recorded about the microplastics themselves were type, count, and length, though we hope to expand data collection in future iterations!
4. The make of the dissecting microscope should be given (in line 112).
1. Added: Amscope 3.5X-180X Trinocular Stereo Zoom Microscope
5. Looking at Figure 3B, it seems that the data is mostly affected by the variation of MP length ingested by fish of 10-12 cm in length. Is the breakdown of fish size equal over all sites or do fish of a specific length come from mostly one site?
1. All fish, from all sites were juveniles varying is total length between 5.6-12.8mm.  There were no apparent patterns in fish length among sites. All of the streams we sampled contained adult fish (common length per FishBase.org https://www.fishbase.se/summary/SpeciesSummary.php?ID=2990&AT=northern+hog+sucker, is 30cm), while it would have been interesting to have sampled these larger fishes, they were not part of the collection of fish we had access to.  Plus, by focusing on early life fish (estimated 1.5-2.5 years old) we assess recent feeding events, less accumulation, which we feel is a separate question.
6. This sentence from appendix A is not readily understood:  A total of 498 microplastic particles were identified in control groups, ... Are these the blank samples? If so, the number is very high.
1. Correct, these are the blank samples and although 500 environmental microplastics sounds like a lot, that is only about 9 particles per control on average compared to about 40 per sample. Unfortunately, a minimal level of contamination was difficult to mitigate in the absence of a clean room but it is accounted for by the controls. 
7. The authors write in line 270: excepting those with extremely stringent protocol [38]. Does this mean to say that globally and in general the reported MP counts are too high because of external MP contamination?  
1. We meant that the studies performed with extremely stringent protocol, ie intensive digestion/chemical analysis processes often find low levels of microplastic contamination as the cited study states. However, this is not consistent with the majority of global research we reviewed and could potentially be due to degradation of plastics because of such aggressive/selective filtration and enumeration techniques. We added an explicit statement about digestion and enumeration (line 294) to clarify, thank you for pointing that out!

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The paper reports an interesting study. The following changes may benefit the manuscript:

a.) a more detailed discussion of the research results - here, it would be important to review the results in context of table 1 (rather than comparing the MP contamination found in the Northern hogsucker with data coming from other parts of the world) - meaning a comparison with published results stemming from the Eastern part of the US. Also, other reports of MP presence from Appalachia and surrounding area may be helpful to give a background to the study. For instance, in the discussion of the possibility of atmospheric entry of MPs, the following paper could be used as a reference: Atmospheric Deposition of Microplastics in South Central Appalachia in the United States (Adam Elnahas, Austin Gray, Jennie Lee, Noora AlAmiri, Nishan Pokhrel, Steve Allen, and Hosein Foroutan, ACS ES&T Air 2025 2 (1), 64-72. DOI: 10.1021/acsestair.4c00189). Also, a more detailed description of the individual sites and their surrounding area would be beneficial for the reader. 

Figures 1 and 2 should be interchanged in the text.

Is there any more information on the MPs - such as color distribution? chemical make-up?

The make of the dissecting microscope should be given (in line 112).

Looking at Figure 3B, it seems that the data is mostly affected by the variation of MP length ingested by fish of 10-12 cm in length. Is the breakdown of fish size equal over all sites or do fish of a specific length come from mostly one site?

This sentence from appendix A is not readliy understood:  A total of 498 microplastic particles were identified in control groups, ... Are these the blank samples? If so, the number is very high.

The authors write in line 270: excepting those with extremely stringent protocol [38]. Does this mean to say that globally and in general the reported MP counts are too high because of external MP contamination?  

 

Author Response

We thank you and the three reviewers, all of the reviewers were constructive and will increase the clarity of our paper.  For your convenience, each reviewer suggestion was cut and pasted below, then our response is immediately below each reviewer suggestion.  Additionally, we have attached a copy of the responses in pdf form if that is easier to look over. We appreciate you extending the deadline a week, that has given us time to thoroughly and meaningfully address each comment.  We look forward to your decision.  

 

Thank you,

 

Isabella Tuzzio, on behalf of the author team

Reviewer 1

1. The authors quantify microplastic (MP) contamination in hogsucker fish in central Appalachia. MPs were counted visually after digestion in 10% KOH solution. Observations of MPs were correlated with agricultural land use and waste water discharges. My major comments are related to the statistical approaches used and the laboratory methods. Count data, especially when zero-inflated or highly skewed, should be handled with the appropriate distribution on model error. Broadly, the best is to select a statistical approach that fits the structure of the data rather than transforming the data to fit the assumptions of a statistical approach, more details are in the minor comments. Visual identification is a qualitative measure of MPs, in my experience each sample is counted by two individuals and the results are often different (high observational error). The state of the science now is to include more quantitative measurement techniques, e.g., Raman, infrared spectroscopy, or FTIR. For this manuscript, the use of only visual methods is enough to test the authors’ hypotheses if some measure of observer error could be included in the text and a more robust discussion of the results of laboratory control samples.
1. We address statistical methodology concerns in more detail in the following responses. Regarding microplastic identification methodology,  we agree that visual identification is subject to error, and in order to mitigate observational error due to visual inspection we filtered the stomach contents through a sieve tower with five mesh sizes to reduce the number of plastics on any one filter and improve count accuracy. We also used a single observer (lead author) to count every sample so that interobserver bias is not present in addition to count bias. We feel that this methodology combined with the control data and hot needle test ensure relatively accurate ratios to determine microplastic levels per site.
2. The introduction uses adequate sources and presents a clear narrative for the manuscript. An addition that should be considered is a direct text reference to other studies which used fish GI tracks as indicators of pollutant sources such as heavy metals. The authors could briefly discuss the connection between MPs and other emerging contaminants such as PFAS. Additionally, the use of fish as bioindicators of aqueous pollution sources which if analyzed directly from water samples would have concentrations too low for adequate coverage.
1. Thank you for the recommendations, they have been very helpful in diversifying the introduction!  We incorporated a study utilizing benthic fish GI tracts to evaluate bioaccumulation of heavy metals (line 64) as well as discussing microplastics as vectors for PFAS/chemical transport throughout ecosystems (line 33). We also incorporated explicit mentions of bioaccumulation in fishes to more clearly illustrate the methodological choice to use fish as a bioindicator instead of taking direct water samples. 
3. Methods: Lines 133-134 are written referring to the results of this study, meaning they should be in the results or discussion section. Descriptions of the dispute over total length and contaminant concentration should be in the introduction.
1. Thank you for catching that; we moved lines 133-134 into the Appendix where they will fit more clearly with the discussion of controls (line 382-383). Regarding the total length and contaminant concentration dispute we feel that it is important to discuss that relationship within section 2.5 because it provides the rationale behind our data analysis methodology/process, though we acknowledge that it is not traditional to highlight diverging hypotheses in a methods section. If this does not aid in the flow of the paper or the reader’s understanding as we had hoped, please follow up, we have no problems reworking the structure for increased clarity and greatly appreciate your thorough review!
4. Stats – details below, in general, the data type should be considered carefully in these analyses. MPs can be counts or concentrations, both are left censored at zero and require non-gaussian distributions. Even MP length is continuous both left censored at zero, these details should be considered carefully in selecting what statistical approaches will be used.
1. Our understanding is that log-transforming data then running Gaussian statistics is equivalent to using a lognormal distribution, however, we also acknowledge that using an appropriate distribution is preferred / the better approach.  Therefore, we reran all statistics using a negative binomial distribution. Surprisingly, we did not have zero inflated data, but it is 0-truncated and over-dispersed, so, in alignment with the reviewers suggestion we opted for the negative binomial distribution.
5. 1) Linear model; MP ~ fish length: # Are the microplastic data in this regression represented as a count or concentration?
1. The microplastic data are represented as a count, we have included the data in the supplementary files.
6. 2) ANCOVA; log transform MP count and MP length ~ watershed + fish length: This is not an optimal approach to handling these data, log transformed data can introduce bias in the analysis and model residuals may still fail to meet assumptions. A generalized linear model using a Poisson or negative binomial error distribution will allow for zeros, non-negative nature of count data, and perform better.
1. We reran all statistics using a negative binomial distribution. Surprisingly, we did not have zero inflated data, but it is 0-truncated and over-dispersed, so, in alignment with the reviewers suggestion we opted for the negative binomial distribution. There were no qualitative changes, however, we did update the statistics to reflect the revisions using the negative binomial distributions.
7. 3) Pearson correlation analysis of environmental variables and MPs: # What are the environmental variables you examined? I missed these if they were listed previously. How are the MP data represented in this analysis; counts or concentrations?
1. We examined specific conductivity, watershed area, land cover types, and E. Coli concentrations at each site (Section 2.1 Environmental Variables) we have included an environmental data table (lines 110-111) to clarify. The microplastic data is in counts. 
8. 4) Mixed effects model; MP ~ land cover types, E. coli, water conductivity, watershed area, fish length + (1|sampling location) : # Again, with MPs as counts a transformation is not the optimal approach, using a distribution that describes the inherent properties of count data should be assessed.

1. Thanks for the observation. We re-ran the models using a negative binomial distribution. We included the results of the most parsimonious models in the results section and assessments of the residuals in the appendix. The results were consistent with previous analyses, except that in the new model selection process, E. coli was selected in the second best model, and watershed area was not among variables in the two best models. 

 We note, however, that while the models converged in the new analyses using a negative binomial distribution and DHARMa's residual vs. predicted analyses and plots demonstrate "no significant problem was detected” (please, see residuals Appendix A3 and A4), comparison of residuals distributions among sets of models indicate to us that the first set of models (first version of the MS) based on transformed data (lognormal distribution) fitted better.

1. 5) Likelihood Ratio Test: # Add a sentence that you did a step-wise iterative model approach (either iteratively removing or adding covariates), otherwise readers may assume that your testing the models in 1, 2, and 4 of this section against each other.

1. We included the explanation. Thanks.

1. Was there a method description for how the lab controls were used?
1. Yes, section 2.4 Contamination Protocol and Appendix A contain control details. A brief overview is as follows: Each sample was paired with a corresponding procedural blank in order to better quantify background contamination levels. Blanks were petri dishes opened during dissection of specimens, filled with equal amounts of 10% KOH, sealed, filtered, and enumerated in the same manner as their corresponding specimen simultaneously. Appendix A details the type, count, mean length, and total length of control microplastic particles for comparison with the sample particles. 
2. Results: Minor point; should the site map come before the initial results figure? Readers might expect to see a site map in the introduction or methods sections.
1. We agree, we moved the site map into the methods section (line 99)
3. Lines 209 to 227, presenting the best and second best model results is a little confusing here, could just the best model be presented?
1. We updated the wording, the results of the best and second best model are separate paragraphs, which we feel makes things less confusing - we hope reviewers agree!  We wish to retain both models, our goal is not predictive (which we agree we should focus on the best model), but instead, we are hypothesis testing and the second model indicates a role sewage pollution (i.e. E. coli), which we feel is a valuable piece of the story, which would be missing without inclusion of the second best model.  Plus the difference in AIC between the top two models was relatively small.  We feel inclusion is important as it helps complete the story, the full suite of drivers, that will ultimately encourage further research.  The connection between microplastic content in the hog sucker stomachs and environmental drivers, we believe is what makes our paper a stronger contribution and separates it from other microplastic fish papers that largely just document occurrence and variation (e.g. temporal or size-related).
4. Line 221, for hogsuckers, would it be assumed that larger rivers (increased watershed area) would have larger fish? So, would fish length be more appropriate as a random effect?
1. The streams we sampled represent typical mid-order Appalachian streams, which hold all life stages of northern hog suckers, though our study included only juveniles (~5-12cm, which may not have been clear in the earlier draft, but we strove to make clear in the revised draft).  This species has a common adult length of ~30cm and an asymptotic length of ~61cm (https://www.fishbase.se/summary/SpeciesSummary.php?ID=2990&AT=northern+hog+sucker).  Northern hog suckers are adapted for cooler waters and are generally confined to low- and mid-order streams, not in large rivers. We do not anticipate any systematic relationship between fish size and stream or watershed size for this species in the streams studied

We opted to reduce complexity in our models and therefore used a random intercept model with sites treated as random effects, while length as fixed factors. This assumes length has the same effect across all watersheds but still accounts for baseline differences between them. This is particularly necessary in our analyses where the dataset is relatively small and there is not enough variation in lengths across watersheds (although there is some variation as shown by the ANCOVA), all fishes are juveniles. By using a simpler random effect structure  i.e. sites as a random effect, we were still able to account for variation in responses among different sites/watersheds. 

We ran a set of models to explore the reviewer’s suggestion. We ran models with fish length as a random effect and compared these models with the two best models presented in the MS (both with fish length as fixed effect). The models with length included as a fixed term had a significantly better performance. Importantly, all of the models indicated that proportion of agricultural lands was the dominant driver of microplastic contamination, or qualitatively all the models produced the same results. As a note, we expect the model with random slope (length | watershed) could indeed be appropriated in our case to account for effects of length variations across watersheds. However, the random slope structure model  (where the effect of a length could vary across watersheds) did not converge possible due to the nature of the data set (unbalanced data across watersheds, small sampling numbers/watershed numbers). Thus, simplifying the random effects structure while still accounting for length seemed a reasonable solution. 

1. Were the results of the lab controls presented?
1. Yes, they can be found in Appendix A (line 380).
2. Discussion: Line 239, if all your fish were juvenile it should be mentioned somewhere in the methods. Additionally, some quick life history data would help readers; do they always live in these rivers? Do they feed on the benthos or pelagically? Are the omnivores/predators/detritivores? (If I missed this somewhere, please ignore).
1. We agree, we added information about feeding ecology, juvenile status, and fish length range in the introduction and methods (lines 55-64, 93-94)

Reviewer 2

1. The problem of microplastics in the modern world is very acute, and it is getting worse. Non-biodegradable materials are present everywhere, enter organisms, accumulate in tissues, but it is unclear what happens to them later in organs, tissues and cells. This is a problem of problems. Due to insufficient epidemiological and etiological studies, there is no complete understanding of the possible symptoms or diseases caused by microplastics. Therefore, the fight against excess microplastics requires the attention and efforts of every citizen, chemical enterprises, scientific institutions and governments. Of course, it is necessary to monitor the spread of microplastics in all biotopes – in soil, air, in inland freshwater bodies, as well as in seas and oceans. In this regard, this article is quite relevant. I found the authors' conclusion convincing that the local intake of microplastics in streams in north central West Virginia is the dominant source of microplastics compared to global factors such as atmospheric deposition.
1. Thank you for your feedback and support! We have incorporated the above and below information from other reviewers’ suggestions to improve clarity and contextualization, please let us know if you have any additional questions/suggestions.

 

Reviewer 3

1. The paper reports an interesting study. The following changes may benefit the manuscript: a.) a more detailed discussion of the research results - here, it would be important to review the results in context of table 1 (rather than comparing the MP contamination found in the Northern hogsucker with data coming from other parts of the world) - meaning a comparison with published results stemming from the Eastern part of the US. Also, other reports of MP presence from Appalachia and surrounding area may be helpful to give a background to the study. For instance, in the discussion of the possibility of atmospheric entry of MPs, the following paper could be used as a reference: Atmospheric Deposition of Microplastics in South Central Appalachia in the United States (Adam Elnahas, Austin Gray, Jennie Lee, Noora AlAmiri, Nishan Pokhrel, Steve Allen, and Hosein Foroutan, ACS ES&T Air 2025 2 (1), 64-72. DOI: 10.1021/acsestair.4c00189). Also, a more detailed description of the individual sites and their surrounding area would be beneficial for the reader. 
1. Thank you for your feedback! We agree that a review of eastern US microplastic pollution within fishes would be ideal but as we point out there is a research gap in the investigation of microplastics in freshwater fish so we were unable to find many comparable studies in similar environments and conducted a more holistic review globally. We included the paper you suggest in the discussion (line 322) which greatly improves experimental context, thank you for the tip! We have also included an environmental variable table (lines 110-111) to give more context to the individual sites and their surrounding areas.
2. Figures 1 and 2 should be interchanged in the text.
1. Thank you for catching that! We moved figure 1 to the methods section for increased clarity.
3. Is there any more information on the MPs - such as color distribution? chemical make-up?
1. For this study the only information we recorded about the microplastics themselves were type, count, and length, though we hope to expand data collection in future iterations!
4. The make of the dissecting microscope should be given (in line 112).
1. Added: Amscope 3.5X-180X Trinocular Stereo Zoom Microscope
5. Looking at Figure 3B, it seems that the data is mostly affected by the variation of MP length ingested by fish of 10-12 cm in length. Is the breakdown of fish size equal over all sites or do fish of a specific length come from mostly one site?
1. All fish, from all sites were juveniles varying is total length between 5.6-12.8mm.  There were no apparent patterns in fish length among sites. All of the streams we sampled contained adult fish (common length per FishBase.org https://www.fishbase.se/summary/SpeciesSummary.php?ID=2990&AT=northern+hog+sucker, is 30cm), while it would have been interesting to have sampled these larger fishes, they were not part of the collection of fish we had access to.  Plus, by focusing on early life fish (estimated 1.5-2.5 years old) we assess recent feeding events, less accumulation, which we feel is a separate question.
6. This sentence from appendix A is not readily understood:  A total of 498 microplastic particles were identified in control groups, ... Are these the blank samples? If so, the number is very high.
1. Correct, these are the blank samples and although 500 environmental microplastics sounds like a lot, that is only about 9 particles per control on average compared to about 40 per sample. Unfortunately, a minimal level of contamination was difficult to mitigate in the absence of a clean room but it is accounted for by the controls. 
7. The authors write in line 270: excepting those with extremely stringent protocol [38]. Does this mean to say that globally and in general the reported MP counts are too high because of external MP contamination?  
1. We meant that the studies performed with extremely stringent protocol, ie intensive digestion/chemical analysis processes often find low levels of microplastic contamination as the cited study states. However, this is not consistent with the majority of global research we reviewed and could potentially be due to degradation of plastics because of such aggressive/selective filtration and enumeration techniques. We added an explicit statement about digestion and enumeration (line 294) to clarify, thank you for pointing that out!

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

The authors have addressed the points of the reviewer. For the future, a spectroscopic analysis in regard to the nature of the plastic of the MPs is recommended.