the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 24 Feb 2026
Future mercury levels in fish: model vs. observational predictions under different policy scenarios
Abstract. Mercury (Hg) poses a global threat due to its long-range transport and transformation into methylmercury (MeHg), a potent neurotoxin that bioaccumulates in aquatic food webs. While global and regional efforts to reduce anthropogenic Hg emissions are ongoing, the implications of these policies for future Hg deposition and consequent MeHg levels in fish remain uncertain. This study synthesizes published modeling studies to examine projected relationships among Hg emissions, atmospheric deposition, and lake fish MeHg concentrations in 2050 under various policy scenarios. While models reveal a strong linear relationship between emissions and deposition (R² = 0.79), and a moderate correlation between Hg deposition and fish MeHg (R² = 0.63), these trends contrast with observational data, which often show nonlinear or more complex responses. Modeled atmospheric deposition and lake area emerged as key predictors, with higher deposition and smaller lakes associated with higher modeled fish MeHg levels. Notably, despite wide variation in model structures, including differences in atmospheric chemistry, emission inventories, legacy emissions, meteorological drivers, methylation-demethylation kinetics, and food web dynamics, the linear trends persisted. This apparent linearity underscores robust large-scale cause–effect patterns but also calls for caution: do current models truly capture the complexity of real atmospheric and ecosystem processes, or might they oversimplify mercury’s nonlinear cycling and ecological responses? These findings highlight the need to remain open to processes and interactions not yet fully represented in models, ensuring that future mercury assessments and policy strategies reflect the true complexity of natural systems.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(2193 KB)
-
Supplement
(323 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2193 KB) - Metadata XML
-
Supplement
(323 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2026-730', Anonymous Referee #1, 02 Mar 2026
- AC1: 'Reply on RC1', Henna Gull, 27 Mar 2026
-
RC2: 'Comment on egusphere-2026-730', David Amptmeijer, 24 Mar 2026
The manuscript highlights the important and timely issue that MeHg bioaccumulation models for lakes underestimate natural variability and tend to predict overly linear responses to changes in Hg deposition compared to what is observed in nature.
There are, however, several points where the manuscript could be improved. Most notably, several terms and descriptions would benefit from clearer wording, and because the manuscript focuses strongly on identifying linearity in models versus observations, I believe the discussion would be strengthened by a more in‑depth examination of why this linearity arises in current models and whether this is necessarily problematic.
As such I would recommend that the paper undergoes major revision before publication. I have attached a PDF with my detailed comments and suggestions for revision.
While my suggestions are substantial, I genuinely enjoyed reading the manuscript and consider its core point a valuable and timely observation.
-
AC2: 'Reply on RC2', Henna Gull, 16 Apr 2026
We sincerely thank the reviewer for the detailed, thoughtful, and constructive comments, and for the positive assessment of our work. We greatly appreciate the time and effort invested in the review. Please find attached our detailed point-by-point response to the review comments.
-
AC2: 'Reply on RC2', Henna Gull, 16 Apr 2026
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2026-730', Anonymous Referee #1, 02 Mar 2026
General overview of the manuscript:
This manuscript synthesizes outputs from ten modeling studies to estimate future mercury (Hg) emissions, deposition, and (in four studies) projected fish MeHg concentrations in 2050 under sixteen climate policy scenarios. The effort to integrate across models and policy scenarios is valuable and timely, particularly given ongoing international assessments of mercury mitigation effectiveness.
While the study provides useful comparative insights, several areas would benefit from clarification, particularly regarding model evaluation, geographic scope, and presentation of the tables and figure. While many of the models considered produce global outputs, the analysis focuses on a relatively small number of lakes in the United States due to the regional limitations of the fish MeHg bioaccumulation models. Additionally, comparisons between modelled and observed trends in fish MeHg are not presented for most of the select lakes, nor is there any comparison offered in the other environmental media considered in the bioaccumulation models. Finally, the compiled datasets are not provided in the Supplementary Information, which may limit their utility for future large-scale synthesis efforts (e.g., those associated with the international Minamata effectiveness evaluations).
Comments on the Abstract:
The abstract reads well. One particularly interesting result is the strength of the relationship between projected changes in fish Hg and atmospheric deposition. However, since all fish Hg concentrations are model-derived, it would be helpful to clarify to what extent this similarity reflects structural assumptions within the coupled atmospheric–bioaccumulation models (for example, assumptions regarding response times of fish MeHg to changes in atmospheric inputs). Without comparisons to observed deposition and fish MeHg relationships (e.g., using isotope approaches or long-term monitoring datasets), it is difficult to assess how independently this relationship emerges from the models versus being constrained by their structure. Some brief discussion of this point would strengthen the interpretation.
Additionally, the final two sentences adopt a cautionary tone regarding model limitations that feels somewhat abrupt. While it is important to acknowledge that models cannot capture the full complexity of atmospheric and ecosystem processes, the abstract does not specify which processes or interactions are currently missing or how future work might address these gaps. As such, the conclusion reads slightly general and somewhat disconnected from the specific findings of the study. The authors may wish to either briefly specify the key processes that require improved representation, or reframe the concluding sentences to more directly emphasize the primary contribution of the study and avoid commenting on these limitations within the abstract. This would help the abstract end on a more clear and cohesive note.
Comments on the Introduction:
The Introduction reads clearly and covers much of the relevant literature. However, additional detail regarding the atmospheric modeling frameworks would be helpful. For example, do the emissions inventories primarily reflect anthropogenic sources, or do they also incorporate natural re-emissions under changing temperature and precipitation conditions in the climate scenarios? A brief clarification would improve transparency for readers less familiar with these modeling systems.Major comments on the body of the text:
Lines 109–110:
The authors note that fish MeHg projections were geographically restricted. It would be helpful to clarify whether this restriction reflects limitations in available published model outputs or inherent constraints of the bioaccumulation models themselves. My understanding is that models such as BASS and SERAFM can be parameterized for different regions if sufficient local ecological and biogeochemical data are available. A brief explanation of why projections were limited to these specific regions would improve clarity.Line 111:
The manuscript refers to a “detailed evaluation” of the models used to simulate Hg deposition and fish MeHg bioaccumulation. Could the authors elaborate on what this evaluation involved beyond generating the summary values presented in Tables 1–3? For example, were any statistical comparisons conducted between model outputs and historical observational datasets (atmospheric Hg, deposition, soil/sediment Hg, or fish MeHg)? If such comparisons were not possible due to the forward-looking 2050 projections, explicitly stating this limitation would be helpful.Related to this, Lines 121–122 indicate that multiple linear regression models were used to test whether lake characteristics explain variation in projected fish MeHg changes. Are these lake characteristics projected for 2050 as well? If so, this suggests that the regression examines relationships among modeled 2050 variables. Clarifying this would help readers interpret the independence and robustness of the statistical analysis.
Table 2 (Policy-in-action scenario):
There are two rows for Walleye from the New York Adirondacks with different % changes in fish MeHg in 2050, despite identical base years, references, region, and species. Could the authors clarify what distinguishes these two rows? Additionally, deposition change is listed as −29% for the Adirondacks overall, yet a value of +12% appears in the bottom two rows for Walleye. Some clarification of how these values relate would improve interpretability.Also related to Tables 1–3: some models listed have geographic coverage outside the primary U.S. study area (e.g., India, China, South Pacific, Arctic). It would be helpful to explain how outputs from these regions were incorporated into the present analysis and why they are relevant to the study’s geographic scope.
Lines 157–164:
The number of references listed for the different policy scenarios somewhat interrupts the flow of the paragraph. It may improve readability to refer readers primarily to Tables 1–3 and streamline the in-text citations. However, some references cited here (e.g., Sunderland and Selin 2013; Geyman et al. 2024, 2025) do not appear in the tables. If these are particularly important for contextual interpretation, perhaps a brief explanation of their specific relevance would clarify their inclusion.Lines 243–257:
This is a strong section, and it is helpful to see discussion of discrepancies between modeled and observed values. For the specific U.S. regions examined in this study, are there atmospheric deposition or elemental Hg observations available that could be used for comparison? Even brief contextual comparison of observed vs. model predicted values would strengthen confidence in the modeled outputs and their interpretations.Lines 258–266:
The discussion of hemispheric discrepancies appears to rely largely on a single Southern Hemisphere example. If observational data from the U.S. or broader Northern Hemisphere are available, incorporating them (even briefly) would provide a more balanced perspective to this section.Lines 267 and 279:
Section titles 3.2 and 3.3 are somewhat non-descriptive, and the placement of “, 2050” at the end reads awkwardly. Revising the titles for clarity and flow would improve readability.Line 334:
The phrase “two integrated data points” is unclear. Are these integrated data points associated with specific lakes or regions? A brief clarification would help readers who are unfamiliar with the cited study (Perlinger 2018) to understand the distinction. Related to this sentence, the mention of an additional five integrated data points on line 226 under the high deposition reduction scenario do not have a citation – are these also from the same Perlinger study or from the Knightes 2009 study that is referenced in the sentence following this statement?Lines 357–358 and Line 408:
Line 408 indicates that systems with larger lake area and higher wetland coverage show greater declines in fish MeHg. Meanwhile, lines 357–358 suggest that fish from smaller lakes may respond more rapidly to atmospheric Hg changes. These statements appear somewhat at odds. Clarifying the reason for these discrepancies would help with the interpretation, as well as some evaluation of how well robust these PCA results are.Lines 431–434 (and Line 408):
The negative relationship between wetland extent and predicted fish MeHg change is intriguing and contrasts with many field observations that associate wetlands with enhanced methylation. Given that this relationship appears consistently across multiple models, it would be valuable to explore potential mechanistic explanations in greater detail. For example, do model assumptions regarding hydrology, dilution, or loading pathways drive this outcome? Additional discussion and statistical analysis would strengthen confidence in the PCA results and clarify how they should be interpreted relative to observed trends in empirical studies.Notes on Tables and Figures
Tables 1–3:
These tables contain important compiled information but are currently difficult to interpret. Suggestions:- Define SRES in each table caption, as opposed to a row header in Table 1 only.
- Consider streamlining columns to improve readability (for example, removing the “Scenario” column since scenario is already defined in row headers).
- Instead of grouping the model projections by policy pathways (minimal control, intermediate mitigation, maximum mitigation), I suggest that the projections are separated into two tables, with one table containing models that provide deposition/ emission only outputs and another table containing models with both deposition and fish MeHg outputs. All policy pathways with projected data should be included in each of the two tables, respectively.
- I am unsure whether formatting changes occurred during manuscript upload, but I am having difficulty interpreting rows where a single reference appears to correspond to both a deposition model and one or more fish MeHg models. For example, in Table 3 under the Aspirational scenario, “Hendrick’s LMB model” is listed in the first row corresponding to the Great Lakes region; however, no species or % change in MeHg values are listed within that row. All fish species and associated % MeHg changes appear to be listed only for the Michigan Upper Peninsula lakes. Is it correct to interpret this as meaning that the Perlinger et al. (2018) dataset does not include fish MeHg outputs for the broader Great Lakes or New York Adirondacks regions? If so, this could be clarified explicitly in the table. If not, additional explanation is needed to distinguish how the deposition model output relates to the fish MeHg outputs presented below.
Figure 1:
Clear and informative. It may improve flow if presented before the tables so readers can orient themselves spatially prior to reviewing tabulated data. Aligning legend order with caption order (e.g. placing NPS above MFR in the figure legend) may also improve clarity.Figure 2:
Interesting figure, though its centrality to the manuscript could be reconsidered. The authors may wish to consider whether it belongs in the main text or Supplementary Material.Table 4:
The technical comparison of model structures is valuable but may be more appropriate for Supplementary Material, particularly since the main text (Lines 180–195) summarizes the most relevant distinctions.Figure 5:
This figure is somewhat difficult to interpret in its current form, and the caption does not provide enough detail for it to stand alone. The associated text (Lines 356–361) communicates the key message clearly without needing to reference the figure. The authors may wish to reconsider whether this figure adds sufficient clarity in its present format. If the exact values plotted in Figure 5 are captured in your Tables 1-3 , the authors may want to reference the appropriate table within the associated text.Very Minor Notes
Lines 200–203:
It may help to very briefly clarify that emissions and deposition estimates are derived from the same reference for each region.Line 290:
“Dynamic Mercury Cycling Mode” should likely read “Model.”Citation: https://doi.org/10.5194/egusphere-2026-730-RC1 - AC1: 'Reply on RC1', Henna Gull, 27 Mar 2026
-
RC2: 'Comment on egusphere-2026-730', David Amptmeijer, 24 Mar 2026
The manuscript highlights the important and timely issue that MeHg bioaccumulation models for lakes underestimate natural variability and tend to predict overly linear responses to changes in Hg deposition compared to what is observed in nature.
There are, however, several points where the manuscript could be improved. Most notably, several terms and descriptions would benefit from clearer wording, and because the manuscript focuses strongly on identifying linearity in models versus observations, I believe the discussion would be strengthened by a more in‑depth examination of why this linearity arises in current models and whether this is necessarily problematic.
As such I would recommend that the paper undergoes major revision before publication. I have attached a PDF with my detailed comments and suggestions for revision.
While my suggestions are substantial, I genuinely enjoyed reading the manuscript and consider its core point a valuable and timely observation.
-
AC2: 'Reply on RC2', Henna Gull, 16 Apr 2026
We sincerely thank the reviewer for the detailed, thoughtful, and constructive comments, and for the positive assessment of our work. We greatly appreciate the time and effort invested in the review. Please find attached our detailed point-by-point response to the review comments.
-
AC2: 'Reply on RC2', Henna Gull, 16 Apr 2026
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 1,135 | 666 | 87 | 1,888 | 181 | 58 | 130 |
- HTML: 1,135
- PDF: 666
- XML: 87
- Total: 1,888
- Supplement: 181
- BibTeX: 58
- EndNote: 130
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Henna Gull
Ju Hyeon Lee
Hoin Lee
Hélène Angot
Sae Yun Kwon
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2193 KB) - Metadata XML
-
Supplement
(323 KB) - BibTeX
- EndNote
- Final revised paper
General overview of the manuscript:
This manuscript synthesizes outputs from ten modeling studies to estimate future mercury (Hg) emissions, deposition, and (in four studies) projected fish MeHg concentrations in 2050 under sixteen climate policy scenarios. The effort to integrate across models and policy scenarios is valuable and timely, particularly given ongoing international assessments of mercury mitigation effectiveness.
While the study provides useful comparative insights, several areas would benefit from clarification, particularly regarding model evaluation, geographic scope, and presentation of the tables and figure. While many of the models considered produce global outputs, the analysis focuses on a relatively small number of lakes in the United States due to the regional limitations of the fish MeHg bioaccumulation models. Additionally, comparisons between modelled and observed trends in fish MeHg are not presented for most of the select lakes, nor is there any comparison offered in the other environmental media considered in the bioaccumulation models. Finally, the compiled datasets are not provided in the Supplementary Information, which may limit their utility for future large-scale synthesis efforts (e.g., those associated with the international Minamata effectiveness evaluations).
Comments on the Abstract:
The abstract reads well. One particularly interesting result is the strength of the relationship between projected changes in fish Hg and atmospheric deposition. However, since all fish Hg concentrations are model-derived, it would be helpful to clarify to what extent this similarity reflects structural assumptions within the coupled atmospheric–bioaccumulation models (for example, assumptions regarding response times of fish MeHg to changes in atmospheric inputs). Without comparisons to observed deposition and fish MeHg relationships (e.g., using isotope approaches or long-term monitoring datasets), it is difficult to assess how independently this relationship emerges from the models versus being constrained by their structure. Some brief discussion of this point would strengthen the interpretation.
Additionally, the final two sentences adopt a cautionary tone regarding model limitations that feels somewhat abrupt. While it is important to acknowledge that models cannot capture the full complexity of atmospheric and ecosystem processes, the abstract does not specify which processes or interactions are currently missing or how future work might address these gaps. As such, the conclusion reads slightly general and somewhat disconnected from the specific findings of the study. The authors may wish to either briefly specify the key processes that require improved representation, or reframe the concluding sentences to more directly emphasize the primary contribution of the study and avoid commenting on these limitations within the abstract. This would help the abstract end on a more clear and cohesive note.
Comments on the Introduction:
The Introduction reads clearly and covers much of the relevant literature. However, additional detail regarding the atmospheric modeling frameworks would be helpful. For example, do the emissions inventories primarily reflect anthropogenic sources, or do they also incorporate natural re-emissions under changing temperature and precipitation conditions in the climate scenarios? A brief clarification would improve transparency for readers less familiar with these modeling systems.
Major comments on the body of the text:
Lines 109–110:
The authors note that fish MeHg projections were geographically restricted. It would be helpful to clarify whether this restriction reflects limitations in available published model outputs or inherent constraints of the bioaccumulation models themselves. My understanding is that models such as BASS and SERAFM can be parameterized for different regions if sufficient local ecological and biogeochemical data are available. A brief explanation of why projections were limited to these specific regions would improve clarity.
Line 111:
The manuscript refers to a “detailed evaluation” of the models used to simulate Hg deposition and fish MeHg bioaccumulation. Could the authors elaborate on what this evaluation involved beyond generating the summary values presented in Tables 1–3? For example, were any statistical comparisons conducted between model outputs and historical observational datasets (atmospheric Hg, deposition, soil/sediment Hg, or fish MeHg)? If such comparisons were not possible due to the forward-looking 2050 projections, explicitly stating this limitation would be helpful.
Related to this, Lines 121–122 indicate that multiple linear regression models were used to test whether lake characteristics explain variation in projected fish MeHg changes. Are these lake characteristics projected for 2050 as well? If so, this suggests that the regression examines relationships among modeled 2050 variables. Clarifying this would help readers interpret the independence and robustness of the statistical analysis.
Table 2 (Policy-in-action scenario):
There are two rows for Walleye from the New York Adirondacks with different % changes in fish MeHg in 2050, despite identical base years, references, region, and species. Could the authors clarify what distinguishes these two rows? Additionally, deposition change is listed as −29% for the Adirondacks overall, yet a value of +12% appears in the bottom two rows for Walleye. Some clarification of how these values relate would improve interpretability.
Also related to Tables 1–3: some models listed have geographic coverage outside the primary U.S. study area (e.g., India, China, South Pacific, Arctic). It would be helpful to explain how outputs from these regions were incorporated into the present analysis and why they are relevant to the study’s geographic scope.
Lines 157–164:
The number of references listed for the different policy scenarios somewhat interrupts the flow of the paragraph. It may improve readability to refer readers primarily to Tables 1–3 and streamline the in-text citations. However, some references cited here (e.g., Sunderland and Selin 2013; Geyman et al. 2024, 2025) do not appear in the tables. If these are particularly important for contextual interpretation, perhaps a brief explanation of their specific relevance would clarify their inclusion.
Lines 243–257:
This is a strong section, and it is helpful to see discussion of discrepancies between modeled and observed values. For the specific U.S. regions examined in this study, are there atmospheric deposition or elemental Hg observations available that could be used for comparison? Even brief contextual comparison of observed vs. model predicted values would strengthen confidence in the modeled outputs and their interpretations.
Lines 258–266:
The discussion of hemispheric discrepancies appears to rely largely on a single Southern Hemisphere example. If observational data from the U.S. or broader Northern Hemisphere are available, incorporating them (even briefly) would provide a more balanced perspective to this section.
Lines 267 and 279:
Section titles 3.2 and 3.3 are somewhat non-descriptive, and the placement of “, 2050” at the end reads awkwardly. Revising the titles for clarity and flow would improve readability.
Line 334:
The phrase “two integrated data points” is unclear. Are these integrated data points associated with specific lakes or regions? A brief clarification would help readers who are unfamiliar with the cited study (Perlinger 2018) to understand the distinction. Related to this sentence, the mention of an additional five integrated data points on line 226 under the high deposition reduction scenario do not have a citation – are these also from the same Perlinger study or from the Knightes 2009 study that is referenced in the sentence following this statement?
Lines 357–358 and Line 408:
Line 408 indicates that systems with larger lake area and higher wetland coverage show greater declines in fish MeHg. Meanwhile, lines 357–358 suggest that fish from smaller lakes may respond more rapidly to atmospheric Hg changes. These statements appear somewhat at odds. Clarifying the reason for these discrepancies would help with the interpretation, as well as some evaluation of how well robust these PCA results are.
Lines 431–434 (and Line 408):
The negative relationship between wetland extent and predicted fish MeHg change is intriguing and contrasts with many field observations that associate wetlands with enhanced methylation. Given that this relationship appears consistently across multiple models, it would be valuable to explore potential mechanistic explanations in greater detail. For example, do model assumptions regarding hydrology, dilution, or loading pathways drive this outcome? Additional discussion and statistical analysis would strengthen confidence in the PCA results and clarify how they should be interpreted relative to observed trends in empirical studies.
Notes on Tables and Figures
Tables 1–3:
These tables contain important compiled information but are currently difficult to interpret. Suggestions:
Figure 1:
Clear and informative. It may improve flow if presented before the tables so readers can orient themselves spatially prior to reviewing tabulated data. Aligning legend order with caption order (e.g. placing NPS above MFR in the figure legend) may also improve clarity.
Figure 2:
Interesting figure, though its centrality to the manuscript could be reconsidered. The authors may wish to consider whether it belongs in the main text or Supplementary Material.
Table 4:
The technical comparison of model structures is valuable but may be more appropriate for Supplementary Material, particularly since the main text (Lines 180–195) summarizes the most relevant distinctions.
Figure 5:
This figure is somewhat difficult to interpret in its current form, and the caption does not provide enough detail for it to stand alone. The associated text (Lines 356–361) communicates the key message clearly without needing to reference the figure. The authors may wish to reconsider whether this figure adds sufficient clarity in its present format. If the exact values plotted in Figure 5 are captured in your Tables 1-3 , the authors may want to reference the appropriate table within the associated text.
Very Minor Notes
Lines 200–203:
It may help to very briefly clarify that emissions and deposition estimates are derived from the same reference for each region.
Line 290:
“Dynamic Mercury Cycling Mode” should likely read “Model.”