Spatial differentiation of carbon-control structures in urban coastal ecosystems under nutrient enrichment

Kawahito, Kohei; Okada, Daichi; Sohma, Akio

doi:10.5194/egusphere-2026-2710

Preprints

https://doi.org/10.5194/egusphere-2026-2710

Preprints

04 Jun 2026

| 04 Jun 2026

Spatial differentiation of carbon-control structures in urban coastal ecosystems under nutrient enrichment

Kohei Kawahito, Daichi Okada, and Akio Sohma

Abstract. Urban coastal ecosystems, which are strongly influenced by human activities and elevated nutrient inputs, can contribute to climate-change mitigation through carbon uptake, fixation, and storage. However, it remains unclear how the underlying carbon-control structures governing these functions respond to nutrient enrichment, owing to the interactions of multiple biogeochemical processes across pelagic and benthic systems. In this study, we applied the benthic–pelagic coupled ecosystem model EMAGIN-B.C. ver. 2 to Tokyo Bay to examine how nutrient loading influences carbon cycling in urban coastal environments. To interpret these responses mechanistically, carbon cycling was organised into three paired flux balances representing carbon uptake (A/R), carbon fixation (F/U), and carbon storage (S/D). These functional pairs were analysed within a conceptual framework of a Dual Carbon Loop consisting of organic and carbonate pathways. Model simulations demonstrate that increased nutrient loading enhances atmospheric CO₂ uptake and pelagic carbon fixation. However, the dominant mechanisms controlling carbon cycling differ substantially across regions. In the estuarine region, nutrient enrichment amplified pelagic primary production, resulting in simultaneous increases in carbon uptake, fixation, and organic carbon burial. In the central bay, production and remineralization are intensified in tandem, indicating strong internal coupling of carbon fluxes. In contrast, tidal flats exhibited a transformation-dominated response wherein externally supplied organic matter was rapidly processed by benthic communities, thereby limiting net pelagic fixation while maintaining relatively stable carbonate storage and producing a system characterised more by transport, transformation, and redistribution than by new production. These contrasting responses indicate that identical nutrient forcing can produce distinct carbon-cycling behaviours depending on the regional ecosystem structure. We interpret these patterns as spatial differentiation of carbon-control structures governed by the relative balance of opposing carbon fluxes within the Dual Carbon Loop system. The proposed framework provides a mechanistic basis for understanding how nutrient management influences climate-mitigation functions in urban coastal ecosystems and offers a perspective for analysing shallow, nutrient-enriched coastal systems characterised by strong benthic–pelagic biogeochemical coupling.

Received: 13 May 2026 – Discussion started: 04 Jun 2026

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Kohei Kawahito, Daichi Okada, and Akio Sohma

Status: final response (author comments only)

RC1:
'Comment on egusphere-2026-2710', Anonymous Referee #1, 21 Jul 2026

Comments to Authors
This is an interesting paper examines how nutrient loading influences carbon cycling in Tokyo Bay, which is one of the typical examples of eutrophic, urban coastal water in Japan, by using a novel ecological model developed by authors. The authors define three carbon functions as carbon uptake, fixation, and storage and established a conceptual framework of a “Dual Carbon Loop” consisting of organic and carbonate pathways. They analytically demonstrated the dominant mechanisms of carbon cycling differ substantially across three regions: the estuarine regions directly influenced by river discharge, the central bay, and the tidal flat area rich in benthic organisms.
Main results of the contrast response in these regions to nutrient enrichment are remarkable.
Such findings are quite valuable and unique for the general readers of EGUsphere. The manuscript is well written and quite understandable. Thus, the reviewer suggests that this is acceptable, after slightly minor points to be corrected or incorporated as below.
The followings are some major or specific points that are to be considered for further revision.
1. General information of Tokyo Bay
The reviewer supposes that general readers of EGUsphere are unfamiliar with Tokyo Bay. So, its geographical/physical information such as the mean water depth, mean retention time, tidal range, and population of its watershed, as well as biological activity are to be presented in an appropriate place in the manuscript. Especially, the tidal flat information that the bay has relatively large sandy tidal flat rich in bivalve at least up to around 2000 would be important, because it explains why the authors consider tidal-flat region as one of the characteristic regions. It is strongly related to the relative preference of the “carbonate pathway” for the storage function in tidal-flat system, which is one of the most important results of this analysis.
2. Model functional forms and parameters
The authors extended the original EMAGIN-B.C. model with some improvements as shown in Line 139-142. In this manuscript, however, information on functional forms of physicochemical processes and the values of model parameters is not presented. At least, reference would be necessary.
Additionally, the reviewer was not able to catch carbon cycling after mortality of benthos. Flesh of suspension feeders will turn to be labile organic carbon and shell CaCO₃ will be produced autonomously?
Additionally, how to evaluate partial pressure of CO₂ in the air appeared in Eq (1). Constant value or seasonally different?
3. Introduction
Description on the historical regulation policy including total volume control for pollutants (TN and TP were added as the target since 2001, the beginning year of the of 5^th total volume control) and recent social experiments on artificial seasonal nutrient supply from sewage treatment plants from Line 46 to Line 51 seems unclear. In this study, around the year 2000 was the baseline of the analysis. In this sense, more detailed description would be necessary, in order to make clearer the objectives and target period of this analysis. Also, up to our knowledge, nutrient management plan recently established in Hyogo and Aichi prefectures are not directly aimed to climate-change mitigation as described in Line 53.
For general readers of EGUsphesre, additional description on recent nutrient management policy in Japan would be informative, e.g., Uehara and Hidaka (Ocean and Coastal Management 244 (2023)).
4. Chapt.4: Model validation
In Line 206, the authors describe that “the correlations for PP and POC in the bottom layer are relatively low”. However, PP (Bottom) are missing both for Estuarine region and Central bay region in Figure 5. The reviewer wonders that such discrepancies may lead to inaccurate estimation of the burial rate and storage function.
5. Analysis
In this study, results of a kind of the sensitivity analysis with varying nutrient loading such as DIN and DIP are described precisely. How about organic loadings? They are kept constant in the analysis?
6. Temporal variation of model state variables
Main results of the temporal variations of model variables expressed as 1-day moving average are shown in Figures 11, 14, and 17. Each figure shows almost monthly or semi-monthly variations. Why was such a somewhat regular fluctuating pattern calculated?
Interesting features are found for the carbon storage flux, as shown in Figures 13 (Estuarine region) and 16 (Central-bay region). The amount of “organic carbon storage” seems almost constant temporally, whereas “CaCO₃ storage” slightly varies corresponding to the timing of abrupt decline of suspension-feeders and deposit feeders, which may be the results of damages caused by hypoxia. These differences among the two storage pathways may be the results that the total depth of the sediment layer is deep enough for organic carbon distribution, but not for CaCO₃. Different characteristics in the seasonal variations of these storage rates are found for the tidal-flat system. Both carbon and CaCO₃ burial rates shows seasonal, almost sinusoidal variations as shown in Figure 19. Are these different temporal characteristics are originated from the effect of bio-turbation, or other mechanics?
7. Results of the tidal-flat region
As for carbon fixation function in tidal-flat region, the authors described in Line 772-774 that fixation is affected by hypoxia, as well as in the summary section Line 798-799. However, judging from DO concentration shown in figures 17 and 19, bottom DO seems high enough even in the bottom layer. Therefore, the reviewer does not understand such description. Additionally, the authors summarize that the tidal-flat region is characterized as transformation-dominated type: externally supplied organic matter is transformed and remineralized within the benthic system (Line 834). The reviewer well agreed this kind of general characteristics of sandy tidal flat rich in bivalve. However, the reviewer has difficulty to find out from which part of the manuscript can the authors conclude that the “externally supplied organic matter is transformed” is originated.
8. Minor points
(1) Unit of model variables and parameters
Model state variables and their units are listed in Table 1. However, related units in Figures 10 to 19 are not always the same. For example, units of state variables are in molar basis in Figures 11, 14, and 17, whereas these are in g basis for Table 1. Unification of units would be strongly recommended.
Additionally, units of biomass of benthos are expressed as m-2 in Figure 17, but cm-1 in Figures 16 and 19. These are all in cm-2 units? Please check them carefully.
(2) Section 2.1.4: Fractions of organic matter
Three fractions of organic matter are described, based on their different degradability. The term “Multi-G model for organic carbon degradation” would be more popular.
(3) Fixation-enhancement (-reduction) or Fixation-enhancing (-reducing)
“Fixation-enhancement” or “Fixation-reduction” is used in the semi-headings of 6.2.1 and 6.2.2, whereas “Fixation-enhancing” or “Fixation-reducing” is used in Figure 8. The same mixed terminology is found for Figure 8 and the semi-headings of 6.3.1 and 6.3.2. Please unify the terminology.

Citation: https://doi.org/10.5194/egusphere-2026-2710-RC1
- AC1: 'Reply on RC1', Akio Sohma, 02 Sep 2026
  
  General response
  We sincerely thank the reviewer for the positive and encouraging assessment of our manuscript and for the constructive suggestions for improving its clarity and scientific presentation.
  1. General information of Tokyo Bay
  We thank the reviewer for this valuable suggestion. In the revised manuscript, we will add concise geographical, physical, and ecological information on Tokyo Bay in the Introduction, including its surface area, mean water depth, water residence time, tidal range, and watershed population. We will also describe the spatial heterogeneity of the bay, including its river-influenced estuarine waters, seasonally stratified central-bay waters, and shallow sandy tidal flats.
  In particular, we will clarify that the tidal-flat region represented in the model corresponds to Banzu Tidal Flat, which supported substantial populations of the Manila clam Ruditapes philippinarum during the baseline period of this study. We will also add information on the active filtration and ingestion of suspended particulate matter by this bivalve population, together with the relevant references.
  We will further connect this ecological setting explicitly to the interpretation of the tidal-flat results. In the revised discussion of carbon storage, we will explain that the carbonate-dominant storage structure of the tidal-flat region is associated primarily with suspension-feeder biomass, shell formation, and the subsequent transfer of pre-existing shell CaCO₃ following mortality and benthic turnover. This provides the ecological basis for selecting the tidal-flat region as one of the three contrasting environments analyzed in this study.
  2. Model functional forms and parameters
  We thank the reviewer for pointing out that the sources of the model formulations and parameter values, the fate of benthic-faunal carbon following mortality, and the treatment of atmospheric pCO₂ were not sufficiently clear in the original manuscript.
  First, in the revised manuscript, we will clarify the sources of the functional forms and parameter values used in EMAGIN-B.C. ver. 2.0. The functional forms and fitted parameters describing the pH-dependent physiological and metabolic responses are provided in Omachi and Sohma (2022), and their application to Tokyo Bay is described in Omachi and Sohma (2023). The ATP-based formulation of organic-matter remineralization, including its governing formulation and parameter definitions, is provided in Ishizuka and Sohma (2025). We will also clarify that the basic formulations of the original pelagic-benthic ecosystem model follow Sohma et al. (2008), whereas the carbonate-system extension follows Sohma et al. (2018).
  Second, we will explicitly describe the fate of benthic-faunal carbon following mortality. The soft-tissue organic carbon of benthic fauna consists of fast-labile, slow-labile, and refractory cellular organic fractions. Upon mortality, these fractions are transferred to the corresponding detrital state variables while retaining the prescribed fractional composition of the living biomass. Thus, the soft tissue is not transferred entirely to the fast-labile organic-carbon pool. For suspension feeders, shell CaCO₃ associated with living organisms is diagnostically related to soft-tissue carbon biomass using a prescribed soft-tissue-carbon-to-shell-CaCO₃-carbon ratio. Upon mortality, the shell CaCO₃ that was already formed during the lifetime of the organisms is transferred to the benthic CaCO₃ pool as dead shell material. Mortality therefore does not generate new CaCO₃.
  Finally, we will clarify the treatment of atmospheric pCO₂. In the present simulations, atmospheric pCO₂ is prescribed as a spatially and temporally constant value of 730 µatm throughout the annual cycle, and the same value is applied to all nutrient-loading scenarios. The air-sea CO₂ flux is calculated from the difference between this prescribed atmospheric pCO₂ and the simulated surface-water pCO₂.
  3. Introduction
  We thank the reviewer for this important comment. We agree that the original description did not clearly distinguish among the historical pollutant-load reduction policies, recent nutrient-management initiatives, and the climate-related perspective examined in this study.
  In the revised manuscript, we will provide a clearer historical and policy context. We will explain that the Total Pollutant Load Control System was introduced in 1979, initially targeting chemical oxygen demand (COD), and that total nitrogen (TN) and total phosphorus (TP) were subsequently added as target substances under the fifth Total Pollutant Load Control program initiated in 2001.
  We will also clarify that recent trials of seasonal nutrient-management operations at wastewater treatment plants and other nutrient-supply measures in parts of the Seto Inland Sea and Ise-Mikawa Bay primarily aim to reconcile water-quality conservation with the maintenance or recovery of biological and fisheries productivity, rather than directly to mitigate climate change. Accordingly, the present study will be framed as an evaluation of the potential climate-mitigation co-benefits and trade-offs associated with the changes in nutrient loading, rather than as an evaluation of nutrient supply itself as a climate-change mitigation measure. We will also add the reference suggested by the reviewer, Uehara and Hidaka (2023), together with relevant policy references.
  We will further clarify the temporal interpretation of the analysis. The annual forcing cycle was constructed from observational data for 1998-2002 and repeatedly applied until the model reached a periodic steady state. The resulting state is interpreted as representative of the typical seasonal ecosystem conditions of Tokyo Bay around the year 2000. This observational period spans the introduction of TN and TP control under the fifth Total Pollutant Load Control program but predates the more recent nutrient-supply initiatives described above.
  Finally, we will explicitly state that the simulations are not intended to reproduce a specific calendar year or the effects of a particular nutrient-management policy. Instead, idealized nutrient-loading scenarios are applied to this common baseline, while the other forcing conditions are held common among scenarios, to isolate ecosystem responses to changes in external nutrient inputs.
  4. Chapter 4: Model validation
  We thank the reviewer for pointing out the absence of bottom-layer phytoplankton comparisons in the original Fig. 5 and for raising the important question of how the relatively weak reproduction of bottom-layer biological and particulate-carbon variables may affect the estimated carbon-storage function.
  In the revised manuscript, we will first clarify the terminology by replacing the abbreviation “PP” with “PHY” wherever phytoplankton carbon biomass is intended, to avoid confusion with primary production. We will also clarify that the quantity referred to as detrital POC represents the sum of the fast-labile, slow-labile, and refractory detrital organic-carbon pools.
  We will revise Fig. 5 so that bottom-layer PHY is shown for both the estuarine and central-bay regions, together with bottom-layer detrital POC where corresponding observations are available. Table 2 will also explicitly report the statistical performance for these variables. The relatively low correlations for bottom-layer PHY (R = 0.14) and detrital POC (R = 0.24) will be acknowledged rather than interpreted as strong validation.
  We agree that the relatively weak reproduction of these bottom-layer biological and particulate-carbon variables introduces uncertainty into the absolute magnitude of the estimated organic-carbon storage flux. We will therefore state this limitation explicitly in Sect. 4. At the same time, the modeled storage flux is not calculated directly from a single bottom-layer PHY or POC value. It emerges from the coupled effects of particle settling and deposition, oxygen-dependent remineralization of multiple sedimentary organic-matter fractions, and downward transport within the sediment.
  To provide additional constraints on these benthic processes, we will supplement the pelagic validation with available benthic and sediment observations in the Appendix, including comparisons of sedimentary detrital organic carbon, porewater nutrients, benthic-faunal biomass, and sediment DO profiles. Because these benthic datasets are sparse and were collected at different sites, years, and seasons, we will use them primarily to evaluate order of magnitude, broad seasonal tendency, vertical structure, and oxygen penetration depth rather than as formal temporally matched statistical validation.
  Accordingly, the revised manuscript will explicitly distinguish between uncertainty in the absolute magnitude of carbon storage and the use of the model to examine mechanistic responses and relative differences among regions and nutrient-loading scenarios under a common model framework.
  5. Analysis: organic loading
  We thank the reviewer for requesting this clarification. External organic-matter loads were not altered among the nutrient-loading scenarios.
  In the revised manuscript, we will clarify that the sensitivity experiments vary only the total aggregated external DIN and DIP loads assigned to the fourteen loading points. The baseline temporal patterns of the external loadings are prescribed as repeating annual functions, and these seasonal variations are retained in all scenarios. For DIN and DIP, the baseline load time series are multiplied by the prescribed scenario factors while maintaining their spatial allocation and seasonal pattern.
  In contrast, other material loads, including organic-matter loads, retain the same baseline time-varying forcing in all scenarios and are not scaled with DIN and DIP. The remaining external forcing conditions are likewise kept unchanged among the scenarios. Therefore, the simulated changes in organic-carbon production, remineralization, and storage arise from changes in inorganic nutrient loading rather than from scenario-dependent changes in external organic-matter loading.
  6. Temporal variation of model state variables
  We thank the reviewer for the careful examination of the temporal variations and for suggesting possible explanations for the contrasting behaviors of the organic-carbon and CaCO₃ storage pathways.
  First, we will clarify the origin of the regular temporal variations shown in Figs. 11, 14, and 17. We also apologize for an inconsistency in the description of the moving-average period in the original manuscript. The time-series results shown in Figs. 11-19 are based on 1-day moving averages, whereas some text and figure descriptions incorrectly referred to 10-day moving averages. We will correct these descriptions throughout the revised manuscript. This correction concerns the stated averaging period and does not change the plotted results.
  The remaining regular variability after applying the 1-day moving average arises from the periodic physical and biological forcing represented in the model. The underlying model variability includes diel changes associated with the solar-radiation cycle together with diurnal and semidiurnal tidal variability. The hydrodynamic forcing includes the M₂, S₂, K₁, and O₁ tidal constituents. Their superposition generates spring-neap modulation on approximately semimonthly timescales and associated lower-frequency ecosystem responses. Thus, although the 1-day moving average attenuates diel and semidiurnal variability, spring-neap and broader seasonal-scale variations remain visible.
  Second, we agree that the contrasting temporal behaviors of organic-carbon and CaCO₃ storage are related to differences in their representation within the sediment model. However, the difference does not arise because the modeled sediment depth is insufficient for CaCO₃. Detrital organic carbon is vertically resolved within the upper 10 cm of sediment, whereas benthic CaCO₃ is represented as an areal pool without a vertically resolved sediment profile.
  Seasonal changes in organic-matter supply, mortality, and remineralization are strongest near the sediment surface. As these signals are transferred downward through the vertically resolved sediment, they are modified by remineralization, changes in organic-matter reactivity, downward transport toward the lower sediment boundary, and sedimentary redistribution. Consequently, the seasonal signal expressed in the organic-carbon storage flux across the lower boundary of the actively resolved sediment can be attenuated and delayed.
  In contrast, shell CaCO₃ associated with living suspension feeders is linked diagnostically to their biomass. When suspension feeders die, the shell CaCO₃ already formed during their lifetime is transferred to the benthic CaCO₃ pool; mortality itself does not generate new CaCO₃. Because this benthic CaCO₃ pool is represented as an areal pool rather than as a vertically resolved profile, recurring shell-transfer events can be expressed more directly in the modeled CaCO₃ storage flux. This helps explain the closer correspondence between abrupt changes in suspension-feeder biomass and seasonal changes in CaCO₃ storage in the estuarine and central-bay regions.
  In the tidal-flat region, both organic-carbon and CaCO₃ storage fluxes show clearer seasonal variability. We will interpret these patterns as the combined result of seasonal changes in organic-matter supply, benthic-faunal growth and mortality, shell formation, remineralization, and sedimentary redistribution. Bioturbation may contribute to this redistribution, but its contribution was not isolated from the other processes in the present analysis; we therefore do not attribute the seasonal pattern specifically or exclusively to bioturbation.
  These clarifications will be incorporated into the descriptions of the periodic temporal variability and the contrasting representations of the organic-carbon and CaCO₃ storage pathways in the revised manuscript.
  7. Results of the tidal-flat region
  We thank the reviewer for pointing out these two important issues in our interpretation of the tidal-flat region. We agree that the original manuscript overemphasized hypoxia as a controlling mechanism in the tidal flat and that the attribution of the transformation-dominated response specifically to externally supplied organic matter was not sufficiently supported by the analyses shown.
  First, regarding hypoxia, we will revise the interpretation of the tidal-flat carbon-fixation response. As the reviewer correctly noted, bottom-water DO remains relatively high in the tidal-flat region in Figs. 17 and 19, and the modeled benthic fauna do not exhibit the abrupt hypoxia-induced biomass collapse observed in the estuarine and central-bay regions. We will therefore no longer interpret hypoxia-induced benthic-faunal loss as a dominant control of carbon fixation or carbonate storage in the tidal-flat region.
  Instead, the revised interpretation will be based directly on the component fluxes. Phytoplankton and benthic-algal photosynthesis increase with nutrient loading, and suspension-feeder shell formation also increases. However, suspension-feeder respiration and benthic organic-matter remineralization increase more strongly. Consequently, the annual mean organic-carbon component of the net carbon-fixation balance remains negative and becomes more negative with increasing nutrient loading. We will therefore interpret the tidal-flat response as reflecting strong benthic carbon turnover, rather than hypoxia-induced suppression of primary production.
  Second, we agree that the original statement attributing this behavior specifically to “externally supplied organic matter” was not directly supported by the figures. The model control volume exchanges carbon laterally with adjacent waters, so the regenerated carbon may include locally produced and laterally supplied fixed carbon. However, the present analysis does not partition the benthic organic carbon according to its local versus external origin. We will therefore remove the unsupported inference that the transformation-dominated structure is specifically driven by externally supplied organic matter.
  In the revised manuscript, the term transformation-dominated will instead refer to the process balance directly supported by the simulations: fixation-enhancing processes increase with nutrient loading, but benthic respiration and remineralization increase more strongly, producing an increasingly negative net carbon-fixation balance, while both organic-carbon and CaCO₃ storage fluxes increase and CaCO₃ storage remains dominant. Thus, the term describes intensive benthic processing and turnover of fixed carbon rather than the inferred provenance of that carbon.
  We will revise the tidal-flat Results and the corresponding regional synthesis accordingly.
  8. Minor points
  (1) Units of model variables and process quantities
  We thank the reviewer for these helpful comments.
  We agree that the units were not presented consistently in the original manuscript. In the revised manuscript, we will standardize the model-variable and process units and clarify the basis on which benthic quantities are expressed. Table 1 will be revised to use element-specific molar units. Pelagic variables will be expressed per unit water volume, whereas benthic biomass and benthic CaCO₃ will be expressed per unit sediment surface area, sediment detritus per unit volume of sediment solids, and dissolved benthic variables per unit porewater volume.
  We will also standardize the corresponding units and axis labels in Figs. 10-19. In particular, benthic suspension- and deposit-feeder biomass will be expressed consistently per unit sediment surface area, eliminating the inconsistent cm-based and m-based expressions in the original figures. Corresponding panels among the three regional figure sets will use consistent units and notation.
  The model-observation comparisons in Fig. 5, Table 2, and the Appendix will retain the conventional mass-based units used in the observational datasets, because these figures are intended for direct comparison with measurements. In each case, simulated and observed quantities will be converted to the same units before comparison. We will make this distinction between model/process units and observational validation units explicit.
  (2) Fractions of organic matter
  We thank the reviewer for this suggestion. We agree that “Multi-G model” is the more widely recognized terminology for representing organic matter with different degrees of degradability. We will therefore revise Sect. 2.1.4 to describe the fast-labile, slow-labile, and refractory fractions as a three-component Multi-G model.
  We will also clarify that the corresponding fractions shown within living-organism compartments represent cellular organic constituents rather than detritus. Upon mortality, these fractions are transferred to the corresponding detrital pools, linking living-biomass composition to subsequent degradation, remineralization, and burial within the Multi-G framework.
  (3) Terminology for enhancing and reducing mechanisms
  We thank the reviewer for identifying this inconsistency. We will standardize the terminology throughout the manuscript using the adjectival forms “fixation-enhancing,” “fixation-reducing,” “storage-enhancing,” and “storage-reducing” when referring to mechanisms and pathways. The corresponding section headings, figure labels, captions, and main-text references will be revised consistently.
  
  Citation: https://doi.org/10.5194/egusphere-2026-2710-AC1
RC2:
'Comment on egusphere-2026-2710', Anonymous Referee #2, 24 Jul 2026

Kawahito et al. described results of sensitivity analysis of carbon cycle model of urban coastal ecosystem by nutrient enrichment. It nicely shows the responses of carbon cycle to the change of nutrient flux from land and interestingly indicated the monotonic increases of carbon uptake, fixation and storage with different time scales. It is an interesting manuscript showing important responses of carbon cycle in the urban coastal ecosystem to the change of nutient flux; however, there are many unclear parts which make it difficult to read and understand, and it requires major revision.

First of all, there are so many figures (19 figures with many panels), and many of them are not even explained in the text. Some of figures are very complicated. It is very difficult to follow what authors want to say with some figures. I suggest authors reduce the number of figures and move to appendix if it is available, and explain the figures more thoroughly in the text.

Secondly, I am confused with model setting and extraction of masses/processes which discussed in the text. I believe the model components/paths are combined to understand simplified carbon cycles, but authors mixed the names model components/paths with combined carbon cycles. They should be more carefully defined and explained thoroughly in the text and expressed in Figures and Tables. Figures should be also simplified to focus the important points. I could not find the description of how to model and evaluate the storage to deep ocean which may be important.

Thirdly, verification of the model is very weak. Sensitivity analysis itself is interesting, but different ecosystems and the setting may be able to give variety of responses. Specifically, verification of carbonate system is necessary. Following to the second point, clearer definitions of measured data with modeled results are necessary to compare.

Specific points
79 “carbon uptake (A/R), carbon fixation (F/U), and carbon storage (S/D)”: It is better to use easier abbreviations with full explanations in the text. Carbon Assimilation”?”/Release or Respiration”?” (A/R), Carbon Fixation/”?” (F/U), Carbon Storage/Dissolution (S/D).

Fig. 1 How does this model express and evaluate the carbon storage to deep ocean and deep sediments? It is known that large portions of POC/DOC are exported to deep sea and stored.

108 “Dissolved Inorganic Carbon (DIC)” usually means total carbonate (CO₂ + HCO₃⁻ + CO₃²⁻) not (H₂CO₃, HCO₃⁻, and CO₃²⁻).

118 Seagrass/Seaweed also? Is “Benthic algae” and “Seagrass/Seaweed” micro and macro, respectively? Benthic fauna include Suspension feeders and Deposit feeders? It is better to explain pelagic and benthic systems separately.
122 Both detritus (POC?) and DOC?

Fig. 2 This figure is too complicated. Three detritus components (labile, semi-labile, refractory) and CaCO3 are in living organism boxes, and there are other detritus and CaCO3 boxes. Are detritus and CaCO3 in living organism boxes necessary? What is the box included detritus and DOM in benthic system?

Fig. 4 Is Pelagic system calculated with only 2 m interval to bottom? Is this enough?

197 Model validation
It is required to describe more about the observations and how those are compared with model components.
Why is the verification only limited to pelagic system and less verification for carbonate and benthic processes? They are very important.
How to compare POC and DOC to model components, Detritus and DOM? Does POC include phytoplankton, zooplankton and detritus?
How to obtain phytoplankton biomass (PP)? Convert from Chl-a? PP usually means Primary Production.
212 Underestimate bottom-layer PP and POC? Need more explanation.

Table 2 TOC=POC+DOC?

238 DIN, DIP: Do you need to increase the number of names?

Fig. 5 There are so many panels, but few were explained in the text.

Fig. 7 Is photosynthesis only by phytoplankton, not by seagrass/seaweed/benthic algae?
What is H2CO3? Is it different from DIC? For A2, DIC increased and H2CO3 decreased? Is it CO2 similar to pCO2 (decreased for A2)? Green is not box but in the outside. Usually boxes are components (mass) and arrows are processes (fluxes), but they seem to be mixed in this figure.

Table 2 Some of components are different from Table1. Is PP same as Phytoplankton (mg C/L)? How the field data was taken? Chl-a? POC=pelagic detritus?
Need more explanations for field data also.

426 “particulate inorganic carbon” should be PIC? Use once defined words as much as possible.

Is it necessary to show the time series (seasonal change) in Fig. 11, 12, 13, 14, 15, and 16. There are almost no descriptions in the text.
Fig. 11, (B)(C) The color is not easy to see.

Citation: https://doi.org/10.5194/egusphere-2026-2710-RC2
- AC2: 'Reply on RC2', Akio Sohma, 02 Sep 2026
  
  General response
  We sincerely thank the reviewer for the constructive and detailed assessment of our manuscript. We appreciate the reviewer’s comments on the figure organization, model definitions, and model validation, which helped us identify several areas where the original manuscript was insufficiently clear. We will revise the manuscript substantially to improve its structure, methodological transparency, validation, and mechanistic interpretation.
  General comment 1. Number and explanation of figures
  We thank the reviewer for this important suggestion. We agree that the original manuscript contained a large number of multi-panel figures and did not provide sufficient guidance in the text regarding the purpose and interpretation of several of them.
  In the revised manuscript, we will first streamline the figure organization. The former Fig. 3, which mainly illustrates the model calculation flow rather than a scientific result, will be moved to the Appendix as Appendix Fig. B1. We will also reorganize the validation material so that Fig. 5 focuses on representative seasonal model-data comparisons for the three study regions, while additional benthic and sediment evaluations will be presented separately in Appendix Figs. A2–A4 and their observational datasets will be summarized in Appendix Table A1.
  We will also clarify the analytical roles of the remaining main-text figures. Figs. 7-9 present the candidate mechanisms governing carbon uptake, fixation, and storage, respectively. Fig. 10 provides an integrated cross-region comparison of the three carbon functions and the associated carbon budgets. Figs. 11-19 then provide the process-level and seasonal evidence used to interpret the uptake, fixation, and storage responses of the estuarine, central-bay, and tidal-flat regions.
  We have carefully reconsidered whether the detailed regional figures in Figs. 11-19 should also be moved to the Appendix. However, because the central objective of this study is to identify spatially differentiated carbon-control structures from the underlying process balances, we consider these figures to provide the direct evidence required for the regional mechanistic interpretations rather than ancillary information. We will therefore retain this systematic three-region-by-three-function figure set in the main text.
  At the same time, we agree that retaining these figures requires substantially clearer guidance for the reader. We will revise Sect. 7 so that the relevant figures and panels are explicitly referenced and interpreted in the text, and we will revise the figure captions, labels, axes, legends, and graphical conventions to make the role of each panel clearer. Fig. 10 and Table 3 will provide compact cross-region syntheses, whereas Figs. 11-19 will serve as the supporting process-level evidence.
  Thus, rather than retaining the original figure structure unchanged, we will reduce methodological material in the main text, strengthen the cross-region synthesis, and improve the linkage between each retained figure and the corresponding mechanistic interpretation.
  General comment 2. Model setting, model components/processes, and definition of carbon functions
  We thank the reviewer for this important comment. We agree that, in the original manuscript, the distinction among model state variables, individual biological and biogeochemical processes, and the three diagnostic carbon functions was not sufficiently clear.
  In the revised manuscript, we will reorganize the framework so that these levels are explicitly separated. The model state variables and individual process pathways will be described in Sect. 2, whereas the three carbon functions will be defined separately in Sect. 5 using explicit operational metrics.
  Specifically, carbon uptake will be evaluated as the net air-sea CO₂ flux, with flux from the atmosphere to the ocean defined as positive. Carbon fixation will be evaluated using a net carbon-fixation balance, calculated from the combined biological and chemical DIC-consuming and DIC-producing processes in the pelagic and benthic systems. This diagnostic includes photosynthesis, respiration and remineralization, CaCO₃ formation, and CaCO₃ dissolution. We will also clarify that this diagnostic balance is not the complete prognostic DIC budget, because physical transport and exchange, together with air-sea CO₂ exchange, are treated separately in the full DIC balance. Carbon storage will be evaluated as the transfer flux of organic carbon and CaCO₃ across the lower boundary of the actively resolved upper sediment into deeper permanent sediment layers.
  We will also revise the mechanistic framework so that the labeled mechanisms are clearly presented as candidate causal pathways that enhance or reduce one of these three diagnostic functions, rather than as independent model components or independent carbon cycles. The mechanisms are coupled and may operate simultaneously through the organic and carbonate pathways of the Dual Carbon Loop.
  Regarding deep-ocean storage, we agree that the original manuscript did not clearly define the model scope. The present model does not explicitly represent export to or storage in the deep ocean. We will therefore state this limitation explicitly. In this study, carbon storage refers only to the modeled transfer of organic carbon and CaCO₃ from the actively resolved upper sediment to deeper permanent sediment layers. Carbon export to the deep ocean is outside the scope of the present analysis.
  We will revise the relevant text, figures, and captions accordingly so that model components and process fluxes are clearly distinguished from the diagnostic carbon functions used in the analysis.
  General comment 3. Model verification/validation
  We thank the reviewer for this important comment. We agree that the validation presented in the original manuscript was too limited, particularly for the carbonate system, and that the correspondence between observed quantities and model variables was not sufficiently documented.
  In the revised manuscript, we will substantially expand Sect. 4 and the associated validation material. In addition to dissolved oxygen, inorganic nutrients, and organic-carbon variables, we will include quantitative comparisons for the carbonate-system variables DIC, pH, and pCO₂ in Table 2 and representative seasonal comparisons in Fig. 5. For these carbonate-system comparisons, we will use Tokyo Bay observations reported by Kubo et al. (2017), collected from March 2007 to December 2010. These observations were also used for model–data comparison of the carbonate system in Sohma et al. (2018).
  Because the simulations represent a periodically repeating annual cycle rather than individual calendar years, the observations will be compared with the control-case simulation according to the corresponding model location, vertical layer, and seasonal timing rather than by matching the observation year itself. All simulated values presented in the revised validation will be recalculated from the present EMAGIN-B.C. ver. 2.0 control simulation and will not be taken from the previous model studies.
  We will also clarify explicitly how observational quantities correspond to model variables. Phytoplankton carbon biomass (PHY) will be estimated from observed chlorophyll-a using the carbon-to-chlorophyll-a ratio employed in the model. Modeled detrital POC will be calculated as the sum of the fast-labile, slow-labile, and refractory detrital organic-carbon pools; modeled DOC as the sum of the labile and refractory dissolved organic-carbon pools; total POC as PHY + zooplankton carbon + detrital POC; and TOC as total POC + DOC. We will clearly state how these modeled quantities are matched operationally to the available observational data.
  We will further extend the evaluation beyond the pelagic water column. Additional comparisons with available observations of benthic-faunal biomass, sedimentary detrital organic carbon, porewater nutrients, and sediment DO profiles will be provided in Appendix Figs. A2-A4. The observation periods, locations, data sources, and previous uses of all validation datasets will be summarized in Appendix Table A1.
  We also agree that the limitations of the validation should be stated explicitly. Suitable observations of TA were not available for quantitative validation, direct pelagic CaCO₃ observations suitable for model validation were unavailable, and directly matched observations of air-sea CO₂ flux were limited. We will therefore not claim direct validation of these quantities. Likewise, agreement is weaker for some biological and particulate-carbon variables, particularly bottom-layer PHY and detrital POC.
  Accordingly, the revised manuscript will assess model performance from the combined behavior of nutrient, oxygen, organic-carbon, carbonate-system, and benthic/sediment variables, while explicitly acknowledging the remaining observational limitations. The subsequent sensitivity analysis will therefore emphasize mechanistic responses and relative differences among regions and nutrient-loading scenarios under a common model framework, rather than implying high precision in all absolute modeled quantities.
  Specific point 1. Line 79: A/R-F/U-S/D notation
  We thank the reviewer for this helpful suggestion. We agree that the original A/R-F/U-S/D notation was unnecessarily difficult to follow and that the individual letters were not sufficiently intuitive.
  In the revised manuscript, we will therefore discontinue the A/R-F/U-S/D paired notation. The three diagnostic functions will instead be referred to directly throughout the Introduction and the definition of the analytical framework as carbon uptake, carbon fixation, and carbon storage.
  For the mechanistic analysis in Sect. 6, we will introduce a simpler and internally consistent notation based directly on these three functions. Mechanisms that enhance or reduce carbon uptake will be denoted Uᵢ⁺ and Uᵢ⁻, respectively; those that enhance or reduce the carbon-fixation balance will be denoted Fᵢ⁺ and Fᵢ⁻; and those that enhance or reduce carbon storage will be denoted Sᵢ⁺ and Sᵢ⁻. The subscript identifies the individual mechanism, while the superscript “+” or “−” indicates whether the mechanism acts to enhance or reduce the corresponding carbon function; it does not denote the algebraic sign of an individual model flux.
  We will also accompany these symbols with descriptive mechanism names in the text and figures rather than relying on the symbols alone, and Table 3 will summarize the dominant uptake-, fixation-, and storage-related mechanisms for each region.
  We will not adopt terms such as “assimilation/release” or “storage/dissolution” because they would not correspond exactly to the operational definitions used in this study. Carbon uptake is specifically defined as net air-sea CO₂ exchange rather than biological assimilation, and processes opposing carbon storage include not only CaCO₃ dissolution but also organic-carbon remineralization and other processes that reduce transfer to permanent sediment storage. We therefore consider the revised function-based notation to be both simpler and more consistent with the quantities actually evaluated.
  Specific point 2. Fig. 1: deep-ocean and deep-sediment storage
  We thank the reviewer for raising this important point. We agree that the original Fig. 1 did not clearly distinguish between carbon storage in deeper sediment layers and carbon export to the deep ocean.
  In the revised manuscript, we will clarify that the present EMAGIN-B.C. application evaluates sedimentary carbon storage, not deep-ocean carbon storage. Carbon storage is operationally defined as the transfer flux of organic carbon and CaCO₃ from the actively resolved upper sediment domain across its lower boundary into deeper permanent sediment layers. This transfer is distinct from the initial sedimentation of particulate material into the surface sediment.
  Here, “permanent” is an operational model designation for carbon transferred below the actively resolved upper sediment and does not imply explicitly modeled geological-scale preservation or residence time.
  We will revise Fig. 1 and its caption so that the storage pathway is explicitly shown as transfer to permanent sediment layers, rather than implying export to the deep ocean.
  The present model does not explicitly represent export of POC or DOC to the deep ocean or their subsequent deep-ocean storage. We will therefore state this limitation explicitly in Sect. 5.2.3 and clarify that deep-ocean carbon storage is outside the scope of the present study.
  Specific point 3. Line 108: definition of DIC
  We thank the reviewer for identifying this error. We agree that the definition of DIC in the original manuscript was incomplete because it referred to H₂CO₃ without explicitly including dissolved CO₂.
  In the revised manuscript, we will define DIC as the sum of CO₂*, HCO₃⁻, and CO₃²⁻, where CO₂* represents the combined concentration of dissolved CO₂ (CO₂(aq)) and carbonic acid (H₂CO₃). We will use this definition consistently throughout the manuscript and in the relevant figure captions.
  Specific point 4. Lines 118 and 122: biological groups and organic-matter pools
  We thank the reviewer for requesting clarification of the biological functional groups and organic-matter pools represented in the model. We agree that these distinctions were not sufficiently clear in the original manuscript.
  In the revised manuscript, we will distinguish the pelagic and benthic biological components more explicitly. EMAGIN-B.C. includes separate compartments for benthic algae (microphytobenthos) and seagrass/seaweed (macroscopic benthic primary producers). In the present Tokyo Bay application, the benthic-algae compartment is activated, whereas the seagrass/seaweed compartment is not activated. Therefore, the carbon fluxes analyzed in this study include benthic microalgal production, particularly in the shallow tidal-flat region, but do not include carbon fluxes mediated by seagrass or seaweed.
  We will also clarify that phytoplankton and zooplankton are each represented as single aggregated functional compartments, while benthic fauna are represented by suspension feeders and deposit feeders. In the present application, biological shell CaCO₃ formation is assigned only to the calcifying suspension-feeder compartment.
  Regarding organic matter, we will distinguish clearly between particulate detrital organic matter and dissolved organic matter. Detrital organic carbon is represented by three particulate pools—fast-labile, slow-labile, and refractory detritus—whereas dissolved organic matter is represented separately by labile and refractory DOM pools. Thus, detritus and DOM are separate model components rather than a single organic-matter pool.
  We will also clarify the interpretation of the organic fractions shown within the living-organism compartments in Fig. 2. These fractions represent cellular organic constituents of living biomass, not detritus contained within the organisms. Upon mortality, they are transferred to the corresponding detrital pools.
  These distinctions will be described explicitly in Sects. 2.1.3-2.1.4, Fig. 2 and its caption, and Table 1.
  Specific point 5. Fig. 2: complexity and interpretation of boxes/pools
  We thank the reviewer for pointing out that the original Fig. 2 was overly complicated and that the representation of organic-matter fractions and CaCO₃ within the biological compartments could be misinterpreted.
  In the revised manuscript, we will replace the original Fig. 2 with a model-structure diagram based on the corresponding formulation of Sohma et al. (2018), adapted to the terminology used in the present study. The revised figure will more clearly distinguish among living biological compartments, detrital organic-matter pools, dissolved organic-matter pools, CaCO₃ pools, and the material-transfer pathways connecting the pelagic and benthic systems.
  We will also clarify explicitly that the fast-labile, slow-labile, and refractory fractions shown within the living-organism compartments do not represent detritus contained within the organisms. They represent the cellular organic constituents composing the living biomass. Upon mortality, these cellular fractions are transferred to the corresponding fast-labile, slow-labile, and refractory detrital pools while retaining their prescribed fractional composition.
  For suspension feeders, we will clarify that the living biomass state variable represents soft-tissue organic carbon. The associated shell CaCO₃ is diagnostically related to the soft-tissue biomass using a prescribed soft-tissue-carbon-to-shell-CaCO₃-carbon ratio. Upon mortality, the shell CaCO₃ that was already formed during the lifetime of the organism is transferred to the benthic CaCO₃ pool as dead shell material; mortality itself does not generate new CaCO₃.
  The revised Fig. 2 and its caption will therefore distinguish more clearly between living biomass, nonliving detrital and dissolved organic-matter pools, and carbonate pools, while retaining the full process structure needed to represent the coupled pelagic-benthic carbon cycle.
  Specific point 6. Fig. 4: 2-m pelagic vertical resolution
  We thank the reviewer for raising this question. Yes. In the present Tokyo Bay application, the pelagic water column is vertically discretized at 2-m intervals, with the number of pelagic layers varying according to the local water depth.
  In the revised manuscript, we will clarify the rationale and scope of this vertical resolution. The 2-m configuration follows the previous Tokyo Bay applications of the same modeling framework (Sohma et al., 2008, 2018) and was retained to ensure methodological consistency and comparability. It is intended to represent the seasonal and regional ecosystem dynamics relevant to the present analysis, including vertical gradients associated with seasonal stratification, bottom-water hypoxia, nutrient cycling, and carbon-system responses.
  We agree, however, that a 2-m pelagic resolution is not intended to resolve fine-scale turbulence or the detailed benthic boundary layer. We will state this limitation explicitly. Processes within the sediment are represented at much finer vertical resolution: the upper 10 cm of sediment are divided into 30 layers, with layer thicknesses ranging from approximately 0.1 mm near the sediment surface to 1.2 cm at depth, allowing steep oxygen and redox gradients within the sediment to be resolved separately.
  The model-observation comparisons indicate that this configuration captures major seasonal-scale features relevant to the present analysis, including the seasonal development and recovery of bottom-water hypoxia. However, we do not regard the agreement in bottom-water DO as validation of all fine-scale vertical or benthic processes. We therefore use the 2-m configuration for the regional and seasonal mechanistic comparisons undertaken here while explicitly acknowledging that it does not resolve fine-scale turbulence or benthic boundary-layer structure.
  Specific point 7. Sect. 4: details and scope of validation
  We thank the reviewer for these detailed comments on the model validation. We agree that the original manuscript did not provide sufficient information on the observational datasets, the construction of Fig. 5, or the correspondence between observed quantities and model variables.
  In the revised manuscript, we will substantially expand Sect. 4 and summarize the observational datasets in Appendix Table A1. For the conventional pelagic variables, we will identify the FY1998-FY2002 Public Water Quality Monitoring programs conducted by the Tokyo Metropolitan Government and Kanagawa and Chiba prefectures, together with the Comprehensive Survey on Regional Water Quality conducted by the Ministry of the Environment of Japan. For carbonate-system validation, we will identify the Tokyo Bay observations reported by Kubo et al. (2017), collected from March 2007 to December 2010. We will also document the additional site-specific observations used for Banzu Tidal Flat and for the benthic and sediment comparisons.
  We will clarify how the observations were compared with the model. Observed values will be assigned to the corresponding model grid cells and vertical layers. Because the model represents a periodically repeating annual cycle rather than individual calendar years, comparisons will be made according to location, vertical layer, and seasonal timing rather than by matching the observation year itself. The monitoring stations used for the statistical validation will be shown in Fig. 4. In Fig. 5A and B, observations assigned to the representative estuarine and central-bay grid cells will be grouped by calendar month and compared with the corresponding monthly model outputs. We will clarify that “surface” and “bottom” denote the uppermost and lowermost pelagic model layers, respectively, rather than unspecified fixed observational depths. The corresponding grid-cell water depths will also be shown in Fig. 4.
  We will also clarify the biological and organic-carbon quantities used in the comparison. We will replace “PP” with PHY for phytoplankton carbon biomass to avoid confusion with primary production. Observed PHY will be estimated from chlorophyll-a using a carbon-to-chlorophyll-a ratio of 30 g C g Chl-a⁻¹, consistent with the model formulation. Modeled detrital POC will be calculated as the sum of the fast-labile, slow-labile, and refractory detrital organic-carbon pools. Total POC will be defined as PHY + zooplankton carbon (ZP) + detrital POC, modeled DOC as the sum of the labile and refractory dissolved organic-carbon pools, and TOC as total POC + DOC. For Table 2, the available field POC data will be treated operationally as the observational counterpart of modeled detrital POC, and this limitation will be stated explicitly.
  We agree that pelagic validation alone is insufficient for a study in which benthic processing and carbonate chemistry are central. We will therefore add comparisons for DIC, pH, and pCO₂ to the pelagic validation and provide additional benthic and sediment evaluations in Appendix Figs. A2-A4, including benthic-faunal biomass, sedimentary detrital organic carbon, porewater nutrient profiles, and sediment DO profiles. The Banzu Tidal Flat comparison will also be included in Fig. 5C using the available site-specific observations. Because the Banzu and benthic datasets are sparse and cover broader or different time periods, these comparisons will be interpreted primarily in terms of order of magnitude, broad seasonal tendency, and vertical structure rather than as formal temporally matched statistical validation.
  Finally, we will revise the interpretation of bottom-layer phytoplankton and particulate carbon. Rather than making a general claim of systematic “underestimation,” we will report the actual statistical performance and acknowledge that agreement is weak for bottom-layer PHY (R = 0.14) and detrital POC (R = 0.24). We will explicitly recognize that these discrepancies introduce uncertainty into the absolute magnitude of estimated organic-carbon storage, while the subsequent analyses will focus primarily on mechanistic responses and relative differences among regions and nutrient-loading scenarios under the same model framework.
  Specific point 8. Table 2: definition of TOC
  We thank the reviewer for requesting this clarification. Yes. In this study, TOC is defined as total POC + total DOC.
  In the revised manuscript, we will state the definitions explicitly. Modeled detrital POC is calculated as the sum of the fast-labile, slow-labile, and refractory detrital organic-carbon pools. Total POC is then defined as phytoplankton carbon (PHY) + zooplankton carbon (ZP) + detrital POC. Total DOC is calculated as the sum of the labile and refractory dissolved organic-carbon pools, and TOC = total POC + total DOC.
  We will also revise Table 2 so that the former “POC” row is labeled “Detrital POC,” because the field POC measurements used in that comparison are treated operationally as the observational counterpart of the modeled nonliving detrital POC component. This distinction will be stated explicitly in the table note and Sect. 4.
  Specific point 9. Line 238: DIN and DIP terminology
  We thank the reviewer for this comment. To avoid introducing unnecessary additional terminology, we will retain the standard abbreviations DIN and DIP, with their full names, dissolved inorganic nitrogen and dissolved inorganic phosphorus, given at first occurrence.
  We will also clarify their operational representation in the present model. In EMAGIN-B.C., dissolved inorganic nitrogen is represented by the NH₄ and NO₃ pools, whereas nitrite (NO₂) is not represented as an independent state variable. Accordingly, for the purposes of the present model analysis, DIN is represented as NH₄-N + NO₃-N, while DIP is represented by PO₄-P. These terms will then be used consistently throughout the manuscript without introducing additional abbreviations or alternative names.
  Specific point 10. Fig. 5: number of panels and explanation in the text
  We thank the reviewer for this comment. We agree that the original manuscript presented many panels in Fig. 5 without sufficiently explaining their purpose or guiding the reader through the main validation results.
  In the revised manuscript, we will reorganize the description of Fig. 5 and explain more explicitly what each regional panel set is intended to evaluate. Panels A and B will present representative seasonal comparisons for the estuarine and central-bay regions, including DO, phytoplankton carbon biomass (PHY), detrital POC, inorganic nutrients, and the available carbonate-system variables. Panel C will provide the corresponding site-specific comparison for Banzu Tidal Flat using the variables for which suitable tidal-flat observations are available.
  We will revise Sect. 4 to guide the reader through the principal results shown in these panels rather than simply presenting the figure without interpretation. In particular, we will discuss the model’s reproduction of the seasonal development and recovery of bottom-water hypoxia, the weaker agreement for bottom-layer PHY and detrital POC, and the broad consistency of the tidal-flat simulation with the available observed ranges and seasonal tendencies.
  We will also expand the Fig. 5 caption to identify the observational datasets used in each panel set, clarify the meaning of surface and bottom model layers, and define the plotted quantities. Quantitative performance across the full observational dataset will be summarized separately in Table 2.
  Because Fig. 5 is intended as a compact visual overview of model performance across multiple interacting biogeochemical variables, we will retain the selected multi-variable comparisons rather than describe every individual subpanel separately in the text. The revised text will instead focus on the major strengths, discrepancies, and limitations that are relevant to the subsequent mechanistic analysis.
  Specific point 11. Fig. 7: photosynthesis, carbonate-system notation, and graphical logic
  We thank the reviewer for these helpful comments on Fig. 7. We agree that the original figure did not sufficiently distinguish among photosynthetic compartments, carbonate-system variables, and biological or biogeochemical processes.
  First, photosynthesis in the model is not restricted to phytoplankton. EMAGIN-B.C. includes both phytoplankton photosynthesis in the pelagic system and benthic-algal (microphytobenthic) photosynthesis in the benthic system. In the present Tokyo Bay application, the benthic-algae compartment is activated, whereas the seagrass/seaweed compartment is not activated. We will clarify this explicitly in Sect. 2.1.3. In Fig. 7, the U₁⁺ pathway specifically represents pelagic photosynthetic DIC drawdown by phytoplankton, because this pathway illustrates the direct effect of enhanced pelagic production on surface-water pCO₂ and air-sea CO₂ uptake. Benthic-algal photosynthesis remains included in the model and in the carbon-fixation analysis, particularly in the tidal-flat region.
  Second, we will replace the potentially confusing use of H₂CO₃ with the standard carbonate-system notation CO₂*. DIC will be defined as CO₂* + HCO₃⁻ + CO₃²⁻, where CO₂* denotes the combined concentration of dissolved CO₂, CO₂(aq), and carbonic acid, H₂CO₃. Thus, CO₂* is one component of DIC and is directly related to seawater pCO₂; it is not synonymous with total DIC.
  Third, we will clarify the carbonate-system response in the former A2 pathway, now represented as the U₂⁺ mechanism. CaCO₃ dissolution increases both DIC and TA, by approximately 1 mol of DIC and 2 equivalents of TA per mole of CaCO₃ dissolved. Under the carbonate-system conditions simulated here, the relatively larger increase in TA shifts carbonate speciation away from CO₂* and toward HCO₃⁻ and CO₃²⁻. Consequently, total DIC can increase while CO₂* and pCO₂ decrease. When this lower-pCO₂ water is transported toward the surface, it can secondarily enhance atmospheric CO₂ uptake. We will state explicitly that this interpretation applies to the modeled carbonate-system conditions rather than presenting it as an unconditional consequence of CaCO₃ dissolution.
  Finally, we agree that the original Fig. 7 mixed the visual representation of state variables and processes in a way that could be confusing. We will clarify that Fig. 7 is a conceptual causal-pathway diagram, not the model mass-balance or state-variable diagram; the latter is presented separately in Fig. 2. In the revised Fig. 7, boxes will identify forcing or state/carbonate-system variables, whereas process or flux changes will be indicated by labeled arrows. The caption will explicitly explain the graphical conventions and will state that the individual panels isolate candidate causal pathways for interpretation, although these pathways can share processes and operate concurrently in the full model.
  Specific point 12. Table 2: PHY, chlorophyll-a, and POC correspondence
  We thank the reviewer for requesting clarification of the quantities compared in Table 2.
  First, the abbreviation “PP” in the original manuscript referred to phytoplankton carbon biomass, not primary production. Because “PP” can easily be interpreted as primary production, we will replace this abbreviation with PHY throughout the revised manuscript.
  The available field observations for phytoplankton were based on chlorophyll-a concentrations. For comparison with the carbon-based model state variable, observed chlorophyll-a will be converted to phytoplankton carbon biomass using a fixed carbon-to-chlorophyll-a ratio of 30 g C g Chl-a⁻¹, corresponding to the chlorophyll-a-to-carbon ratio used in the model.
  We will also clarify the definition of particulate organic carbon. In the model, detrital POC is calculated as the sum of the fast-labile, slow-labile, and refractory detrital organic-carbon pools. Total POC is separately defined as PHY + zooplankton carbon (ZP) + detrital POC.
  For the statistical comparison in Table 2, the available field POC observations will be treated operationally as the observational counterpart of the modeled detrital POC component. Because the observational POC measurements do not resolve the individual model compartments, we will state this correspondence explicitly rather than implying that the observed and modeled quantities are compositionally identical.
  We will also provide the observational data sources, periods, and comparison procedures in Sect. 4 and Appendix Table A1, and revise the Table 2 terminology and notes accordingly.
  Specific point 13. Line 426: particulate inorganic carbon terminology
  We thank the reviewer for pointing out the inconsistent terminology. We agree that defined terminology should be used consistently throughout the manuscript.
  In the revised manuscript, rather than introducing an additional abbreviation “PIC,” we will use the more specific term CaCO₃ (or shell CaCO₃, where appropriate) consistently when referring to the particulate inorganic-carbon pool represented in the present model. This is preferable because the particulate inorganic carbon explicitly represented in the present EMAGIN-B.C. analysis is CaCO₃ associated with the carbonate pathway.
  We will therefore remove the inconsistent generic use of “particulate inorganic carbon” and standardize the corresponding text, figure captions, and mechanism descriptions using CaCO₃ terminology.
  Specific point 14. Figs. 11-16: need for seasonal time series and readability of Fig. 11(B,C)
  We thank the reviewer for this important comment. We agree that the original manuscript did not sufficiently explain why the seasonal time-series panels were needed or how they supported the mechanistic interpretation.
  In the revised manuscript, we will clarify the complementary roles of the annual-mean and seasonal results. The annual-mean values provide the primary basis for comparing the magnitude and direction of the responses among nutrient-loading scenarios. The seasonal time series, however, provide additional process-level information that cannot be obtained from the annual means alone. In particular, they allow us to examine the timing and phase relationships among primary production, carbonate-system responses, bottom-water oxygen conditions, benthic-faunal biomass changes, remineralization, and carbon-storage fluxes. These temporal relationships are important for evaluating the mechanisms proposed for the contrasting estuarine, central-bay, and tidal-flat responses.
  We will therefore retain the seasonal information that directly supports the mechanistic interpretation, while revising Sect. 7 to refer explicitly to the relevant figures and panels and to explain the principal seasonal features used in the analysis. We will also state clearly in Sect. 7.1 that Figs. 11-19 provide the process-level and seasonal evidence underlying the regional interpretations, whereas Fig. 10 and the annual-mean components of Figs. 11-19 provide the more compact comparison among nutrient-loading scenarios.
  We will also clarify that the seasonal curves are 1-day moving averages, not 10-day moving averages. The remaining regular variability reflects the periodic solar and tidal forcing represented in the model, including spring-neap modulation associated with the M₂, S₂, K₁, and O₁ tidal constituents.
  We have also carefully considered the reviewer’s comment on the readability of Fig. 11(B) and (C). In the revised figure, the panel titles and caption will explicitly distinguish DIC and TA production-consumption components, positive and negative bars will indicate production and consumption fluxes, respectively, and the red line will indicate their algebraic sum. The same graphical conventions will be used consistently in the corresponding regional figures.
  We also re-examined the color coding in Fig. 11(B,C) and the corresponding panels of Figs. 14 and 17. We will retain a common color assignment for the same process components across the three regional figure sets so that they can be compared directly, while improving the legends, labels, panel descriptions, and graphical consistency to make the individual components easier to distinguish.
  
  Citation: https://doi.org/10.5194/egusphere-2026-2710-AC2
RC3:
'Comment on egusphere-2026-2710', Anonymous Referee #3, 17 Aug 2026

This manuscript applies EMAGIN-B.C., a benthic–pelagic coupled ecosystem model, to Tokyo Bay to examine coastal nutrient enrichment, carbon cycling, and benthic–pelagic coupling, framing the results around a "Dual Carbon Loop" concept and a comparison of estuarine, central-bay, and tidal-flat regimes. The topic is well within the scope of Biogeosciences and addresses processes, such as nutrient loading, carbon cycling, and climate mitigation in coastal systems, that are of clear scientific interest. The modeling approach is appropriate to the question, and the systematic comparison across the three regimes is the most valuable contribution.
However, several major issues in methodological documentation, interpretation, and evidential support require substantial revision. The Introduction doesn't establish the broader Tokyo Bay-specific knowledge gap strongly enough, making it difficult to judge how significant the regional contribution is. The broad conclusion that nutrient enrichment can produce spatially different carbon cycle responses is possible and of clear interest. However, several of the specific mechanistic interpretations used to define the proposed regional response types are not yet convincingly supported by the analyses shown.
The manuscript is clearly written and the figures, tables, and terminology are generally clear in isolation, but the manuscript is too long and repetitive: similar mechanisms and regional summaries are explained multiple times, and the progression from methods to hypotheses to results (with synthesis for each region) to general synthesis is not always efficient. Overall, I consider this an interesting and potentially valuable study built on an appropriate model and question, but the manuscript doesn’t yet provide sufficient evidence to support all of its conclusions at the level required for publication in its current form. I recommend major revision addressing methodological documentation, model validation, consistency between the reported results and their mechanistic interpretation, support for the proposed regional mechanisms, and development of a substantive Discussion, alongside the manuscript's organization and concision.
Major comments
Major comment: Insufficient documentation of model configuration for reproducibility (Sect. 3, lines 180–190)
Sect. 3 doesn't provide enough specific detail for the model configuration to be reproduced independently. Lines 180–182 list the categories of external forcing ("freshwater inflow, nutrient loading, solar radiation, wind stress, and open boundary conditions at the mouth of the bay") and state that these are "prescribed as annual periodic functions based on observational data from 1998–2002," but not only the source of the external forcings is not described but also the construction of these functions is not described (is it climatological means? fitted sinusoidal functions?). Line 185 states that simulations are run "until the system reaches a periodic steady state," without quantifying how long this spin-up time was. The manuscript defers "detailed model settings" to Sohma et al. (2018), which is fine for the generic model formulation and equations, but not for details specific to the present experiments. I would ask the authors to add the following:
-For freshwater inflow, river nutrient loads, solar radiation, wind stress, and open boundary conditions (physical and biogeochemical), please state the original data source and how the annual periodic functions were constructed from the 1998–2002 observations (a good figure of an example of these periodic functions in the Supplementary, could be very informative). Please also describe the initialization strategy for physical and biogeochemical variables. Were the nutrient-loading scenarios initialized independently or from the equilibrated control state?
-Advection and diffusion are described as being "considered," but it is not clear how the exchange and transport terms between boxes and layers are determined (and this is relevant for mechanism S3). Besides, I don’t fully understand “across offshore boundaries” (just one boundary of the box? Just the offshore boundary of the Bay?). It would be nice to actually present the physical transport equations for at least one tracer in enough detail to know how exchanges between all neighboring boxes are calculated. Separately, DIN and DIP are introduced into the surface layer of 14 computational boxes, but it is not clarified how total river loads are partitioned among these boxes, which river sources correspond to which boxes, and how the resulting load time series are determined.
-We know from Table 1 (and not in the main text), that bivalves and polychaetes are represented as the suspension and deposit feeder groups in the model. And suspensions are the only ones that can form CaCO3. That makes me think about what biological functional groups the aggregated phytoplankton (relevant for the later validation), zooplankton, and benthic pools are intended to represent (see also my minor comment about non-inclusion of other calcifiers in Minor Comments).
-What integration time step(s) are used for transport, pelagic ecosystem processes, sediment processes, and carbonate chemistry, and are these the same or different? At what frequency are model outputs saved (e.g., hourly, daily)?
Major comment: Incomplete description and scope of model validation (Sect. 4)
-The central conclusions concern carbon uptake, carbonate chemistry, and carbon storage, yet the validation presented in Sect. 4 doesn’t extend to the carbonate system variables that directly build these conclusions. I would prioritize evaluation of DIC, TA, pH, or pCO2 as most important, with CaCO3 and air–sea CO2 flux as desirable. Where observations are not available, comparison against published Tokyo Bay measurements or ranges could offer at least a benchmark. Where no suitable data exist at all, this should be acknowledged explicitly.
-Sect. 4 states that simulated PP, POC, DO, NO3, NH4, PO4, DOC, and TOC were compared with "observational data," but doesn't identify the source or program, years, stations, or depths of these observations. That should be acknowledged by citing the dataset or monitoring program, and the number and location of stations. For example, the manuscript doesn’t sufficiently explain how Fig. 5 was constructed. What years the observations span?, which station(s) contribute to each boxplot?, were multiple stations pooled?, Does the modeled curve represent the uppermost and deepest model layer exactly or some depth-averaged value?. If the observations span a range of depths and/or locations while the model curve represents a single grid cell and layer, the comparison may not be entirely fair. I would also ask the authors to report the actual depth values rather than only "surface" and "bottom", since I suspect these may not correspond to the same depths as "surface" and "bottom" in Table 2 (see my Minor Comment on Table 2). Given that we don't know which functional group P represents in the model, against which "phytoplankton biomass" observation is it being compared? And which model state variables were summed in the model to calculate and validate POC, DOC, and TOC ?
-Finally, why does the tidal-flat region, the third region central to the study, not receive comparable validation attention in Fig. 5 ?.
Major comment: Inconsistencies in the 16-mechanism A/R–F/U–S/D
Some mechanisms are not consistent with the paper's own operational definitions of uptake, fixation, and storage, and in several cases the written mechanism and its corresponding schematic (Figs. 7–9) describe different processes. As a result, the labeling is not internally coherent and confuses the coupled structure it is meant to describe, adding complexity rather than clarifying model behavior. Having to analyze 16 different panels, doesn't make it more friendly for the reader. The examples below are not an exhaustive audit, and I would encourage the authors to review the full labeling:
-Line 75 (Introduction): The manuscript introduces the notation A/R, F/U, and S/D without explicitly defining what each individual letter stands for; the functional meaning only becomes clear much later, in Sect. 6. With 16 mechanism labels in total (A1–A3, R1–R3, F1–F3, U1–U2, S1–S3, D1–D2), the notation becomes a significant burden for the reader once figures and captions combine them with other symbols (e.g., Fig. 11's caption already mixes panel letters, A/R, A1–A3, and R1–R3). It is also easy to confuse U and D, since both represent decomposition/remineralization losses but within different balances. I would suggest either clearly explaining and visually introducing the full letter hierarchy early (e.g., in Fig. 1, see my Minor Comment below), or considering more intuitive labels in place of single letters. Another option is to create a new table summarizing the mechanisms, so then the reader can just refer to that table when reading the results. More importantly, while the authors do mention that carbon uptake, fixation, and storage are mutually coupled processes, but governed by different dominant mechanisms (Line 305), separating these coupled causal pathways into labeled mechanisms, make it feel more independent than they actually are. For example, A1, F1, and S1 largely describe successive stages of one chain: photosynthetic drawdown of DIC, organic matter fixation, and burial. Or for example “U” appears in 7.2.1, 7.3.1, 7.4.1 when discussing the A/R uptake family. All of these make the current labeling not very friendly and complicate the coupled structure of the underlying carbon cycle processes. Maybe a hierarchy with environmental forcing --> ecosystem processes (photosynthesis, remineralization, etc) --> diagnostic carbon functions (uptake, fixation, storage), seems more natural. Nevertheless, I leave the specific restructuring to the authors, as it should depend on their model results.
-F2 doesn't fit the paper's own definition of fixation. The manuscript states that trophic transfer of organic carbon from phytoplankton to zooplankton and higher trophic levels "strengthens fixation capacity." But this transfer moves already fixed carbon rather than a new consumption of DIC. This is an ambiguity between carbon fixation, retention of already fixed carbon, and net ecosystem accumulation.
-R1 mentions that fewer calcifying organisms leads to less CaCO3 formation, which leads to TA decreases, and so pCO2 rises. However, calcification consumes approximately 2 equivalents of TA per mole of CaCO3 precipitated. All else equal, less CaCO3 formation (or less calcification) therefore means less TA removal and would tend to leave TA relatively higher, not lower, and hence lower pCO2. Now, the schematic in Fig. 7 seems to imply fewer bivalves, less shell formation, lower CaCO3 stock, less CaCO3 dissolution, hence lower TA, higher pCO2. That make sense, but then it is the reduced dissolution, not reduced shell formation itself, that lowers the TA. I recognize that A2 simultaneously introduces dissolution and remineralization effects, so the actual modeled net TA response may be more complex than this pathway suggests, but the direction stated in R1 needs clarification.
-A3 attributes moderate hypoxia to reduced benthic grazing that increases phytoplankton production, while U2 attributes severe hypoxia to a weakening of biological production via a vaguely defined "ecosystem instability" (Line 475). A nonlinear response of this kind is possible, but the process causing the change in the sign of the biological response between these two regimes (with proper supporting references) is not described.
-The main text (Sect. 6) describes S2 as "carbon storage through CaCO3 burial," S3 as transformation and storage of externally supplied/laterally transported organic carbon, and D2 as “dissolution of CaCO3 under low-pH conditions”. However, Fig. 9 assigns very different processes to these same labels: S2 is shown as sediment organic matter --> remineralization --> hypoxia --> a temporary increase in sediment organic matter, with nothing about the CaCO3 pathway. S3 is shown as hypoxia suppressing organic matter remineralization, with no lateral/external organic carbon pathway. D2 is shown as reduced shell formation (reduced CaCO3 production) rather than enhanced dissolution.
Major comment: inconsistencies between text results and figures, and within figures
-Fig. 12(A) shows net organic matter fixation increasing (so DIC transformed to O.M.) with nutrient loading, while in 12(B) whole water column DIC concentration also increases (red line labeled “DIC conc”). This is not necessarily contradictory; DIC could be replenished by transport, exchange, or other unshown sources, but the figure doesn’t allow the reader to see how these are reconciled. In general, I would ask the authors to clarify, for each region, how the diagnosed biological and chemical process terms relate to the physical transport, exchange, and boundary terms not shown in the figures, and to confirm that the complete DIC budget closes.
-Sect. 7.3.1: “Bottom-layer (remineralization) is strongly amplified” / “remineralization-related fluxes increase substantially” / “bottom-layer DIC concentration increases” is not demonstrated clearly by Fig. 14. Fig. 14(B) is labeled as the annual DIC production/consumption budget for the pelagic surface, not the bottom layer: bottom-layer DIC is simply not plotted in Fig. 14. The text says “the A/R structure does not constitute an F-dominated monotonic amplification pattern”, but per paper’s own definition, F belongs to their fixation F/U framework, not A/R. The most troublesome to me is that the net uptake response itself looks very monotonic (CO2 uptake up, pCO2 down, DIC_surf down) as in Fig. 11, while the text claims it is not. If anything, perhaps the positive and negative DIC process terms in 14B become larger than in 11B (not very clear, different scales), so here maybe the “co-amplification”, but the net result still changes monotonically toward uptake.
-Fig. 17(B)'s red "Total production and consumption" line becomes increasingly positive with nutrient loading. At the same time, surface DIC decreases and atmospheric CO2 uptake increases. Same as above, this reconciliation requires an additional sink coming from physical export, transport or some other omitted term. The text in Sect. 7.4.1 complicates things further when it says that enhanced primary production “promotes DIC consumption in the surface water” and then benthic recycling "partly offset[s] the increase in net DIC consumption", but according to Fig. 17(B), there isn’t net DIC consumption to “partly offset”; production already exceeds consumption. Also, I would keep the "(U1 AND R3)" at the last sentence rather than “benthic recycling processes (U1–R3)”, because U1 and R3 belong to different mechanism families.
-Tidal flat (Fig. 18) Sect. 7.4.2 states that "carbon fixation flux increases with increasing nutrient loading." However, Fig. 18(A) net organic matter fixation term becomes progressively more negative with increasing loading, while CaCO3 is only a small positive contribution. An increasingly negative net term should indicate declining, not increasing, net fixation.
-Sect. 7.4.3: Fig. 19 doesn't show an organic carbon transport flux or distinguish locally produced organic matter from laterally/external supplied organic matter. So the jump to “externally supplied organic carbon…” (Line 785, Sect. 7.4.3) without showing where it came from or how much was externally/laterally supplied, is not supported.
There are a couple more I was able to identify in the results in Sect. 7 (see Minor Comments), but please review thoroughly. Also, summaries in Sect. 7.3.4 or 7.4.4, carry forward the unsupported or internally inconsistent interpretations identified. The same happens in Sects. 7.5, 7.5.1–7.5.4, 8.1–8.2. I would ask the authors to address the concerns raised, since resolving them will necessarily affect these later synthesis sections as well.
Major comment: The manuscript's length and structure
Beyond the specific figure and content concerns raised in my other comments, the manuscript's overall structure creates substantial redundancy. Nine main-text multi-panel figures (three per region [uptake, fixation, storage]) repeatedly re-plot the same set of state variables in different combinations. Many individual panels within these figures are not clearly interpreted or used to support the argument in the text. Each region's results are summarized once at the end of its own subsection (Sects. 7.2.4, 7.3.4, 7.4.4), summarized again across regions in Sect. 7.5, condensed efficiently in Table 3, and then restated again in Sects. 8.1 and 8.2 before the Discussion reaches genuinely new material (see Minor Comment below).
This redundancy makes the manuscript considerably longer than its scientific content requires, and makes it harder for the reader to identify what is genuinely new at each stage. It also makes it difficult to actually see the spatial differentiation among the three regions: comparing uptake currently requires the reader to move between Figs. 11, 14, and 17, fixation between Figs. 12, 15, and 18, and storage between Figs. 13, 16, and 19. A side-by-side regional comparison for each diagnostic would be better for the comparative objective. In general though, for Figs. 11–19 I would suggest the authors consider retaining only the panels necessary for the regional interpretation and moving supporting diagnostics not directly used in the main text to the Supplement, which would also help address the overall length. Also, it would be nice if the text could guide the reader through figures with, for example: ”see Figure 11Aj”, “as shown in Figure 12A2”, etc. Finally, Sect. 5 defines the three functions (uptake, fixation, storage) and explains how each is quantified and Sect. 6 immediately reintroduces those same three functions to explain their controlling mechanisms. I would combine both sections into one.
I know that the specific restructuring should reflect the authors' own judgment about what best serves the paper's argument. But I do think the overall length and repetition across Sects. 5–8 and Figs. 11–19 deserve reconsideration, independent of the more specific content issues raised elsewhere in this review.
Major comment: Lack of Discussion
The end of the manuscript (Sect. 8), consists almost entirely of summarizing and re-summarizing the regional results already presented in Sect. 7. It currently doesn't present a proper Discussion section that progressively shifts from what happened toward why it matters. The authors don’t compare the findings with prior work on coastal nutrient enrichment, carbon cycling, or benthic–pelagic coupling in other estuarine and coastal systems, or with other Tokyo Bay studies. This would help the reader judge whether the reported regional patterns are specific to Tokyo Bay or reflect more general coastal dynamics. Several limitations raised elsewhere in this review (see Major and Minor Comments below) are not acknowledged or discussed anywhere in the manuscript, including: the fixed N:P ratio in the nutrient scenarios (would results change if N and P were varied independently, and is there any sensitivity to this?), the absence of carbonate system validation, the ambiguity around what "long-term carbon storage" actually represents, etc. Sect. 8.3 and 8.4's were going in the right direction, but for example Sect 8.4 extends the findings toward recommendations for evaluating coastal carbon mitigation strategies more broadly. However, the nutrient experiment is an idealized simultaneous scaling of riverine DIN and DIP at fixed stoichiometry, not a simulation of an actual nutrient-management policy. The Discussion should also identify follow-up work that would test or strengthen the conclusions. For example, these can be: independent N and P perturbation experiments, full-domain or other locations analysis, more different loading increments if nonlinearity or thresholds are to be discussed (Sect 8.3), testing under interannual forcing, etc.
Minor commnets
-Abstract: The sentence “In contrast, tidal flats...new production.” reads too long and dense. I suggested shorten or split in two.
-Introduction: Tokyo Bay is mentioned early on as one of several Japanese enclosed seas affected by nutrient-management policies, and again at the end of the Introduction as the study site, but the Introduction otherwise gives little regional background before the model application (the first real description only appears in Sect. 3). So it is difficult to judge why Tokyo Bay is an appropriate case study and what gap the present work fills. I suggest adding a short paragraph earlier in the Introduction covering, for example, some of the seasonal productivity and hypoxia in the region, dominant river nutrient inputs, previous Tokyo Bay ecosystem or carbon modeling, observational carbon cycle studies relevant to the climate and/or carbon management framing, etc.
-Line 60: “Moreover, CaCO3 formation consumes TA and lowers pH, thereby altering pCO2 and influencing air-sea CO2 exchange”. Since the previous sentences were talking about anaerobic conditions in sediments, I suggest making a better connection here by also specifying where calcification happens. I’m a bit uneasy with the word “consumes TA” but I think it is fine. But “altering” is unnecessarily vague. Because CaCO3 formation decreases TA twice as much as DIC on a molar basis, the net effect of calcification is an increase in seawater pCO2 and a decrease in pH.
Line 65: “...while river water characterized by low pH and high pCO2 may enhance CO2 release to the atmosphere.” Is this specifically a characteristic of river waters entering Tokyo Bay, or is it intended as a general statement about estuaries? Riverine pH and pCO2 can vary depending on watershed geology, alkalinity, organic matter inputs, season, and biological activity.
-Figure 1. I am not sure this figure is necessary for the target audience of Biogeosciences. But it would add more value if the “carbon uptake,” “carbon fixation,” and “carbon storage” are better highlighted, if it visually incorporated the Dual Carbon Loop concept and if the three functional balances that structure the rest of the paper are added: A/R, F/U, S/D (though, see the Major comment about the labeling)
Sect. 2: When I read this the first time, I initially thought it was a full coupled 3D hydrodynamic-biogeochemical model. So, I would suggest stating the spatial architecture of the model upfront in Sect. 2, e.g., saying here that this is a vertically resolved, spatially interconnected multi-box ecosystem model.
Sect. 2: In general, I find that the large number of sections (2), subsections (2.1), and subsubsections (2.1.1), [ or in Sects. 5–7], makes the manuscript somewhat difficult to follow. Several subsubsections contain only a short paragraph or a few sentences. In these cases, I think the text would read more smoothly if it were incorporated into the broader subsection rather than given a separate heading. For example, I would suggest making all of Section 2.1 a short paragraph, recognizing that much of the existing model formulation is documented in Sohma et al. (2018), and detailed generic formulation could be moved to the Supplement. From there Sect. 2.2 should clearly distinguish what is actually new in this model version from v1. Besides, since v2 adds two completely new modifications, I would include the equations and how they are implemented in the model in the Supplement, as well. So anyone can take v1 and reconstruct v2 or at least validate the v2 modifications.
-Sect. 2.1.3 “CaCO3 formation by suspension feeders is also explicitly incorporated.”. So only bivalves can produce CaCO3 shells here; no coccolithophores, foraminifera, or pteropods. This is fine, and perhaps easy to justify, but because the conclusions explicitly depend on the carbonate loop, I think the authors should state and justify this assumption explicitly. This distinction matters if later once want to generalize the Dual Carbon Loop beyond Tokyo Bay.
-Figure 2: maybe a legend with what each different type of arrow means, or different colors of arrows, or different colors of boxes, or solid and dashed rectangles, should be added. Besides, it is not necessary to spell out the abbreviations in the same figure; they can be placed in the caption (ODU, TA, TEA, etc).
-Figure 3: I would suggest moving Fig. 3 to the Supplement. It mainly shows the order in which existing equations are evaluated, which is useful documentation but not necessary to follow the scientific results.
Table 1: Both pelagic AND benthic state variables are given in units per water volume (mg L-1), not only pelagic. But for the benthic dissolved variables, it is not clear what volume this refers to. Are these porewater concentrations (per liter of porewater), or concentrations per liter of bulk sediment volume?
-Line 170. Please specify which rivers are represented, and briefly characterize their watersheds (e.g., agricultural, urban/industrial, or dominated by outfall discharges), as this context would help the reader interpret the nutrient loading driving the model.
-Line 175. I’m really interested in the sediment component of the model, especially because of the vertical resolution to the upper sediment. Since benthic processes play an important role in the mechanistic interpretation, it would be nice to see depth-resolved sediment diagnostics. Maybe a sediment profile or seasonal comparison could help demonstrate how the modeled benthic processes differ among regions and nutrient-loading scenarios. However, this is totally optional and only if it adds value to the results (it can go even in the Supplement).
-Figure 4: The caption should mention that the numbers inside each box are the maximum depth of that box. I would actually combine Fig. 4 with Fig. 6 since both are showing the same domain and spatial distribution.
-Line 205: The manuscript infers from the good bottom DO performance that the model appropriately represents the coupling among primary production, particle sinking, decomposition, and vertical mixing. However, multiple combinations of production, respiration, mixing, and sediment oxygen demand can lead to similar DO concentrations, so high correlation in an integrated variable like bottom DO doesn't independently validate each process. I would also encourage discussing RMSE alongside R, since RMSE for bottom DO (1.68 mg L-1) is proportionally similar to that for surface DO relative to their respective observed means (~31% vs ~25%), indicating that the high bottom DO correlation doesn't necessarily correspond to smaller absolute error. I would also report mean bias or mean modeled values alongside the mean observed values already given in Table 2. Even better, a normalized Taylor diagram could also provide a compact summary of correlation, variability, and centered RMSE across variables, though this could go in the supplementary.
-Sect. 4, Line 205 to 215: From “In this context,...”. I’m okay with the content of the paragraph but it currently has like a back-and-forth structure: it shows the limitation, then the defense, then a validation claim. Then it goes with a general reassurance, but then explain the limitation again, to then another defense and ends with a strong reassurance. I suggest consolidating and reorganizing it better. Then the final sentence of Sect. 4 “Overall, the model reproduces the seasonal-scale structure of carbon cycling and nutrient dynamics in Tokyo Bay and is therefore suitable for analysing mechanistic responses to variations in external loading,” feels repetitive again after reading the above highlighted paragraph.
-Line 215: Fig. 5 (A) and (B) show only the model validation, not the “differentiation in carbon-control structures, including relative changes in benthic carbon storage”
-Table 2: It is not clear whether these statistics are averaged over all 26 boxes or only the boxes shown in Fig. 5, and this should be stated explicitly here or in the methods. If it is an average over the 26 boxes, it is then unclear what depths "surface" and "bottom" refer to here. I would also suggest incorporating the “note” into the main table caption since it doesn’t add something new and some abbreviations are repeated.
-Line 245: The nutrient sensitivity experiments scaling DIN and DIP is a reasonable and useful choice for isolating the effect of loading magnitude, but it means the experiments cannot capture scenarios where N and P change disproportionately. For example, differential watershed retention, denitrification, changes in fertilizer composition, or land-use change could shift nutrient limitation and, in turn, affect the primary production and carbon cycle responses. I would suggest the authors briefly acknowledge this (1) as a limitation in the Discussion, (2) as a different approach to managing nutrients in the Bay, and/or (3) as a potentially useful direction for future work.
-Sect. 5.2. I think it would be useful for each function to state what is plotted, what is summed and excluded, and the sign convention. For example, carbon uptake is clearly defined as the air–sea CO2 flux, FCO2, though later discussion and schematic (Fig. 7) sometimes treat pCO2 itself as the uptake metric rather than a driver of FCO2. Please keep this distinction consistent. For carbon fixation, no diagnostic equation is given, despite the decomposition figure already showing the relevant component fluxes. It would be nice to state explicitly with an equation (e.g., DIC-consuming minus DIC-producing biological and chemical terms) the single quantity plotted as "fixation". Note also that Sect. 5.2.2 describes the DIC budget as integrated over the entire pelagic water column, whereas the fixation figures describe the plotted quantity as vertically averaged, please clarify. It should also be clarified that this represents only the biological and chemical contribution to the DIC tendency, because I think it doesn’t include transport and boundary exchange terms (see Major Comments about missing physical terms). For carbon storage, the manuscript describes it two different ways: A burial flux (“The carbon storage function is defined based on the burial flux of organic matter and CaCO3 that becomes permanently sequestered in deeper sediment layers.”) and a residual of supply minus remineralization/dissolution losses (“is determined by the dynamic balance between sedimentary supply fluxes (S) and decomposition or loss fluxes (D)”....Note that no physical transport, resuspension, sediment-water exchange, etc is mention to affect the budget, which sounds incomplete). Those might be equivalent under a closed steady-state sediment budget, but see my comment Sect. 5.2.3 below. For the second, it would be nice to state explicitly which quantity is plotted as "storage" and how organic carbon and CaCO3 contributions are combined.
-Sect. 5.2.3: The manuscript describes carbon storage as an annual-to-decadal function, with the two definitions stated above. However, since each scenario is run to a repeating annual equilibrium cycle, it is not obvious how this permanent sequestration is implemented: a permanently accumulating carbon stock could not itself be among the state variables required to reach periodic steady state, unless burial is effectively zero, material is removed from the active domain once buried, or accumulation is tracked as a separate diagnostic quantity outside the equilibrated system. Fig. 13 reinforces this ambiguity, as panel A appears to show annual burial and storage fluxes while panel C separately shows benthic organic matter and CaCO3 stocks. I would ask the authors to reconcile this “permanently sequestered” inconsistency and also clarify what quantities are plotted in Figs. 13/16/19: fluxes, annual integrated totals, or stocks?.
Sect. 6: The material at the start of Sect. 6, up through the description of the DIC–TA–pH–pCO2 system, largely describes what the model simulates. I would suggest moving this to the model description in Sect. 2, since it fits there more naturally. Additionally, this section makes some fairly strong claims and definitions that would benefit from supporting references. Examples: “In some coastal systems, large amounts of organic matter are laterally transported from external sources and processed by benthic organisms and microbial communities (REF)“, “Calcifying organisms produce CaCO3 shells that accumulate in sediments after death...This mechanism constitutes a long-term carbon-storage pathway within the carbonate loop (REF)”, among others. Finally, Sect. 6 organizes uptake, fixation, and storage around a temporal hierarchy (short-term, seasonal-to-annual, and annual-to-decadal, respectively), but the same diagnostics (10-day moving-average time series and annual mean) are used for all three functions, so the analysis doesn't demonstrate their different adjustment timescales. Because the analysis focuses on the periodic annual equilibrium, I would suggest the authors either present this hierarchy as a process-based expectation and acknowledge that its timescales are not directly tested, or, if feasible, support it with a transient analysis (e.g., tracking uptake, fixation, and storage as the system relaxes from one equilibrium to another following a change in loading).
-R3 mechanism: This is a perfect example of why more background in the introduction or Sect. 3 is needed. Without background, when Sect. 6 suddenly says “blooms collapse when nutrients become depleted,” the reader has no way to know whether this is Tokyo Bay ecology, a model behavior, or a generic hypothesis. What generally terminates blooms: nutrient limitation, grazing, mixing, light, or combinations?. The reader needs to know what the normal seasonal phytoplankton cycle is in Tokyo Bay. For example, Line 185 says “This state is interpreted as representing the typical seasonal ecosystem conditions of Tokyo Bay...” but we don’t know what the typical conditions are. Then R3 should either demonstrate from the simulations that nutrient depletion actually precedes the modeled bloom decline, or rephrase it more generally.
-Figure 8 and 9: These figures introduce new colors, type of arrows (red arrow), dashed rectangles, and changes within panels not explained in the caption. For example, why is F1 DIC at the surface blue and then in F2 is black? What’s the small red arrow in F2 ?. This makes it confusing what each panel is trying to communicate. In Figure 9, I don’t understand the inclusion of bottom hypoxia or suspension feeders in D1 and D2, if D1 is strong remineralization and D2 talks about lower pH. Sure, both are associated with hypoxia because of remineralization, but that’s not the main point here.
-S3 mechanism: Could laterally transported organic matter also affect remineralization, net biological DIC consumption, and air-sea CO2 exchange? In other words, could the underlying process represented by S3 also contribute to the fixation and uptake responses?.
-Figure 11: Sect. 5.2 states that 10-day moving averages are used for carbon uptake (Sect. 5.2.1) and carbon storage (Sect. 5.2.3), whereas the corresponding figures (Figs. 11, 14, 17 for uptake; Figs. 13, 16, 19 for storage) describe the time series as 1-day moving averages. Please, clarify the final moving-average day. In any case, why is there so much high-frequency variability in these time series?. I would expect a smoother seasonal cycle and not phytoplankton going up and down every ~1 month. Finally, the thin vertical reference lines in the time-series panels are not explained in the caption (the same happens in the other figures).
Figure 12: “(A) Vertical average of pelagic [mmolC m-3 h-1¹]”... of pelagic what? I assume it is Organic M and CaCO3 components in A1 and total carbon-fixation flux in A2. In general, everything in a figure should be well explained in the titles, axes, and caption. Panels should use a consistent format: either variable name and units on the y-axis or just in the titles (Panel B and C). Please also describe in the caption the DIC concentration shown in panel C. Finally, the title in B says “carbon storage flux,” but this is fixation (the same happens in Fig. 15 and 18)
Figure 13: Bottom DO appears in Fig. 11(g,l) in mmol L-1 and again in Fig. 13 in mg L-1. I think they represent exactly the same model layer. I would suggest using consistent units, ideally mg O2 L-1, following the convention already established in Table 1 and the validation figures and tables. In general, the same diagnostic variable should not switch units between adjacent figures without explanation. Where DO is repeated primarily to provide mechanistic context, a brief cross-reference to Fig. 11 rather than a full repeated panel may help with the manuscript's overall length.
-Line 650: “Although remineralization (U) also increases...”. U is the broader family of fixation-reduction mechanisms: Just U1 specifically represents enhanced remineralization, while U2 is hypoxia-induced suppression.
-Line 660: “fixation flux (i.e., production rate)”, but this is not consistent with the paper’s own definition. In Sect. 5.2.2, fixation is explicitly net biological DIC consumption, after accounting for production and respiration, remineralization, and dissolution. So fixation flux is not simply the production rate. The claim that mortality “increases” doesn't match Fig. 13(g) unless they specifically mean mortality period or severity rather than annual mortality flux.
-Line 680: “...while carbonate supply remains limited, and no pronounced amplification of dissolution (D1–D2) is observed”. Per the paper's own definition, D1 is enhanced organic matter remineralization exceeding sedimentary supply and D2 is CaCO3 dissolution under low pH conditions, so together they are broader storage loss processes, not just “dissolution”.
-Sect. 7.3.2: The paragraph itself starts by saying exactly that: “carbon fixation flux increases markedly with increasing nutrient loading.” But a few sentences later it says: “the F–U difference does not increase monotonically…”. Fig. 15(A) looks monotonic to me: the annual-mean carbon-fixation flux increases from 0.5x --> control --> 2x --> 5x. Indeed, both gross production and gross decomposition are co-amplified, but net fixation still increases monotonically because production increases more strongly than the losses.
-Sect. 7.4.2: The claim that mortality “increases during the summer hypoxic period” is not directly shown, because Fig. 19 gives only an annual-mean mortality bar, not a mortality time series. I guess the summer timing is inferred from the coincidence of low bottom DO and the collapse of suspension and deposit feeders in panels k–m, and is marked by those two vertical lines (which are not described in the caption).
-Sect. 7.4.3: The claim of “repeated emersion–inundation cycles that sustain oxygen supply” makes me wonder how this process is actually represented in this box model?
-Sect 8: As mentioned in the Major Comment, to avoid repetition with Sect 7 summaries, I would suggest condensing the summaries into a single results summary at the beginning of Sect. 8, which could make use of Table 3, rather than restating it across multiple subsections. I would also suggest that the authors consider adding, either at the beginning or end of Sect. 7, a modest whole-bay synthesis: for example, a summary figure or table showing bay-wide responses in uptake, fixation, and storage across the four nutrient scenarios. Figure and tables can even go to the supplementary. It would be nice to see if the strongly responding estuarine region occupies enough of the bay's total area to influence the whole Tokyo Bay carbon budget, or whether the central bay dominates it.
-Sect. 8.3: I would not be so definitive with “nonlinearity in coastal carbon cycling arises not from abrupt threshold behavior but from continuous modulation..” since here only 4 experiments have been tested. You can have a sudden transition or “bifurcation” in between.
-Conclusion: "Accordingly, the concept of carbon-control structure proposed in this study provides a fundamental framework for process-based coastal ecosystem modelling and for the design of coastal carbon-management strategies." The statement is broader than what the study demonstrates. The framework is developed and tested using one multi-box model applied to Tokyo Bay under idealized, fixed-stoichiometry nutrient-loading scenarios. This doesn’t establish it as a "fundamental" framework applicable generally to coastal ecosystems. Similarly, the study doesn't evaluate management strategies directly (it varies riverine nutrient loading at a fixed N:P ratio), so it may inform understanding of how nutrient loading affects carbon processes, but doesn't demonstrate how to design management strategies. I would ask the authors to soften this conclusion and specify more precisely the systems, processes, or management context to which they believe the framework applies.
Technical comments:
-Line 180: I counted 32 state variables
-Check the typos in Figures 12, 15, and 18 about carbon “storage”. It should be carbon fixation.
-Check consistency in units throughout figures and text
-Figure 10: Red says “CO”.

Citation: https://doi.org/10.5194/egusphere-2026-2710-RC3
- AC3: 'Reply on RC3', Akio Sohma, 02 Sep 2026
  
  General response
  We sincerely thank the reviewer for the detailed and constructive assessment of our manuscript. We appreciate the reviewer’s recognition of the value of the three-region comparison and agree that the original version required clearer methodological documentation, stronger validation, tighter consistency between operational definitions and mechanistic interpretation, and a more substantive Discussion. In the revised manuscript, we will substantially reorganize the presentation, strengthen the application-specific model and validation documentation, replace the original A/R–F/U–S/D mechanism notation with a function-based framework, correct the identified inconsistencies in the regional interpretations, and state the scope and limitations of the analysis more explicitly.
  Major comments
  Major Comment 1. Model configuration and reproducibility
  Response:
  We thank the reviewer for identifying the need for substantially greater application-specific documentation. We agree that the original manuscript relied too heavily on references to earlier Tokyo Bay applications and did not provide enough information to reproduce the present numerical experiment.
  In the revised manuscript, we will expand Sects. 2 and 3 to describe the spatial architecture and numerical configuration explicitly. EMAGIN-B.C. ver. 2.0 will be described as a spatially interconnected, vertically resolved benthic–pelagic ecosystem model. In the present Tokyo Bay application, the model is configured as a multi-box system consisting of 26 horizontal computational cells, 2-m pelagic layers whose number varies with local water depth, and 30 benthic layers resolving the upper 10 cm of sediment. Adjacent horizontal computational cells and vertical pelagic layers are linked through physical transport. The physical transport, pelagic ecosystem, benthic sediment, and carbonate-system calculations use a common time step of 0.2 h (12 min), and model output is saved at the same interval.
  We will also provide the general advection–diffusion–reaction equation for a pelagic tracer and clarify that the same formulation represents exchange among adjacent horizontal cells and vertical pelagic layers. The hydrodynamic fields are obtained from the hydrodynamic calculation and aggregated to the ecological cells while maintaining water-volume continuity. We will replace the ambiguous phrase “across offshore boundaries” with a clear statement that the open boundary is the mouth of Tokyo Bay, where exchange with the adjacent marine domain is imposed through the prescribed boundary conditions.
  The revised Sect. 3 will identify the principal data sources used to construct freshwater inflow, material loads, meteorological forcing, and open-boundary physical and biogeochemical conditions. The one-year periodic functions are constructed from the 1998–2002 observations using linear interpolation in time and are repeated as a representative seasonal cycle rather than as a reconstruction of a particular calendar year. Initial ecological variables are set from the mean 1 April values derived from the same observational period. All four nutrient-loading cases are initialized from the same initial conditions and integrated independently for approximately 50 years with their scenario-specific DIN and DIP loads applied from the start, until a stable periodic annual cycle is obtained.
  We will further clarify that the fourteen loading points are spatially aggregated external-input locations rather than fourteen individual rivers. Their baseline inputs include contributions associated with major river inflows, rainfall-derived terrestrial runoff, and direct coastal point discharges. The nutrient scenarios multiply the resulting total aggregated DIN and DIP load time series while retaining the spatial allocation and seasonal pattern.
  Finally, we will clarify the biological aggregation used in the present application. Phytoplankton and zooplankton are each represented as a single aggregated functional compartment, benthic fauna are represented by suspension feeders and deposit feeders, and biological shell CaCO₃ formation is assigned only to the calcifying suspension-feeder compartment. Together, these revisions will provide the application-specific information needed to understand and reproduce the present experimental setup while referring to Sohma et al. (2008, 2018) for the previously published generic model formulations.
  Major Comment 2. Scope and documentation of model validation
  Response:
  We thank the reviewer for this detailed assessment. We agree that the original validation was insufficiently documented for a study whose conclusions depend on carbonate chemistry and benthic–pelagic coupling.
  In the revised manuscript, Sect. 4 and Appendix Table A1 will identify the observational programs, periods, locations, and variables used in the validation. The conventional pelagic variables will be based on the FY1998–FY2002 public-water monitoring datasets and the Comprehensive Survey on Regional Water Quality, while available Tokyo Bay carbonate-system observations reported by Kubo et al. (2017) will be used for DIC, pH, and pCO₂. Banzu Tidal Flat and additional benthic/sediment datasets will also be documented explicitly.
  We will clarify how observations are assigned to the corresponding model grid cells and vertical layers. Because the model represents a repeating annual cycle rather than individual calendar years, the comparison will be made by location, model layer, and seasonal timing. “Surface” and “bottom” will be defined as the uppermost and lowermost pelagic model layers, respectively. Fig. 4 will show the relevant model-cell depths and validation locations, and Fig. 5 will provide representative seasonal comparisons for the estuarine, central-bay, and tidal-flat sites.
  The biological and organic-carbon quantities will also be defined explicitly. The ambiguous abbreviation PP will be replaced by PHY for phytoplankton carbon biomass. Observed chlorophyll-a will be converted to phytoplankton carbon using the carbon-to-chlorophyll-a ratio used in the model. Modeled detrital POC will be the sum of the fast-labile, slow-labile, and refractory detrital pools; total POC will be PHY + ZP + detrital POC; modeled DOC will be the sum of the labile and refractory dissolved pools; and TOC will be total POC + DOC. For Table 2, field POC will be treated operationally as the observational counterpart of modeled detrital POC, and this limitation will be stated explicitly.
  We will extend the evaluation beyond the pelagic water column. DIC, pH, and pCO₂ will be included where suitable observations are available, while the lack of suitable observations for TA and pelagic CaCO₃, as well as directly matched observations of air–sea CO₂ flux, will be stated explicitly. Appendix figures will compare available benthic-faunal biomass, sedimentary detrital organic carbon, porewater nutrients, and sediment DO profiles. Because some benthic observations are sparse and not temporally matched to the periodic simulation, these comparisons will be interpreted primarily in terms of order of magnitude, broad seasonal tendency, vertical structure, and oxygen penetration depth.
  The revised text will also avoid treating a high bottom-water DO correlation as independent validation of every underlying process. Model performance will be assessed from the combined behavior of nutrient, oxygen, organic-carbon, carbonate-system, and benthic/sediment variables, while the weaker agreement for bottom-layer PHY and detrital POC and the resulting uncertainty in absolute organic-carbon storage will be acknowledged. The subsequent scenario analysis will therefore emphasize mechanistic responses and relative differences among regions under a common model framework rather than implying uniformly high accuracy in all absolute quantities.
  Major Comment 3. Internal consistency of the mechanism framework
  Response:
  We thank the reviewer for this important critique. We agree that the original A/R–F/U–S/D notation created unnecessary cognitive load and, more importantly, allowed several candidate pathways to become inconsistent with the operational definitions of the three carbon functions.
  We will therefore discontinue the paired A/R–F/U–S/D notation. The three diagnostic functions will be referred to directly as carbon uptake, carbon fixation, and carbon storage. Candidate mechanisms will be labeled according to the function they affect: U_i+ and U_i− for uptake-enhancing and uptake-reducing mechanisms, F_i+ and F_i− for fixation-enhancing and fixation-reducing mechanisms, and S_i+ and S_i− for storage-enhancing and storage-reducing mechanisms. The superscript indicates the direction of influence on the diagnostic function and not the algebraic sign of an individual model flux. Descriptive mechanism names will accompany the symbols so that the interpretation does not depend on notation alone.
  We will also revise the underlying mechanism definitions. Trophic transfer of already fixed organic carbon will no longer be treated as additional DIC fixation; rather, it will be described as redistribution that can affect retention before carbon returns to DIC. The carbonate pathway associated with loss of calcifying suspension feeders will distinguish reduced shell formation from the separate consequence of reduced CaCO₃ dissolution-derived alkalinity supply, avoiding the incorrect implication that reduced calcification itself lowers TA. The previous unsupported switch between moderate- and severe-hypoxia biological responses will be replaced by pathways tied directly to processes represented and supported by the simulations.
  The storage mechanisms will be rebuilt so that Sect. 6 and Fig. 9 describe the same causal pathways. Carbon storage will have a single operational definition—the transfer flux across the lower boundary of the actively resolved upper sediment—and the candidate mechanisms will describe processes that enhance or reduce that flux. Unsupported attribution to externally supplied organic carbon will be removed because the present analysis does not partition benthic carbon by local versus lateral origin.
  Figs. 7–9 will be explicitly presented as conceptual causal-pathway diagrams rather than mass-balance diagrams. Their captions will explain the graphical conventions and note that the candidate pathways can share individual processes and operate concurrently. Table 3 will provide the compact regional synthesis. This restructuring is intended to preserve the useful mechanistic interpretation while making it fully consistent with the three quantities actually evaluated in Sect. 5.
  Major Comment 4. Inconsistencies between results, figures, and carbon budgets
  Response:
  We thank the reviewer for the careful cross-check between the figures and the Results. We agree that several statements in the original manuscript either compared quantities that were not the same budget or carried mechanistic interpretations beyond what the plotted variables demonstrated.
  First, we will make an explicit distinction between the diagnostic biological/chemical carbon-fixation balance and the complete prognostic DIC budget. The fixation diagnostic sums the defined biological and chemical DIC-consuming and DIC-producing terms, whereas the complete DIC tendency additionally includes physical transport and exchange, together with air–sea CO₂ exchange. We will state this distinction in Sect. 5.2.2 and in the relevant figure captions, so that an increase in a biological/chemical production–consumption sum is not interpreted as requiring the vertically averaged or surface DIC concentration to change in the same direction.
  We also re-examined the complete DIC budget for the three analysis regions and confirmed that it closes in the model calculations. Because Figs. 12, 15, and 18 are intended to diagnose the biological and chemical carbon-fixation balance rather than to present the complete DIC mass balance, we will not add a separate full-budget figure.
  Second, we will revise the regional Results to refer only to quantities actually shown in the corresponding figures. Statements about bottom-layer remineralization or bottom-layer DIC will not be inferred from panels that show surface-layer component fluxes. The central-bay uptake response will be described as a monotonic shift toward stronger atmospheric CO₂ uptake where that is what the air–sea flux shows; co-amplification will refer to the simultaneous strengthening of opposing component processes, not to a non-monotonic net uptake response.
  Third, the tidal-flat carbon-fixation interpretation will be corrected. The annual mean organic-carbon component of the net fixation balance becomes increasingly negative with nutrient loading, while the CaCO₃ contribution remains smaller. We will therefore describe the tidal-flat response as increasingly dominated by carbon return to DIC despite stronger direct fixation processes, rather than stating that net fixation increases.
  Finally, we will remove the unsupported claim that the tidal-flat transformation-dominated structure is driven specifically by externally supplied organic carbon. The model permits lateral carbon exchange, but the present diagnostics do not partition the source of benthic organic carbon. The revised interpretation will be based on the directly resolved process balance: photosynthesis and shell formation increase, but benthic respiration and remineralization increase more strongly, producing an increasingly negative net fixation balance while both organic-carbon and CaCO₃ storage fluxes increase and CaCO₃ storage remains dominant.
  We will carry these corrections through the regional summaries, cross-region synthesis, Discussion, Table 3, and figure captions so that the later interpretation no longer reproduces the inconsistencies identified in the original Results.
  Major Comment 5. Manuscript length, repetition, and figure organization
  Response:
  We thank the reviewer for this structural recommendation. We agree that the original manuscript repeated regional conclusions too often and did not sufficiently distinguish the roles of the definition, mechanism, evidence, and synthesis sections.
  We will streamline the manuscript while retaining the process-level evidence needed for the central mechanistic argument. The former Fig. 3, which documents calculation flow rather than a scientific result, will be moved to Appendix Fig. B1. Additional benthic and sediment validation material will be placed in the Appendix. Fig. 10 will provide a compact side-by-side cross-region comparison of the three carbon functions and associated balances, and Table 3 will provide a concise mechanism matrix.
  We have reconsidered moving Figs. 11–19 out of the main text. Because the study's main objective is to identify region-specific carbon-control structures from the underlying process balances, we consider the systematic three-region-by-three-function figure set to be direct evidence rather than ancillary material. We will therefore retain these figures in the main text, but revise Sect. 7 so that only the panels needed for the mechanistic interpretation are explicitly discussed and cited at panel level. Captions, axes, legends, and graphical conventions will be made consistent so the figures are easier to compare across regions.
  We will also reduce repeated summaries. The regional subsections will focus on the evidence for each region, the cross-region section will synthesize the contrasts, and the Discussion will move from those results to interpretation, comparison with previous work, limitations, and implications instead of restating the same regional summaries.
  We do not plan to merge Sects. 5 and 6 because they serve different analytical functions. Sect. 5 defines the operational metrics used to quantify uptake, fixation, and storage; Sect. 6 provides the candidate causal mechanisms used to interpret those metrics; and Sect. 7 tests which candidate mechanisms are expressed in the simulations. We will make this division explicit and remove overlapping explanatory material so that the separation improves, rather than duplicates, the logic of the paper.
  Major Comment 6. Need for a substantive Discussion
  Response:
  We thank the reviewer and agree that the original Sect. 8 was not sufficiently developed as a Discussion. In the revised manuscript, Sect. 8 will be reorganized so that it does not simply repeat the regional Results.
  The revised Discussion will first interpret the major regional contrasts in relation to prior observations and modeling studies of Tokyo Bay carbon cycling, nutrient enrichment, hypoxia, and benthic–pelagic coupling, and then consider which aspects are likely to be more generally relevant to heterogeneous urban coastal systems. The discussion will explicitly distinguish direct simulation results from broader process-based interpretation.
  We will also add a dedicated limitations component. This will acknowledge that DIN and DIP were scaled together at a fixed external N:P ratio, so the simulations do not test independent N and P perturbations or shifts in nutrient limitation; that the forcing is an idealized repeating annual cycle rather than interannual forcing; that suitable observations are unavailable for direct validation of some quantities, including TA, pelagic CaCO₃, and directly matched air–sea CO₂ flux; that the storage diagnostic represents transfer below the actively resolved upper sediment rather than geological-scale permanence; that other calcifying planktonic groups are not independently represented; and that the analysis focuses on three representative regions rather than a full-domain attribution of every regional contribution to the total Tokyo Bay carbon budget.
  The policy framing will also be moderated. The nutrient-loading experiments are designed to isolate ecosystem responses to changes in total inorganic nutrient input and are not simulations of a specific nutrient-management policy. We will therefore discuss potential climate-mitigation co-benefits and trade-offs rather than presenting nutrient supply itself as a climate-change mitigation measure or claiming that the study directly designs management strategies.
  Finally, we will identify future tests that could strengthen and generalize the framework, including independent N and P perturbations, additional loading levels where threshold behavior is of interest, full-domain synthesis, application to other coastal systems, and simulations under interannual or transient forcing. These revisions will make Sect. 8 a genuine Discussion of significance, limitations, and generalizability rather than another summary of Sect. 7.
  Minor comments
  Minor Comment 1. Abstract: long tidal-flat sentence
  Response:
  We thank the reviewer for this suggestion. We agree that the original tidal-flat sentence in the Abstract was too dense. In the revised manuscript, we will divide it into shorter sentences and retain only the key contrast: the tidal-flat region exhibits strong benthic carbon turnover, an increasingly negative net carbon-fixation balance with nutrient enrichment, and carbonate-dominant sedimentary storage. The unsupported attribution to externally supplied organic carbon and the earlier overemphasis on hypoxia will not be retained.
  Minor Comment 2. Introduction: Tokyo Bay background
  Response:
  We thank the reviewer and agree that the regional context should be established before the model application. In the revised Introduction, we will add a concise Tokyo Bay background describing its semi-enclosed geometry, strong river-to-central-bay gradients, seasonal stratification and recurrent bottom-water hypoxia, shallow sandy tidal flats with active benthic communities, and previous observational and modeling evidence for spatially and seasonally heterogeneous carbon cycling. We will also explain why the estuarine, central-bay, and Banzu tidal-flat regions provide contrasting settings for testing region-specific carbon-control structures under nutrient enrichment.
  Minor Comment 3. Line 60: calcification, TA, pH, and pCO₂
  Response:
  We thank the reviewer for this clarification. In the revised Introduction, we will state explicitly that biological CaCO₃ formation represented in the present Tokyo Bay application occurs in the calcifying suspension-feeder compartment. We will also replace the vague wording “altering pCO₂” with the stoichiometrically explicit description that CaCO₃ formation decreases DIC by approximately 1 mol and TA by approximately 2 equivalents per mole of CaCO₃ formed and, under otherwise comparable carbonate-system conditions, tends to lower pH and increase pCO₂. We will avoid presenting this as an unconditional ecosystem-scale response because the realized pCO₂ response also depends on the coupled DIC and TA budgets.
  Minor Comment 4. Line 65: low-pH/high-pCO₂ river water
  Response:
  We agree that the original wording was overly general because riverine carbonate chemistry varies among watersheds. We will therefore remove the generalized characterization of river water as having low pH and high pCO₂ and revise the surrounding text to avoid assigning a universal carbonate-system effect to river inputs. Where riverine carbonate chemistry is relevant, we will describe it as system- and watershed-dependent and distinguish the specific Tokyo Bay forcing used in the model from general estuarine behavior.
  Minor Comment 5. Figure 1: role of the conceptual overview
  Response:
  We thank the reviewer. We will retain Fig. 1 because it serves as the single high-level conceptual entry point to the paper, but we will simplify and refocus it. The revised figure will explicitly highlight carbon uptake, carbon fixation, and carbon storage and show how they are embedded in interacting organic and carbonate pathways constituting the Dual Carbon Loop. The obsolete A/R–F/U–S/D notation will not be added; instead, the figure will introduce the three functions in the same terminology used throughout the revised manuscript. Detailed candidate mechanisms will remain in Figs. 7–9 so that Fig. 1 does not duplicate them.
  Minor Comment 6. Sect. 2: model architecture upfront
  Response:
  We agree and will clarify this at the beginning of Sect. 2. EMAGIN-B.C. ver. 2.0 will be described explicitly as a spatially interconnected, vertically resolved benthic–pelagic ecosystem model. In the present Tokyo Bay application, the model is configured as a multi-box system in which adjacent horizontal computational cells and vertical pelagic layers are linked through physical transport. The application-specific spatial configuration will then be described in Sect. 3.
  Minor Comment 7. Sect. 2 organization and distinction between ver. 1 and ver. 2.0
  Response:
  We thank the reviewer. We considered consolidating the short subsections. However, we will retain the seven subsections in Sect. 2.1 because they distinguish conceptually different structural components of EMAGIN-B.C. that are directly relevant to the present analysis. We will add a brief introductory sentence to clarify this organization. More importantly, Sect. 2.2 will explicitly identify the two process modifications in ver. 2.0 relative to Sohma et al. (2018): pH-dependent physiological/metabolic responses and the oxygen-dependent ATP-efficiency formulation for remineralization. The sources containing the functional forms and parameter definitions will be cited directly. We will retain the previously published generic formulations by reference, add the general physical tracer-transport equation in the main text, and move the former calculation-flow figure to Appendix Fig. B1. These revisions will provide the information needed to distinguish ver. 2.0.
  Minor Comment 8. Sect. 2.1.3: scope of biological calcification
  Response:
  We agree. In the revised manuscript, we will state explicitly that biological shell CaCO₃ formation in the present Tokyo Bay application is assigned only to the calcifying suspension-feeder compartment. Coccolithophores, foraminifera, pteropods, and other calcifying groups are not independently resolved. We will identify this as a model-scope limitation when discussing generalization of the carbonate loop beyond the present application. The conclusion will therefore refer to the carbonate pathway represented here rather than imply complete resolution of ecosystem calcification.
  Minor Comment 9. Figure 2 graphical conventions
  Response:
  We thank the reviewer. We will revise Fig. 2 and its caption so that the graphical conventions are explicit. Boxes will be identified as model variables or functional compartments and arrows as biological, chemical, or material-transfer pathways as labeled in the figure, including the exchanges linking pelagic and benthic systems. Abbreviations and units will be consolidated in the caption and Table 1 rather than redundantly spelled out within the diagram where this improves readability. The figure will also distinguish living biomass, nonliving detrital/DOM pools, and carbonate pools more clearly.
  Minor Comment 10. Figure 3: move calculation flow out of main text
  Response:
  We agree that the former Fig. 3 is methodological documentation rather than a central scientific result. Because this Biogeosciences workflow uses an Appendix rather than a separate Supplement for the present manuscript, we will move the figure to Appendix B as Appendix Fig. B1. The main text will retain only a concise reference to the coupling among DIC and TA production–consumption processes, physical transport, carbonate equilibrium, and state-variable updating.
  Minor Comment 11. Table 1: units of benthic dissolved variables
  Response:
  We thank the reviewer for identifying this ambiguity. In the revised Table 1 and its note, dissolved benthic variables will be stated explicitly on a porewater-volume basis. Benthic biomass and benthic CaCO₃ will be expressed per unit sediment surface area, sediment detritus per unit sediment-solids volume, and dissolved benthic variables per unit porewater volume. We will standardize the corresponding terminology and units in the figures and text.
  Minor Comment 12. Line 170: rivers and watershed characterization
  Response:
  We agree that the loading description needs more context, but we will also correct the implication that the fourteen loading points correspond to fourteen individual rivers. In the revised Sect. 5.1 and Fig. 6 caption, we will explain that the points are spatially aggregated external-input locations. Their baseline loads combine contributions associated with major river inflows, rainfall-derived terrestrial runoff, and direct coastal point discharges using source-specific loading data. We will identify the data sources and briefly describe the strongly urbanized/industrialized watershed setting relevant to the Tokyo Bay nutrient load. Where an individual river is specifically relevant to an analysis region, such as the Obitsu River near Banzu Tidal Flat, it will be named explicitly.
  Minor Comment 13. Line 175: optional sediment-profile diagnostics
  Response:
  We appreciate this suggestion and agree that sediment profiles are particularly useful for evaluating the benthic component. In the revised manuscript, rather than adding another large set of scenario-result figures to the main text, we will use the available benthic and sediment observations in the Appendix. These will include porewater nutrient profiles and sediment DO profiles, together with selected benthic biological and sedimentary organic-carbon comparisons. The profiles will be used as supplementary process-level validation of the benthic representation; we will not add additional scenario diagnostics unless they provide clear process-level information beyond the validation material already included.
  Minor Comment 14. Figure 4 caption and possible combination with Figure 6
  Response:
  We thank the reviewer. We will revise the Fig. 4 caption to state explicitly what the grid-cell depth numbers represent; in the present model map they denote the mean water depth assigned to each computational cell. We have also reconsidered combining Figs. 4 and 6. We will keep them separate because they serve different purposes: Fig. 4 documents model grid configuration and validation locations, whereas Fig. 6 identifies the aggregated external-input points and the three representative analysis regions. To reduce redundancy, both captions and map conventions will be coordinated and unnecessary repeated explanation will be removed.
  Minor Comment 15. Line 205: overinterpretation of bottom-water DO validation
  Response:
  We agree with this important caution. In the revised Sect. 4, bottom-water DO will be treated as a process-integrating validation variable rather than evidence that each underlying production, sinking, respiration, mixing, and sediment-demand process is independently validated. Table 2 will present correlation together with RMSE, mean observed values, significance, and sample size, and the text will discuss performance using multiple variables rather than emphasizing R alone. We will therefore remove the stronger process-specific validation claim and state more conservatively that the model reproduces major seasonal-scale features while uncertainty remains in individual process representations. We do not consider an additional Taylor diagram necessary because the revised Table 2 and validation figures provide the required quantitative context.
  Minor Comment 16. Sect. 4: repetitive limitation/defense structure
  Response:
  We agree. The revised Sect. 4 will be reorganized so that it first reports the principal validation results, then states the main discrepancies and observational limitations, and finally defines the appropriate scope of inference once. Repetitive reassurance statements will be removed. The section will end with a measured statement that the subsequent analyses emphasize relative and mechanistic responses among regions and nutrient-loading scenarios under a common model framework, while acknowledging uncertainty in some absolute modeled quantities.
  Minor Comment 17. Line 215: Fig. 5 does not demonstrate carbon-control differentiation
  Response:
  We agree and will correct this wording. Fig. 5 will be described only as model–data validation at representative sites. Claims about spatial differentiation of carbon-control structures will be reserved for the scenario results and synthesis in Fig. 10, Figs. 11–19, Sect. 7, and Table 3. This will separate model evaluation from the subsequent mechanistic inference.
  Minor Comment 18. Table 2: scope of observations and meaning of surface/bottom
  Response:
  We thank the reviewer for requesting this clarification. In the revised Sect. 4 and Table 2 caption/note, we will state that the statistics use all available paired observations included in the pelagic validation datasets after assignment to their corresponding model cells and layers; they are not restricted to the representative cells plotted in Fig. 5. “Surface” and “bottom” will denote the uppermost and lowermost pelagic model layers at the corresponding model location rather than one fixed observational depth across the bay. Fig. 4 will provide the grid-cell water depths and validation locations. Repeated abbreviation definitions will be consolidated so that the table caption and note remain concise.
  Minor Comment 19. Line 245: fixed N:P ratio in nutrient scenarios
  Response:
  We agree. The revised manuscript will state explicitly that DIN and DIP are multiplied by the same scenario factor and therefore the N:P ratio of the aggregated external inorganic-nutrient loads is held constant. The experiment is designed to isolate the effect of total nutrient-loading magnitude and cannot represent disproportionate changes in N and P or shifts in nutrient limitation. This limitation will be discussed in Sect. 8, and independent N and P perturbation experiments will be identified as an important direction for future work. We will not imply that the present scenarios reproduce every realistic nutrient-management pathway.
  Minor Comment 20. Sect. 5.2: operational definitions, included terms, exclusions, and signs
  Response:
  We thank the reviewer for this important suggestion. We agree that the operational definitions of the three carbon functions should state more explicitly what quantity is evaluated, how component terms are combined, what is excluded, and the corresponding sign convention.
  In the revised manuscript, we will clarify these definitions in Sect. 5.2 as follows.
  For carbon uptake, the diagnostic quantity is the net air–sea CO₂ exchange flux, F_CO₂, with flux from the atmosphere to the ocean defined as positive. Surface-water pCO₂ is therefore a driver of F_CO₂ through the air–sea CO₂ gradient, rather than the uptake metric itself. We will maintain this distinction consistently in the text and figures.
  For carbon fixation, we will define the operational metric explicitly as the net carbon-fixation balance and add a simple diagnostic equation expressing it as the sum of biological and chemical DIC-consuming terms minus the sum of biological and chemical DIC-producing terms. The included processes are photosynthetic DIC consumption, respiration and remineralization, CaCO₃ formation, and CaCO₃ dissolution in the pelagic and benthic systems. For each analysis grid cell, these component fluxes are combined and normalized by the volume of the entire pelagic water column. A positive value therefore indicates that conversion of DIC into organic carbon and CaCO₃ exceeds the return of fixed carbon to DIC, whereas a negative value indicates net regeneration of DIC. We will clarify that this diagnostic is not the complete prognostic DIC budget: physical DIC transport and exchange and air–sea CO₂ exchange are excluded from the fixation diagnostic and are treated separately in the complete DIC balance.
  For carbon storage, we will use a single operational definition: the transfer flux of organic carbon and CaCO₃ across the lower boundary of the actively resolved upper sediment into deeper permanent sediment layers. Positive storage flux will denote transfer toward these deeper sediment layers. Organic-carbon and CaCO₃ storage fluxes will be shown as separate components in Figs. 13, 16, and 19. Sedimentation into the upper sediment, temporary changes in upper-sediment carbon stocks, remineralization, and CaCO₃ dissolution will not be treated as alternative definitions of storage; rather, they are processes that influence the magnitude of the transfer flux to permanent sediment layers.
  We will also revise the corresponding figure captions and terminology so that the normalization and sign conventions are stated consistently throughout the manuscript.
  Minor Comment 21. Sect. 5.2.3: periodic steady state versus “permanent” storage
  Response:
  We thank the reviewer for identifying this ambiguity. We agree that the original wording did not clearly distinguish between the periodically equilibrated carbon stocks within the actively modeled sediment domain and the carbon-transfer flux used as the storage diagnostic.
  In the revised manuscript, we will clarify that carbon storage is not represented as a permanently accumulating state variable within the actively resolved upper sediment. Instead, it is operationally defined as the transfer flux of organic carbon and CaCO₃ across the lower boundary of the actively resolved sediment domain toward deeper sediment layers. The periodic steady-state condition applies to the state variables within the actively modeled domain, which return to the same seasonal cycle each year. A nonzero downward transfer flux across the lower sediment boundary is therefore compatible with periodic steady state of the actively modeled carbon stocks.
  We will also clarify the meaning of “permanent” in this context. The term refers operationally to transfer below the actively resolved upper-sediment domain; the model does not explicitly simulate subsequent deeper-sediment diagenesis, geological-scale preservation, or carbon residence time. We will therefore avoid interpreting this diagnostic as a direct measure of geological-scale permanence or deep-ocean sequestration and will state this limitation explicitly.
  We will further clarify the quantities shown in Figs. 13, 16, and 19. The carbon-storage quantities plotted in these figures are fluxes, not accumulating carbon stocks or annual integrated totals. Panel A shows annual mean organic-carbon and CaCO₃ storage fluxes to the deeper sediment layer, whereas the corresponding seasonal panels show 1-day moving averages of these storage fluxes. Other panels show biological state variables and process diagnostics that help interpret the mechanisms controlling the storage flux but do not themselves constitute the carbon-storage metric.
  These revisions will distinguish clearly between temporary carbon stocks within the actively modeled sediment, transfer across the lower sediment boundary, and longer-term preservation processes that are outside the scope of the present model.
  Minor Comment 22. Sect. 6: organization, supporting references, and temporal hierarchy
  Response:
  We thank the reviewer for these helpful comments on the organization and interpretation of Sect. 6. We agree that the original version mixed model description, mechanistic interpretation, and statements about characteristic timescales more than was necessary.
  In the revised manuscript, we will separate these roles more clearly. Material describing the model structure itself, including carbonate-system calculations and the coupling among biological, chemical, and physical processes, will be placed in Sect. 2. Sect. 6 will instead be restricted to the interpretive framework used in the present analysis: the Dual Carbon Loop and the candidate mechanisms through which nutrient loading may enhance or reduce carbon uptake, the net carbon-fixation balance, and carbon storage. We will clarify that these candidate pathways are not independent model components and that Sect. 7 evaluates which of them are supported by the simulated process balances in each region.
  We will also revise or remove overly general statements that are not directly supported by the present analysis. In particular, we will avoid attributing the tidal-flat response specifically to externally supplied organic carbon because the present analysis does not partition fixed carbon according to local versus lateral origin. Likewise, carbonate-related mechanisms will be described in terms of the calcifying suspension-feeder compartment explicitly represented in the present model rather than generalized to calcifying organisms more broadly. Where general process statements remain necessary, we will provide appropriate supporting references.
  We also agree with the reviewer that the present simulations do not directly demonstrate distinct adjustment timescales for uptake, fixation, and storage. Each nutrient-loading scenario is analyzed after reaching a periodic steady state, and the annual-mean and seasonal diagnostics characterize processes within that equilibrated annual cycle rather than the transient relaxation following a change in loading. We will therefore present the short, intermediate, and long adjustment timescales only as process-based expectations informed by characteristic process and turnover timescales, and will state explicitly that these adjustment timescales were not quantified using transient perturbation experiments.
  Because quantifying transient adjustment times is outside the scope of the present periodic steady-state comparison, we will not infer quantitative response times from the current simulations. A transient perturbation analysis would be a useful subject for future work but is not required for the mechanistic comparisons undertaken here.
  Minor Comment 23. Former R3 mechanism: bloom decline and nutrient depletion
  Response:
  We thank the reviewer for identifying this unsupported causal statement. We agree that the original wording did not establish whether nutrient depletion was a demonstrated model result, a Tokyo Bay-specific observation, or a generic hypothesis.
  In the revised manuscript, we will not claim that nutrient depletion is the cause of modeled bloom termination unless that sequence is directly demonstrated. The former R3 pathway will be recast under the revised notation as U_3−, “post-production DIC accumulation through respiration and remineralization.” It will state more generally that periods of nutrient-enhanced production can be followed by declining photosynthetic DIC consumption while respiration and remineralization of accumulated organic matter continue or intensify, increasing DIC and pCO₂ and thereby reducing atmospheric CO₂ uptake or promoting release.
  We will also strengthen the Introduction with concise background on the normal seasonal productivity, stratification, and hypoxia of Tokyo Bay. This will provide the regional ecological context without assigning bloom termination to a single mechanism. Thus, the revised pathway will be a process-based candidate mechanism supported by the simulated balance rather than an unsupported claim that nutrient depletion specifically precedes every bloom decline.
  Minor Comment 24. Figures 8 and 9: unexplained graphical conventions and pathway clutter
  Response:
  We thank the reviewer. We agree that the original Figs. 8 and 9 introduced graphical conventions that were not adequately explained and, in some panels, included secondary steps that obscured the direct mechanism.
  In the revised figures, we will standardize the visual logic across Figs. 7–9 and explain it explicitly in the captions. Boxes will denote forcing/state or carbon-pool elements, black arrows will denote causal or material-transfer links, highlighted arrows will identify the direct diagnostic process step where needed, and dashed outlines will be used only when their meaning is stated. A given variable will use a consistent color unless a change is explicitly defined.
  We will also simplify the pathways so that each panel emphasizes the process that directly affects the corresponding diagnostic function. In the fixation figure, trophic transfer will be shown as redistribution of already fixed carbon rather than additional fixation. In the storage figure, hypoxia or suspension-feeder changes will be retained only where they are causally necessary to explain changes in sedimentary supply, shell formation, mortality-related shell transfer, or dissolution. The obsolete D1/D2 structure will be replaced by the revised S_i− mechanisms so that the figure and Sect. 6.3 describe the same processes.
  Minor Comment 25. Former S3 and cross-function role of lateral organic carbon
  Response:
  We thank the reviewer and agree that the original S3 formulation conflated carbon provenance, physical transport, and the diagnostic functions. Lateral transport of already fixed organic carbon does not itself constitute new carbon fixation, and the present analysis does not partition benthic organic matter according to local versus external origin.
  In the revised manuscript, we will therefore remove the former storage-specific S3 mechanism based on “externally supplied organic carbon.” Physical transport will be treated as part of the complete prognostic budgets and as a process that can redistribute carbon among model cells. Once transported organic carbon is respired or remineralized locally, that local return of fixed carbon to DIC can reduce the net carbon-fixation balance and, through its influence on DIC/pCO₂, can also affect air–sea CO₂ uptake. However, these effects will be described through the relevant U_i− or F_i− pathways rather than by labeling lateral transport itself as fixation or storage. This revision keeps provenance, transport, and functional diagnostics conceptually distinct.
  Minor Comment 26. Moving-average terminology and regular temporal variability
  Response:
  We thank the reviewer for drawing attention to this presentation issue. The seasonal curves are based on 1-day moving averages, not 10-day moving averages, and we will correct this terminology consistently throughout the text and captions.
  We will also explain the regular residual variability. A 1-day moving average suppresses most diel and semidiurnal variability, but the model retains periodic solar and tidal forcing, including spring–neap modulation generated by the M2, S2, K1, and O1 tidal constituents and associated lower-frequency ecosystem responses. These forcings produce the remaining regular variability visible in the seasonal curves.
  Any vertical reference lines retained in the final figures will be defined explicitly in the corresponding caption; if they do not carry a necessary analytical meaning, they will be removed. This will prevent graphical markers from being interpreted as unreported events or thresholds.
  Minor Comment 27. Figure 12 labels and DIC quantity
  Response:
  We thank the reviewer. We will revise Figs. 12, 15, and 18 so that all panels and axes are explicitly labeled as carbon-fixation diagnostics rather than carbon-storage quantities. Panel A will distinguish the annual mean organic-carbon and CaCO₃ components of the net carbon-fixation balance from the seasonal 1-day moving-average time series. Panels B and C will be labeled as component fluxes contributing to the net carbon-fixation balance.
  We will also state explicitly that the DIC line in panel C is the vertically averaged pelagic DIC concentration for the corresponding analysis grid cell. The carbon-fixation balance and its component fluxes are normalized by the volume of the entire pelagic water column. The same definitions, units, and graphical conventions will be applied consistently to Figs. 12, 15, and 18.
  Minor Comment 28. Figure 13 DO units and repeated DO panel
  Response:
  We thank the reviewer. We will standardize the process-analysis figures so that bottom-water DO is expressed consistently in mol O₂ m⁻³, while model–observation validation figures may retain the conventional mass-based units of the observational datasets for direct comparison. Corresponding panels in Figs. 13, 16, and 19 will use the same unit and notation.
  We will retain bottom-water DO in the storage figures because oxygen conditions are mechanistically relevant to benthic-faunal biomass, mortality, remineralization, and CaCO₃ storage pathways. However, the captions and Sect. 7 will make clear that DO is a contextual process diagnostic and not part of the operational storage metric itself. This should make the repeated panel informative rather than redundant.
  Minor Comment 29. “Remineralization (U)” notation
  Response:
  We thank the reviewer for identifying this notation error. We agree that the original phrase “remineralization (U)” was incorrect because “U” referred to the broader family of fixation-reducing mechanisms rather than specifically to remineralization.
  In the revised manuscript, we will remove this usage as part of the broader replacement of the original A/R–F/U–S/D notation with the function-based U_i+/U_i−, F_i+/F_i−, and S_i+/S_i− framework.
  In the revised Sect. 7.2.1, respiration and remineralization will be described directly as DIC-regenerating processes that exert an uptake-reducing influence, consistent with the post-production DIC-accumulation pathway U_3−. When the same processes are discussed in relation to the net carbon-fixation balance in Sect. 7.2.2, their direct return of fixed organic carbon to DIC will instead be identified as the fixation-reducing mechanism F₁⁻.
  This revision will avoid using a family-level symbol as a synonym for an individual process and will clarify that the same underlying biogeochemical process can influence different diagnostic carbon functions through different causal pathways.
  Minor Comment 30. Sect. 7.2.2: fixation is not production rate; mortality interpretation
  Response:
  We thank the reviewer for identifying both inconsistencies. We agree that the phrase “fixation flux (i.e., production rate)” was inconsistent with the operational definition of carbon fixation used in Sect. 5.2.2.
  In the revised manuscript, we will clearly distinguish gross direct fixation from the net carbon-fixation balance. Direct fixation occurs through photosynthesis and shell formation, whereas the operational carbon-fixation function is evaluated from the net balance between DIC-consuming processes and processes returning fixed carbon to DIC, including respiration, remineralization, and CaCO₃ dissolution. Accordingly, Sect. 7.2.2 will state that the net carbon-fixation balance increases with nutrient loading primarily because strengthened pelagic photosynthetic fixation exceeds the opposing carbon-return processes, rather than equating the fixation function with gross production rate.
  We also agree that the original statement that benthic-faunal mortality “increases” was not supported by the annual-mean mortality flux shown in Fig. 13. In fact, the annual-mean benthic-faunal mortality flux decreases with increasing nutrient loading as the standing benthic biomass available for mortality declines. We will therefore distinguish this annual-mean flux from the seasonal biomass response. The seasonal trajectories show summer declines in suspension- and deposit-feeder biomass and longer low-biomass periods under higher nutrient loading, coincident with low bottom-water DO, but they do not directly demonstrate an increase in the seasonal mortality flux. We will describe these patterns as being consistent with hypoxia-related benthic-faunal suppression without interpreting them as evidence for an increased annual-mean mortality flux.
  These revisions will make the interpretation in Sect. 7.2.2 consistent with both the operational definition of carbon fixation in Sect. 5.2.2 and the quantities actually shown in Fig. 13.
  Minor Comment 31. Sect. 7.2.3: “dissolution (D1–D2)” groups distinct loss processes
  Response:
  We thank the reviewer for identifying this inconsistency. We agree that the original phrase “dissolution (D1–D2)” was incorrect because the former D1 and D2 mechanisms represented different storage-reducing processes: enhanced organic-matter remineralization and CaCO₃ dissolution, respectively.
  In the revised manuscript, we will remove this wording as part of the broader reorganization of the storage-mechanism framework. The former D1–D2 classification will be replaced by function-based storage-reducing mechanisms S_i−, each representing a distinct causal pathway. These include hypoxia-induced reduction in fauna-mediated organic-matter supply, hypoxia-induced reduction in shell formation and CaCO₃ storage, and remineralization-driven acidification and CaCO₃ dissolution.
  For the estuarine region specifically, we will revise Sect. 7.2.3 so that the decline in annual mean CaCO₃ storage is interpreted primarily from the concurrent decreases in suspension-feeder biomass and shell-formation flux, consistent with the S₂⁻ pathway. CaCO₃ dissolution through S₃⁻ may provide an additional opposing influence, but because its individual contribution is not isolated in Fig. 13, we will not present it as the dominant or quantitatively demonstrated cause.
  This revision will avoid grouping distinct storage-loss processes under the term “dissolution” and will align the Results terminology with the revised mechanism definitions in Sect. 6.3 and Fig. 9.
  Minor comments 32–37
  Minor Comment 32. Sect. 7.3.2: monotonic central-bay carbon-fixation response
  Response:
  We thank the reviewer for identifying this inconsistency. We agree that the statement that the F–U difference “does not increase monotonically” was inconsistent with Fig. 15(A) and with the preceding description of the central-bay response.
  In the revised manuscript, we will correct this interpretation. The annual mean net carbon-fixation balance in the central-bay region increases monotonically from the 0.5× to the 5.0× nutrient-loading scenario. Nutrient enrichment strengthens both photosynthetic fixation and the opposing return of fixed carbon to DIC through respiration and remineralization. However, the increase in the fixation-enhancing processes is larger, so the net carbon-fixation balance continues to increase with nutrient loading.
  We will therefore use the term co-amplification to describe the simultaneous strengthening of production and carbon-return processes, not a non-monotonic response of the net carbon-fixation balance. In the revised Sect. 7.3.2, we will state explicitly that F₁⁺ remains dominant despite the strengthening of F₁⁻ and other opposing processes, resulting in a monotonically increasing net carbon-fixation balance.
  This revision will make the interpretation fully consistent with Fig. 15(A) and with the revised definition of the central-bay co-amplification structure.
  Minor Comment 33. Sect. 7.4.2: unsupported summer-mortality interpretation in the tidal-flat region
  Response:
  We thank the reviewer for identifying this unsupported interpretation. We agree that Fig. 19 does not show a seasonal mortality-flux time series and therefore does not support the original statement that mortality “increases during the summer hypoxic period” or that the duration of such a mortality period increases with nutrient loading.
  We will remove this interpretation from Sect. 7.4.2. More importantly, after re-examining the tidal-flat results, we will revise the mechanistic interpretation of this region so that it does not invoke hypoxia-induced benthic-faunal mortality as a dominant control. Bottom-water DO remains relatively high in the tidal-flat region compared with the estuarine and central-bay regions, and the benthic-faunal trajectories do not show the abrupt hypoxia-related collapse characteristic of those regions.
  The revised Sect. 7.4.2 will therefore focus on the process fluxes that are directly supported by the simulations. Phytoplankton and benthic-algal photosynthesis and suspension-feeder shell formation increase with nutrient loading, while suspension-feeder respiration and benthic organic-matter remineralization increase more strongly. The increasingly negative net carbon-fixation balance will consequently be interpreted as evidence of intensive benthic carbon turnover rather than increased summer hypoxic mortality.
  We will retain the annual-mean benthic-faunal mortality flux only where it is directly shown and relevant to the storage analysis. In the tidal-flat region, its increase with nutrient loading will be interpreted in relation to the larger standing benthic biomass and associated mortality and benthic turnover, rather than as evidence of a hypoxia-driven seasonal mortality event.
  We will also clarify or remove the unexplained vertical reference lines in the seasonal panels, consistently with our response to the reviewer’s earlier comment on the time-series figures.
  Minor Comment 34. Sect. 7.4.3: representation of emersion–inundation and oxygen supply
  Response:
  We thank the reviewer for asking us to clarify this point. Emersion and inundation are explicitly represented in the tidal-flat cells of the model rather than being imposed only as a conceptual interpretation.
  In the revised model description, we will explain that the emersion/inundation state of each tidal-flat cell is determined from the relationship between the local bathymetric depth and the time-varying tidal level calculated by the hydrodynamic component, following Sohma et al. (2008). During emersion, pelagic–benthic exchange processes that require inundation, including suspension-feeder feeding and sediment–water exchange of dissolved substances, are suspended and resume upon inundation. Atmospheric oxygen supply to the sediment through aeration is represented during emersion. This treatment was implemented and evaluated in the previous Tokyo Bay tidal-flat application of Sohma et al. (2008) and is retained in EMAGIN-B.C. ver. 2.0.
  We agree, however, that the original wording in Sect. 7.4.3 could be interpreted as attributing the simulated carbonate-storage response specifically to repeated emersion–inundation cycles without directly isolating their contribution. We will therefore remove this strong causal attribution from the Results. The revised interpretation of tidal-flat carbon storage will instead focus on the responses directly supported by the present scenario simulations: CaCO₃ storage increases together with suspension-feeder biomass and shell-formation flux, while bottom-water DO remains relatively high and marked hypoxia-induced faunal collapse is not observed.
  Thus, the revised manuscript will distinguish clearly between how emersion and inundation are represented in the model and which processes can be identified as dominant controls of the storage response from the present nutrient-loading analysis.
  Minor Comment 35. Sect. 8: reduce repetition and consider a whole-bay synthesis
  Response:
  We thank the reviewer for these helpful suggestions. We agree that the original Sect. 8 repeated several regional summaries already presented in Sect. 7 and did not sufficiently distinguish synthesis from Discussion.
  In the revised manuscript, we will streamline this structure. The cross-regional results will be consolidated at the end of Sect. 7, where Fig. 10 will provide a direct side-by-side comparison of the representative estuarine, central-bay, and tidal-flat regions and Table 3 will summarize the dominant uptake-, fixation-, and storage-related mechanisms and the resulting carbon-control structures. Sect. 8 will then be developed as a substantive Discussion rather than another regional Results summary. It will focus on the mechanistic significance of the regional differentiation, comparison with previous observations and modeling studies of Tokyo Bay and other coastal systems, limitations of the present analysis, future directions, and implications for coastal carbon management.
  We also appreciate the suggestion of a whole-bay synthesis. The EMAGIN-B.C. application represents the full Tokyo Bay domain using 26 spatially interconnected computational cells; however, the detailed process analysis in the present study was specifically designed around three representative grid cells that characterize contrasting estuarine, central-bay, and tidal-flat environments. A bay-wide integrated carbon budget would therefore constitute an additional full-domain analysis beyond the regional mechanistic comparison that is the focus of the present study.
  We have consequently chosen not to introduce a new bay-integrated uptake, fixation, and storage budget in this revision. Instead, Fig. 10 and Table 3 will provide a compact synthesis of the regional contrasts supported directly by the analyses presented here. We will also state explicitly in the Discussion that the three carbon-control types describe these representative environments under the simulated conditions and should not be interpreted as an area-weighted classification of the entire bay.
  We agree that determining how the different regional responses combine to control the integrated Tokyo Bay carbon budget is an important next step. We will therefore identify full-domain analysis, including evaluation of the relative contributions of different regions to bay-wide carbon uptake, fixation, and storage, as a specific direction for future work.
  Minor Comment 36. Sect. 8.3: threshold and bifurcation claims
  Response:
  We thank the reviewer for this important caution. We agree that the original statement was too definitive. Because only four nutrient-loading levels were examined, the present simulations cannot exclude an abrupt threshold, sharper nonlinear transition, or bifurcation occurring between the tested loading levels.
  We will therefore revise this interpretation substantially. Rather than stating that coastal carbon-cycle nonlinearity arises not from abrupt threshold behavior, we will restrict our conclusion to what is directly supported by the simulations. Across the four tested loading levels (0.5×, 1.0×, 2.0×, and 5.0×), no abrupt transition was resolved, but threshold-like behavior or sharper nonlinear transitions between these levels cannot be excluded.
  Accordingly, we will describe the present result as non-proportional modulation of process dominance across the tested scenarios. Nutrient enrichment changes the relative strengths of co-occurring and opposing processes, and these changes differ among the representative regions; however, the present experiment does not determine the detailed response curve or demonstrate the absence of thresholds.
  We will also state this limitation explicitly in the revised Discussion. Simulations using finer nutrient-loading increments, together with independent perturbations of DIN and DIP, would be required to determine whether the identified changes in dominant process balance are gradual or include more localized nonlinear or threshold-like responses.
  Thus, the revised manuscript will retain the mechanistic conclusion that nutrient loading can modify the relative dominance of carbon-cycle processes while avoiding any inference that the present four-scenario experiment rules out abrupt transitions or bifurcations.
  Minor Comment 37. Conclusion: scope and management implications
  Response:
  We thank the reviewer for this important comment. We agree that the original conclusion was too broad in describing the proposed carbon-control concept as a “fundamental framework” for coastal ecosystem modeling and carbon-management strategy design.
  In the revised manuscript, we will substantially narrow this statement. The present study evaluates the framework using EMAGIN-B.C. ver. 2.0 applied to Tokyo Bay under idealized nutrient-loading scenarios in which total external DIN and DIP loads are scaled together at a fixed N:P ratio. We will therefore not present the resulting carbon-control structures as universally applicable ecosystem categories or claim that the present simulations directly establish a general management framework.
  Instead, the revised Conclusion will state that the carbon-control framework provides a process-based approach for distinguishing among carbon uptake, carbon fixation, and carbon storage and for interpreting their spatially differentiated responses to nutrient loading under the simulated conditions. The identified monotonic-amplification, co-amplification, and transformation-dominated structures will be described as mechanistic descriptors of the dominant process balances in the representative Tokyo Bay environments examined here, rather than as fixed classifications of coastal ecosystems.
  We will also revise the management implications accordingly. The nutrient-loading experiments were idealized sensitivity analyses and were not designed to reproduce or optimize a specific nutrient-management policy. We will therefore avoid stating that the framework directly supports the “design” of carbon-management strategies. Instead, we will explain that it can help identify potential climate-mitigation co-benefits and trade-offs associated with nutrient-management actions undertaken for water-quality conservation or biological and fisheries productivity objectives by distinguishing how such changes may affect carbon uptake, fixation, and storage differently across heterogeneous coastal environments.
  Finally, we will explicitly acknowledge that broader applicability requires testing in other coastal systems, under different nutrient stoichiometries, functional-group compositions, and environmental forcing regimes.
  Technical comments 1–4
  Technical Comment 1. Number of model variables
  Response:
  Thank you for pointing this out. The number “33” in the original manuscript was incorrect.
  We have recounted the variables listed in Table 1 and will revise the manuscript to state that the present Tokyo Bay application includes 32 model variables, comprising 30 prognostic variables and two diagnostic carbonate-system variables (pH and pCO₂). We will also use the term “model variables” rather than referring to all 32 quantities as state variables, because pH and pCO₂ are diagnosed from the prognostic DIC and TA variables through the carbonate-equilibrium calculation.
  Technical Comment 2. “Carbon storage” typos in Figs. 12, 15, and 18
  Response:
  Thank you for identifying these typographical errors. We will correct the erroneous references to “carbon storage” in Figs. 12, 15, and 18 and ensure that the corresponding panel titles, axis labels, legends, and captions consistently refer to the carbon-fixation balance and its component fluxes. We will also check the three corresponding figures together to ensure consistent terminology throughout.
  Technical Comment 3. Unit consistency throughout figures and text
  Response:
  We thank the reviewer for pointing this out. We agree that the units were not presented consistently throughout the original manuscript.
  In the revised manuscript, we will standardize the units used for model variables and process quantities throughout the text, tables, and model/process figures. Pelagic variables will use element-specific molar units per unit water volume, whereas benthic-faunal biomass and benthic CaCO₃ will be expressed per unit sediment surface area, sediment detritus per unit volume of sediment solids, and dissolved benthic variables per unit porewater volume. Corresponding panels among the estuarine, central-bay, and tidal-flat figures will use the same units and notation.
  In particular, carbon-storage fluxes will be expressed in mmol C m⁻² d⁻¹, pelagic plankton biomass in mol C m⁻³, bottom-water DO in mol O₂ m⁻³, benthic-faunal biomass in mmol C m⁻², and the benthic-faunal mortality and shell-formation fluxes in µmol C m⁻² h⁻¹. We will also standardize time-unit notation, including the use of d⁻¹ rather than mixed forms such as day⁻¹.
  Model–observation validation figures and tables will retain the conventional mass-based units of the observational datasets where appropriate, because their purpose is direct comparison with measurements. In each case, the simulated and observed values will be expressed in the same units before comparison. We will make this distinction between the standardized molar units used for model/process analysis and the observational units used for validation explicit.
  Finally, we will perform a manuscript-wide check of axis labels, captions, tables, and numerical conversions to ensure that the units and their reference bases are consistent throughout.
  Technical Comment 4. Fig. 10 “CO” typo
  Response:
  Thank you for identifying this typographical error. We will correct “CO” to “CO₂” in Fig. 10 and check the entire figure to ensure that CO₂-related terminology and notation are used consistently throughout.
  
  Citation: https://doi.org/10.5194/egusphere-2026-2710-AC3

Kohei Kawahito, Daichi Okada, and Akio Sohma

Viewed

Total article views: 148 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
95	32	21	148	9	10

HTML: 95
PDF: 32
XML: 21
Total: 148
BibTeX: 9
EndNote: 10

Views and downloads (calculated since 04 Jun 2026)

Month	HTML	PDF	XML	Total
Jun 2026	27	4	5	36
Jul 2026	37	17	8	62
Aug 2026	16	6	4	26
Sep 2026	15	5	4	24

Cumulative views and downloads (calculated since 04 Jun 2026)

Month	HTML	PDF	XML	Total
Jun 2026	27	4	5	36
Jul 2026	37	17	8	62
Aug 2026	16	6	4	26
Sep 2026	15	5	4	24

Viewed (geographical distribution)

Total article views: 134 (including HTML, PDF, and XML) Thereof 134 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 05 Sep 2026

Short summary

This study examined how nutrient enrichment affects carbon uptake, storage, and release in urban coastal waters. Using a process-based ecosystem model, we found that the response differs among shallow tidal-flat areas, the middle of the bay, and areas near river mouths. These differences arise from changes in biological activity, sediment processes, oxygen conditions, and shell formation or dissolution. The results help explain where and when coastal waters may store or release carbon.


Total:	0
HTML:	0
PDF:	0
XML:	0