the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 17 Feb 2025
NutGEnIE 1.0: nutrient cycle extensions to the cGEnIE Earth system model to examine the long-term influence of nutrients on oceanic primary production
Abstract. Understanding the nuances of the effects of nutrient limitation on oceanic primary production has been the focus of many bioassay experiments by oceanographers. A theme of these investigations is that they identify the currently limiting nutrient at a given location, or in other words they identify the proximate limiting nutrient (PLN). However, the ultimate limiting nutrient (ULN; the nutrient whose supply controls system productivity over extensive timescales) can be different from the PLN. Our motivation is to investigate the identity of the ULN. To facilitate this the carbon-centric Grid Enabled Integrated Earth system model (cGEnIE) nutrient cycles have been extended to create NutGEnIE. NutGEnIE incorporates the nutrients nitrogen, phosphorus, and iron. The impacts of diazotrophs, capable of fixing nitrogen, are represented alongside those of other phytoplankton. NutGEnIE is capable of extended model simulations necessary to investigate the ULN while, at the same time, including iron as a potentially limiting nutrient. NutGEnIE will be described, with particular focus on the biogeochemical cycles of iron, nitrogen and phosphorus. Model results are compared to ocean observational data to assess the degree of realism. Model-data comparisons include physical properties, nutrient concentrations, and process rates (e.g., export and nitrogen fixation). These comparisons support the conclusion that NutGEnIE is appropriate for the investigation of the ULN.
Competing interests: One of the co-authors is a members of the editorial board of journal Geoscientific Model Development
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 preprint. The responsibility to include appropriate place names lies with the authors.- Preprint
(2713 KB) - Metadata XML
-
Supplement
(1665 KB) - BibTeX
- EndNote
Status: open (until 22 Apr 2025)
-
CEC1: 'Comment on egusphere-2025-436 - No compliance with the policy of the journal', Juan Antonio Añel, 21 Mar 2025
reply
Dear authors,
Unfortunately, after checking your manuscript, it has come to our attention that it does not comply with our "Code and Data Policy".
https://www.geoscientific-model-development.net/policies/code_and_data_policy.htmlYou have provided in your Code and Data Availability sections two links that does not contain the code and data for your manuscript. We can not accept this, and your manuscript should have not been accepted in Discussions because of it. Therefore, the current situation with your manuscript is highly irregular. We are granting you a short span of time to address this situation (and obviously before the Discussions stage is closed, as these assets are necessary for review) and replying to this comment with the information (links and DOIs) for the new repositories containing all the code and data that you use in your manuscript. Otherwise, we will have to reject your manuscript for publication in our journal.
Juan A. Añel
Geosci. Model Dev. Executive Editor
Citation: https://doi.org/10.5194/egusphere-2025-436-CEC1 -
AC1: 'Reply on CEC1', David Stappard, 24 Mar 2025
reply
Dear Juan A. Añel
Thank you for your comment. I apologise that the Zenodo repository had not been published, this has now been done. Below are proposed revised Code availability and Data availability sections along with two additional references. Hopefully, these changes address the issues raised. A revised copy of the manuscript with these changes in place can also be provided if required.
Code availability
The exact version of NutGEnIE 1.0 code used to produce the results used in this paper is archived on Zenodo under 10.5281/zenodo.14766197 (D Stappard, 2025) (NutGEnIE_v1_0_Code.zip), as are input data and scripts to analyse model outputs and produce the plots for all the simulations presented in this paper (D A Stappard et al., 2025).
Data availability
The exact version of NutGEnIE 1.0 model output and data used in this paper is archived on Zenodo under 10.5281/zenodo.14766197 (D Stappard, 2025) (NutGEnIE_v1_0_Data.zip). The repository includes model outputs and analysis including plots for all the figures presented in this paper (D A Stappard et al., 2025).
Additional References
Stappard, D. (2025), Code and data for "NutGEnIE 1.0: nutrient cycle extensions to the cGEnIE Earth system model to examine the long-term influence of nutrients on oceanic primary production". Zenodo, doi:https://doi.org/10.5281/zenodo.14766197.
Stappard, D. A., J. D. Wilson, A. Yool, and T. Tyrrell (2025), NutGEnIE 1.0: nutrient cycle extensions to the cGEnIE Earth system model to examine the long-term influence of nutrients on oceanic primary production, EGUsphere, 2025, 1-33, doi:https://doi.org/10.5194/egusphere-2025-436.
Yours sincerely,
David Stappard.
Citation: https://doi.org/10.5194/egusphere-2025-436-AC1
-
AC1: 'Reply on CEC1', David Stappard, 24 Mar 2025
reply
-
RC1: 'Comment on egusphere-2025-436', Anonymous Referee #1, 25 Mar 2025
reply
General comments:
Stappard and co-authors detailed the NutGEnIE biogeochemistry module extension to an existing Earth system model. Their motivation was to create a modelling system capable of extended simulations in order to investigate controls on ocean primary productivity over long timescales. While the model would represent a substantial contribution to the field, I found several issues with the manuscript in its current form, which I’m detailing below, that warrant addressing by Stappard et al. before continued evaluation. In particular, the methods section should be heavily revised to improve clarity, and the discussion section should be greatly expanded to include a more thoughtful analysis of the model biases as compared to observational products.
Specific comments:
(1). I’m still uncertain of the total number of explicit biogeochemical tracers the model carries. Typically, model development manuscripts include a full list of explicit tracer equations (often in the supplemental material). I recommend this exercise here to improve clarity, especially as it pertains to the iron cycle in the model.
(2). Upon first read, I was confused how POM was treated in the model, since it was discussed heavily in Sections 2.3.1 to 2.3.3. It was not clear until page 8 of the manuscript (Section 2.3.4) that it is implicitly represented. I recommend a reorganization of section 2.3 to (1) first describe the equations for nutrient uptake, (2) its portioning into DOM/POM, and (3) the remineralization scheme of DOM/POM.
(3). It is unclear why NutGEnIE necessitates a burial fraction parameter (k_BF). Shouldn’t some portion of the POM flux make it to the bottom of the deepest grid cell, and shouldn’t that represent the portion that is buried? Explaining the rationale for such an enhanced burial flux is needed.
(4). The authors briefly introduce top-down control of autotrophs by grazers in Section 1.2, yet do not discuss any model caveats by ignoring this process in NutGEnIE. In particular, this can be an important control on surface primary production in HNLC regions (i.e., some suggest it is ‘bottom-up’ via iron limitation, while others suggest top-down controls lead to higher observed surface nutrients compared to other regions). This caveat should be outlined in greater detail within the discussion section, especially as it relates to biases in model performance (e.g., surface nitrate and phosphate comparisons in HNLC regions).
(5). I recommend elaborating more, or at least clarifying, the iron cycle in the model. Including the equations for each explicit iron tracer would especially helpful in clarifying the underlying dynamics of iron within the model.
(6). L. 184 - 185: Based on the wording here, it is unclear if anammox is represented in NutGEnIE at all. In Figure 2., it does not seem represented. If the authors choose to omit anammox, this should be discussed as a caveat to the model, since anammox is responsible for a considerable portion (~28% based on remineralization stoichiometry, see Babbin et al. 2014 - Science) of fixed-N removal in oxygen-minimum-zones.
(7). L. 224: The authors briefly mention ‘particle concentration’ (Cp) here, but it wasn’t clear to me what they meant by this. POM is only implicitly represented here, so how do they quantify particle concentration?
(8). Eq (12 - 14): I’m confused by these equations. For example, in Eq 12, this is just a fraction (Michaelis-Menten function ranging from 0 - 1), not the actual rate of aerobic remineralization. It would be helpful to first define the total depth-dependent remineralization rate as the divergence of POM flux in each grid cell (i.e., R_remin) . Then, for example, equation 12 would be better represented as [O2]/(K_O2 + [O2]) * R_remin. Also, it is unclear how the sum of the rates in Equations 12 - 14 equals the total remineralization rate of POM.
(9). Eq (15 - 17): Is particulate organic iron (POFe) not also remineralized in the model? Also, it is unclear where and how the Gibbs free energy yield values are used (I also recommend adding subscripts to the distinct Gibbs values, e.g. DeltaG^{o}_{O2}). Are they folded into the calculation of the inhibition constants? If so, it would be important to include their formulas, either here or in the supplementary material.
(10). Figure 4: I believe the authors are missing the nitrification source term for the nitrate panel. Also, here they specify nitrogen fixation as an input to the nitrate budget, but it is not represented in the tracer equation (Equation (4)).
(11). In general, text below equations are often missing units for parameters. It would helpful to include the units, and reference tables when mentioning parameters throughout the methods section.
(12). In Section 2.3.7, only Fe is represented in equations (9) and (10), yet the authors mention that complexed iron (FeL) is also available for biological uptake. Did they mean to write FeL in equations (9) and (10)? If so, wherever the authors mention Fe in the text, should they instead write FeL for clarity?
(13). In Section 2.3.7, it is unclear if ligands are an explicit tracer. If they are, then Figure 3 suggests particulate iron can be created in depth cells below the euphotic zone, yet equation (11) states POM flux is only set by euphotic zone values. Please clarify.
(14). L. 459: There are missing details regarding the initial conditions of the model runs.
(15). L. 477: “ocean physics is not the focus here, so the properties are not discussed in detail”. While I generally agree with this statement, there are some notable biases (compared to WOA, treated here as reality) in both temperature and salinity that could cause stratification and other errors in NutGEnIE. For example, in the Pacific, there is anomalously low temperature in between the subtropics (Figure 5c). Would that not have an influence on the delivery of deep nutrients to the surface there? For instance, there appears to be a very similar bias distribution for surface phosphate and nitrate (Figures 7c and 8c). Similarly, the transect biases in temperature match the patterns in the biases of nutrients. For example, the model appears to be too warm in the deep Atlantic, and is too low in PO4 and NO3. In the deep Pacific, the model is too cold, yet also high in PO4 and, to a lesser extent, NO3. The authors could potentially strengthen their validation exercises by relating some of the nutrient biases to stratification differences.
(16). Similarly, in Section 3.3.4, I believe some of these biases in surface oxygen are related to co-located biases in surface temperature. For example, Figure 5c suggests there is a cold bias in the surface of the subtropical Pacific, where there is also anomalously high oxygen values (Figure 10C).
(17). Figures 7c and 8c: These surface biases are quite large, especially considering the model is designed to study limiting surface nutrients. For example, in Table 6, the authors report a surface mean nitrate value from WOA of 6.0 umol/kg, whereas NutGEnIE reports nearly half that value (2.9 umol/kg). I would have liked to see a more thoughtful discussion on why these biases don’t impact the authors’ confidence in the model’s performance.
(18). L. 673: I’m not sure the model results support this conclusion, especially considering the model greatly underestimates both N-fixation (L. 645) and denitrification (L. 654) when compared to other studies (Section 3.1.1.). Why do the authors think that is the case in their model runs? If they believe this does not impact model performance, this should be elaborated on in the discussion.
(19). I would have liked to have seen some additional validation figures. Comparing model AOU (apparent oxygen utilization) to WOA estimates could help improve confidence in the representation of remineralization in the model. Similarly, N* can be extracted from both the model and WOA to assess model performance in generating spatially-varying N-fixation and denitrification signatures. Finally, oxygen-minimimum-zone (OMZ) thickness comparisons (e.g., thickness of waters within each cell-profile that are less than 60 umol/kg O2) would improve overall confidence in denitrification within the model, since OMZs are crucial regions for balancing the global N-budget.
Technical comments and corrections:
L.30: “Net primary production (NPP) represents the total rate...” (suggested edit)
L.32: “phytoplankton produce biomass” (typo?)
L.37: “Nutrient supply to the euphotic zone acts as a fundamental control on ocean PP levels”. (typo, I also recommend adding a semicolon before “this supply and subsequent growth limitation”)L.41: “grazing reduces the total amount of photosynthesis” (this could be rewritten for clarity).
L.48: “Again, studies have proposed methods of modeling temperature limitation” (I suggest rewording or merging with previous sentence for improved flow of manuscript).
L.50: “Elements (C, H, N, P, O and S)” (I suggest defining these explicitly or omit )
L.51: “such as proteins and nucleic acids.” (I suggest removing ‘etc.’)
L.55: “Addition of the proximate limiting nutrient (PLN) stimulates immediate growth” (I suggest removing the comma after (PLN))
L. 72: “Their work suggests the stratified subtropical gyres” (suggested text addition)
L.85: ‘The modelling’ should be ‘the modeling’ (no capitalization after semicolon). Or, could be rewritten as “Deutsch et al. (2007) only conducted model simulations over short timescales to a modern ocean steady-state”
L.90: Recommend removing “but at the same time”
L.95 - 99: I’m not sure this paragraph is necessary.
L. 102: Consider rewriting the sentence starting with “Such investigations” to place the information and citations outside of the parentheses.
L. 105: Remove comma after “(cGEnIE)”
L. 110: Fix bad reference to section.
Figure 1: Consider adding longitude and latitude tick labels to this and other map figures.
L. 138: “biogenically-induced chemical fluxes (ref) and is capable of” (suggest adding hyphen and ‘is’)
L. 142 - 151: Please be consistent when using chemical abbreviations here and in the rest of the manuscript. Typically, it is best to define chemical abbreviations before using abbreviations throughout the rest of the manuscript (i.e., “that include dinitrogen (N2), ammonia (NH4), calcium (Ca), and sulphate (SO4)”). Also, N2 is italicized here, when other forms in this paragraph are not. Please also include any negative or positive charges on NO3, PO4, NH4, SO4, etc, wherever they appear in the document.
L.149: “By default, results are output as annual averages for each grid cell” (suggest removing ‘figures’, since models only output numerical data, not visuals).
L.150: “giving the possibility of results output relating to shorter timeframes” (suggest rewording)
L.153 - 156: Please be consistent when using parentheses to split sentences. For example, here the authors use ‘a)’ ‘b)’ when elsewhere they use ‘(a)’ and ‘(b)’.
L. 156: “we detail the most pertinent features” (suggested edit)
L. 157: “nutrients taken by phytoplankton are instantly converted to POM and DOM in the surface ocean” (suggested edit for clarity)
L. 166: ”The nutrient uptake terms (…) only have a value in the surface layer”. Is this by design? Does the model restrict uptake to the top grid cell, or is this a result of the model? Please clarify.
L. 183: “Like denitrification, anammox converts…” (add comma, remove capitalization of Anammox)
L. 188: Please rewrite this sentence for clarity (there are some typos), and please define RP, RN, DDFe, and BBFe in the text before using their associated abbreviations. The authors could also reference the associated supplemental figures in the caption of Figure 2.
Figure 2: Please define POP, PON, and POFe in the caption, since they are labeled on the Figure panels. I also recommend redesigning these figure panels such that only explicit tracers, and the fluxes that couple them, are represented. For example, here the authors use green circles to define other phytoplankton and diazotrophs, which may confuse readers into thinking these are explicit tracers.
L. 204: Please consider adding units when describing the components of equations (i.e., mmol / m3 d).
L. 208: “bit” should be “but”
Eqs (2) and (4): How are ’S_{PO4}’ and ’S_{NO3}’ different from ‘RP’ and ‘RN’? If they are identical, consider just using one term. Also, the authors mention ’S_{PO4} being configurable by a parameter on line 210? In Figure S1, surface nutrient inputs are supplied via a forcing field rather than a parameter. Please clarify what is meant here. Also, I recommend adding supplementary figures (similar to Figure S3) that detail the magnitude of the surface forcing for both nitrate and phosphate.
Eq (4): I recommend using consistent terminology for reaction rates. For example, why use the delta symbol for nitrification and the ‘R’ symbol for denitrification, when they are both reactions in the model? Perhaps it is easier to use ‘R_nit’ and ‘R_den’ for clarity?
Eq (6); ‘B_{Fe}’ is labeled as ‘BFe’ in Figure 2. Please be consistent between figure labels/captions and equations. Also, is ‘free dissolved iron III’ another explicit tracer in the model?
L. 217: Please reference Figure S3 after mentioning ‘DDFe’, either here for after describing the re-gridding of Mahowald et al.
L. 232: It would be helpful to define units for these terms in the equation. For example, the units of V^{OPhy}_{max} are unclear since the model does not explicitly represent biomass. The authors mention the parameter values on line 255, but please move them to earlier in the text when they are first defined.
L. 234: I’m confused by the light limitation term. In Figure S37 the text mentions that “lower values of ‘K_light’ indicate that light is more limiting to nutrient uptake”. Where does ‘K_light’ fit into the equation on line 234? Also, this relates to my previous comment about light limitation. Is it restricted to the surface grid cell?
L. 239: ‘Therefore, DIN and Fe uptake are scaled’? (Did the authors mean to include ‘uptake’ here?)
L. 242: Please remove the comma after ‘atmospheric transfer’, and the authors have already used ‘N2’ earlier in the text but are just now defining it as dinitrogen.
L. 245: I recommend moving the discussion of N_thresh (Section 2.3.5) here to improve clarity.
L. 256 - 257: I was confused by this last sentence at first. I recommend reorganizing the text to specifically say that the V^{*}_{max} terms are temperature-dependent (i.e., can reach values both higher and lower than what are reported here). Then, temperature-dependent uptake is scaled by the combined limitation terms. Also, it is unclear why the authors state that “the maximum percentage of the grid cell nutrient concentration taken up by other phytoplankton each time step is 80%”. That would assume that the other limitation terms are equal to 1, but the authors state that the other maximum values are 0.7 and 0.5. Please reword or consider omitting this sentence from the text to improve clarity.
Eq (11): Please include units for the flux. I also think the authors can remove the (z = z_o) from the equation and just use z_o. Also, from this equation, it does seem like the model restricts uptake to the top cell of the model. Please detail this earlier in the manuscript so it is more clear.
L. 266: This is the first time the authors have mentioned that sulfate is an explicit tracer in the model. It would help improve clarity to have mentioned this much earlier in the methods section, and to provide a tracer equation for sulfate (and all other tracers) in the supplemental material.
L. 277: Is the model configurable to represent non-Redfield stoichiometry?
L. 286: Since bacteria are not an explicit tracer, it is not necessary to say ‘by bacteria’ here.
Eq (18): It might improve clarity to separate the two Michaelis-Menten functions rather than showing the product. Also, please include the units for maximum rate of nitrification, and consider using a different symbol (i.e., R_{nit}) to match the style of other reactions in the manuscript.
L. 301: Please reword this sentence.
Figure 3: Please be consistent with terminology elsewhere in the manuscript. Does ‘PartFe’ represent ‘POFe’ in Figure 2?
L. 310: It was confusing where the caption ends and the next sentence begins (minor comment).
L. 312: Please move this text on the equilibrium between Fe, ligands, and complexed iron to line 304 for clarity.
Tables 1 - 5: References to these tables should be placed in the appropriate locations in the text when first mentioning these parameters.
L. 340 - 344: “Parameters were adjusted to result in a combination that showed best agreement with observed nutrient distributions...” (suggested edit)
L. 404: I recommend converting +- 26% into Tg N yr to match the other estimates.
L. 405: “Wang et al. (2019) provide location of fixed nitrogen due to…”. Do the authors mean fixed nitrogen loss?
L. 420: Here the authors italicized PO4, NO3, and O2, whereas in other areas they are not italicized. Please be consistent.
Figures 5 - 8: Please provide labels on the maps and transects so that it is easier to identify which panels represent model results vs. which panels represent validation products. In all Figures, it would also help to extend the colorbar limits slightly to better represent values beyond their current ranges, since the values often reach the maximum/minimum limits. This is most notable in Figures 7 and 8, since the representation of these nutrients are a central point to the manuscript. Finally, please include latitude ticks (higher priority) and longitude ticks on map figures.
Figure 9: Can the authors please convert longitude labels from 0 - 360 format to -180 to 180 format (with E and W labels)? The inset map is also quite small and could be resized for clarity.
L. 610: “For all variables…” (typo)
Table 6: Please use the same numerical precision between surface and interior reported values (i.e., surface value of 0.58 vs 2 for PO4).
L. 651: “Denitrification can occur throughout the water column”. Ideally, this won’t be the case. Instead, NutGEnIE should restrict denitrification to only very low oxygen regions. Perhaps reword this to be clear. Also, I suggest removing ‘by bacteria’ since the model does not resolve bacterial biomass or their metabolisms.
Figure 15: Consider rewording the caption.
Citation: https://doi.org/10.5194/egusphere-2025-436-RC1
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 87 | 21 | 6 | 114 | 14 | 5 | 5 |
- HTML: 87
- PDF: 21
- XML: 6
- Total: 114
- Supplement: 14
- BibTeX: 5
- EndNote: 5
Viewed (geographical distribution)
| Country | # | Views | % |
|---|---|---|---|
| United States of America | 1 | 55 | 43 |
| United Kingdom | 2 | 19 | 15 |
| China | 3 | 12 | 9 |
| France | 4 | 7 | 5 |
| undefined | 5 | 7 | 5 |
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
- 55