the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 16 Feb 2026
Particle flux dynamics amplify marine carbon cycle differences between climate states
Abstract. The ocean represents the largest and most rapidly exchanging carbon reservoir on Earth’s surface and the marine carbon cycle response to changing climate is a matter of continuous investigation. Here, we added dynamic environmental controls on the remineralization and dissolution rates of particulate organic matter, carbonate and silicate minerals (Dinauer et al., 2022) to the Earth system model Bern3D to explore feedbacks between biogenic particle fluxes and marine carbon cycling under different climate conditions. The new representation of marine particle dynamics improves the model’s ability to capture the marine biogeochemical response to long-term cooling, almost doubles the sensitivity of global export production, and amplifies the change in marine carbon storage by a factor of about 1.5. In a model configuration where carbon exclusively cycles between the atmosphere and ocean, this corresponds to an additional atmospheric CO2 drawdown or increase of approximately 20 ppm in response to a -9.1 °C cooling or +6.8 °C warming, respectively.
- Preprint
(4025 KB) - Metadata XML
-
Supplement
(6214 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-275', Anonymous Referee #1, 25 Mar 2026
-
RC2: 'Comment on egusphere-2026-275', Anonymous Referee #2, 08 Apr 2026
Adloff et al. present a study exploring the impact of dynamic particulate organic matter cycling on climate states. The manuscript tackles several substantial advances simultaneously: dynamic particle remineralisation, an updated Fe cycle, and climate-state sensitivity experiments, each of which could warrant a dedicated paper. This comes at the expense of a clear central focus, and some sections feel compressed or under-realised in terms of output or discussion. The central findings on the sensitivity of climate states are potentially interesting but are not explored in enough detail to substantiate whether there are model-specific caveats or notable features like non-linear dynamics. I encourage the authors to prioritise the key message and streamline their scope.
MAIN COMMENTS
Scope of the manuscript:
The modifications to the previous Bern3D implementation, along with the tuning required are described extensively in the Supplement and are, in my view, substantial enough to warrant a more focused discussion in their own right within a model development-focused manuscript, e.g., Geoscientific Model Development.
While I consider the study has a number of interesting findings, I feel that it is currently trying to cover too much ground for a single article. As a result, the different sections do not always follow a coherent flow, and some of the most interesting aspects are constrained by space or moved to the supplement, which also hinders the visibility of their work. I would encourage the authors to sharpen the central narrative: is the main aim to evaluate MSPACMAM in Bern3D, to demonstrate the climate-state dependence of particle flux feedbacks, or to present a comprehensive new Bern3D biogeochemistry configuration?
To illustrate this and better convey my point, the current scope could quite naturally be divided into several manuscripts, for example:
• Focus 1 – Model Description: A comparison of the new POM + particle flux dynamics (MSPACMAM) against the previous Martin-style implementation under a pre-industrial steady state, drawing more extensively on the existing Supplementary Information. This could include a deeper discussion of model biases and improved diagnostics of POC export under PI conditions or a GMD-focused work.
• Focus 2: A study focused on productivity changes arising from the updated Fe budget and how these relate to sediment components that were omitted in the PI steady-state simulations, ultimately examining how the combined Particle + Fe modifications alter POC export and storage relative to the original Bern3D framework.
• Focus3: An in-depth model-diagnostics paper focused on the cooling experiment, discussing how the new framework (Particle + Fe) compares with a range of LGM particle flux and degradation proxies and evaluating the strengths and weaknesses of this new Bern3D framework in simulating LGM-like states (i.e., sediment/ice sheet modules turned off/ changes in wind/wind stress not considered in the energy balance module, etc).
• Focus 4: A paper centred on the responses in POC export from PI → cooling and from PI → warming under the new framework, with particular emphasis on whether transient responses are mirrored across these states when approaching steady state and whether different responses could be traced back to different contributions of processes resulting in Csoft increases or decreases (decomposing the delta DIC changes in Table1 to identify relative contributions).Presentation of Results:
A number of figures are presented as delta-delta plots, e.g., Figure 3c, which is (particle – (martin_cold – martin_pi)). I find this style quite difficult to interpret because there are effectively two dimensions, a change in model and a change in boundary conditions, which are collapsed. It might be informative to visualise as two dimensions (col = change in model, row = change in boundary condition) with the initial state also shown. To some extent, these figures are harder to interpret because of the multiple model changes so a focusing on the manuscript would also benefit this issue.
A lot of the results that seem central to the analysis are presented as supplementary material. The main figures presented are also formatted and presented in different ways such that’s it hard to compare between figures. Again, I think some of this is related to the manuscript having too big a scope and would be addressed by focusing the manuscript.
Overall, I recommend a major revision, mainly to better define the scope, and encourage the authors to consider resubmission after refocusing the manuscript. I hope these comments are helpful, as I see strong potential in this work and would very much like to see its key advances presented in a way that allows them to be fully appreciated.
SPECIFIC COMMENTS
Introduction
L.14. “Volk and Hoffert (1985).”
Citation should not be of in-text type.L.16. “food chain”.
Consider “food web” .L.18. “degradations where temperature and redox conditions are conducive”.
Consider “under favourable temperature and redox conditions”.L.22 “Remineralisation of sinking organic particles and dissolution of carbonate and opal below the well-mixed surface ocean layer transfer carbon, cations, ALK, and nutrients from the photic zone to the intermediate and deep ocean. which is largely balanced in steady-state by corresponding upward fluxes by physical transport (advection, convection, diffusion).”
This sentence is too long and could be broken down to improve clarity and use some citations.
L. 25. These transfer fluxes by sinking particles and the circulation lead to vertical gradients in DIC, ALK, and nutrients, and are referred to as the biological soft tissue and carbonate pumps (Volk and Hoffert, 1985).
This is a bit of a mixed bag sentence. Which is linked to which pump?
Carbonate pump is linked to ALK gradient with influence on DIC whereas soft tissue more linked to Nutrients and DIC gradients.L. 30. The effectiveness of the soft tissue and carbonate pumps in maintaining low surface ocean carbon concentrations depends on the rates of particle (and DOM) production and export, particle concentration profiles, and circulation.
Particle Concentration profile needs to be better explained and differentiated from Particle Export.
L. 49. “Some reconstructed changes in regional export production during the last glacial have also been attributed to changed availability of nitrogen and iron (Fe) (Martin, 1990; Deutsch et al., 2004). Yet, no model produces proxy-consistent marine carbon cycle changes without prescribing further changes in biogeochemistry,”
I kind of get what you are saying but it opens scope for an interpretation that does not sell the use of models, meaning that models are only able to match proxy-based observations when the toggles for BGC modelling and explicit parameterization are exaggeratedly tweaked with. Maybe you are trying to say that models need to incorporate more processes to better represent LGM states? In that case rephrase to highlight your study.
L.55. “columnar”
This is an unusual term, can you define what this means or re-word. (Does this mean 1-D water column?)
L. 53 – 57. “Here, we test how the marine carbon cycle and atmospheric CO2 are affected by dynamical particle cycling in different climate states by adopting the columnar explicit particle flux model MSPACMAM (Dinauer et al., 2022) as a fully integrated module in the intermediate complexity Earth system model Bern3D, alongside an updated Fe cycle. We evaluate the model performance against proxy evidence of marine carbon cycling in a glacial state and assess how the added processes alter the model’s response to idealised warming.”
This sentence should be the last one in your Introduction where you lay out your objectives and main research questions. This would also benefit from further explanation into the wider context of looking at some of these climate feedbacks (basically how you get to the numbers presented in Table 2, and how these could be decomposed into different components – as a suggestion).
L. 58. “The Bern3D model has been used to simulate carbon cycle dynamics in various scenarios of environmental change of the past and future, but only with globally-uniform, static particle attenuation profiles, relying on climate state-dependent parameter adjustments as a simplified representation of the climatic controls on export production and remineralization and a simple Fe cycle without a dynamic sedimentary Fe source (Parekh et al., 2004, 2008).”
This sentence seems out of place. Consider combining it with where you first introduce the use of Bern3D, line 55.
Methods
L.75. “biogeochemcial”
Typo. “biogeochemical”L. 138. “Atmospheric CO2 decreases from its PI value in response to the cooling but is in these sensitivity simulations not influencing radiative forcing and climate.”
This sentence needs to be expanded further to better explain why CO2atm changes do not influence radiative forcing in these simulations. Do you mean that a reduction in radiative forcing, which was done to induce the cooling state, is not accompanied by a reduction in CO2atm and therefore there is no shortwave trapping (i.e., complete absence of such process)? In other words, the energy budget is not sensitive to CO2atm, and is this then a type of ‘atmospheric dust release’ analogue?
I think this is also a key component that needs to be addressed in the discussion as to how you could detangle certain components and better describe/contextualise these simulations to the reader. The same reasoning goes to your warming state simulations and when you sort of drop a clue that winds are constant (right? L.318). On the constant wind: this leads to no simulated changes in wind stress and air-sea gas exchanges. How do you expect this to influence your response in CO2atm, DIC and the cascading effects on the soft tissue and carbonate pumps?
For instance: (i) Indirectly, constant winds could prevent enhanced upwelling of corrosive deep water (low Ω), so simulated CaCO₃ export/dissolution rates may be biased (less undersaturation-driven dissolution). (ii) Weaker/unchanged mixing sustains nutrient limitation by not promoting wind-driven changes in MLD in the euphotic zone, potentially overestimating export production (POC flux).
Results
L. 154. “especially true for Fe [biases] in the surface ocean”?
Can you elaborate on that? Is this an overall decrease but then even more so in the Southern Ocean where Fe availability hinders POC export? You have explored this in your Supplementary Material to some extent but I feel like you are describing your results in the main text too bluntly. Maybe constrained by space/word-limit ? (which again is why I think your work could benefit from a re-focus that would allow you to have more room).
Table 2.
- the oldFe experiment seems missing from the warming
- Why is deltaDIC (Pg C) always decreasing but the concentrations increasing in the cooling column? (Same for deltaO2)
- The changes in export and DIC don’t seem to vary consistently, which makes sense given the changing water-column cycling, but regenerated DIC is likely not the only component of DIC changing. What about the carbonate pump, solubility and disequilibrium carbon, are these contributing to the offset between export and DIC?DeltaCO2/DeltaT for the particle warming experiment is notably larger than the other values. This is really interesting (is it non-linear with warming? If so, is there an inflection point at a particular warming?) but there’s only one experiment!
Figure 3. no units listed
Figure 5. I haven’t seen transfer efficiency shown as a function of depth before. Transfer efficiency is essentially a metric to collapse depth variability in POC flux down to a single number so you can either have a useful metric for changes or look at spatial variability. I wonder if a horizontal plot of TE at a given depth (or a few depths) might be more intuitive?
L. 220. is it possible to show the contribution of different processes to the change in particle fluxes? Currently it’s not clear if this is an interpretation or guided by evidence from the model outputs.
L. 232. I don’t think you can interpret the change in DIC to this extent because a) changes in DIC are not fully local to the change in export and remineralisation and b) other components of DIC may also be changing.
Discussion
L.291. “Instead, similar to the findings by Liu et al. (2024), in our simulations increased preservation of POM in the water column in low and mid-latitudes increases the portion of POM in the composition of sinking particles which reduces the average density of the sinking particle mix and thus their sinking speed.”
POM is a mix of many particle types, so which type is the one that is most responsible for changing buoyancy in your particle mix? Also, please expand as to why POM remains relatively ‘untouched’ under your simulations. You draw a parallel to previous work that found similar preservation in high latitudes attributed to a ballasting effect from dust particles on particle sinking speeds. However, you saw POM preservation in low and mid-latitudes (not in high latitudes as seen in previous work, and you do not have dust simulated). So why is it that you saw a similar preservation response even though you do not simulate the same underlying processes and why is it in a different zonal region?
This whole paragraph needs to be expanded and ideas better organised so it is shaped as a discussion rather than just semi-parallels without going further into the underlying processes that led to this response.
L. 329 .“Like in previous studies with Bern3D, marine oxygen concentrations decrease in our steady state warming simulations as solubility changes and increased export production overcompensate the effect of improved ventilation due to a stronger overturning circulation (Battaglia and Joos, 2018a).”
This sentence is a bit convoluted. First, refrain from using ‘marine oxygen’ and stick to “dissolved oxygen” for clarity. Explain succinctly why you have stronger overturning and ventilation in a warmer steady-state simulation to then link that with enhanced export production dominating over increased ventilation and resulting in an overall decrease in dissolved oxygen.
L. 336. “A decreased carbon storage in a warmer ocean is consistent with other steady state simulations of warming (Kleinen et al., 2016; Kessler et al., 2018) but unlike simulations of near-future marine carbon storage changes in which anthropogenic C emissions drive C uptake by the ocean on a larger scale than the warming-driven degassing (Kwiatkowski et al., 2020).”
This sentence is too long and convoluted. Consider rephrasing, especially the contrasting idea that you are trying to convey with the use of ‘but unlike’. Do you want to highlight that warming affects the solubility pump and thus induces more degassing while highlighting that near future simulations are primarily anthropogenic-C driven? If so, how does your simulation framework compare? Or even, are you trying to say that your simulation does not compare since you adopted different underlying processes (radiative forcing instead of explicit changes in CO2atm) to induce climate change, but still, you arrive at the same C storage signal? Signal-wise there is similarity but, what about spread in terms of the magnitude across studies?
Also refer to the data you show in Table 2. Use your numbers in your Discussion to your advantage and contextualise the relative changes against other studies. You barely point to your figures and table in the Discussion.
Future studies?
One area that could further strengthen the manuscript is a brief discussion of limitations and future directions. A focused manuscript would also provide more space to discuss remaining model-data mismatches (e.g. Southern Ocean Fe limitation, AABW ventilation biases) and exciting avenues for future work, such as transient simulations or full LGM experiments with sediments and dust changes enabled. A dedicated ‘Outlook’ section highlighting key limitations (e.g. disabled sediment module) and promising extensions would greatly enhance the paper’s impact and forward-looking value.
Citation: https://doi.org/10.5194/egusphere-2026-275-RC2
Data sets
Publication resources for Adloff et al. 2026 Markus Adloff, Frerk Pöppelmeier, Ashley Dinauer, Charlotte Laufkötter, Fortunat Joos https://doi.org/10.5281/zenodo.18314202
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 197 | 90 | 11 | 298 | 42 | 11 | 24 |
- HTML: 197
- PDF: 90
- XML: 11
- Total: 298
- Supplement: 42
- BibTeX: 11
- EndNote: 24
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Summary
Using an Earth system model of intermediate complexity (EMIC), the authors examine the potential role of the particle dynamics of sinking detrital material. They do so by improving this process from a simple empirical submodel based on Martin et al. (1987) to a more sophisticated treatment that includes the roles of temperature and oxygen concentration in remineralisation (among other changes). They illustrate an improvement in model performance across a range of conventional biogeochemical metrics when simulated under piControl conditions. Simulating under climate conditions cooled and warmed with opposite sense radiative forcing produces typically opposite-sense responses in the simulated carbon cycles.
Major comments:
My recommendation is Major Revisions as the manuscript would benefit from significant reorganisation to improve clarity and straighten out its narrative. Ideally, extra work would include simulations that gap-fill current omissions (e.g. the Martin simulation is really Martin_oldfe), but I appreciate this may be challenging.
Specific comments:
Ln. 2: “most rapidly exchanging” might not be entirely accurate; large volumes of the ocean are very slowly exchanging, and the terrestrial carbon cycle is arguably more interactive given it has larger gross fluxes
Ln. 7: “long-term cooling” – it would be helpful if the abstract just stated glacial-interglacial cycles are of interest here rather than confuse readers with reference to the exact opposite from that which they might expect from an ESM study (i.e. global warming is the default expectation)
Ln. 7: “almost doubles the sensitivity” – to what?
Ln. 7-8: “amplifies the change in marine carbon storage by a factor of about 1.5” – again, this is missing context; change of what specifically in response to what specifically
Ln. 8-9: “where carbon exclusively cycles between the atmosphere and ocean” – not the land?; but the model is described as an “Earth system model”; if land is excluded, this suggests that it’s not really an ESM; maybe an EMIC is a better description
Ln. 9-10: “approximately 20 ppm in response to a -9.1C cooling or +6.8C warming” – this is a bit clunky for an abstract; I guess what you mean is that for an air-sea exchange difference of 20 ppm, these are the temperature limits; but the “approximately 20 ppm” is somewhat jarring when set against more precise temperature changes; I suggest rewriting this to be simpler and clearer about the sensitivity of the model – I’m not sure what yet, but this is opaque
Ln. 23: “cations” – is this an oblique reference to Calcium?
Ln. 29: something to consider here is a reference to the fact that the carbonate pump drives an alkalinity flux that decreases the buffering capacity of surface seawater; this puts the carbonate pump – in part – pointing in the opposite direction to the soft tissue pump
Ln. 39: “increasingly complex biogeochemical representations” – it would be helpful to point to some examples of these models here
Ln. 55: “intermediate complexity Earth system model” – perhaps call it an EMIC, an “Earth system Model of Intermediate Complexity”?
Ln. 64: “Bern3D” – I would expect a model version number to be used at this point (I appreciate that it’s mentioned a few lines later); it would certainly help if any comparisons between the original and revised model versions are made
Ln. 77: I might be inclined to break to a new subsection here to clearly separate out what’s different in this version compared to default Bern3D (i.e. separate base and revised model subsections); I might also be inclined to give a model name / version number for the variant described; perhaps finishing subsection 2.1 with a statement about the best source(s) for a model description and validation would be helpful?
Ln. 79: “z0” is the flux at 0 metres?; or is this the flux at the base of the euphotic zone?; and if the base of the euphotic zone, what depth is this?
Ln. 84: “columnar”?; do you just mean “water column”?
Ln. 90: CaCO3 dissolution is introduced here but it is ambiguous whether it has any connection to the remineralisation of POM; as it’s introduced in-between descriptions of remineralisation and particle sinking speed, this seems implied, but there is no formal connection directly mentioned
Ln. 84-104: this is the focus of this manuscript so should be completely clear and unambiguous; currently, per my previous comment, I’m not certain of the relationship between POM and CaCO3 (and opal, for that matter); further the manuscript refers to large and small particles, but consigns details to the supplementary material; this is unsatisfactory given it’s the core of the novel work in this manuscript
Ln. 101-102: what might help is a diagram showing the vertical profiles of sinking POC, opal and CaCO3 for the original and revised models here (alongside, say, Martin et al., 1987); while there are plots almost like this in the Supplementary Material (SM), there’s nothing quite like this
Ln. 110-118: again, there’s something of a lack of detail here in the manuscript’s main body; I appreciate it’s in the SI, but it would be good to be clearer on what “oldfe” is and how it differs from the refined model
Ln. 120: while it’s good to have all of this material in SM, in the case of specific model developments (e.g. the iron submodel; Ln. 105) it’d be helpful to point readers to specific figures in this so that they know they exist and where to find them; this generic reference to SM misses a trick in guiding the reader
Ln. 131: “the model was spun up under preindustrial conditions” – it would be helpful to explain what’s different between this spin-up and “normal” use under the EMBM atmosphere; I assumed that the model would always be using the EMBM and I’m not sure what it means not to
Ln. 134: first reference to 13C; I’d have expected to hear something about this before in the model description; even a passing reference would be helpful; is 13C to help with validation?
Ln. 138-139: “Atmospheric CO2 decreases from its PI value in response to the cooling but is in these sensitivity simulations not influencing radiative forcing and climate.” – first, rewrite this to something like “Atmospheric CO2 decreases from its PI value in response to the cooling but does not influence radiative forcing or climate in these sensitivity simulations”; second, this feels like a strange choice given that such feedback seems important in this specific context; given the preceding sentence about the simulated cooling being at the observational limit, it feels like this is a choice to avoid the model being even cooler and beyond the observational range
Ln. 139-140: the choice to not allow the model’s own atmospheric CO2 to influence climate seems even stranger when 4xCO2 experiments are mentioned; these experiments specifically allow this feedback so mentioning them is very odd; I think I understand what you’re doing (applying a climate cooling / warming but avoiding feedback effects), but I think a clearly stated explanation would help readers
Table 1: can’t say I’m a fan of long experiment names when EXP numbers or model version numbers are simpler and clearer to use on plots; but this is aesthetic
Table 1: it feels unsatisfactory to not use the updated Fe scheme in the main meat of the work here; many comparisons are between the Martin and Particle model versions but the differences extend beyond the particle dynamics; as the model is low resolution it is presumably relatively inexpensive to run (though this it is not made clear in the main text), so the absence of clean comparison simulations is difficult to justify; is there an issue of model tuning that complicates simulation?; this is mentioned, but not fully articulated
Table 1: if “oldfe” is to be included, its status as an intermediate step in the work here would be clearer if it was positioned as an intermediate step in this table; also, why not describe Martin as Martin_oldfe?
Table 1: why not put all of the simulations (piControl, cool climate, warm climate) on the same table?
Ln. 145: provide a cite for the preformed / regenerated methodology
Ln. 146-148: this feels very much like a minor sensitivity experiment given the missing context; maybe it’s important later on?; but it might be the sort of thing you ignore here only to introduce it at a relevant point in the discussion
Ln. 149: given the topic of the work the ordering and structure of the results section could perhaps be better; as the focus of the paper is the addition of particle dynamics, starting with the effect of this addition for the default climate (i.e. neither warmed nor cooled) would seem like a good section 3.1; the current separation of warming and cooling effects into separate subsections seems difficult to miss a trick by not contrasting their effects side-by-side; furthermore the separation of cooling effects from particle effects (subsection 3.1.1) seems very strange
Figure 1: strange units (mol C / km2 / y); why not umol / m2 / y?; from my experience, observational scientists do not usually report things in per square kilometre
Figure 1: the results of the Martin experiment are compared to those of the Particle experiment despite the latter including the new Fe scheme; this makes it challenging to separate the effects
Figure 2: does the model have about the right amount of sea-ice for the present-day?
Figure 2: these are sometimes referred to as “thermohaline transects” as they try to crudely capture the thermohaline circulation from younger waters in the North Atlantic through to the oldest waters in the North Pacific
Ln. 166: would one expect North Atlantic ideal age to decrease with warming?; possibly erroneously, I tend to expect warmer climates to have more stratified oceans and older ideal ages; in this vein, Li et al. (2024; see below) report younger ages in LGM simulations while Figure 2 would suggest older ages; I guess different models will simulate AABW and its ventilation differently, so perhaps this manuscript would benefit from discussing why its response might differ
[Li, L., Liu, Z., Du, J., Wan, L., and Lu, J.: Mechanisms of global ocean ventilation age change during the last deglaciation, Clim. Past, 20, 1161–1175, https://doi.org/10.5194/cp-20-1161-2024, 2024.]
Ln. 257: “Response to warming” should be another “Results” subsection but it is numbered as a whole new section; please amend this
Ln. 257: the brevity of this final section on warming favours merging it with the preceding section on cooling; discussing the difference in model behaviour due to opposite sign climate changes seems more sensible to me than presenting them almost as unrelated sensitivity experiments
Ln. 267: would there be any value in reporting how cooling and warming simulations approach equilibrium and on what timescales?; I appreciate that the focus here is on the steady state, but I’d certainly be interested to know whether warming or cooling took longer to reach equilibrium
Ln. 294-295: this study reports weakened AMOC at the LGM, but my (admittedly limited) suggests that there’s considerable uncertainty here; two recent papers I’m aware of on the subject are listed below … (I’m sure there are others)
[Gu, S., Liu, Z., Oppo, D.W., Lynch-Stieglitz, J., Jahn, A., Zhang, J. and Wu, L., 2020. Assessing the potential capability of reconstructing glacial Atlantic water masses and AMOC using multiple proxies in CESM. Earth and Planetary Science Letters, 541, p.116294. https://doi.org/10.1016/j.epsl.2020.116294]
[Zhengyu Liu; Evolution of Atlantic Meridional Overturning Circulation since the last glaciation: model simulations and relevance to present and future. Philos Trans A Math Phys Eng Sci 11 December 2023; 381 (2262): 20220190. https://doi.org/10.1098/rsta.2022.0190]
Ln. 353: the code and data statement seems a bit remiss to me; why is this not available in a Zenodo archive or similar?; I can’t see it attached to the manuscript record