the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 04 Apr 2022
Physiological flexibility of phytoplankton impacts modeled biomass and primary production across the North Pacific Ocean
Abstract. Phytoplankton growth, and hence biomass responds to changing light and nutrient conditions in the near-surface ocean. Although a wide variety of physiological photoacclimation models have been developed and tested against laboratory results, their application and testing against oceanic observations remain limited. Hence the biogeochemical implications of photoacclimation in combination with ocean circulation have yet to be fully explored. We compare modeled phytoplankton biomass and primary production from a recently developed flexible phytoplankton functional type model (FlexPFT), which incorporates photoacclimation and variable carbon (C) : nitrogen (N) : chlorophyll (Chl) ratios, to that obtained with an inflexible control model (InFlexPFT), which assumes fixed C : N : Chl ratios. We couple each plankton model with a 3-D eddy-resolving ocean circulation model of the North Pacific and evaluate their respective performance versus observations of Chl, nutrients, and primary production. These two models yield different horizontal and vertical distributions of Chl and primary production. The FlexPFT reproduces observed subsurface Chl maxima, although it overestimates Chl concentrations. In the subtropical gyre, where light is sufficient, even at low nutrient concentrations, the FlexPFT yields faster growth rates, as well as high Chl concentration and primary production in the subsurface layer. Compared to the FlexPFT, the InFlexPFT yields slower growth rates, and lower Chl and primary production. In the subpolar gyre, the FlexPFT also predicts faster growth near the surface, where light and nutrient conditions are most favorable. Compared to the InFlexPFT, the key differences that allow the FlexPFT to better reproduce the observed patterns are its assumption of variable, rather than fixed, C : N : Chl ratios and inter-dependent, rather than strictly multiplicative, effects of light- and nutrient-limitation.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(9746 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(9746 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-91', Anonymous Referee #1, 15 May 2022
[comments formatted in Markdown, find attached the pdf version]
The manuscript *Physiological flexibility of phytoplankton impacts modeled biomass and primary production across the North Pacific Ocean* by Y. Sasai and colleagues assess the importance of optimal nutrient uptake, photoacclimation and variable stoichiometry on the emergence of large scale chlorophyll and primary production patterns in the ocean, using the North Pacific as a testbed. To do that, the authors designed an experiment based on the comparison of two strategic NPZD biogeochemical (BGC) models coupled to a state of the art, eddy-resolving model of ocean physics (OFES2). One of the BGC models implements optimal uptake kinetics ([Smith et al 2009](https://doi.org/10.3354/meps08022)), whereas the other incorporates interactive, optimal photoacclimation (Chl:C ratios) and variable stoichiometry (C:N deviating from Redfield ratios) as proposed in the context of the FlexPFT theory of [Smith et al 2016](https://doi.org/10.1093/plankt/fbv038)). The experiments revealed clear deviations between predicitions from both models, with a clear gain in accuracy when using the more complex FlexPFT model (bulk properties like Chl and fluxes like PP closer to observations, better resolution of the deep chlorophyll maximum (DCM), etc.). The authors conclude recommending the adoption of similar approaches by the BGC community.
The topic, approach and results are of great appeal and the manuscript is already an important contribution. The authors designed a clean test of their hypothesis about the emergence of surface and vertical gradients in phytoplankton growth and biomass, and the result clearly support their ideas. However, the manuscript is not easy to follow as it stands, especially the combined *Results and Discussion* section. The narrative in this section is quite descriptive, and it fails to provide a clear picture of the ability of each model to reproduce observed patterns. As commented below, these issues seem to arise in part from the lack of motivation and rationale for model assessment in *Methods*. There are other aspects missing like the limitations of the current model (*e.g.* what about nitrogen fixation?), alternative explanations to the emergence of DCM ([Cullen 2015](https://doi.org/10.1146/annurev-marine-010213-135111); there may be more recent reviews), or about the performance of similar BGC models in simulating Chl and PP in the Pacific. Together, these issues led me recommend a major revision of the manuscript. I provide some major concerns, and a long list of minor issues and suggestions below. I hope the authors find both lists useful.
-----------------------------
#### *Major comments*
1. The manuscript is in general well written and structured, but there are two sections that in my opinion deserve another thought, namely *Methods and Materials* and the *Results and Discussion* (see next three points). Although the description of the models is in general easy to follow (see however some minor suggestions below), the fact that the text moves from a complex model to a simpler one is not an optimal choice. I recommend the authors to present first their general approach with the components that are common to both models, and then detail first the simpler model featuring just optimal uptake followed by the more complex model featuring also photoacclimation and variable stoichiometry. I am aware that this might read as a minor issue, but I think it is quite important to ensure that readers realize that, despite their names, FlexPFT is something more complex than InFlexPFT. It is not clear whether one model is a nested version of the other or not (in the sense of a simpler formulation or the result of setting from variables as a constant). For instance, there is certain temptation to just check Table 1 and conclude that InFlexPFT results in reduced Chl and PP when compared with FlexPFT just because $\mu_{\text{max}}$ is lower in the former. There is also some confusion about whether the model implements only photoacclimation or photoacclimation and variable stoichiometry, and about whether one or both of them are simple NPZDs or not.2. Merging *Results* and *Discussion* has certain risks. In my opinion, that section of the manuscript needs to dig a bit more into the results and provide more quantitative tests that enable readers to assess the relative merits of each model and to frame the results in the context of similar work. The text reads well, but it lacks any quantitative comparison except toward the end, when the authors comment on the huge variability of available NPP estimates and provide large scale estimates for the overall production of the North Pacific. The manuscript would benefit from a more systematic assessment featuring regional averages (say, at the biome scale?) and some kind of statistical metric or test.3. There may be other things to say about the choice of the data for the comparison, and about how model output was preprocessed (for instance, how did you process Chl profiles, was there any attempt to mimic the way the ocean color satellites operate?). Since the simulations were forced using JRA-do reanalysis data, one would expect that the target for the models would be to reproduce or match available data. It is not clear what was the aim and objectives of the study, and perhaps that explanation is the only thing missing. The objectives and the rationale for choosing some data and patterns over potential alternatives needs to be justified. The models seems to be doing more than decent job, but the authors need to clarify to what extent some of the apparent biases observed both in surface and subsurface fields reflect are due to biases in simulated physical and chemical conditions or to differences on the phytoplankton model.4. Finally, a key aspect that the authors need to make clear earlier in the paper is the feasibility that proposed and discussed mechanisms may be actually working in the field. There is room to discuss alternative mechanisms currently ignored by the two models assessed. For instance, interactions between grazers and phytoplankters, potential biases in export and recycling, the metabolic diversity of phytoplankton (*e.g.* nitrogen fixers, picophytoplankton), etc.
5. As a bonus question, although it does not seem central to the study at hand, the formulation of zooplankton grazing was quite intriguing for two reasons the deserve further comment;1. the numerical response seems to be nonstandard and deserves further comment, as well as the closure term2. Eq (A2) in L429 includes a quadratic mortality term for phytoplankton. That effectively means that phytoplankton dynamics follow logistic growth, which seems redundant with the formulation of phytoplankton growth as a function of available nutrients, and underwater light and temperature conditions, and with the common assumption of a population controlled though grazing by zooplankton. Perhaps I am missing something?
-----------------------------
#### *Minor comments*
-----------------------------
*Abstract*
L001 - active voice? [Light and nutrient conditions ... ]
L002 - define photoacclimation?
L002 - at the end your model features both photoacclimation and variable stoichiometry, perhaps it is worth highlihgting it
L003 - break the sentence at the comme (it is already a bit twisted), and perhaps join with the next one?
L004 - as commented above, I recommend to go from simple to complex [say optimum nutrient uptake PFT to full FlexPFT], and provide a one sentence description with general details about the two models
L007 - mention OFES2 by name [and acronym]?
L010 - [...] subsurface Chl maxima *in the subtropical gyre* [to provide context]. Otherwise please detail where exactly that happens (especially the overestimation of Chl). As commented above, a figure detailing the magnitude of deviations with satellite data would be very useful.
As a general comment about the abstract; data used for validation is not mentioned at all. Readers might just assume you are testing your model against "oceanic observations" [L003], which may be too vague
-----------------------------
*Introduction*
L035 - not sure if there is something else besides pursuing efficiency and simplicity ;) ...
L077 - if InFexlPFT is a typical NPZD, then call it NPZD, or state here too that it is an NPZD implementing optimal uptake kinetics as per Smith et al 2009 [L135ff]?
L080 - perhaps deter giving the full name and details of OFES2 to MatMet?
-----------------------------
*Methods and Materials*
L085 - I understand you extensively modified the simple NPZD to implement either OU kinetics or FlexPFT. I mean, perhaps it is worth mention it and state that the default configuration consisted in a simple NPZD? [or maybe just provide those details later when talking about the BGC component of the model]
L089 - I would put all details about the configuration of the experiments together [L101]. The sentence in L93 is especially intriguing and disconnected from the rest of the explanation [L104].
L113 - Eq (1): perhaps Q(I,T) instead of just Q?
L119 - Eq (2): I really did not like the symbol $\mu_{Flex}(I,T)$, it seems potentially confusing ... what about just $\mu_{\mathrm{max}} \, S(I,T) \, F(T)$ or $\mu_{Flex}^{*}(I,T)$ [where I would suggest the former] ... otherwise it may lead users to think you need some functional or to iteratively solve the equation?
L124 - Please detail how do you determine maximum affinity [if it is optimized on a daily basis, etc]
L131 - please detail how the optimal value of $\theta$ is updated [I mean, that it is not a constant]
L137 - ideally, it would be nice to see how one can go from Eq (7) to Eq (1) [if that is possible], Otherwise it may be worth stating whether the models are truly nested or they just feature different terms for nutrient the dependent growth [though they propagate to the other terms in FlexPFT]
L153ff - perhaps explicitly include formulas for Chl and PP [$Chl = P \, \theta / Q \quad \text{in FlexPFT}$, etc]
L155 - I also miss some details here about what kind of outputs were compared to observations. In principle, since JRA-do is a renaalysis dataset, you might expect a direct match between simulated fields and observations at sea.
L165 - perhaps follow the order physics, chemistry, biology? Again, missing some rationale for the choices and the way the model was evaluated
-----------------------------
*Results and Discussion*
-> *general comment* as an author myself I can understand the preference for pooling the two sections, as a reader I am not such a fan.
L174ff - it seems that the physical component of the model was evaluated elsewhere; if that is the case it would be better to explicitly state so, but it would be ok to go beyond the ability of the model to reproduce major circulation features to mention at least its skill in reproducing temperature and nutrient fields.
L185 - there is certain paradox here since the initial focus of the manuscript on phytoplankton biomass and productivity mutates here on a large section devoted to two sections devoted to chlorophyll (which, needless to say in the context of a photoacclimation paper, is not biomass)
L185 and 200 - perhaps the titles of these sections should somehow detail that they refer to model to data intercomparisons (w.r.t. the section starting on L321)
L190 - please provide some quantitative summary of deviations between models and obs
L200ff - the narrative here becomes a bit difficult to follow to me ... perhaps an alternative structure [grouping results per biome for instance], or just a diagram or table summarizing the main findings [obs, simulated patters, most important regulating factor, etc] might make the section easier to follow
L277 - why? Is it possible to partition the amount of variation due to each to I, N and T?
L327 - horizontally and vertically? Could you develop a bit more what you mean?
L346ff - I think this paragraph belongs to the previous section
-----------------------------
*Conclusions*
L375 - perhaps *acclimation* instead of *adaptive response*?
L409 - perahps hte IA abbreviation can be omitted here to detail instantaneous acclimation
-----------------------------
*Appendix A*
L429 - please note comment above about Eq (A2)
L435 - please detail the type of numerical response (*i.e.* no need to force interested readers to go to Sasai et al 2016). My excursion to that paper suggest it is a Gompertz function with a threshold ... did not seem entirely standard (I mean, a commonly used formulation).
-----------------------------
*Figures*
Figure 1 - nice maps! Some suggestions doubts to comment in teh text;
- add transect lines to panels c and k?- what happened in the Gulf of Alaska and at Bering Sea?- the distinct shape of the gyre suggest there may be underlying biases in ocean physics propagating to chl [what about simulated MLD?]
Figures 2 and 3 - again nice figures and amazing results
- physics, chemistry, biology? [order of columns]- why not directly comparing data for 2002/2003? [it would be nice to check whether the model reproduces small scale heterogeneity]
Figures 6 and 7
- I like the figures but still feel they fail to clearly convey whether N and I are more important than T ... How would an equivalent figure with T in the abscissa look like? How can you partition which variable contributes more variability?
Figure 8
I think the results for FlexPFT would compare well with satellite based NPP products. Indeed, it would be great if, beyond biases in InFlexPFT the authors can show that actually the simpler model fails to capture large scale gradients [or at least, that is the impression I got].
Figure 9
Is it possible to complement these profiles with a time series plot? [perhaps the monthly climatology at each site]-
AC1: 'Reply on RC1', Yoshikazu Sasai, 15 Jul 2022
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2022/egusphere-2022-91/egusphere-2022-91-AC1-supplement.zip
-
AC1: 'Reply on RC1', Yoshikazu Sasai, 15 Jul 2022
-
RC2: 'Comment on egusphere-2022-91', Anonymous Referee #2, 04 Jun 2022
---------------------------
General Comments
The manuscript "Physiological flexibility of phytoplankton impacts modeled biomass and primary production across the North Pacific Ocean" by Y. Sasai and colleagues compares modeled phytoplankton biomass and primary production from a flexible plankton community model accounting for photoacclimation and variable C:N:Chl, with an inflexible plankton community model assuming constant C:N:Chl ratios. These models are coupled to a 3-D eddy-resolving ocean circulation model of the North Pacific. The authors compare the performance of these models by using Chl, nutrient, and primary production observations and find that primary production and chlorophyll were better predicted/modeled by incorporating photoacclimation and variable C:N:Chl ratios.
This manuscript provides valuable results that are important for the future implementation of plankton community models. However, as the manuscript stands, I suggest major revisions to outlay a more clear motivation and revise the methods, results, and discussion sections to allow readers to more easily follow this manuscript.
-----------------------------
Specific comments:
- There should be a more clear description of the structural differences between models. Although the description of the models is easy to follow, there is some confusion about what the key differences between models are. For example, throughout the manuscript, the text deviates on whether only the complex model implements photoacclimation or both models do. In Table 1, the differences in potential maximum growth rates can create confusion on whether it is the same model simply having a higher growth rate, or understanding where the main differences between models are coming from.
- The results section can be hard to follow in some parts, and quantitative information backing up the results stated will allow readers to better understand the variation between models and models and observations.
- The aims and objectives of the study are lacking throughout the manuscript, especially when stating what observations are being used. There needs to be a better explanation of why this data was used, and why comparing the last 20 years of the model run with observations from different years instead of exact comparisons?
- Lastly, an explanation of limitations and what still needs to be improved from these models can be useful.
------------------------------
Technical corrections:
Abstract:
L005 - Does InFlexPFT also incorporate photoacclimation?
L008 - Briefly Specify where these observations are coming from.
L009 - What about nutrients? They are mentioned in the earlier line.
L010 - Specify where this subsurface Chl maximum is reproduced, and the Chl concentrations are overestimated.
L014 - You should also state the role of FlexPFT incorporating photoacclimation.
Introduction:
L029-L030 - Provide further details on how they are debated.
L072 - cite some of the few tests that have been conducted.
L075 - FlexPFT is also an NPZD model no?, I would recommend rephrasing this sentence to more clearly depict the differences between the control and flexible C:N:Chl model.
Methods and Materials:
L085 - Very descriptive, but this sentence is a bit hard to follow, I would recommend restructuring to make it more clear.
L101 - state the value of this initial nitrogen N field if possible, otherwise be more specific on what you mean here.
L102-L03 - is there a reason why these values were used? Add citation, reasoning, or state that it is part of model calibration?
L104 - This sentence feels a bit out of place. I would add this to your previous description in L093.
L115 - If Q is a function of I, N, and T, I would add that in Eq1. Q(I,N,T).
L116-L117 - Add citation directing to Eq.4. Fv is repeated in L125.
L124 - Explain how you determine potential maximum affinity for N. Also cite table 1.
L131- cite table 1 after the theta explanation.
L132 - Is there a reasoning behind the activation energy Ea used? If so, cite it. Is it derived from observations?
L135 - I understand why $\mu_{InFlex}$ and $\mu_{Flex}$ are used, but they are quite lengthy, if possible I would abbreviate them to have shorter names.
L138 - since you already explained the potential maximum uptake rate and the potential maximum affinity for N above, I don't think you need to explain them again here, but do add the last part of this sentence and citations (L139) in L124.
L154 - This part is difficult to follow. Expand further on this paragraph. All these parameters are introduced, but no equation explains where they come from.
L158-L163 More explanation/rationale is needed here on model evaluation and why these observational datasets were selected.
L162-170 - It would be nice to map the observations and add them as a supplementary figure. It will be easier to understand what observations you are using.
Results and Discussion:
L174 - cite the satellite imagery and in-situ observations.
L174-177 - Should this physical evaluation go on the results. Was this part of this project or evaluated elsewhere? If so, state that.
L182 - Throughout the manuscript, the focus is on comparing biomass and primary production between these two models, but now through the results the focus changes to comparing the chlorophyll pattern which is a proxy to biomass, but not biomass.
L185 - The title should state this is a comparison since the paragraph concentrates on the model to satellite imagery comparison.
L187- Are there any biased statistics to see how well the seasonal variations compare and what the deviations are?
L190 - More quantitative information on this model to satellite imagery comparison would be useful to understand the degree of variation.
L200 - Same comment as L185 (state that it is a comparison in the title).
L200 - This section is difficult to follow. I would suggest restructuring and incorporating tables or diagrams summarizing the major findings, and categorizing the different areas you are comparing.
L217 - These last two sentences are a bit hard to follow, I suggest utilizing more quantitative comparisons between model and observations, to understand the degree of variation.
L169-171 - Is there reasoning why you think both models predict higher growth rates here?
L277-L278 - By what degree more so for FlexPFT?
L327 - Do you mean that the spring bloom occurs across latitudes and longitudes?
L332- L335 - Explain why FlexPFT predicts this.
L345 - Chl:C instead of Chl;C
L346 - I think this paragraph should go earlier.
Conclusions:
L376 - I think you should say you compared Chlorophyll instead of biomass.
Figures:
Figure 1.
- Minor point, but why not average from 2003-to 2019 to make the time comparison the same?
Figure 2.
- State what the white areas represent in panel a.
- Why not use just the 2006 model year for comparison instead of 2000-2019?
Figure 3.
- I would add the text again from Figure 2. Instead of saying "same as for Fig. 2.).
Figure 5.
- Same comment as figure 3. I would restate the information of the figure here.
Citation: https://doi.org/10.5194/egusphere-2022-91-RC2 -
AC2: 'Reply on RC2', Yoshikazu Sasai, 15 Jul 2022
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2022/egusphere-2022-91/egusphere-2022-91-AC2-supplement.zip
-
RC3: 'Comment on egusphere-2022-91', Anonymous Referee #3, 08 Jun 2022
This paper is a nice update on a line of work that aims to bring a modern representation of physiological plasticity in phytoplankton into the mainstream of ocean biogeochemical modelling. I have followed FlexPFT from a distance for a number of years, and it is useful to have a concrete illustration of how it behaves, and how it behaves differently from standard models, in a realistically complex 3D ocean simulation. I have a number of comments about how the discussion and expression of results could be improved, but I would class these as minor revisions.
- The review of evolving representations of phytoplankton physiology in the Introduction is especially nice.
- line 111: W per m^2, not W per m^3
- line 145-47: The role of parameter tuning in the comparison of the two model formulations is potentially very important. If the tuning of InFlexPFT had been done differently—for example, leaving mu_max the same, or lowering it further—would there have been a different pattern of similarities and differences between the 3D model runs? Which model shows a higher or lower growth rate at a particular point in space and time could be as much a matter of specific parameter choices as the structure of the equations. I would appreciate some comments on this point in the Discussion—what differences between Flex and InFlex are truly inherent and not contingent upon particular parameter choices.
- line 191: the success of FlexPFT at reproducing chl patterns seems to be largely a matter of dynamic range. Van Oostende et al. 2018 (https://www.sciencedirect.com/science/article/pii/S0079661117302586) also addressed this challenge in the North Pacific and found a solution by extending a standard 2-phytoplankton NPZD-style model to 3 phytoplankton compartments. So perhaps the poor relative performance of InFlexPFT is really highlighting the limitations of a 1-phytoplankton model. I think this requires some discussion (in the Discussion). If one is going to improve on inflexible plankton models by adding state variables, why add them in the form of flexible physiology instead of additional fixed-response phytoplankton compartments / functional groups? There is more at stake than simple statistical performance; to me the real issue is whether we think that the ocean achieves its wide dynamic range through acclimation and plasticity, or competitve exclusion.
- line 245: is there any way to make this comparison with observed variation in C:N more quantitative, or at least more specific? FlexPFT seems to show about four-fold variation in C:N over a vertical profile, if I am reading the results correctly—based on the references given in these lines, does this seem like roughly the right amount of variation, or too much?
- line 369: this feels like a weak comparison. What fraction of global PP _should_ the North Pacific account for? There is no additional information here, relative to Fig 9, on whether FlexPFT is a quantitative improvement over InFlex. Surely there are published estimates somewhere of North Pacific PP?
Citation: https://doi.org/10.5194/egusphere-2022-91-RC3 -
AC3: 'Reply on RC3', Yoshikazu Sasai, 15 Jul 2022
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2022/egusphere-2022-91/egusphere-2022-91-AC3-supplement.zip
-
AC3: 'Reply on RC3', Yoshikazu Sasai, 15 Jul 2022
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-91', Anonymous Referee #1, 15 May 2022
[comments formatted in Markdown, find attached the pdf version]
The manuscript *Physiological flexibility of phytoplankton impacts modeled biomass and primary production across the North Pacific Ocean* by Y. Sasai and colleagues assess the importance of optimal nutrient uptake, photoacclimation and variable stoichiometry on the emergence of large scale chlorophyll and primary production patterns in the ocean, using the North Pacific as a testbed. To do that, the authors designed an experiment based on the comparison of two strategic NPZD biogeochemical (BGC) models coupled to a state of the art, eddy-resolving model of ocean physics (OFES2). One of the BGC models implements optimal uptake kinetics ([Smith et al 2009](https://doi.org/10.3354/meps08022)), whereas the other incorporates interactive, optimal photoacclimation (Chl:C ratios) and variable stoichiometry (C:N deviating from Redfield ratios) as proposed in the context of the FlexPFT theory of [Smith et al 2016](https://doi.org/10.1093/plankt/fbv038)). The experiments revealed clear deviations between predicitions from both models, with a clear gain in accuracy when using the more complex FlexPFT model (bulk properties like Chl and fluxes like PP closer to observations, better resolution of the deep chlorophyll maximum (DCM), etc.). The authors conclude recommending the adoption of similar approaches by the BGC community.
The topic, approach and results are of great appeal and the manuscript is already an important contribution. The authors designed a clean test of their hypothesis about the emergence of surface and vertical gradients in phytoplankton growth and biomass, and the result clearly support their ideas. However, the manuscript is not easy to follow as it stands, especially the combined *Results and Discussion* section. The narrative in this section is quite descriptive, and it fails to provide a clear picture of the ability of each model to reproduce observed patterns. As commented below, these issues seem to arise in part from the lack of motivation and rationale for model assessment in *Methods*. There are other aspects missing like the limitations of the current model (*e.g.* what about nitrogen fixation?), alternative explanations to the emergence of DCM ([Cullen 2015](https://doi.org/10.1146/annurev-marine-010213-135111); there may be more recent reviews), or about the performance of similar BGC models in simulating Chl and PP in the Pacific. Together, these issues led me recommend a major revision of the manuscript. I provide some major concerns, and a long list of minor issues and suggestions below. I hope the authors find both lists useful.
-----------------------------
#### *Major comments*
1. The manuscript is in general well written and structured, but there are two sections that in my opinion deserve another thought, namely *Methods and Materials* and the *Results and Discussion* (see next three points). Although the description of the models is in general easy to follow (see however some minor suggestions below), the fact that the text moves from a complex model to a simpler one is not an optimal choice. I recommend the authors to present first their general approach with the components that are common to both models, and then detail first the simpler model featuring just optimal uptake followed by the more complex model featuring also photoacclimation and variable stoichiometry. I am aware that this might read as a minor issue, but I think it is quite important to ensure that readers realize that, despite their names, FlexPFT is something more complex than InFlexPFT. It is not clear whether one model is a nested version of the other or not (in the sense of a simpler formulation or the result of setting from variables as a constant). For instance, there is certain temptation to just check Table 1 and conclude that InFlexPFT results in reduced Chl and PP when compared with FlexPFT just because $\mu_{\text{max}}$ is lower in the former. There is also some confusion about whether the model implements only photoacclimation or photoacclimation and variable stoichiometry, and about whether one or both of them are simple NPZDs or not.2. Merging *Results* and *Discussion* has certain risks. In my opinion, that section of the manuscript needs to dig a bit more into the results and provide more quantitative tests that enable readers to assess the relative merits of each model and to frame the results in the context of similar work. The text reads well, but it lacks any quantitative comparison except toward the end, when the authors comment on the huge variability of available NPP estimates and provide large scale estimates for the overall production of the North Pacific. The manuscript would benefit from a more systematic assessment featuring regional averages (say, at the biome scale?) and some kind of statistical metric or test.3. There may be other things to say about the choice of the data for the comparison, and about how model output was preprocessed (for instance, how did you process Chl profiles, was there any attempt to mimic the way the ocean color satellites operate?). Since the simulations were forced using JRA-do reanalysis data, one would expect that the target for the models would be to reproduce or match available data. It is not clear what was the aim and objectives of the study, and perhaps that explanation is the only thing missing. The objectives and the rationale for choosing some data and patterns over potential alternatives needs to be justified. The models seems to be doing more than decent job, but the authors need to clarify to what extent some of the apparent biases observed both in surface and subsurface fields reflect are due to biases in simulated physical and chemical conditions or to differences on the phytoplankton model.4. Finally, a key aspect that the authors need to make clear earlier in the paper is the feasibility that proposed and discussed mechanisms may be actually working in the field. There is room to discuss alternative mechanisms currently ignored by the two models assessed. For instance, interactions between grazers and phytoplankters, potential biases in export and recycling, the metabolic diversity of phytoplankton (*e.g.* nitrogen fixers, picophytoplankton), etc.
5. As a bonus question, although it does not seem central to the study at hand, the formulation of zooplankton grazing was quite intriguing for two reasons the deserve further comment;1. the numerical response seems to be nonstandard and deserves further comment, as well as the closure term2. Eq (A2) in L429 includes a quadratic mortality term for phytoplankton. That effectively means that phytoplankton dynamics follow logistic growth, which seems redundant with the formulation of phytoplankton growth as a function of available nutrients, and underwater light and temperature conditions, and with the common assumption of a population controlled though grazing by zooplankton. Perhaps I am missing something?
-----------------------------
#### *Minor comments*
-----------------------------
*Abstract*
L001 - active voice? [Light and nutrient conditions ... ]
L002 - define photoacclimation?
L002 - at the end your model features both photoacclimation and variable stoichiometry, perhaps it is worth highlihgting it
L003 - break the sentence at the comme (it is already a bit twisted), and perhaps join with the next one?
L004 - as commented above, I recommend to go from simple to complex [say optimum nutrient uptake PFT to full FlexPFT], and provide a one sentence description with general details about the two models
L007 - mention OFES2 by name [and acronym]?
L010 - [...] subsurface Chl maxima *in the subtropical gyre* [to provide context]. Otherwise please detail where exactly that happens (especially the overestimation of Chl). As commented above, a figure detailing the magnitude of deviations with satellite data would be very useful.
As a general comment about the abstract; data used for validation is not mentioned at all. Readers might just assume you are testing your model against "oceanic observations" [L003], which may be too vague
-----------------------------
*Introduction*
L035 - not sure if there is something else besides pursuing efficiency and simplicity ;) ...
L077 - if InFexlPFT is a typical NPZD, then call it NPZD, or state here too that it is an NPZD implementing optimal uptake kinetics as per Smith et al 2009 [L135ff]?
L080 - perhaps deter giving the full name and details of OFES2 to MatMet?
-----------------------------
*Methods and Materials*
L085 - I understand you extensively modified the simple NPZD to implement either OU kinetics or FlexPFT. I mean, perhaps it is worth mention it and state that the default configuration consisted in a simple NPZD? [or maybe just provide those details later when talking about the BGC component of the model]
L089 - I would put all details about the configuration of the experiments together [L101]. The sentence in L93 is especially intriguing and disconnected from the rest of the explanation [L104].
L113 - Eq (1): perhaps Q(I,T) instead of just Q?
L119 - Eq (2): I really did not like the symbol $\mu_{Flex}(I,T)$, it seems potentially confusing ... what about just $\mu_{\mathrm{max}} \, S(I,T) \, F(T)$ or $\mu_{Flex}^{*}(I,T)$ [where I would suggest the former] ... otherwise it may lead users to think you need some functional or to iteratively solve the equation?
L124 - Please detail how do you determine maximum affinity [if it is optimized on a daily basis, etc]
L131 - please detail how the optimal value of $\theta$ is updated [I mean, that it is not a constant]
L137 - ideally, it would be nice to see how one can go from Eq (7) to Eq (1) [if that is possible], Otherwise it may be worth stating whether the models are truly nested or they just feature different terms for nutrient the dependent growth [though they propagate to the other terms in FlexPFT]
L153ff - perhaps explicitly include formulas for Chl and PP [$Chl = P \, \theta / Q \quad \text{in FlexPFT}$, etc]
L155 - I also miss some details here about what kind of outputs were compared to observations. In principle, since JRA-do is a renaalysis dataset, you might expect a direct match between simulated fields and observations at sea.
L165 - perhaps follow the order physics, chemistry, biology? Again, missing some rationale for the choices and the way the model was evaluated
-----------------------------
*Results and Discussion*
-> *general comment* as an author myself I can understand the preference for pooling the two sections, as a reader I am not such a fan.
L174ff - it seems that the physical component of the model was evaluated elsewhere; if that is the case it would be better to explicitly state so, but it would be ok to go beyond the ability of the model to reproduce major circulation features to mention at least its skill in reproducing temperature and nutrient fields.
L185 - there is certain paradox here since the initial focus of the manuscript on phytoplankton biomass and productivity mutates here on a large section devoted to two sections devoted to chlorophyll (which, needless to say in the context of a photoacclimation paper, is not biomass)
L185 and 200 - perhaps the titles of these sections should somehow detail that they refer to model to data intercomparisons (w.r.t. the section starting on L321)
L190 - please provide some quantitative summary of deviations between models and obs
L200ff - the narrative here becomes a bit difficult to follow to me ... perhaps an alternative structure [grouping results per biome for instance], or just a diagram or table summarizing the main findings [obs, simulated patters, most important regulating factor, etc] might make the section easier to follow
L277 - why? Is it possible to partition the amount of variation due to each to I, N and T?
L327 - horizontally and vertically? Could you develop a bit more what you mean?
L346ff - I think this paragraph belongs to the previous section
-----------------------------
*Conclusions*
L375 - perhaps *acclimation* instead of *adaptive response*?
L409 - perahps hte IA abbreviation can be omitted here to detail instantaneous acclimation
-----------------------------
*Appendix A*
L429 - please note comment above about Eq (A2)
L435 - please detail the type of numerical response (*i.e.* no need to force interested readers to go to Sasai et al 2016). My excursion to that paper suggest it is a Gompertz function with a threshold ... did not seem entirely standard (I mean, a commonly used formulation).
-----------------------------
*Figures*
Figure 1 - nice maps! Some suggestions doubts to comment in teh text;
- add transect lines to panels c and k?- what happened in the Gulf of Alaska and at Bering Sea?- the distinct shape of the gyre suggest there may be underlying biases in ocean physics propagating to chl [what about simulated MLD?]
Figures 2 and 3 - again nice figures and amazing results
- physics, chemistry, biology? [order of columns]- why not directly comparing data for 2002/2003? [it would be nice to check whether the model reproduces small scale heterogeneity]
Figures 6 and 7
- I like the figures but still feel they fail to clearly convey whether N and I are more important than T ... How would an equivalent figure with T in the abscissa look like? How can you partition which variable contributes more variability?
Figure 8
I think the results for FlexPFT would compare well with satellite based NPP products. Indeed, it would be great if, beyond biases in InFlexPFT the authors can show that actually the simpler model fails to capture large scale gradients [or at least, that is the impression I got].
Figure 9
Is it possible to complement these profiles with a time series plot? [perhaps the monthly climatology at each site]-
AC1: 'Reply on RC1', Yoshikazu Sasai, 15 Jul 2022
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2022/egusphere-2022-91/egusphere-2022-91-AC1-supplement.zip
-
AC1: 'Reply on RC1', Yoshikazu Sasai, 15 Jul 2022
-
RC2: 'Comment on egusphere-2022-91', Anonymous Referee #2, 04 Jun 2022
---------------------------
General Comments
The manuscript "Physiological flexibility of phytoplankton impacts modeled biomass and primary production across the North Pacific Ocean" by Y. Sasai and colleagues compares modeled phytoplankton biomass and primary production from a flexible plankton community model accounting for photoacclimation and variable C:N:Chl, with an inflexible plankton community model assuming constant C:N:Chl ratios. These models are coupled to a 3-D eddy-resolving ocean circulation model of the North Pacific. The authors compare the performance of these models by using Chl, nutrient, and primary production observations and find that primary production and chlorophyll were better predicted/modeled by incorporating photoacclimation and variable C:N:Chl ratios.
This manuscript provides valuable results that are important for the future implementation of plankton community models. However, as the manuscript stands, I suggest major revisions to outlay a more clear motivation and revise the methods, results, and discussion sections to allow readers to more easily follow this manuscript.
-----------------------------
Specific comments:
- There should be a more clear description of the structural differences between models. Although the description of the models is easy to follow, there is some confusion about what the key differences between models are. For example, throughout the manuscript, the text deviates on whether only the complex model implements photoacclimation or both models do. In Table 1, the differences in potential maximum growth rates can create confusion on whether it is the same model simply having a higher growth rate, or understanding where the main differences between models are coming from.
- The results section can be hard to follow in some parts, and quantitative information backing up the results stated will allow readers to better understand the variation between models and models and observations.
- The aims and objectives of the study are lacking throughout the manuscript, especially when stating what observations are being used. There needs to be a better explanation of why this data was used, and why comparing the last 20 years of the model run with observations from different years instead of exact comparisons?
- Lastly, an explanation of limitations and what still needs to be improved from these models can be useful.
------------------------------
Technical corrections:
Abstract:
L005 - Does InFlexPFT also incorporate photoacclimation?
L008 - Briefly Specify where these observations are coming from.
L009 - What about nutrients? They are mentioned in the earlier line.
L010 - Specify where this subsurface Chl maximum is reproduced, and the Chl concentrations are overestimated.
L014 - You should also state the role of FlexPFT incorporating photoacclimation.
Introduction:
L029-L030 - Provide further details on how they are debated.
L072 - cite some of the few tests that have been conducted.
L075 - FlexPFT is also an NPZD model no?, I would recommend rephrasing this sentence to more clearly depict the differences between the control and flexible C:N:Chl model.
Methods and Materials:
L085 - Very descriptive, but this sentence is a bit hard to follow, I would recommend restructuring to make it more clear.
L101 - state the value of this initial nitrogen N field if possible, otherwise be more specific on what you mean here.
L102-L03 - is there a reason why these values were used? Add citation, reasoning, or state that it is part of model calibration?
L104 - This sentence feels a bit out of place. I would add this to your previous description in L093.
L115 - If Q is a function of I, N, and T, I would add that in Eq1. Q(I,N,T).
L116-L117 - Add citation directing to Eq.4. Fv is repeated in L125.
L124 - Explain how you determine potential maximum affinity for N. Also cite table 1.
L131- cite table 1 after the theta explanation.
L132 - Is there a reasoning behind the activation energy Ea used? If so, cite it. Is it derived from observations?
L135 - I understand why $\mu_{InFlex}$ and $\mu_{Flex}$ are used, but they are quite lengthy, if possible I would abbreviate them to have shorter names.
L138 - since you already explained the potential maximum uptake rate and the potential maximum affinity for N above, I don't think you need to explain them again here, but do add the last part of this sentence and citations (L139) in L124.
L154 - This part is difficult to follow. Expand further on this paragraph. All these parameters are introduced, but no equation explains where they come from.
L158-L163 More explanation/rationale is needed here on model evaluation and why these observational datasets were selected.
L162-170 - It would be nice to map the observations and add them as a supplementary figure. It will be easier to understand what observations you are using.
Results and Discussion:
L174 - cite the satellite imagery and in-situ observations.
L174-177 - Should this physical evaluation go on the results. Was this part of this project or evaluated elsewhere? If so, state that.
L182 - Throughout the manuscript, the focus is on comparing biomass and primary production between these two models, but now through the results the focus changes to comparing the chlorophyll pattern which is a proxy to biomass, but not biomass.
L185 - The title should state this is a comparison since the paragraph concentrates on the model to satellite imagery comparison.
L187- Are there any biased statistics to see how well the seasonal variations compare and what the deviations are?
L190 - More quantitative information on this model to satellite imagery comparison would be useful to understand the degree of variation.
L200 - Same comment as L185 (state that it is a comparison in the title).
L200 - This section is difficult to follow. I would suggest restructuring and incorporating tables or diagrams summarizing the major findings, and categorizing the different areas you are comparing.
L217 - These last two sentences are a bit hard to follow, I suggest utilizing more quantitative comparisons between model and observations, to understand the degree of variation.
L169-171 - Is there reasoning why you think both models predict higher growth rates here?
L277-L278 - By what degree more so for FlexPFT?
L327 - Do you mean that the spring bloom occurs across latitudes and longitudes?
L332- L335 - Explain why FlexPFT predicts this.
L345 - Chl:C instead of Chl;C
L346 - I think this paragraph should go earlier.
Conclusions:
L376 - I think you should say you compared Chlorophyll instead of biomass.
Figures:
Figure 1.
- Minor point, but why not average from 2003-to 2019 to make the time comparison the same?
Figure 2.
- State what the white areas represent in panel a.
- Why not use just the 2006 model year for comparison instead of 2000-2019?
Figure 3.
- I would add the text again from Figure 2. Instead of saying "same as for Fig. 2.).
Figure 5.
- Same comment as figure 3. I would restate the information of the figure here.
Citation: https://doi.org/10.5194/egusphere-2022-91-RC2 -
AC2: 'Reply on RC2', Yoshikazu Sasai, 15 Jul 2022
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2022/egusphere-2022-91/egusphere-2022-91-AC2-supplement.zip
-
RC3: 'Comment on egusphere-2022-91', Anonymous Referee #3, 08 Jun 2022
This paper is a nice update on a line of work that aims to bring a modern representation of physiological plasticity in phytoplankton into the mainstream of ocean biogeochemical modelling. I have followed FlexPFT from a distance for a number of years, and it is useful to have a concrete illustration of how it behaves, and how it behaves differently from standard models, in a realistically complex 3D ocean simulation. I have a number of comments about how the discussion and expression of results could be improved, but I would class these as minor revisions.
- The review of evolving representations of phytoplankton physiology in the Introduction is especially nice.
- line 111: W per m^2, not W per m^3
- line 145-47: The role of parameter tuning in the comparison of the two model formulations is potentially very important. If the tuning of InFlexPFT had been done differently—for example, leaving mu_max the same, or lowering it further—would there have been a different pattern of similarities and differences between the 3D model runs? Which model shows a higher or lower growth rate at a particular point in space and time could be as much a matter of specific parameter choices as the structure of the equations. I would appreciate some comments on this point in the Discussion—what differences between Flex and InFlex are truly inherent and not contingent upon particular parameter choices.
- line 191: the success of FlexPFT at reproducing chl patterns seems to be largely a matter of dynamic range. Van Oostende et al. 2018 (https://www.sciencedirect.com/science/article/pii/S0079661117302586) also addressed this challenge in the North Pacific and found a solution by extending a standard 2-phytoplankton NPZD-style model to 3 phytoplankton compartments. So perhaps the poor relative performance of InFlexPFT is really highlighting the limitations of a 1-phytoplankton model. I think this requires some discussion (in the Discussion). If one is going to improve on inflexible plankton models by adding state variables, why add them in the form of flexible physiology instead of additional fixed-response phytoplankton compartments / functional groups? There is more at stake than simple statistical performance; to me the real issue is whether we think that the ocean achieves its wide dynamic range through acclimation and plasticity, or competitve exclusion.
- line 245: is there any way to make this comparison with observed variation in C:N more quantitative, or at least more specific? FlexPFT seems to show about four-fold variation in C:N over a vertical profile, if I am reading the results correctly—based on the references given in these lines, does this seem like roughly the right amount of variation, or too much?
- line 369: this feels like a weak comparison. What fraction of global PP _should_ the North Pacific account for? There is no additional information here, relative to Fig 9, on whether FlexPFT is a quantitative improvement over InFlex. Surely there are published estimates somewhere of North Pacific PP?
Citation: https://doi.org/10.5194/egusphere-2022-91-RC3 -
AC3: 'Reply on RC3', Yoshikazu Sasai, 15 Jul 2022
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2022/egusphere-2022-91/egusphere-2022-91-AC3-supplement.zip
-
AC3: 'Reply on RC3', Yoshikazu Sasai, 15 Jul 2022
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 307 | 77 | 10 | 394 | 7 | 8 |
- HTML: 307
- PDF: 77
- XML: 10
- Total: 394
- BibTeX: 7
- EndNote: 8
Viewed (geographical distribution)
| Country | # | Views | % |
|---|---|---|---|
| United States of America | 1 | 110 | 30 |
| Japan | 2 | 64 | 17 |
| China | 3 | 38 | 10 |
| Germany | 4 | 33 | 9 |
| United Kingdom | 5 | 23 | 6 |
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
- 110
Sherwood Lan Smith
Eko Siswanto
Hideharu Sasaki
Masami Nonaka
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(9746 KB) - Metadata XML