the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 03 Jul 2024
Toward more robust NPP projections in the North Atlantic Ocean
Abstract. Phytoplankton plays a crucial role in both climate regulation and marine biodiversity, yet it faces escalating threats due to climate change. Projecting the future changes in phytoplankton biomass and productivity under climate change requires the utilization of Earth System Models capable of resolving marine biogeochemistry, and exploits the averaged responses across an ensemble of models (within the Coupled Model Intercomparison Project Phase 6, CMIP6) as the most probable projection. However, in the North Atlantic, this straightforward method falls short in providing robust projections of phytoplankton net primary production (NPP) over the 21st century. This is because the processes controlling NPP strongly differ from one model to another, thus causing model divergence. A new inter-comparison approach was therefore developed and applied to 8 CMIP6 models exhibiting substantial divergence in their NPP projections in the North Atlantic. This approach is based on the identification of the mechanisms causing model divergence and the assessment of their reliability, in order to conduct an informed selection of the most reliable models. The basin was first divided into 3 bioregions tailored to the characteristics of each model using a novel method based on a clustering procedure. Two key mechanisms causing model divergence were then identified in the subtropical and subpolar regions (namely, diazotrophy and the presence of an ammonium pool, respectively). This allowed for an informed selection of models in each region, resulting in reduced uncertainty and a more pronounced decrease in total NPP in the subtropical North Atlantic and a stronger increase of small phytoplanton NPP in the subpolar North Atlantic. Our model selection strengthened carbon export and phytoplankton biomass decreases under climate change, but had no impact on zooplankton biomass. By leveraging the diversity of CMIP6 models, this innovative approach identifies the key mechanisms influencing NPP projections and provides valuable insights into the future trajectory of the Earth’s climate system.
- Preprint
(3631 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-1820', Anonymous Referee #1, 10 Sep 2024
General comments
NPP projections in CMIP6 biogeochemical models strongly diverge during the 21st century in the North Atlantic despite a strong consistency between the physical and biogeochemical variables controlling NPP. This manuscript studies an ensemble of 8 ESMs which embed diverse biogeochemical components with significant differences in their parametrizations of primary production. Two mechanisms causing model divergence are identified: diazotrophy (for the subtropical region) and the presence of an ammonium pool (for the subpolar region).
This manuscript is worth publishing. It is well written and I was quite convinced by the explanations. But some further work is needed to make it easier to read, more convincing and impactful. The variety of models (with different PFTs/limiting nutrients/species naming) makes it difficult to disentangle and to interpret results. Throughout the manuscript, there are some results that are not shown in figures. It is a bit confusing. I suggest that you should either show them, or explicitly state that they are “not shown” in the text. Reference to Figures (and which subplot) and Tables could be improved to better guide the reader. In particular, section 3.3.2 on the Subpolar region needs to be better justified (with at least an additional figure or with a sensitivity experiment demonstrating the influence of the ammonium pool).
Detailed comments
Introduction:
While reading the manuscript, I thought about emergent constraints as another method that could be relevant given the objective of this study. You finally mention and discuss this method and relevance for this specific topic only in the conclusion. The same could apply to a selection of CMIP6 models based on their ability to represent past observed changes in NPP. We also somehow lack validation of the ability of the selected models to represent past NPP changes. I would recommend better framing the different methods that could be used (including emergent constraints) and why you chose this one much earlier in the manuscript, as early as in the introduction.
P3-4 Section 2.1:
- Table 1. It would be of interest to list the equilibrium climate sensitivity of the chosen models as well, to inform on their sensitivity to climate change. I would also suggest mentioning the presence (or not) of an ammonium pool in the BGC models.
- Line 88-89: Please specify the models you are referring to (the simplest and most complex BGC models) and refer to Table 1
- Line 89: typo, "the most complex model” : “"complex” is currently missing
- Line 101-102: For the IPSL model, you might be explicit on the version of diazotrophy that is used (given the Bopp et al. 2022 study).
- L91: Why don’t you combine diatoms and microphytoplankton as one large phytoplankton class ? In the manuscript, you justify this because diatoms rely on silicate limitation, but in the end, it seems to me that you never use this silicate limitation to disentangle some processes between the different models. This would allow you to show an additional curve (CanESM5-CanOE) in Fig. 3, 5 and 6, panel B.
- Could you add 1-2 sentences about the differences between the models regarding PO4 and NH4 pools to introduce your explanations about their influences in the following sections? This would immediately explain why some models are missing in the PO4 panel of Fig. 3.
Section 2.2.2. Please provide the explicit identifier (number) of the product to avoid confusion as Copernicus Marine Service products evolve over time
P5 L152: how were coastal areas excluded?
P6 L175: Please refer to Table 2.
P6 L180. Models disagree on sign, but are projected changes large in these regions in the different models? 2D maps of model spread would be useful to complement Fig 1.
P7 Section 3.2. The reference to Fig 2 comes too late. The same applies to the merging of the 2 subpolar subregions. Their separation might be shown in Fig 2 with a dashed line for instance.
P7 L214: yes, surface NO3 concentrations are decreasing in all models but for the 3 ones that project an increase in NPP, it decreases significantly less (2-3 times less) than the others. Same for PO4. Do you think this is a consequence of the increase in N2 fixation and hence of the increase in NPP ?
P7 L213-214: Why do you show only NO3, PO4 and temperature at sea surface while other variables are vertically integrated? This needs to be better justified…
P8, L219: Fig4 -> Fig4a
P8, paragraph on models without any diazotrophy:
- You refer to an observational dataset “the observed value is 0.4 mmol.m-3” but which obs dataset? Over what time period? Perhaps you could add it to Fig.3
- For ACCESS-ESM1-5, you do not mention the increase of MLD in the projections. Is this not the reason for the increase in small phytoplankton NPP (more than the SST increase mentioned in the text)?
- P8 L228. Replace “nanophytoplankton” by “small phytoplankton” for consistency throughout the manuscript.
P8, paragraph on models with explicit diazotrophy: please precise the subplot of Fig 3 you refer to. L233: primary production = total primary production?
P8 L245: For consistency of this section and previous two paragraphs, you might start this paragraph in a more general sense, addressing models with parameterized diazotrophy.
P8, paragraph on models with implicit diazotrophy: please, precise “not shown” at the end of the sentence “However, unlike in IPSL-CM6A-LR, …”. It is not clear from your explanations why the two CanESM models (CanESM5 and CanESM5-CanOE) give such different projections of NPP. L250, does “large phytoplankton” refer to the microphytoplankton of CanESM5-CanOE?
P9 L268: What about diatoms? All models show a diatom NPP decrease but with groups of models showing substantially different decreases.
P9, paragraph L272-283: here an additional figure showing the absolute value of NO3 and NH4 as a function of time is needed so that the reader can see what you are referring to in the paragraph. I am also wondering how you explain the different behaviour between IPSL-CM6A-LR and CESM2/CESM2-WACCM models regarding the projections of diatom NPP and small PHY NPP.
P10, L291: “were still covered” -> will still be covered
P10, L292-293: sentence “However… NO3 and PO4 surface concentrations decrease in all models (fig. 6)”: But it is not the same process as in the subpolar region that causes NO3 and PO4 to decrease. They are probably more consumed due to the increase in NPP.
P10, L289-293: What about changes in upper ocean stratification? Changes in NPP can also occur in areas that were not and will not be covered by sea-ice.
P10, L296: You refer to the half-saturation constant. Is it that of diatoms?
P10. Section 3.4 lacks a reference to Table 2.
P11, L332: I would delete: “and the extreme would be kept in the subset”
P11, L335: “influence diatoms in the former but not the latter”: is it not the opposite?
P13, L390: Typo: “Emergent” --> “emergent”
Fig. 1: Use the same arrangement as in Fig. 3 for the box at the bottom of the figure. The caption could be made more explicit on using the multi-member mean for each ESM.
Fig. 2: Could you add the boundary between the 2 subpolar subregions (with a dotted line for example) so that the reader is aware of their extent in each model and observations?
Fig. 3f: diazotrophy subplot: Please give more details about this “diazotrophy” term. Or give the equations. Is it equal to diazotroph NPP? Why is the unit different between explicit and implicit diazotrophs? It is confusing as it is.
Fig. 3g and h: Please specify that surface NO3 and PO4 concentrations are shown.
Fig. 1, 3, 5, 6: Please make these important figures larger with larger labels. Difficult to distinguish the blue colour of IPSL-CM6A-LR from that of CESM2-WACCM. Please add a title indicating the variable on each subplot to make it easier to read. Could you find a graphical way to remind the reader which model belongs to which group (no diazotrophy, implicit or explicit one) to make it easier to interpret and follow your explanations?
Fig. 4: Instead of “correlations”, I would call it “scatterplot” and then add the correlation numbers to the subplot. Could you explain in more details how you constructed this graph? What does each cross represent? This figure needs to be clarified.
Fig. 7: Please recall the model selection for each region in the figure caption
Table 1: Replace “Diatoms (param.) with Diazotrophs (param.). Could you add to this table the nutrients limiting the growth of each phytoplankton class, so that we know if there is a NH4/PO4/Si… pool? The words “with a parametrization and without an explicit diazotroph group” should be removed in the Table 1 description.
Citation: https://doi.org/10.5194/egusphere-2024-1820-RC1 -
RC2: 'Comment on egusphere-2024-1820', Anonymous Referee #2, 15 Sep 2024
The paper describes a method for making more robust projections of NPP than the ensemble mean normally used for CMIP6 models. The authors use a sub-ensemble of models that shows a strong coherence in the variables that controls NPP. By dividing the North Atlantic into bioregions, the best models in each region is being identified through comparisons of the differences in sensitivity to some mechanisms (diatoms, small phytoplankton, N2 fixation,...). This selection also results in more certain estimates of future carbon export.
The paper is well written and to the point. My main concern is on the statement on the strong coherence between the models based on large scale averaging of processes controlling NPP. This is an oversimplification. Yes, these are important processes, but NPP is controlled by physics on a scale (both spatial and temporal) that is not shown agree by the authors. When that is said, the authors are perfectly right on their statement on model democracy, thus studies like this to make more robust projections are of great significance. I also agree with the authors that such selection should be based both through a comparison with observations (most common) and the present process-based selection. I recommend the paper for publication after a moderate revision. This revision should include a more throughout discussion on the physical drivers, their impact on NPP, and possibly a more detailed comparison of this physics between the selected models (see e.g. https://doi.org/10.1016/j.seares.2023.102366). As it is now, the assumption on coherence is not convincing and thereby also the hypothesis on differences in model parameterization as the only cause for model discrepancy.
I have added a few minor points below.
• line 45. Perhaps one of the most important effects of climate change will be shifts in community structure. I doubt the CMIP6 models include much of this. To allow for full dynamics will require more focused trait based models (e.g. https://doi.org/10.1093/icesjms/fsy090)
• line 58. I am not convinced by this statement about coherence. Neither Figure 1 nor Kwiatkowski et al., 2020, is to a level of detail that this can be concluded on a regional and temporal level.
• line 60. With this in mind I do not agree that the discrepancies observed necessarily originates from the differences in the representation and parameterization of the biological processes. My postulation would be that the main reason for the observed discrepancies are the physics, but on a more detailed level than investigated by authors.
• line 62. An example on how to identify different physical drivers and investigate their impact on modeled primary production can be found in https://doi.org/10.1016/j.seares.2023.102366
• line 67. The hypothesis comes as a consequence of the authors conclusion about coherence. There is a need to elaborate more on the importance of physics both local and regionally.
• line 181. As said above I am not convinced about this strong coherence. Although I recognize that robustness defined as sign agreement is a reasonable metric, regional and decadal to multi-decadal variability is lost in this averaging.
• Section 4.2. It should be mentioned although NPP is important for ecosystem projections, timing and species composition is also of vital importanceCitation: https://doi.org/10.5194/egusphere-2024-1820-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 243 | 74 | 38 | 355 | 10 | 10 |
- HTML: 243
- PDF: 74
- XML: 38
- Total: 355
- BibTeX: 10
- EndNote: 10
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1