Review

This manuscript analyses climate model results of phytoplankton response in the North Atlantic. It examines a scenario of so-called “negative emissions” which implies artificial removal of CO2 from the atmosphere. The authors focus on differences between a future time (around year 2200) at which CO2 concentrations are back levels of ~400 ppm corresponding to approximately the year 2000 after having peaked at more than 700 ppm around year 2100. They find large differences in phytoplankton composition between those time periods, despite similar levels of CO2, and term this phenomenon “hysteresis”. The manuscript is well written and nicely illustrated and would be interesting for readers of this journal.

I have two more major comments and a few minor technical suggestions. The first major comment concerns the use of the word “hysteresis” in the title and elsewhere in the manuscript. It is different from the original use in physics (electrodynamics), where it describes differences in EQUILIBRIUM states of a system for identical boundary conditions, where the differences arise only from the history of the system. It often implies thresholds that mark switching between those different states. Here, however, the system under consideration is not at equilibrium and the differences between the states are due to delays in the response. I think the use of this term (hysteresis) is not warranted in this context and suggest describing it using a different term. At least, a discussion is warranted of this issue.

The second major comment concerns the use of half-saturation coefficients to model nutrient uptake in their model. It would be good if the authors could clarify if the use constant half-saturation coefficients or variable ones. In the real ocean there is a variety of half-saturation coefficients depending on species and other factors and it has been argued that half-saturation values should be modeled as variable (Aumont et al. 2015). What would be the effects on the results if variable half-saturation coefficients were used, if MARBL uses constant ones?

Line 166: “… thereby promoting the growth of small phytoplankton.” I think this would also apply to diatoms, right? If so, remove “small”.

84: “SO” explain acronym

93: replace “Leibig” with “Liebig”

96: Long reference is repeated.

180: “(NO3, SiO3)” What about PO4? Is it also decreasing? Is it modeled separately? I assume it could affect diazotrophs, who may be P limited.

186: “reported” by whom?

208-209: but in SPNA it shifts from Fe limitation to N-limitation

243-246, 274: the SPNA is a relatively small region of the global ocean. How much does global EP decrease and what is the contribution of the SPNA to it?

250-251: related to previous comment. This effect may not be large enough to affect atmospheric CO2 significantly.

259-260: Not sure this is the whole reason, since diazotrophs also decrease in the subtropical North Atlantic where SSTs do not decrease.

268: why is nutrient recovery delayed?

273-274: should it be “positive” rather than “negative” feedback?

281-283: It may be worthwhile to note here that the GFDL model does not exhibit "hysteresis"

Aumont, O., Eth., C., Tagliabue, A., Bopp, L., and Gehlen, M.: PISCES-v2: an ocean biogeochemical model for carbon and ecosystem studies, Geoscientific Model Development, 8, 2465–2513, https://doi.org/10.5194/gmd-8-2465-2015, 2015.

Summary

The authors present an interesting and generally well-written and well-structured manuscript investigating the response of phytoplankton communities to future climate change and “negative emissions” using model simulations. They find substantial shifts in community composition towards smaller species, which persist after atmospheric CO₂ concentrations are reduced. They attribute these shifts primarily to changes in the Atlantic Meridional Overturning Circulation (AMOC) and nutrient availability, and highlight the negative consequences for carbon export efficiency.

Overarching comments

I have some reservations about the use of “hysteresis” as the key term to describe the main findings. The term is likely unfamiliar to many readers and risks obscuring the central message, while also not being used entirely accurately in this context. Throughout the introduction, the authors refer to “irreversible responses”, “irreversible changes”, and state “we report irreversible shifts”, which sets the expectation that this language will underpin the key message of the paper. I therefore recommend replacing “hysteresis” with “irreversible shifts” or “irreversible changes” throughout the manuscript. Alternatively, if the authors wish to retain “hysteresis”, it should be clearly defined and introduced in the introduction rather than appearing for the first time, with no explanation, in the Experimental Design section (line 114).

Line by line comments

84: The acronym SO is undefined at this point. At line 132 it is defined as Southern Ocean, but that seems inconsistent here given the study’s focus on the SPNA region. Please clarify or correct.

96: Duplicated Long et al. (2021) reference.

120-130: The description of the statistical method (“The statistical method … [ ] … simulation”) would fit better in the Methods section rather than the Results.

186: the statement ‘… it has been reported that…’ requires a reference.

244: ‘prestige’ seems inappropriate in this context; consider replacing it with ‘lose this status’.

262: ‘negative CO2 emission’ should be changed to ‘negative CO2 emissions’.

269-270: ‘exacerbated’ means to intensify or increase, which is the opposite of what you report in your results for carbon export. This word needs changing to ‘reduced’ or a suitable synonym.

Figure 1: The green used for Diatoms in panel (b) is identical to the green used for Ramp-Up in panels (c–e), which could be confusing. I recommend selecting a different colour for the diatom category to avoid ambiguity.