My peer review is attached as a PDF.

General Comments

The paper examines the facilitation and inhibition of CO2 uptake throughout the entire duration of a peak-shaving deployment of SAI. A three-member ensemble simulation of SSP5-3.4-over is compared against SSP5-3.4-sulfur in CNRM-ESM2-1+. The paper finds finds that although there is an enhancement of the carbon cycle during the early phases of the deployment, the later phases see a decline in enhancement, eventually becoming an emission burden in the final phases of peak-shaving and in the years after deployment. 

I believe that this paper has valuable contributions and makes some excellent points about the impermanence of SAI’s carbon cycle enhancement in a peak-shaving scenario. However, there are portions of the paper in which the language can be imprecise, and the points made are overreaching. In particular, the paper states that phasing out SAI may be made less desirable or more difficult by the burden incurred in the later phases. This statement seems to make certain assumptions about the scenario. In particular, either it assumes that those in the scenario take advantage of the entirety of enhancement provided by SAI in the early stages and ignore that some of SAI’s carbon benefit is temporary, or it assumes that those doing CDR in the scenario care more about smaller rate of carbon removal they can do in the later phases than the smaller amount of carbon removal they must do in the later phases. 

In all, I believe the science of this paper to be very good, but it draws conclusions from its findings in a way that should be revised.

Specific Comments

1. Some notes around the concept of making phasing SRM out more difficult:
1. Line 92: Prolonged SRM deployment and higher CDR in response to sink degradation are a sliding scale, not necessarily an “and”. For example, higher CDR to exactly offset the source in phase III of SSP534-sulfur would result in the same phase out, if I’m understanding correctly.
2. The statements (e.g. Line 382) that the phase-out of SAI may be made “undesirable” can be further developed, as certain assumptions about those desires can be be made explicit.
1. If the global community uses up the extra CO2 budget gained from the early phases of peak-shaving by either not mitigating or doing less CDR, then would they have the CO2 concentrations of SSP534-over and thus an extra burden on the tail end. It is relatively clear how the people of Phases II and III might not want to ramp up CDR more than they would have had to. 
2. A less clear example is that if is if they mitigate and do CDR in Phase I as they would have in the baseline SSP534-over case, then they would have lower CO2 concentrations in Phases II and III. They would need more CDR for the same CO2 removal during these phases, but they would also have less CO2 to remove (in net, 60 Gt less). If the speed at which carbon cycle draws out CO2 is the primary desire, they would find difficulty, but if they value getting the CO2 concentration back down to a certain level while maintaining a certain temperature (the premise of peak shaving), I struggle to see how phasing out SAI would be made more undesirable.
3. Lines 324-328 / Figure 3: the extra 38 GT of CO2 burden seems like it would be less than two years of extra CDR as modeled in ssp534-sulfur, ending in 2152 instead of 2150. While I agree with the “non-negligible” statement with respect to CDR that exists today, the CDR in the paper’s scenarios far outpace the examples the paper gives. This should be at least addressed.
2. Line 21: The concept of “buying time” during peak shaving usually refers to time it takes halt temperature rise and reduce it again, and the risks incurred during that temperature peak. It does not usually refer to the extra help from carbon cycle enhancement.
1. E.g. Zarnetske PL, Gurevitch J, Franklin J, Groffman PM, Harrison CS, Hellmann JJ, Hoffman FM, Kothari S, Robock A, Tilmes S, Visioni D, Wu J, Xia L, Yang CE. Potential ecological impacts of climate intervention by reflecting sunlight to cool Earth. Proc Natl Acad Sci U S A. 2021 Apr 13;118(15):e1921854118. doi: 10.1073/pnas.1921854118. PMID: 33876741; PMCID: PMC8053992.
3. The difference in compatible emissions do not seem linked to the three phases of SAI deployment specifically, considering the crossover at 2100. Looking at 3b and 4a (for land at least), it looks as though under SSP534-sulfur, the sink becomes more of a sink and the source becomes more of a source. To say in line 377 that the uptake is increased during SAI roll-out but decreased during phase-out feels off, since it seems to have more to do with when it’s a source or sink, and thus the background GHG emissions pathway and the CO2 concentrations. 
1. In other words, would phasing out SAI while doing no CDR cause the negative difference starting at 2100 in Figure 3b? Alternatively, would not phasing out SAI and maintaining a constant amount of cooling while doing the same amount of CDR as simulated create the 2100 crossover? 
4. Line 26: Although CO2 removal is often considered a form of mitigation (IPCC AR6 WGIII: CDR Factsheet), it is also often considered separate from mitigation (e.g. NASEM, 2021: “This portfolio must involve reducing GHG emissions to the atmosphere (mitigation), and removing carbon from the atmosphere and reliably sequestering it”). This paper should explicitly state whether it is defining mitigation to include CDR or considering CDR to be independent, cite a source to support that definition, and then stick to it for the duration of the paper. 
1. I would recommend separating the two, such that something like the first line of the abstract “...allowing additional time for the implementation of conventional climate mitigation strategies and CO2 removal.” could be read without confusion, regardless of whether the reader defines CO2 removal as mitigation or not.
5. Lines 37-78: Be clear about the baseline that the paper is comparing SAI to with each comparison. For example, the CO2 fertilization effect is likely to the same temperature, no-SAI, but the reduced heat stress is likely to the same CO2 concentration or year, no-SAI. 
6. Line 100: Regarding “increase of 2°C”, say relative to what baseline (pre-industrial, but also discuss how are it is being defined)
7. Line 117: The paper should give SSP126 the same background as it gave SSP534. At present, it is put into the text without definition.
8. Figure 3b: I am curious about the total area over 0 and under 0 (which sum to be 60 Gt), especially since there seem to be two “phases”: where the difference is positive going until ~2100 and where it is negative after.
9. Figure 3b: It would be nice to have a measure of variability – what is significant difference between compatible emissions vs. what is a byproduct of variability. 
10. Figure 4: Some sort of Carbon sink vs. Temperature plot may be useful here. Masking temperature by land and ocean could be included, or use cooling done by SAI vs. difference in sinks, but given that peak-shaving SAI is about control of temperature, it may be enlightening to see what effect it has (or does not have) on the sinks. 
11. Line 253-254: Cite Trisos et al. 2018 or similar paper about the effects of termination shock when talking about a sudden cessation of SRM having different impacts.
1. Trisos, C.H., Amatulli, G., Gurevitch, J. et al. Potentially dangerous consequences for biodiversity of solar geoengineering implementation and termination. Nat Ecol Evol 2, 475–482 (2018). https://doi.org/10.1038/s41559-017-0431-0
12. Line 332: Not necessarily disagreeing with “SAI is not CDR,” but this statement is a little unclear. Will the net 60 Tg carbon benefit become net 0 Tg eventually? If so, then highlight this. If not, then some of SAI’s CO2 removal is permanent, if limited. It is worth noting that the paper says in Line 325 that amounts of CO2 of similar magnitude to 60 Gt are “non-negligible.” 
13. Line 359: Cirrus Cloud Thinning is technically not Solar Radiation Modification.

Technical Corrections

1. Line 20: 1.5°C -> 2.0°C
2. Line 26 / Line 92: CDR is never defined to be CO2 Removal or Carbon Dioxide Removal; it appears for the first time in line 92.
3. Line 33: Injections -> Injection, enhance -> enhances
4. Line 47: Sur- face 
5. Line 62: “the major levers” feels informal -- consider different phrasing
6. Figure 1a: mention pre-industrial somewhere (perhaps Y-label or title)
7. Figure 1b: X-label – Years->Year
8. Figure 2: mention pre-industrial somewhere (perhaps Y-label or title)
9. Line 181: The-> the
10. Line 195: be- tween
11. Line 241: are -> is
12. Line 251-253: “Hence their call . . . under SRM” is a sentence fragment.
13. Line 377: “a third comes from the ocean.”