This study nicely illustrates the impact of NOx levels in determining whether regions within South America and Africa are likely to incur an ozone-climate penalty. This is useful work and should be suitable for publication after accounting for the suggestions below. There is a robust correlation between isoprene emission changes and ozone production. A signal of correlation of increased ozone production with increased NOx change is found, and in one model there is a correlation with absolute NOx levels.

Reasons for the NOx changes are not explored, these could be due to changes in wet/dry deposition, changes in emission (lightning is mentioned, but not explored) or changes in organic nitrate and PAN formation. The latter are mentioned in the introduction, but not followed up in the discussion. In figure 6 (particularly UKESM1, but also GISS) the NOx and Isoprene emission changes seem anticorrelated over S. America, and possibly also over Africa. This would support the increased sequestration of NOx in organic nitrates and PANs.

It would be useful to look at P-L to determine whether the isoprene emissions have a net positive or negative impact on ozone. The introduction implies that the sign of the net effect depends on the NOx background whereas in figure 6 for UKESM1 the areas of increased isoprene emission all seem to have increase ozone production. Would the balance become more obvious when looking at P-L?

Using monthly values in figure 7 might lead to spurious correlations. There will be strongly seasonal variations in isoprene and NOx, and also strong seasonal variations in meteorology (wet vs dry). Some of the correlations in figure 7 might be due to the meteorology -i.e. isoprene emission changes might be stronger in the dry season where the meteorological impacts on ozone might also be more positive if dry gets drier.

Line 58-60: This sentence starts with the effect of ozone on climate, but the references are all to the effect of climate on ozone. Ozone doesn’t lead to a positive forcing through increases in anthropogenic precursors, it is through its absorption and emission of longwave radiation.

Line 82: “Biogenic isoprene is the major O3-forming NMVOC …” This seems to imply that more ozone is formed from isoprene than from other NMVOCs. Is this true? Is this globally or just over forests? Does this mean gross formation i.e. ISOPOO +NO dominates the sum of RO2+NO, and ignoring the sinks. Elsewhere it is not clear even whether isoprene is a net producer of ozone.

Line 86: It is mostly OH+NO2->HNO3 that causes NOx-saturation.

Line 115: This could be described better by explicitly saying that O1D+H2O is the major ozone loss.

Section 2.1: Are the Price and Rind parameterisations in UKESM1 and MRI the same? Thornhill et al. 2021 found different climate responses from schemes that were all supposed to be Price and Rind. Soil NOx is mentioned, do these models include climate-dependent soil NOx emissions?

Line 234: It might be clearer to state that fixing CO2 in pd means that the difference compared to the baseline includes the effect of CO2 inhibition.

Line 241: Are the model levels really “above the canopy” or are they above the orography?

Line 386: Might be clearer to say that Emissions from cities create positive sensitivities to climate change in all months.

Line 414: The plot of loss frequency is very useful.  It would also be useful to show dry deposition as a velocity in the supplement i.e. dividing by ozone concentration in kg/m3.

Line 414: In UKESM1 CO2 will also affect the stomatal closure, I don’t know about the other models.

Figure 7: The correlation patterns here don’t look to be quite the same as judging the correlations by eye from figure 6. This might be because this uses month as another variable. Is the isoprene coefficient really 0.00 for UKESM1? It looks as if higher isoprene leads to higher ozone production, both in figure 7 and in 6.

Line 512: Should say which reactions happen faster at higher temperatures.

Conclusions: This doesn’t discuss why there is an ozone-climate penalty.

Line 572: Strictly you have shown that the ozone-climate penalty is in areas of high NOx rather than ozone.

This manuscript presents an interesting work focusing on the climate change effect on near surface ozone over South America and Africa using three ESMs.

It is generally well structured and presented. However I have a number of points that have to be addressed before acceptance. See my analytical comments below:

Comments

line 51:It can be also produced from the oxidation of CO in the presence of  adequate levels of NOx so that net ozone production (P-L) is positive. Please modify accordingly.

line 68: Apart from chemical and biological processes there are also complex dynamical processes such as transport from the stratosphere to troposphere. See for example Morgenstern et al. 2018, Akritidis et al., 2019.

line 84: There also soil NOx emissions which are not mentioned (e,g. Romer et al., 2018).

line 84: What do the authors mean with severely NOx-limited regions? Please clarify.

line 86-87: Do the authors imply the NO+O3  titration process acting as an ozone sink close to NO emission sources which is common in highly polluted areas being in the VOC limited regime?

line 102: Concerning the isoprene chemistry in models, maybe the authors could also refer here more explicitly the uncertainty due to the different model assumptions on the yields of isoprene nitrates and their subsequent NOx-recycling ratios.

line 223: The authors mention that the simulations are fully coupled but the ocean is not coupled since SSTs are prescribed.

line 230: Apart from the cited reference, this approach has been also applied in Chapter 6 of IPCC AR6 (Szopa et al, 2021).

line 240: The authors mention that tropics as lying between 40 N and 40 S. The tropics are commonly defined as the area between the Tropic of Cancer (roughly 23.5-degrees North latitude) and the Tropic of Capricorn (roughly 23.5 degrees-South Latitude). In the domain of Figure 1, the subtropics of Northern and Southern Hemisphere are also included. This is a comment to be considered as in several places throughout the manuscript the authors refer to the tropics but the analysis includes also subtropics.

line 272: Maybe you can also discuss this result in relation to the results of Turnock et al. (2020).

line 316: Do you have some explanation why NOx surface concentration decreases in UKESM and slightly increase in MRI and GISS in Fig 2b? A discussion with possible reasons is missing.

line 316: There is no discussion of OH changes in Fig. 2c. Could you discuss why MRI and UKESM show an increase while GISS shows a decrease? Theoretically, increases in methane, CO and NMVOCs reduce OH while increases in water vapour and temperature, incoming solar radiation, NOx and tropospheric ozone enhance OH.

line 331: Please discuss also how the models deal with the soil NOx emissions in the simulations.

lines 387-389: It is rather confusing when discussing megacities and urban scale in model results with much coarser resolution. The current datasets produced for CMIP6 were produced at 0.5^∘ (historical anthropogenic and future) (Feng et al., 2020). Capturing the distinctions between urban and rural emissions, and the finer distinctions in between, is an ongoing challenge for emission inventories for global datasets. I think that the discussion should rather point that areas with ozone penalty identified in this study include a number of highly populated regions and megacities.

line 390: The authors introduce in Section 3.4, chemical production and loss terms but they have to describe how these terms are calculated (simply discuss the model diagnostics for these terms).

lines 397-398: There is missing discussion on the reasons for the decrease of dry deposition of ozone due to climate change. Is this a robust signal? There are different physical and biophysical processes affecting dry deposition of ozone on which climate change may induce opposite sign of change. To what level these processes are parameterized in the model simulations is crucial for the understanding.

lines 412-413: This is a comment linked to my previous comment on dry deposition of ozone. Please check to what level biosphere-atmosphere interactions are taken into account in the current simulations (negative and positive feedbacks on ozone through stomatal upatake and deposition velocity reduction). Mind also the links of dry deposition velocity with the boundary layer changes under a warmer climate.

lines 437-438: Could you commend on the possible reasons that determine the net ozone production increase for UKESM? Despite NOx decrease in many regions due to natural emission changes and chemistry under climate change, net ozone production is positive. So it is essential the high NOx levels at polluted regions (NOx regional levels) which play a key role for an increase in net ozone production rate in a warmer and more humid environment. On top of that there also isoprene emission increases which may have an impact on ozone and it could useful to discuss their contribution on net ozone production rates (in which regions have a positive or negative impact and if this is related to the regions NOx levels).

line 502: It is rather misleading when the authors state that "climate change could lead to an ozone–climate penalty in areas which already have a high background O3 concentration". The ozone climate penalty is linked to the regional NOx levels in a highly polluted region where net ozone production rates increase in warmer and more humid environment (even if natural NOx levels are slightly reduced due to climate change).

Figure 6: A letter should be assigned in each one of the 12 sub-figures of the panel.

Minor technical comments

lines 26-27: The whole sentence needs rephrasing as it does not read well.

line 30: Please add a "comma" after "on average".

lines 211-212: The sentence needs rephrasing as it does not read well.

line 226: Please replace "are" with "and".

line 271 : Please replace "the" with "that".

line 291: I would suggest "range between" instead of "fall within".

line 376: It is Fig. 4c instead of Fig.4d.

line 377: It is Fig. 4d instead of Fig.4c.

line 382: I would suggest "larger" than "much more extreme".