Please do not be disheartened by my recommendation of major revisions: this is a promising manuscript that will represent a significant contribution to scientific progress when it is ready for publication. I have selected major revisions because I feel that some of the explanation is currently incomplete and I have suggested some additional analysis that may take longer than would be appropriate for minor revisions. I enjoyed reading this manuscript and I look forward to seeing a revised version. Please see attached PDF for my reviewer comments. 

Content comments

The paper discusses changes in Arctic ocean primary production and its drivers in CMIP6 models with comparison to CMIP5, with some focus on subregional scales.

1. Despite initial indication of subregional analysis there is great emphasis and description of panarctic averages. Given the high variability and large differences among Arctic regions for both drivers and PP, I don’t think a panarctic view is very meaningful. It may be a good benchmark to add, but only becomes scientifically useful if the regional drivers are discussed/understood. While there is regional discussion later on in the paper (mostly spatial maps) and one figure with multi-model mean subregional trends, I think the value of the science would be much higher if the subregional trends were shown/discussed for all models and in context with the initial panarctic mean discussions. This would allow to highlight the differences among models with information on where the models show different PP, if the drivers are consistent among models and regions or if there are certain regions/models that drive the panarctic mean. It would also allow insights into how the model uncertainty varies across region and time for both PP and limitation functions.
2. The mixed use of relative and absolute values leads to inaccurate statements and misleading impressions of PP increase/decrease. Please ensure comparisons are accurate and actually comparable and I highly recommend the use of absolute values to adequately highlight/represent what is going on in the CAO versus the shelf and subarctic seas.
3. Please refrain from making blunt and potentially inaccurate statements about Impacts on Inuit (see detailed note below).

In summary, I think the paper has really good potential to provide a valuable science contribution. A detailed, subregional evaluation of CMIP6/CMIP5 models is overdue!  However, at this stage I recommend major changes to improve the discussion/value of the subregional analysis, ensure accurate comparisons and highlighting, and remove potentially harmful commenting on Inuit impacts.

Detailed content comments:

Figure 1 and text P8L7-15: I assume this is pan-Arctic. I find a pan-Arctic average evaluation not very meaningful as the Arctic is quite diverse, both in drivers and responses (see e.g. Steiner and Reader 2024,  https://doi.org/10.1029/2024JC020970). a panarctic average can represent great correspondence all around or bad correspondence with some areas over and some underestimated. Given that the authors indicated 10 regions, I suggest add/discuss the evaluation for all regions in addition to the pan-Arctic already here. That way it is possible to see which regions may be contributing more or less to model-obs biases (see introductory note).

Also P10L1-10 I think adding the subregional evaluation here would allow to be more quantitative on regional biases and strengthen this section.

Figure 3: Indicate in caption what region is averaged over. Again I suggest doing these plots for all subregions at least in the supplementary material (if available for this journal), it is really the regional changes that are interesting and useful.

P11 L1-5 This would be really interesting to see for the sub regions

Figure 7 Indicate region average, I assume this is pan-Arctic.

Again,  I see even less value in pan-Arctic limitation functions, I think this discussion would be more valuable if subdivided or when the panarctic mean is supported by the  subregional analysis right from the start

P16: “Remarkably, the regions of greatest relative NPP uncertainty (central Arctic Ocean) differ from those exhibiting the largest absolute NPP increases (shelf regions, Fig. 4a), “

As these are relative differences, I assume the very low PP values in the CAO lead to the very high relative differences and the pattern may not be very meaningful. Absolute anomalies would be more meaningful here.  Also, given the use of relative differences, which overemphasize the CAO in this context this is not “remarkable” at all. This statement  is in fact misleading. I suggest to clarify this in this section and either remove or revise to " Given the very low PP values in the CAO it is not surprising that....”

L11 “highlighting that uncertainty patterns do not necessarily coincide with the magnitude of projected changes.”

Given that the authors are comparing relative with absolute, this statement is not accurate. To compare accurately either compare relative with relative or absolute with absolute, otherwise remove the statement.

P17 L10/11 “Relative increases in total NPP are greatest in the central Arctic basin, with >200 % enhancement compared to 1995-2014 values.”

Again, note that with the very low PP this may not be very meaningful, and actually misleading, especially given the color coding which gives an impression of a massive greening of the Arctic as well as giving the impression that the 40% decline in southerly regions is minor, which it definitely is not. I strongly recommend using the absolute values here.

P18 L5 “Nlim is lower”

For Nlim it is important to highlight that some Nlim is because there is a lot of productivity which quickly consumes available nutrients, others are because there are not a lot of N in the first place, so a more useful estimate would be the ratio of limitation to productivity, or alternatively present the Nlim at the begin of the productive season (which is more difficult as this varies by latitude).

L8/9 “40 %, indicating that diatoms face strongest nutrient constraints there.”

Inaccurate phrasing, largest change does not mean largest constraint, largest constraints are where values are lowest (again, ideally referring to the begin of the growing season)

L11/12 “2014). “Light limitation is highest”  - This is light limitation, i.e the higher the factor, the lower is the limitation, i.e light limitation is lowest in the Nordic and Barents

P21 L7 “these ecosystem changes pose serious threats to local Inuit communities, …”

careful with the phrasing here, Inuit don't like to hear this, especially if it is neither clear nor accurate:

1. these are projections and not proven changes, 2. threats have not been evaluated in this study, 3. only phytoplankton productivity has been evaluated, this is not the full ecosystem, 4. primary productivity is projected to increase, why should this negatively impact Inuit?

Please rephrase or even better,  remove this sentence. If there is a need or strong wish to keep it, I suggest using something like "the projected changes on PP are expected to impact the local ecosystems with unclear impacts on Inuit subsistence species and harvesting".

L9 “factors” : maybe also indicate remaining differences in biogeochemical parameterizations, e.g. community structure?  I suggest the authors also add a note on the lack/limited inclusion of sea ice algae in CMIP models which impacts timing of productivity, vertical distribution and carbon export.  (could refer to ice algal model papers, e.g sea ice book Castellani et al 2025)

Technical comments

P2 l26 rm space after scape

P3 L18 one bracket only – Eyring

2.2 Model diagnostics ( rm s)

P4 l5  for light and nutrients

L22 provide actual equation for Eppley function

P7 line 2 - I wonder if there is a newer publication as you are referring to WOA18, several years after the referenced publication

Figure 1: suggest to not use black as a color for individual models (even with circles and triangles) to better highlight the model mean, maybe also make the mean line thicker  also indicate averaging area in caption

P8 L8 394 Tg… : add by *** as for obs

P10L19-21 greater => larger anomalies , greater => higher radiative forcing, smaller model ensemble

P10L11 larger lower????

Fig 4 caption in 2081 => for 2081

P12 L10 losing up to more – losing more

L11 The Barents

L12 rm: in these zones

P13 L0 ensemble, however

L11: “The area-specific Arctic Ocean ΔNPP is relatively low compared to other regions,” unclear what is meant here. what does area-specific mean and what are the other regions ?

P14 L 5 suggest rem “This region” at the beginning – start with Baffin Bay differs

also “the multi-model”

L7 anomaly remains – I think "change" works better here

L9 basins: maybe ocean regions? there are many regions in the Arctic, but only very few basins

P15 L11-12  “relaxed light limitation which is augmented by the influence of warming on phytoplankton temperature functions and slightly offset by enhanced nutrient limitation “ - awkward phrasing, suggest improved light condition or "a reduction in the light limitation", then start a new sentence

L13 However, the deltaLim

Fig 10 caption – correct bracket SSP8

P18 L10: “maximum Nlim decrease of less than 20 % “ - suggest "with Nlim decreases below 20%

L15 The light

L16 The temperature

P19 L14 are contributing, suggest use “amay contribute”

L20 exhibit a different

P20 L3 check citation format Lewis et al

L4 than the scenario

L5 suggest start new sentence after NPP projections

L6 among ( 2 =between, >2 among)

L9 no comma after bracket, - a higher

L18  mean state concentrations of what ?

L20 “ Regions experiencing NPP decline, such as Baffin Bay and the Nordic Seas,”

duplication , suggest just "These regions ...

L24 The Nordic

L32 This production => This increase is

L33 the Nordic

L36 suggest removing the word fundamental, maybe a bit too strong here

L37 remove thus

P21 L3 “implications”  suggest "impacts": ....

also the authors indicate diverse implications, but then don't identify what is diverse. Maybe use "with a variety of impacts:" in lieu

L3-5 suggest splitting into 2 sentences

L9 “ varying sea ice thickness and optical properties,”

awkward phrasing, suggest: through sea ice of varying thickness and optical properties

L17 suggest remove “incidentally”

L19 suggest "unclear impacts"

L20 suggest  “potential implications”