We thank Reviewer #1 for the constructive and encouraging comments and for emphasizing the strengths of the manuscript with respect to the dataset, methodology, and interpretation of physical–biological coupling in the Mediterranean Sea. Below, we address the comments point by point.

General comment

The manuscript presents a well-structured and comprehensive analysis of the spatial and temporal variability of primary production in the Mediterranean Sea over the past 25 years, based on an impressive satellite dataset. The study provides a valuable contribution to our understanding of the coupling between physical forcing and phytoplankton dynamics in this complex basin. The results are coherent and highlight the key role of physical drivers such as mixing, circulation, and nutrient inputs in shaping phytoplankton biomass, diversity, and productivity at both seasonal and interannual scales.

The paper is clearly written, well organized, and offers novel insights through the use of empirical orthogonal function (EOF) decomposition to identify dominant modes of variability. The analysis successfully integrates multiple parameters (SST, MLD, chlorophyll) to characterize the relationships governing primary production. However, some aspects could be further strengthened. In particular, the absence of an explicit analysis of light availability and its variability deserves justification, given its potential influence on phytoplankton dynamics and primary production.

Author’s reply: We fully agree with the reviewer and we plan to add the light field to complement the analysis of the other variables.

Similarly, while the discussion of surface primary production (PPfo) is interesting as a complementary product, it partly duplicates the conclusions derived from surface chlorophyll patterns and could be streamlined to focus on the most relevant findings.

Author’s reply: The revised version will be re-organized to avoid repetitions and to better streamline the discussion on the contributions by the various variables.

Overall, this is an interesting and well-conducted study that significantly contributes to the long-term understanding of Mediterranean primary production. The figures and data presentation are clear and effectively support the conclusions. I recommend publication after minor technical revisions, mainly aimed at clarifying the introduction and improving the description of the applied methodology (e.g., briefly summarizing Volpe et al., 2012, to ensure full reproducibility). Adding a few additional references could also help strengthen the contextual framework.

Technical comments:

Line 28: Both “color” and “colour” are used in the manuscript. Please harmonize the spelling throughout the text for consistency and improved readability.

Author’s reply: We will harmonize this.

Line 63: Although the ¹⁴C method for estimating primary production is widely used, some uncertainties remain regarding its accuracy (e.g., Baltar and Herndl, 2019). This point could be briefly acknowledged in the text.

Author’s reply: Thanks. We will include this in the introduction.

Line 125: A period appears to be missing after the closing parenthesis.

Author’s reply: We will check the punctuation all over the manuscript.

In Figure 1, would it be possible to include the Mixed Layer Depth in the sensitivity analysis of primary production? This could help illustrate its influence more clearly.

Author’s reply: We understand the point and it would be very useful if MLD was an input to the model. Unfortunately, this is not the case as the model “only” uses light, chlorophyll and temperature, for which the sensitivity has been checked.

Line 206: For improved readability, consider adding a short definition of “Case-2 waters.”

Author’s reply: We will add a short description in section 2.1.2.

Line 230: A brief description of the method developed by Volpe et al. (2012) would be helpful here to ensure reproducibility.

Author’s reply: In section 2.2, ee will add a short description of the methodology developed in Volpe et al (2012) and relevant for this work.

Line 250: Please provide a reference to support the statement on the “well-known longitudinal gradient.”

Author’s reply: We will add references to Bosc et al. (2004) and D’Ortenzio and Ribera d’Alcalà (2009), which are already included in the reference list.

Line 255: Could the absence of a longitudinal PPeu gradient be related to an excessively deep integration range?

Author’s reply: The absence of a longitudinal gradient in PPeu is consistent with the findings of Volpe et al. (2012) who, while explaining the longitudinal gradient in surface Chl, wrote “Moreover, although the total chlorophyll values differ by a factor of roughly four (ranging from nearly 3, during summer, to more than 5 in spring, Fig. 3a–b), average surface values diverge by an order of magnitude (Santoleri et al., 2008).”. In other words, integration over the entire euphotic layer mitigates the longitudinal gradient between the two sub-basins. Since PPeu is integrated over the euphotic layer, it is not surprising that it does respond to this dynamics.

Line 267: Might the apparently high productivity in the Gulf of Gabès be influenced by bottom reflectance affecting satellite observations?

Author’s reply: Yes. In the revised version we will underline more clearly that the coastal PP values should be regarded with care as the present version of the PP model is optimized for open ocean waters.

Line 274: In Figure 2, if possible, please consider using a color-blind-friendly palette to enhance accessibility.

Author’s reply: All figures will use a color-blind-friendly palette.

Line 370: As a complement to this analysis, relevant results can also be found in Toseland et al. (2013).

Author’s reply: Toseland et al (2013) will be added as reference.

Line 375: Please refer to Marañón et al. (2021) for discussion of the relationship between light and the Deep Chlorophyll Maximum.

Author’s reply: Marañón et al (2021) will be added as reference.

Line 404: It is unclear what “these two” refers to. Please clarify.

Author’s reply: Sentence will be modified.

Line 440: It may be useful to discuss the PSC (phytoplankton size class) distribution in light of how it is derived (i.e., directly from chlorophyll concentration).

Author’s reply: The way PSC were derived is mentioned at line 289 as caveat to explain the large correlation with total chlorophyll.

Line 468: How confident are you that this pattern results from long-term warming rather than from an increased frequency of marine heatwaves?

Author’s reply: We will add a sentence to mention that the thermal long-term trend could be associated with the increased frequency of marine heatwaves.

We thank Reviewer #2 for the detailed and thoughtful assessment of the manuscript. We appreciate the recognition of the richness of the dataset and the relevance of the topic, as well as the constructive comments aimed at improving the clarity and robustness of the study. We acknowledge the concerns regarding the articulation of the scientific objectives, the description of the methodology, and the structure of the manuscript. In response, we will substantially revise the text to better clarify the aims and key results, to more clearly outline the methodological framework, and to strengthen the discussion. The primary production model will be described in more detail, probably with a dedicated section added to the Appendix. Below, we address the main and specific comments point by point.

General comments

The manuscript presents a basin-scale analysis of Mediterranean phytoplankton biomass, size structure, and primary production using more than 25 years of data from the Copernicus Marine Service products. Seasonal modes of the phytoplanktonic variables with sea surface temperature (SST) and mixed layer depth (MLD) are derived with Empirical Orthogonal Function (EOF). Physical control of phytoplankton is assessed with statistical correlation between the modes.

The dataset is rich and the topic is relevant, however, I find that the scientific objectives and key results are not clearly articulated, I have several concerns about the method which, in general, lacks clear outlining. The text is long with a heavy structure and could largely benefit from a synthesis effort grounded on Volpe et al., (2012) results. Finally, there is no discussion on the used data sets or on the method.

I think the study has potential but needs significant refinement to clearly demonstrate its contributions. As such, I recommend major revision before consideration in the Ocean State Report 10. The main points are addressed below and further detailed through the Specific Comments.

• I found it difficult to identify the key objectives of the Is it to demonstrate the capabilities of recent products? The consideration of the primary production integrated over the first optical depth (PPfo) feels like a late addition. Despite 25+ years of data and while methods are readily available (e.g., filtered EOF in Volpe et al., 2012) the study does not robustly explore long-term trends on the variables.

Author’s reply: We will modify the sentences at the end of the introduction to better account for the general paper objectives. To avoid the possible spurious correlation that can arise from comparing surface variables (Chl, SST) with integrated variables (PP) and to have a comparison reference, we will introduce the necessity of also computing PP over the first-optical-depth (PPfo). We will stress that, to increase the comparison efficiency, PPfo will need to be computed exactly the same way as PPeu. Long-term trends are not included here because constitute the topic of another work in preparation.

• The method should gain in clarity and In the primary production model several key points need to be addressed or at least discussed (case II against case I water, the validation of reconstructed chlorophyll-a vertical profiles). Second, statistical correlation tests should be better introduced and reinforced with significance tests at least for the one long term trend.

Author’s reply: We will add details about PP model especially about the chlorophyll vertical reconstruction and physiology parameterization. Moreover, in the method section we will add all relevant details about the statistical correlation tests and their significance.

• I found that the manuscript form did not help highlighting the key results of the study. I feel that the writing quality is uneven with heavy phrasing, inconsistencies (both in language and statistical metrics) and redundancy (especially through the result section) that add unnecessary complexity. I suggest a full language edit and a restructuring of the results based on a synthesis I believe that clearly stating the result interpretation from Volpe et al.,(2012) early in the paper would allow an increased focus on newly considered variables (CHL, PP and PSC). Finally, defining clearly all used acronyms, variables and mechanisms early on (through the introduction and method) would improve fluidity.

Author’s reply: We will try to make the text "easier" to read, reducing it as much as possible and making it more fluent

• The author discusses the oceanographic interpretation of the modes and causality mechanisms inferred from statistical correlation. They do not mention study limitations (e.g. centered on abiotic control) or on the new products, their limitations (e.g. PP model) nor what they recommend in future products developments (e.g. using PSC in PP calculations).

Author’s reply: We will modify the text to also highlight the limitations of the PP approach and model. Furthermore, we will add a paragraph at the end of the discussion on possible future developments and products that could help improve this type of analysis

Specific comments

Line 9: I suggest to be more explicit on how causality links are assessed. e.g. While EOF is used to find the dominant seasonal modes of variability of the biological variables, their link to physical forcing are assessed through statistical correlation.

Author’s reply: We will do it  

Through the manuscript, there is a lack of continuity in usage of acronyms with SST being defined at line 36, 131, 139 and 211. CHL is first introduced as chlorophyll-a concentration (at Line 34), then Chl is defined as chlorophyll concentration (Line 116) and referred to as ‘biomass’ from the result section. I suggest homogenizing the references to surface chlorophyll-a concentration.

Author’s reply: we will review the text, remove repetitions and ensuring that references to CHL are consistent

I feel that two narrative threads coexist in the introduction: a first one from paragraph 1 to 7 and a second one the last three paragraphs. Many ideas are recalled several times (e.g. the need of basin scale products is the core subject of paragraph 6 but also recalled at line 69, 82, 89, 126). I think the introduction could benefit from a synthesis effort. Many details are stated (or recalled) in the last three sections of the introduction. The comparison of statistical methods in the last paragraph of the introduction might be more fitted to the method section or, considering the word constraint of the OSR, to the supplementary material.

Author’s reply: We will revise the introduction, condensing it, and we will move the details described for the statistical methods to the methods section, if necessary

line 31: This paragraph (or the next) is an opportunity to recall that phytoplankton biomass and diversity is categorized as an EOV by the GOOS.

Author’s reply: Thank you for the suggestion. We will definitely take it into consideration

line 106: The study of Volpe is firstly cited here. This paper seems to be the methodological inspiration of the presented manuscript. I recommend making more explicit the methodological similarities and differences (SST and MLD are used here while SST and sea surface anomalies in Volpe et al., 2012). I also suggest mentioning, here or in the method, what the environmental variables are proxying for. This would fluidify the reading of subsequent sections.

Author’s reply: The revised introduction will include a paragraph about the choice of the variables used in the work and how and why they are fit for purpose. In the method section we will expand on similarities and differences with respect to Volpe et al. 2012.

From line 109: the authors state the objective of the study. I think that this paragraph could gain in clarity in relation to the above comments. Are long term trends considered or only seasonal patterns? To what purpose are PPeu and PPfo considered?

Author’s reply: The objectives of the work will be more explicit and consistent with the new version of the introduction. In the method section we will expand on the PPeu and PPfo concepts and their overall contribution to the paper.

I suggest rephrasing line 114 to 115 especially the goal of the work, and the way it was applied. Line 114: I do not understand the author’s intention with this method’s header. It is long and redundant with what should only be stated in dedicated sub-sections. The sentence starting at line 119 ends at line 127. I suggest simplifying this header by stating only the first sentence and then sorting each idea in the respective subsequent sub section (phytoplankton sizes classes should be described in section 2.1.2, PP in section 2.1.1, line 130 in section 2.1)

Author’s reply: Thank you for the suggestions, we will modify the text accordingly

Line 128: I suggest using the exact value from Morel and Berthon, 1989 instead of one-fifth as it is already an approximation.

Author’s reply: We will substitute “about one-fifth” with “1/4.6”

line 130: Volpe et al., 2012, consider sea level anomaly as a proxy for currents, the study keeps only SST and MLD as proxies of oceanic forcing, what is the reason behind that?

Author’s reply: MLD dynamics is among others the best physical parameter or proxy that shapes phytoplankton dynamics at seasonal or annual scale. In 2012 a model with a good representation of the MLD dynamics was not available yet, and MADT was used instead.

Line 144: What are the spatio-temporal resampling techniques? Is it a monthly average or a snapshot?

Author’s reply: We will add that the nearest neighborough approach was used for the spatial regridding before computing the monthly average.

Line 146: I suggest rephrasing which constitutes the topic of the work.

Author’s reply: We will substitute the sentence with: “The use of monthly resolution minimizes the influence of high-frequency variability, thereby emphasizing the annual and interannual signals that constitute the primary focus of this investigation”

line 146-147: I do not understand what links the spatial and temporal resolution here.

Author’s reply: The phenomena under investigation point out the required spatiotemporal resolution of the data, as spatial and temporal scales are inherently coupled. Small-scale phenomena (on the order of 1–2 km) evolve over days, whereas large-scale phenomena (spanning many kilometres) evolve over months. Therefore, to focus on annual and inter-annual signals using monthly data, a sufficiently coarse spatial resolution must also be chosen. This effectively filters out smaller-scale, high-frequency noise and retains the relevant large-scale patterns. We will substitute the sentences with: “The monthly temporal resolution attenuates high-frequency variability, focusing the analysis on the annual and inter-annual signals that constitute the primary focus of this investigation. By adopting a 4 km spatial resolution and be consistent with the monthly scale variability, we can identify potential mesoscale contributions by filtering out the finest scales, which are also incompatible with monthly time steps.”

Line 152: I fail to see why the author reconstructed a chlorophyll-a profile up to the optical depth instead of directly using the satellite chlorophyll-a concentration which characterizes the FOD. Furthermore, It remains unclear to what purpose the FOD is considered.

Author’s reply: It is correct that satellite-derived chlorophyll concentration represents the mean concentration over the first optical depth (FOD). However, we adopted the same chlorophyll-profile reconstruction method used to estimate euphotic-zone primary production (PPeu), in order to ensure that the two primary production estimates (PPeu and FO-layer PP (PPfo)) are fully consistent and directly comparable. As for the PPfo role in the paper, please see response to general comment 1.

line 161: the presented equation of P=P(z,lambda,t) is not yet column integrated. Depth and wavelength integration are required to get column integrated PP(t). Could the author furnish more information on how the photosynthetic quantum yield evolves in respect of chl(z) and T(z)?

Author’s reply: (Please note that this platform does not support formula rendering) We will revise the formula to include the triple integral over wavelength, time, and depth and we will add a sentence at the end of the paragraph: “The daily, column-integrated primary production (PP) was computed as the triple integral of the production rate with respect to wavelength (over the PAR band, 400–700 nm), time (over 24 hours), and depth (throughout the euphotic zone for PPeu and down to the first optical depth for PPfo).” Following the original model, the quantum yield is scaled relative to its maximum value phi(max) through a dimensionless light-response function:  phi=phi(max)*f(x) with x=PUR/KPUR where PUR is the photosynthetically usable radiation and K_PUR is a scaling irradiance controlling the transition between light-limited and light-saturated photosynthesis. The function f(x) follows the formulation of Platt et al. (1980), accounting for saturation and photoinhibition. Temperature influences photosynthesis through its effect on the scaling irradiance K_PUR, which is parameterized using the Eppley-type formulations: K_PUR = K_PUR (20°) * 1.065^(T-20). As a consequence, temperature modulates the realized quantum yield indirectly by shifting the photosynthesis–irradiance response, without altering either the functional form of f(x) or the definition of phi(max). In the original formulation, the maximum quantum yield phi(max) is assumed to be constant, under the hypothesis that its variability compensates for opposing changes in the maximum chlorophyll-specific absorption coefficient a(max). In the present study, following Bricaud et al. (1998), the chlorophyll-specific absorption spectra are allowed to vary in both magnitude and shape as a function of chlorophyll concentration. Consequently, a(max) is no longer constant but varies with chlorophyll concentration (thus with depth), implying that phi(max) cannot be treated as constant either. Following Morel et al. (1996), the maximum quantum yield has been expressed as a function of chlorophyll concentration as: phi(max)=0.05*[Chl^0.66/(0.44+chl^0.66)]. This formulation accounts for observed photo-physiological adaptations across trophic regimes while preserving the original representation of light and temperature effects on photosynthesis. By introducing the explicit chlorophyll dependencies for both phi(max) and a(max), the present formulation makes explicit the physiological variability that was implicitly embedded in the constant product phi(max)*a(max) in the original model. We will synthesize this in a few sentences in the manuscript to better describe the effect of chlorophyll and temperature on the photosynthetic quantum yield.

Line 171-172: the authors describe how light penetration is modelled through the water column for PP calculation. The used methods from Bricaud et al., 1998 and Morel et al., 2002 are valid for Case I water. I think this point questions the validity of the subsequent PP estimation in case II water (straits, western Adriatic, and the Gulf of Gabes). (Line 251, 266, line 381, line 466)

Author’s reply: The reviewer is right, PP model offers better performance in Case 1 waters, given that its various parameterizations were developed for open ocean waters. We will add statements here and in other sections of the paper to clarify that in coastal waters the model estimates have a higher associated uncertainty and that results require careful interpretation. In any case, the Mediterranean Sea is overall a “Blue water” (i.e. Case 1) basin, and the optically complex waters, where the ad-hoc algorithm is invoked, account for ~1% of the basin’s pixels over the entire time series.

From line 172 to line 182: the authors describe the statistical reconstruction of chlorophyll-a vertical profiles from surface observation using a k-means clustering on the MedBiop dataset. This methodological point is pivotal for the study and seems to be redundant with the method in the QUID of product ref. no. 1 (Colella et al. 2025). I could not find a statistical validation of the reconstructed vertical profiles of chlorophyll-a from surface observations, could the authors forward the corresponding citation?

Author’s reply: Presently, there is no other study that describes this method for reconstructing chlorophyll profiles. Therefore, we will include and detail it, comparing the reconstructed profiles for the Mediterranean Sea with the historical reconstructions by Morel and Berthon (1989) and Uitz et al. (2006).

Line 180 and later on: the reference could be to product ref. no. 3 so to be more concise.

Author’s reply: We will update the reference, thanks.

The sensitivity analysis confirms how PP behaves in respect of phytoplankton biomass, illumination and temperature. I have trouble identifying the need for this sensitivity analysis as to me, it is expected and highlights even further that the reconstructed vertical profiles of chlorophyll-a (discussed above) must be proven to be correct.

Author’s reply: We believe the model's sensitivity analysis is important for demonstrating its behaviour and tackling the responses to the physical environment. Since, as stated above, we aim to provide evidence of the robustness of the Chl profile reconstruction method, we deem also useful to present this part of the analysis.

To improve readability I suggest moving the PP section (2.3.1) after sections 2.1.2 and 2.1.3 which introduce Chl and SST.

Author’s reply: Thanks, we will evaluate the option of moving the PP section to improve the readability of the paper

From line 200: I suggest rephrasing the paragraph on chlorophyll-a to improve reading fluidity.

Author’s reply: We will rephrase the paragraph

Line 217: I suggest rephrasing the sentence to in modulating space-time variability of phytoplankton biomass, production and diversity.

Author’s reply: We will modify the sentence

I suggest rephrasing section 2.2 and to mention earlier that the method is a copy of Volpe et al., 2012 which derived EOF for all variables (PP, CHL, SST, MLD) by centering all time series and recalling that only mean fields and seasonal anomalies will be investigated (no filtered EOF).

Author’s reply: We will follow the reviewer suggestion of setting out more clearly that no filtered EOF is being used in this context.

Section 2.2 did not introduce how mean and anomalous fields of biomass and productivity fields are correlated to environmental variables. In the results section, metrics related to linear correlation (sometimes with temporal lag) are used. The type of correlation subsequently used (spatial, temporal or lagged between EOF modes) should be briefly explained here.

Author’s reply: We will add relevant details about the performed analysis, here.

Line 231 and the header of section 3 (line 237) both announce how the results will be addressed. I think that only one of the two is necessary.

Author’s reply: Yes, we will rephrase the text to avoid redundancy

In the caption of figure 2 the averaged fields are shown up to 2023 (which is fine) while at line 250 the average Chl concentration is stated to be over the entire period. This should be checked for consistency.

Author’s reply: The error is in the caption for Figure 2. The reference period for the datasets is 1998–2025. We will check and correct for any inconsistencies

Line 285: In accordance with a previous comment, the method behind the metric given as percent is not presented. I understand it is the amount of variance explained by the linear model between the two variables, but is it a linear correlation on the mean field (shown in figure 2) or a correlation on each individual time series that is averaged later on? At line 290, r² coefficients are considered and raise identical questions. This should be addressed in the method with a subsection dedicated to statistical correlation.

Author’s reply: As mentioned above, we will add relevant details about the performed analysis in the method section.

Figure 3: For clarity, I suggest to change the caption to : Spatial dominance of PSC size classes (Micro, Nano and Pico in panels a, b and c, respectively) in months. Panel d shows the overall dominance of each size class in percent over the entire time series.

Author’s reply: Thanks, We will modify the caption text, taking the suggestion into account.

I also suggest defining what are the percentages in the legend of panel d ( temporal or spatial dominance?) And to define how to compute a pixel dominance in the method.

Author’s reply: The revised version will include a clear definition of both "pixel dominance" and how the percentages shown in the figure were computed.

Thorough section 3.2.1: I would suggest presenting first, the mode and then, discussing its spatial variability. I found that the reversed order (spatial variability before the EOF mode) was confusing.

Author’s reply: We will evaluate whether to reverse the order of the discussions. If doing so improves the text's readability, we will certainly make the change

From line 318: and up to the end of the manuscript, the term phytoplankton biomass is used instead of Chl (or CHL). This change brought confusion when discussing seasonal evolution of periods (line 355) as seasonal DCM formation could also be invoked to discuss diminution of phytoplanktonic biomass in the surface layer.

Author’s reply: We will clarify earlier in the text that Chl, in absence of seasonal photoacclimation, is a good proxy of phytoplankton biomass and as such the two will be used interchangeably.

Line 330: I suggest rephrasing this sentence as the other temporal evolution is vague (other modes ? of MLD or of the other variables?)

Author’s reply: We will rephrase the sentence to clarify that by "the other temporal evolution," we were referring to the "temporal evolution" of the first mode of the other variables

From Line 335: the author discusses the spatial connection with SST 1st mode which is related to seasonal cooling and warming. I recommend describing this mode as such first.

Author’s reply: We will evaluate whether moving the analysis of the SST leading mode to the beginning of the paragraph improves the text's readability

I suggest rephrasing line 345 and line 353-354.

Author’s reply: Thank you for the suggestion; we will rewrite these sentences to make them clearer

From line 365: the authors discuss the temporal mismatch of PPeu and CHL and link it to the subsequent warming and stratification. Over the course of a year, regimes transition from high nutrient low light (winter) - high nutrient high light (spring) - low nutrient high light (summer) - low nutrient low light (fall) (Bellacicco et al., 2016). Therefore seasonal modulation of illumination should also be invoked when discussing seasonality of PP. I expect illumination to cause the spring onset in PP production which stops when nutrients are consumed.

Author’s reply: We agree that accounting for light variability is crucial. We will include an analysis of light variability, possibly as detailed as the other variables

From line 366: and following the PP model, the use of SST to show temperature control on phytoplankton metabolism could be strengthened with the use of modelled temperature vertical profiles.

Author’s reply: That is indeed correct: the PP model uses the vertical temperature profiles modelled and distributed by CMEMS, as described in lines 179-184. Since this is evidently unclear, we will revise the text to clarify this point.

line 371: Is this the intention with the consideration of the FOD to show that PP calculation should be integrated over the euphotic depth?

Author’s reply: As discussed in General Comment 1, PPfod was selected as the reference variable only because it is strictly linked to surface fields. In contrast, PPeu, is strongly influenced by the vertical distribution of biogeochemical variables (e.g., the deep chlorophyll maximum), which can mitigate the relationship between PPeu and surface properties. As shown by the sensitivity analysis, the model response is primarily driven by chlorophyll concentration, and this dependence is more clearly expressed when analysing PPfod. Nevertheless, PPeu remains the variable of interest to us.

Looking at figure 5, the lag between minimum PP and minimum SST seems to be longer than maximum PP and maximum SST. What are the differences between the correlation of PPeu and SST at line 371 and at lines 390 and 391 ?

Author’s reply: We realized that the way this part of the analysis was presented can be largely improved.

Line 375 defines the First Optical Depth, and is redundant since it is already defined in the method.

Author’s reply: We will remove the definition here, as the one in the methods section is sufficient

From line 380: the authors discuss PPfo referencing figure 6, I find it particular to have this figure at the end of the manuscript while being described here. I suggest finding a way to add the panels and EOF modes in respectively Figure 4 and Figure 5.

Author’s reply: Yes, we understand the reviewer concern. We will reformat them to incorporate PPfo with the other variables.

Line 385: I suggest explaining what being more resilient means and slower variability. Averaging PP for the FOD is known to offer an incomplete view of column integrated PP as rightly stated by the authors beforehand.

Author’s reply: We will better explain that PPfo is less resilient than PPeu because, it is relatively more influenced by the rapid atmospheric dynamics and less influenced by the slow lower layer dynamics. In other words, the effect of rapidly changing atmospheric dynamics is relatively more important for PPfo than for PPeu as the latter benefits of the mitigating effect of the lower layer dynamics.

line 387: I suggest rephrasing to Surface phytoplankton biomass present inverse correlation with SST (phase inversion or 6 month lag) and strong correlation with MLD.

Author’s reply: We will rephrase the sentence to better clarify the concept

Line 390: surface biomass accumulation or biomass accumulation ?

Author’s reply: Biomass accumulation in the euphotic layer. We will modify the sentence.

Line 391: a reference would be expected for the top down control. Furthermore, are the authors now discussing surface biomass or integrated biomass?

Author’s reply: Here, we are discussing PPeu, so we are referring to the biomass of the euphotic layer. We will rephrase this for clarity. We will insert these references for top-down control:

• Calbet, A. and Landry, M. R. (2004). Phytoplankton growth, microzooplankton grazing, and carbon cycling in marine systems. Limnology and Oceanography, 49(1), 51–57. https://doi.org/10.4319/lo.2004.49.1.0051
• Rodríguez-Gálvez, S., Macías, D., Prieto, L., & Ruiz, J. (2023). Top-down and bottom-up control of phytoplankton in a mid-latitude continental shelf ecosystem. Progress in Oceanography, 217, 103083. https://doi.org/10.1016/j.pocean.2023.103083

Line 393: a citation would be expected for how size differential cell sinking contributes to phytoplankton dynamics.

Author’s reply: We will insert these references:

• Durante, G., Basset, A., Stanca, E., & Roselli, L. (2019). Allometric scaling and morphological variation in sinking rate of phytoplankton. Journal of Phycology, 55(6), 1386–1393. https://doi.org/10.1111/jpy.
• Liu, X., et al. (2023). Sinking rates of phytoplankton in response to cell size and carbon biomass: A case study in the northeastern South China Sea. Journal of Marine Systems, 240, 103885. https://doi.org/10.1016/j.jmarsys.2023.103885

from line 393: the summary of findings on the control mechanisms of primary production miss to mention the seasonal DCM and illumination. line 393: If sea water temperature is a proxy for nutrient entrainment, is it sea water temperature that dictates production or winter replenishment of nutrients stocks? As such, MLD isn't a better indicator of nutrient stocks? I think the causal link relating to the statistical correlation could be here better highlighted

Author’s reply: Given the large amount of changes and the inclusion of the light field in the analysis, this entire section will likely be reshaped in a way that the revised version will account for the role of DCM and light conditions.

Line 418 and Figure 5c: The second mode of surface chlorophyll-a seems to correlate well with the first mode of MLD (both temporally and spatially) and less with its second mode. In general I fail to see why the second mode of a variable should be necessarily linked to the second mode of another variable. In Volpe et al., 2012, the seasonal EOF modes are independently derived from one another. (see header of Volpe et al., 2012 4. Results and discussion)

Author’s reply: Reviewer is totally right, there is in principle no good reason to expect the first mode of one variable to necessarily correlate with the first of another and so on. However, one can (qualitatively or intuitively) expect the first modes of surface ocean variables to somehow describe a similar scale of the system variability. This is actually reflected in the section titles and constitute the rationale as to why we were exploring variable correlations this way. Again, given the large amount of changes involved we will likely rephrase this part of the analysis to account for the reviewer comment.

In figure 5b, the second mode of SST explains 1.2% of the variance, is this mode significant ?

Author’s reply: We agree that 1.2% of explained variance casts doubts on the relevance of this mode; still this does not have anything to do with the significance of the mode itself.

Line 432: the author compares the 2nd modes of Peu and Pfod, finding that these two modes are of different shape: PPeu mode stretches longer over summer. As interpretation, the author seems to suggest a higher production in the first optical depth due to warmer surface water compared to the production of the full euphotic depth (i.e. PPfod > PPeu). However, with the column integrated equation of line 161 and assuming the euphotic depth to be deeper than the first optical depth then Peu necessarily sums Pfod in its calculation. Furthermore, the EOF modes are normalized, hence amplitude differences on the origin variables can not be inferred from amplitude difference in these modes. I suggest clarifying this discussion.

Author’s reply: Here we meant that PPfo representing only the surface layer is more susceptible of benefiting of the warmer temperature (heat comes from above) and of the water stratification. We hypothesised this to be the cause of the longer lasting highs.

Line 456 and 460 and the last paragraph of the summary and conclusion section are redundant.

Author’s reply: We will modify the text to avoid repetitions.

Line 465: In similarity between PPfo and Chl fields reinforces the dominant role of phytoplankton biomass as the main driver of production. I am unsure about what the authors are highlighting here. Recalling that biomass is here proxied by surface chlorophyll-a concentration, I think that what is highlighted is the need to account for the full euphotic depth when doing primary production calculation.

Author’s reply: What we meant to highlight, here, was the role of surface satellite-derived Chl in shaping PP which has been demonstrated to be larger when considering PPfo with respect to PPeu. We want to stress that what mentioned here is absolutely consistent and in line with the analysis presented in previous sections. There is no doubt that integrating over the entire euphotic depth provides a clearer picture of phytoplankton production.

line 466-467: The higher biomass and production hotspot of the bloom regions have been quantitatively estimated as climatological fields. I would have liked to see how these biomass hotspots evolve with climate change, are they resilient as suggested hereafter ?

Author’s reply: As mentioned in General comment 1, long-term trends are not included here because constitute the topic of another work in preparation.

Line 468: The trend is only observed on surface chlorophyll which does not permit to infer much about oligotrophication. For that purpose the full euphotic zone biomass should be considered.

Author’s reply: We will clarify that the observed oligotrophication, driven by a surface chlorophyll decrease and SST increase, is associated with the surface layers

From line 472: the authors summarize the findings on the first EOF mode, which is linked to seasonality. However, at line 475 deep convection is mentioned. Deep convection is, from Line 487, linked to the second EOF mode. As a result I have trouble making a clear distinction between the two modes. They both have an annual period and thus are seasonal modes but relate to different mechanisms (stratification and deep convection, respectively). In general, I fail to see how each mode is connected to each other.

Author’s reply: We understand the confusion and will rewrite the paragraph trying to be more consistent with the “labels” given in previous sections to the various modes.

From line 484: The coherence between phytoplankton size classes and total biomass further indicates that the seasonal cycle is expressed consistently across community structure, with microphytoplankton showing the largest amplitude of variability and picophytoplankton the most stable. Again, isn’t the biomass considered only at the surface? I would expect this finding - high correlation between PSC and CHL - to be discussed in relation to the PSC derivation method.

Author’s reply: We will surely better explain in section 2.1.2 that phytoplankton biomass (Chl) and size classes (PSC) are surface quantities as they are derived from satellite observations while phytoplankton production is an integrated quantity in the case of PPeu and a surface quantity in the case of PPfo. Furthermore, we deem unnecessary to mention what already clearly stated at line 289: “As expected by their functional forms and by their being derived as a direct function of Chl (Di Cicco et al., 2017; Di Cicco et al., 2025), main field of PSC (Figure A 2) spatially show patterns very close to Chl”.

Line 504, I think that what has been quantified is that the considered PP model has chlorophyll as the main driver. This point could be discussed in regard to the used PP model.

Author’s reply: At this stage of the paper, readers should be aware that the primary production (PP) analyzed here is model-derived. Consequently, it should be clear that the PP estimates are a direct result of the model's parameterization. However, we will edit the sentence to prevent misunderstanding.

Line 508, I think that the resilience of Mediterranean sea primary production was not shown by the results.

Author’s reply: We will remove the sentence.

Technical corrections

Line 126 The reference to the product could be recalled here (product ref no. 1).

Author’s reply: it will be done

Line 130: I suggest rephrasing the sentence as follow: To assess the role of the ocean dynamics onto the biological compartment as described by Chl, PSC and PP, we use the remotely sensed Sea Surface Temperature (SST) and modeled Mixed Layer Depth (MLD) from the Mediterranean component of the Copernicus Monitoring Forecasting Centre (MFC).

Author’s reply: Sentence will be rephrased.

line 135: please rephrase Data used in this work are mostly directly

Author’s reply: Sentence will be rephrased.

Line 148: i suggest rephrasing to All products derive from Mediterranean-specific processing chains and are composed of

Author’s reply: Sentence will be rephrased.

line 211: to evaluate the role of water temperature instead of to evaluate water temperature role

Author’s reply: Sentence will be rephrased.

line 212: product ref. no. 2 instead of SST dataset. Same comment for the MLD dataset.

Author’s reply: Sentence will be rephrased.

line 227: For all instead of From all

Author’s reply: Sentence will be rephrased.

line 247: I suggest to use average field and anomalies instead of the rule and exception.

Author’s reply: Sentence will be rephrased.

Line 247: induced by rather than induced to.

Author’s reply: Sentence will be rephrased.

Line 287: I suggest rephrasing when it comes to look into the various variables space time variability to when investigating space time variability of the variables through section ....

Author’s reply: Sentence will be rephrased.

Line 284: section 2.2.1 instead of section Data.

Author’s reply: Sentence will be rephrased.

Line 290: a word seems missing being their and derived

Author’s reply: Sentence will be rephrased.

Line 295: in correspondence with instead of in correspondence of line 295: I suggest to rephrase as only over 1 % of the basin area

Author’s reply: Sentence will be rephrased.

Line 318: apart from what is shown in the SST filed

Author’s reply: Sentence will be rephrased.

Line 331: spanning over

Author’s reply: Sentence will be rephrased.

Line 425: Similarly to phytoplankton biomass instead of As of the phytoplankton biomass

Author’s reply: Sentence will be rephrased.