Grasse et al investigated the Si processes regulated by diatoms and  silicoflagellates by pelagic mesocosms in coastal Peru upwelling areas and stable Si isotopes. Overall, this is an interesting study and provide important knowledge of Si isotope fractionation of -3.63 ‰ during silicoflagellate production, implying a potential important application when investigating sedimentary records of biogenic Si. Prior to its publication, I have the following major and minor comments.

Major comments:

Regarding the calculation of Si isotope fractionation factor for silicoflagellates, the authors are missing an in-depth assessment of uncertainties. Given the data, diatoms and silicoflagellates are growing at different rates and exhibiting different abundance over the entire period. During the first 10 days, the production is exclusively dominated by diatoms, while thereafter, silicoflagellates take over within days. This indicates the growth rate of silicoflagellates is higher than diatoms and meanwhile consume more DSi relative to per unit cell of diatoms (this seems possible to be estimated given the experimental data), which in turn raise a question of what is a proper number for the initial d30Si value of DSi for them to grow. It is fine to use Eq 6 to estimate a mixed 30ε value, but this is probably only valid when assuming both diatoms and silicoflagellates keep their growth rate constant and the difference between these two growth rates remain the same. This means, even if they can share the same d30Siinit value derived from DW, the fraction of DSi, f, could be varied at each time point for diatoms and silicoflagellates. Subsequently, Eq 7 calculates the 30ε for silicoflagellates using relative abundances, but since their Si/cell differ much as shown in Fig 5b, should this be taken into account in the f in Eq 7?

Perhaps I make this question overcomplicated, but another thought is M2 and M7 is dominated by silicoflagellates by over 97% in certain days, why we need to bother the mixture with diatoms? It could be straightforward to use those data in those days to directly calculation 30ε for silicoflagellates.

I may miss some important information in this, but since this is the fundamental part and the main conclusion of the ms, such uncertainties should be clarified before publication.

Other suggestions:

For clarity, it is better to say Si isotope fractionation factor in the text, or at least isotope fractionation factor, instead of “fractionation factor”.

Line 47 the superscript “-” should be removed.

Line 70 “dinoflagellates” should start with the captial letters.

Line 184 “at” should be removed.

Line 173-174 vs Line 207-209 Why BSi was digested at different NaOH concentration and temperature? Are there any specific reasons? And for BSi contents and its Si isotope measurement, how the authors assess the contribution from non-biogenic Si particles?

Line 315 and some relevent text in the result section. What is the application of DIP in this study? Are they just used to show DIP is not a limiting nutrient for primary producivity?

Line 600-605 I agree that we should be more careful when using sedimentary bSi to reconstruct dSi utilization. But this is also dependent on the relative abundance of each bSi species in sediment records of the studied area and the purification of these species for isotope analysis.

Grasse et al. provide the first in situ estimate of δ³⁰Si fractionation for silicoflagellates, which is a new and important contribution to understanding Si dynamics in the modern and past marine environments. The inclusion of the 2017 KOSMOS mesocosm experiment adds novelty by offering a controlled environment to study Si isotope fractionation, a major limitation in field-based studies. While the manuscript is well written and informative, several sections would benefit from clearer articulation of the scientific objectives, improved structure, and a more detailed explanation of uncertainties. In addition, some broader implications were underdeveloped given that their short summary indicates that this information would be “providing a novel tool for understanding dSi utilization in the past. The paper could strengthened by more explicitly connecting the results to: (1) global δ³⁰Si budgets, (2) paleoceanographic reconstructions (3) potential biases in interpreting  δ³⁰Si (water column and sediment). Below is a list of comments suggesting both major and minor revisions.

Major comments:

The introduction provides extensive background on the Peruvian Upwelling, Si cycling dynamics, plankton community dynamics and Si isotopic fractionation, but the primary research questions or hypotheses of the study should be more clearly presented. For example, the central scientific question is a little bit buried: “What is the Si isotope fractionation factor for silicoflagellates, and how does it influence δ³⁰Si in dynamic upwelling systems?”. This should be explicitly stated towards the end of the Introduction.

While diatom dynamics are thoroughly explained in the introduction, a discussion of (silico) flagellates is somewhat brief and lacks detail. Given that the study presents the first Si isotope fractionation factor for silicoflagellates, more ecological and physiological context (and references) would be useful. For example, lines 69-73 lack references and there appears to be an error in the paragraph. Some questions to address include: (1) How do silicoflagellate silica structures differ from diatom frustules? (2) What is known about their silica uptake pathways? The emphasis of what is unknown will help the reader understand the importance of the questions being addressed.

Also, I found that the presentation of the KOSMOS mesocosm experiment in the introduction lacked some detail. A clearer statement of the study targets (i.e. specific processes) and why these findings mater for broader oceanographic or paleoceanographic applications should be highlighted.

While nutrient supply mechanisms (vertical advection, mixing, seasonal variability) are described accurately, the manuscript somewhat underplays mesoscale variability, differences in water mass sources, and timescale interactions (physical vs. biological). All of these can strongly shape local δ³⁰Si signatures and community composition. Briefly incorporating these factors would produce a more comprehensive discussion of why field-based δ³⁰Si measurements are difficult to interpret.

The critique of Rayleigh and steady-state models is important, but the manuscript does not fully explain how these limitations influence isotopic interpretations in practice. The mesocosms are semi-closed, but episodic mixing (bottom layer intrusion, biomass movement) introduces potential non-Rayleigh effects. This should be more thoroughly addressed. Specifically: (1) Do oversimplified δ³⁰Si assumptions bias estimates of Si utilization? (2) Are there specific examples from past studies where these biases have been demonstrated and/or discussed?  

Minor comments:

For the silicoflagellate ε calculation, the assumption that silicoflagellates dominate uptake between days 13–17 is reasonable but requires more explicit demonstration (perhaps with size-fraction bSi?). Also, is there a possible role of dissolution (diatom or other organisms) in this zone? How would this influence the interpretation of the ε30Si?

Line 106 - “admixture from multiple sources”  Do you mean “a mixture..”?

Lines 84-92 - Also, see Frings et al. 2024 for newer information. .https://doi.org/10.1016/j.quascirev.2024.108966meyr

Line 540 – perhaps cite Cotard et al. 2025 (https://doi.org/10.1002/lno.70243) for the lithogenic input ? They have some interesting evidence supporting that lithogenic input could affect the dSi composition. Is there any other evidence that could support this argument? For instance, rainfall, sediment load, turbidity, wind direction, proximity to dust sources?