Dear,

Please find my comments and suggestion as a pdf file.

Yours sincerely

The manuscript presents a new set of relative pollen productivity estimates (RPPs) for a selected number of taxa across continental Spain using a REVEALS-based optimisation framework combined with modern pollen and vegetation datasets. The study addresses an important gap in Mediterranean palaeoecology, where quantitative vegetation reconstruction remains limited by the lack of regional RPP datasets. The manuscript compiles a substantial amount of data, including more than 1100 modern pollen samples, multiple vegetation datasets, validation coretops, and comparisons with previously published European RPP studies. I particularly appreciate the effort to discuss the methodological challenges associated with Mediterranean ecosystems and the attempt to validate the obtained RPPs. The study is potentially valuable for future palaeoecological reconstructions in Iberia and southern Europe.

However, several important methodological and interpretative issues should be addressed before publication. In its current form, some conclusions appear overstated, and several aspects of the optimisation and validation framework require clearer justification and more critical discussion.

One of my main concerns relates to the comparison with other European RPP studies based on Gaussian Plume Models (GPM), while the present study relies on a Lagrangian Stochastic Model (LSM). The manuscript should discuss more critically how directly comparable recalculated GPM-based RPPs and LSM-derived RPPs actually are. Even after harmonisation, substantial methodological differences remain between studies. One of the key questions is therefore how these new RPP estimates can be meaningfully compared with previously published “traditional” RPP studies. Much more detail is required in the Methods section to explain how GPM-based RPPs were transformed or harmonised for comparison with LSM-derived values. At present, the manuscript refers to Abraham and Fortova (in prep.) for key methodological details. Since this work is not yet published, it is difficult to assess the robustness of the approach presented here. Much more methodological detail should therefore be included directly in the manuscript itself.

The authors consider the existing commonly used methods as unsuitable for the study region. This clearly requires strong justification. In particular, the manuscript should better explain, why a more traditional RPP approach would not be preferable, how the proposed optimisation framework improves RPP estimation compared to standard methods for RPP calculation, and why an alternative framework is needed when traditional approaches have already been successfully applied in many regions worldwide.

Furthermore, the discussion concerning the Gaussian Plume Model (GPM) is currently not sufficiently convincing. The manuscript argues that GPM-based approaches are inappropriate while favouring the Lagrangian Stochastic Model (LSM). However, GPM remains the most widely used dispersal model in pollen modelling studies globally. At present, the manuscript dismisses GPM somewhat too easily, without sufficiently demonstrating why it should be rejected in this context. I strongly encourage the authors to calculate and compare two sets of RPPs: one based on LSM, and one based on GPM. Such a comparison would considerably strengthen the manuscript and would allow the authors to test their assumptions empirically rather than relying primarily on theoretical arguments. Currently, the manuscript presents several statements regarding the limitations of GPM without quantitative demonstration. A more scientifically robust approach would therefore be to compare the two frameworks directly and discuss their respective strengths and weaknesses based on empirical results rather than categorical statements.

Figure 6 is also not convincing. The relationships shown remain relatively weak, with substantial scatter and clear deviations from the 1:1 relationship for most taxa. Overall, these results appear mixed rather than strongly convincing. I would not consider this a particularly strong validation figure. The figure supports the idea that the approach may have some utility, but it does not convincingly demonstrate robust predictive performance across taxa.

This is also the case for figure 8 that suggests that many of the obtained RPP values do not match particularly well with previously published studies. This makes it difficult to assess whether the proposed framework improves RPP estimation. Based on Figures 6 and 8, the overall results are not particularly convincing.

The manuscript repeatedly emphasises validation as a major strength. However, only 8 arboreal taxa are effectively validated, despite presenting RPPs for 21 taxa. This creates confusion in several sections, including the Abstract and Discussion, where broader claims are made about “validation of the Iberian RPPs”. In reality, herbaceous and shrub taxa are not independently validated, and validation is restricted to selected arboreal taxa. The manuscript should therefore clearly distinguish between validated taxa and taxa without independent validation.

Another important concern relates to the optimisation framework itself. The optimisation algorithm explicitly searches for RPP values that minimise the discrepancy between REVEALS-derived vegetation estimates and observed vegetation cover. Consequently, the later agreement between REVEALS outputs and vegetation data is at least partly expected from the optimisation procedure itself, rather than representing a fully independent validation outcome. This issue appears several times throughout the manuscript, especially where improved agreement between REVEALS estimates and vegetation cover is interpreted as evidence of methodological success. For example, the manuscript states that the optimisation procedure identifies the RPPs that best reproduce vegetation composition, but later discusses this agreement as a result of the study rather than as a direct consequence of the optimisation target.

Some conclusions also appear overstated. In particular, the discussion suggesting that Iberian landscapes may have been substantially more open than previously reconstructed should be moderated. The manuscript demonstrates that some taxa may be over- or underrepresented in pollen records due to differential productivity, but this alone does not directly demonstrate past landscape openness. Similarly, statements regarding future disturbance reconstructions and ecosystem dynamics should be presented more cautiously, especially given that shrub and herb taxa remain largely unvalidated. The paper would be stronger if the discussion remained more tightly linked to the actual findings.

The study necessarily combines datasets with different spatial resolutions and taxonomic detail. While this is understandable given the complexity of Iberian landscapes, the implications of these mismatches are not discussed critically enough. In particular, the manuscript should discuss more explicitly RSAP-related issues, local versus regional vegetation representation, and how patchy Mediterranean vegetation may affect the optimisation procedure.

The rationale for excluding 51 modern coretops from open areas and using them as a validation dataset also requires better justification. Why were only open-area sites selected for validation? Would it not be preferable to include a wider range of vegetation settings in order to evaluate whether the framework performs consistently across different landscape types?

Finally, the manuscript relies heavily on statistical summaries, but the original REVEALS outputs and untransformed pollen relationships are not shown clearly. It would be useful to provide more transparent comparisons between raw pollen data, transformed REVEALS estimates, and observed vegetation cover.