Comparing Pre- and Post-Fire Strategies to Mitigate Wildfire-Induced Soil Erosion in Two Mediterranean Watersheds

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

In this study, the authors coupled fire simulations with soil erosion modelling to estimate annual wildfire-induced soil loss for two watersheds in Portugal. they identified optimal treatment locations with the aim of maximizing the reduction of soil loss, and estimated treatment effectiveness using treatment leverage and cost-effectiveness. Both mitigation strategies were predicted to reduce post-fire soil loss, with effects increasing with treatment extent. Treatments had a strong mitigation effect particularly in extreme fire years. The most effective soil loss mitigation strategy varied among the two watersheds.

Here some suggestions.

1. The premise of this study is the reliability of fire intensity simulation. The authors applied the fire behavior simulation and minimum travel time (MTT) spread algorithm，But the authors did not analyze these output results in the results.
2. After a fire occurs, it will burn different types of landscapes, such as grasslands and forests, causing soil erosion. Which time period did the author analyze based on, the year of the fire, the second or third year, or until vegetation restoration? Here needs to be explained clearly.
3. I think the author assumed fire occurred instead of a real fire, and the results are highly uncertain. However, the author's given results are all deterministic, which is not in line with the situation. I suggest increasing the number of simulations and providing the range of soil loss values.

Author Response

Comments 1: The premise of this study is the reliability of fire intensity simulation. The authors applied the fire behavior simulation and minimum travel time (MTT) spread algorithm，But the authors did not analyze these output results in the results.

 

Response 1: We applied the fire algorithm to simulate the spread and behavior of thousands of hypothetical wildfires in the two study watersheds. We agree with the reviewer that the results depend on accurate fireline intensity prediction, but also on accurate spatial and temporal fire patterns (e.g. burn probability). The fire spread simulation system was thoroughly calibrated (section 2.3) and has proven to provide reliable results in other studies (e.g. Carlson et al. 2025). 

 

We agree with the reviewer that we did not analyze the direct fire spread simulation outputs because this analysis was beyond the scope of the work. Our objectives were defined in Lines 89-92 and focused on soil loss mitigation rather than, for example, burned area mitigation. Analysing the fire outputs in addition to the soil erosion outputs would likely divert the reader from the main messages of the work, misaligning it with its objectives.

 

In addition, we have added information regarding the study of Carlson et al. (2025) that evaluated MTT fire spread simulation in L159-160.

 

 

 

Comment 2: After a fire occurs, it will burn different types of landscapes, such as grasslands and forests, causing soil erosion. Which time period did the author analyze based on, the year of the fire, the second or third year, or until vegetation restoration? Here needs to be explained clearly.

 

Response 2: We agree with the reviewer that this information is important and must be clearly indicated in the Methods section. Our soil erosion estimates refer to the first post-fire year and are supported by several studies that show a major decrease in soil loss from the first to the subsequent post-fire years. This information is contained in Lines 209-213 of the manuscript that read “RUSLE was used in this study to estimate average annual soil erosion and not time series of annual erosion, thus the effect of wildfires on the C-factor and, hence, on soil loss was assumed to be limited to the first post-fire year. This assumption seemed reasonable, since various field studies in central Portugal have reported a major decrease in soil loss from the first to the second post-fire year [27,62,63].”. 

 

Acknowledging that this is an important issue, we dedicated the first paragraph of the section 4.2 (Discussion - Limitations) to it, highlighting the limitations and the need for further improvements in the future (L567-574): “We applied RUSLE that assumes long-term average conditions, which may not apply to short-term extreme post-fire erosion dynamics [6]. Although there are more sophisticated models that could be applied [76–78], RUSLE has proven to be reasonably accurate and suitable for risk analysis in Mediterranean areas [58,79]. We acknowledge that improvements need to be made to better model burn severity. We used a very simple relation between flame length and burn severity with known limitations [13], that could for example underestimate the effect of extremely intense wildfires on soil burn severity, and thereby, soil loss.”

 

 

 

Comment 3: I think the author assumed fire occurred instead of a real fire, and the results are highly uncertain. However, the author's given results are all deterministic, which is not in line with the situation. I suggest increasing the number of simulations and providing the range of soil loss values.

 

Response 3: We fully agree with the reviewer that there is uncertainty associated with our results. We simulated thousands of hypothetical wildfires and estimated associated soil erosion. We then estimated annual soil erosion maps for 500 fire seasons, for each study area. This is described in detail throughout the Methods section. As a result, each pixel has a distribution of soil erosion values. Figure 2 shows the average annual soil loss, where the pixel-by-pixel distribution is summarized to an average value. Figures 4 and 7 show a stochastic analysis that represents the probability of exceedance of a given average annual soil loss. For the remaining analysis Figures (2, 3, 5 and 6), we opted to simplify the results to make the message clearer. Nevertheless, in the Results section we provided information about the distribution of our results. For example, lines 343-346 read “Under the 2023 fuel scenario, for example, the 95% percentile of soil loss corresponded to 5.88 t ha-1 yr-1 for Castelo de Bode and to 1.95 t ha-1 yr-1 for Odelouca, values seven-fold larger than average.”.

 

To better show the annual variability of treatment effectiveness we added a new figure, two new paragraphs (Results + Discussion sections) and additional line to the Conclusions. Here, we break the average-level analysis of treatment effectiveness to an annual scale and show the magnitude of year-to-year variability, and how variability and magnitude of effectiveness vary across annual soil loss. We thank the reviewer for the comments that allowed us to explore an interesting (unexplored) part of the Results.

 

References:

Carlson, A. R., Hawbaker, T. J., Bair, L. S., Hoffman, C. M., Meldrum, J. R., Baggett, L. S., & Steblein, P. F. (2025). Evaluating a simulation-based wildfire burn probability map for the conterminous US. International Journal of Wildland Fire, 34(1).

González-Pelayo, O.; Prats, S.A.; AMD, V.; DCS, V.; Maia, P.; Keizer, J.J. Impacts of Barley (Hordeum Vulgare L.) Straw Mulch on Post-Fire Soil Erosion and Ground Vegetation Recovery in a Strawberry Tree (Arbutus Unedo L.) Stand. Ecol. Eng. 2023, 195, 107074, doi:10.1016/j.ecoleng.2023.107074.

Keizer, J.J.; Silva, F.C.; Vieira, D.C.S.; González-Pelayo, O.; Campos, I.; Vieira, A.M.D.; Valente, S.; Prats, S.A. The Effective-ness of Two Contrasting Mulch Application Rates to Reduce Post-Fire Erosion in a Portuguese Eucalypt Plantation. Catena 2018, 169, 21–30, doi:10.1016/j.catena.2018.05.029.

Prats, S.A.; Wagenbrenner, J.W.; Martins, M.A.S.; Malvar, M.C.; Keizer, J.J. Mid-Term and Scaling Effects of Forest Residue Mulching on Post-Fire Runoff and Soil Erosion. Sci. Total Environ. 2016, 573, 1242–1254, doi:10.1016/j.scitotenv.2016.04.064.

Vieira, D.C.S.; Borrelli, P.; Jahanianfard, D.; Benali, A.; Scarpa, S.; Panagos, P. Wildfires in Europe: Burned Soils Require Attention. Environ. Res. 2023, 217, 114936, doi:10.1016/j.envres.2022.114936.

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript presents a numerical simulation of exploring pre- and post-fire strategies to mitigate wildfire-induced soil erosion in two Mediterranean watersheds. Overall it is well organized and the methodology sound. There are a few minor technical issues that the author may want to improve or reconsider:

(1) In the soil loss RUSEL modeling used in the present study, are Factors R, K, LS affected by wildfires, especially Factor K, erodibility factor? The manuscript seems to indicate only C-factor (and perhaps P?) is altered by wildfires. Obviously there have been many studies discussing the effect of wildfires on the soil erodibility factor. If it is considered in the present study, how is the wildfire effects modeled in the simulations? If not possible to offer every detail, at least briefly with some summarized information to help the readers. Otherwise in the opinion of this reviewer, at least some elaboration or discussion on the rationale or implication of the absence of such considerations should be warranted.

(2) Line 425, “Results for both watersheds were similar for the 2030 fuel scenario (Figure A4b).”It seems to the reviewer that this statement is not entirely convincing, perhaps slightly debatable, especially for low treatment range (<10%).

(3) Fig. 6 and Line 429~441, what is the “cost-effectiveness ratio”, is it the same as the “cost-effectiveness” defined earlier (Fig. 5)? If so, why use a different term, it should be avoided or explained. Otherwise a new parameter should be defined clearly. This reviewer made an assumption that they are the same parameters in the context.

(4) Line 460, “the synergist effect” or “synergistic”?

(5) Given the discussion presented in Line 466~478, it might be helpful to present the y-axis value in %.

(6) Since the authors decided to break the appendix into A (associated with Materials and Methods) and B (associated with Results), which is helpful though probably unnecessary, then it might be useful to number Figs. A2~A5 as Figs. B1~B4.

(7) There are some occasional grammar, wording, formatting issues that could be corrected or improved, such as:

Line 64, “had a significant mean effect size”, not sure about the meaning of “mean effect size”;

Line 74, “the stochastic nature of BOTH high-severity wildfire and post fire extreme rainfall events”;

Line 83, “risk-based”, not sure the framework/approach can be considered “risk-based”, at least not without further elaborations;

Line 85, “integrated”, “integrate”;

Line 303, “which each scenario comprising…”, revise;

Line 329, incomplete sentence or incorrect punctuation;

The format in km2, yr-1, ha-1… in the main text need to be corrected.

Author Response

Comments 1: In the soil loss RUSEL modeling used in the present study, are Factors R, K, LS affected by wildfires, especially actor K, erodibility factor? The manuscript seems to indicate only the C-factor (and perhaps P?) is altered by wildfires. Obviously there have been many studies discussing the effect of wildfires on the soil erodibility factor. If it is considered in the present study, how is the wildfire effects modeled in the simulations? If not possible to offer every detail, at least briefly with some summarized information to help the readers. Otherwise in the opinion of this reviewer, at least some elaboration or discussion on the rationale or implication of the absence of such considerations should be warranted.

Response 1: In L203-204 we state that only the C-factor was changed to mimic the effect of wildfires:  “the effect of wildfires on soil loss was estimated by modifying the values of the C-factor values depending on three burn severity levels“. We have added the following sentence to further clarify the readers “The C-factor was modified assuming that under higher burn severity, a large fraction of the surface cover was removed.”

 

Regarding the “how is the wildfire effects modeled in the simulations?”. The Methods section clearly states that the C-Factor was changed according to the burn severity, which in turn was defined based on the fireline intensity (L203-207). 

 

Regarding the modification of the C-factor, we followed the approach by Vieira et al. (2018) for being the only study having evaluated the performance of RUSLE for recently burnt areas in Portugal against field erosion data. At the same time, we did not adjust the P-factor following Vieira et al. (2018), first and foremost because we lacked the necessary ground cover measurement data. At the same time, however, we believe that fire impacts on protective soil cover should be represented through the C factor, while the P-factor should be restricted to represent the impacts of soil conservation measures and, in particular, those that target (micro-)topography. We could have used the P factor and not the C-factor to represent erosion mitigation but decided to apply directly a treatment effectiveness percentage (table 2) to the estimated soil erosion for very large wildfires (>500ha).

 

In relation to the impact of wildfire on the K factor, field erosion studies in Portugal (e.g., Prats et al., 2016,Keizer et al., 2018; Gonzalez-Pelayo et al., 2023) have shown that soil burn severity is typically moderate, i.e., does not involve visual changes in key topsoil properties such as structure and colour. High soil burn severity has been observed occasionally but only very locally (apparently associated to pre-fire litter accumulation, for example of logging residues) so that it is not expected to play a key role in plot and field scale erosion rates as modelled by RUSLE. Furthermore, Vieira et al. (2023) made an extensive literature search and found several incoherences regarding the K-factor which reduced the original database from 39 to only 6 observations (see Supplement 3 in Vieira et al. 2023).

In conclusion, literature shows that we can accurately mimic the effect of wildfires by modifying only the C-factor of RUSLE. For consistency, we have also added the reference from Vieira et al 2023 to the Methods section where the RUSLE modification is described (section 2.4).

 

 

Comments 2: Line 425, “Results for both watersheds were similar for the 2030 fuel scenario (Figure A4b).”It seems to the reviewer that this statement is not entirely convincing, perhaps slightly debatable, especially for low treatment range (<10%).

Response 2: We agree with the reviewer and have corrected the sentence that now reads “Results for both watersheds were similar for the 2030 fuel scenario (Figure B3b), except for the lower treatment extents (5 and 10%).”

Comments 3:  Fig. 6 and Line 429~441, what is the “cost-effectiveness ratio”, is it the same as the “cost-effectiveness” defined earlier (Fig. 5)? If so, why use a different term, it should be avoided or explained. Otherwise a new parameter should be defined clearly. This reviewer made an assumption that they are the same parameters in the context.

Response 3: We thank the reviewer for noticing this inconsistency. In fact, both terms are equivalent. We have revised Figure 6 and changed the y-axis that now reads “Cost-effectiveness ratio”.

Comments 4:  Line 460, “the synergist effect” or “synergistic”?

Response 4: We have changed to “A synergistic effect”

Comments 5:  Given the discussion presented in Line 466~478, it might be helpful to present the y-axis value in %.

Response 5: We thank the reviewer for noticing this inconsistency. To make it consistent with Figure 4 and the associated text, we prefer to maintain the y-axis of Figure 7 and change the text in the Lines 477-489 (in revised manuscript). 

 

Comments 6: Since the authors decided to break the appendix into A (associated with Materials and Methods) and B (associated with Results), which is helpful though probably unnecessary, then it might be useful to number Figs. A2~A5 as Figs. B1~B4.

Response 6: We thank the reviewer for the suggestion. We have changed the figures to B1~B4.

Comments 7:  There are some occasional grammar, wording, formatting issues that could be corrected or improved, such as: 

• Line 64, “had a significant mean effect size”, not sure about the meaning of “mean effect size”;
• Line 74, “the stochastic nature of BOTH high-severity wildfire and post fire extreme rainfall events”;
• Line 83, “risk-based”, not sure the framework/approach can be considered “risk-based”, at least not without further elaborations;
• Line 85, “integrated”, “integrate”;
• Line 303, “which each scenario comprising…”, revise;
• Line 329, incomplete sentence or incorrect punctuation;

Response 7: Following the same order as above:
- L64: Changed to “had a significant mean soil loss reduction effect”.

• L74: changed accordingly
• L83: It is “risk-based” because we analyze our results combining the (1) Probability that a particular negative event will occur a severe wildfire) and (2) Potential consequences (i.e. Impact) if the event does occur (a certain level of soil loss). The best example of the risk-based approach are the exceedance probability curves shown in Figures 4 and 7.
• L85: changed accordingly
• L303: revised accordingly
• L329: Sentence deleted.

Comments 8: The format in km2, yr-1, ha-1… in the main text need to be corrected.

Response 8: All units format have been revised.

 

References:

Carlson, A. R., Hawbaker, T. J., Bair, L. S., Hoffman, C. M., Meldrum, J. R., Baggett, L. S., & Steblein, P. F. (2025). Evaluating a simulation-based wildfire burn probability map for the conterminous US. International Journal of Wildland Fire, 34(1).

González-Pelayo, O.; Prats, S.A.; AMD, V.; DCS, V.; Maia, P.; Keizer, J.J. Impacts of Barley (Hordeum Vulgare L.) Straw Mulch on Post-Fire Soil Erosion and Ground Vegetation Recovery in a Strawberry Tree (Arbutus Unedo L.) Stand. Ecol. Eng. 2023, 195, 107074, doi:10.1016/j.ecoleng.2023.107074.

Keizer, J.J.; Silva, F.C.; Vieira, D.C.S.; González-Pelayo, O.; Campos, I.; Vieira, A.M.D.; Valente, S.; Prats, S.A. The Effective-ness of Two Contrasting Mulch Application Rates to Reduce Post-Fire Erosion in a Portuguese Eucalypt Plantation. Catena 2018, 169, 21–30, doi:10.1016/j.catena.2018.05.029.

Prats, S.A.; Wagenbrenner, J.W.; Martins, M.A.S.; Malvar, M.C.; Keizer, J.J. Mid-Term and Scaling Effects of Forest Residue Mulching on Post-Fire Runoff and Soil Erosion. Sci. Total Environ. 2016, 573, 1242–1254, doi:10.1016/j.scitotenv.2016.04.064.

Vieira, D.C.S.; Borrelli, P.; Jahanianfard, D.; Benali, A.; Scarpa, S.; Panagos, P. Wildfires in Europe: Burned Soils Require Attention. Environ. Res. 2023, 217, 114936, doi:10.1016/j.envres.2022.114936.

Reviewer 3 Report

Comments and Suggestions for Authors

The paper presents two strategies for protecting soils from erosion. Two different areas in Portugal were selected for the study. The costs of erosion protection on slopes with varying inclinations were calculated. Additionally, the costs associated with the fire reduction strategy were analyzed in comparison to the expenses incurred for post-fire remediation. The study concludes that it is essential to consider complex actions, which often involve a combination of both strategies.

Comments and Suggestions:

• Please standardize the formatting of the following literature references to ensure consistency with the rest of the list: lines 669, 712, 740, 786, 827, and 839.

• Line 603: The sentence ends with an excessive number of dots. Please correct the punctuation.

• In the "Object and Methods" section, please specify the types of soils involved in the experiments. Additionally, indicate their classification according to the World Reference Base for Soil Resources (WRB 2022).

Author Response

Comments 1: Please standardize the formatting of the following literature references to ensure consistency with the rest of the list: lines 669, 712, 740, 786, 827, and 839.

Response 1: We used the MDPI reference style. We believe that all pending issues regarding the references will be solved during the proof-reading stage. 

Comments 2: Line 603: The sentence ends with an excessive number of dots. Please correct the punctuation.

Response 2: Corrected.

Comments 3: In the "Object and Methods" section, please specify the types of soils involved in the experiments. Additionally, indicate their classification according to the World Reference Base for Soil Resources (WRB 2022).

Response 3: According the Harmonized World Soil Database (v2.0), the predominant soils in our study áreas are Regosols and Podzols in CAstelo de Bode and Regosols in Odelouca. We have added this information to the Study Area description in the Materials and Methods sub-section 2.2

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors quantified the impacts and effectiveness of pre-versus post-fire treatment strategies on soil loss mitigation. They coupled fire simulations with soil erosion modelling to estimate annual wildfire-induced soil loss for two watersheds in Portugal. They identified optimal treatment locations with the aim of maximizing the reduction of soil loss, and estimated treatment effectiveness using treatment leverage and cost-effectiveness. Both mitigation strategies were predicted to reduce post-fire soil loss, with effects increasing with treatment extent.

The authors have responded to all comments accordingly, and I have no other suggestions.

It is recommended that the author make further detailed revisions to the article format to meet the publication requirements.

For example, there is no Eq.1. 

Author Response

Comments1: The authors have responded to all comments accordingly, and I have no other suggestions. It is recommended that the author make further detailed revisions to the article format to meet the publication  requirements. For example, there is no Eq.1. 

 

Response 1: We have fixed the issue with Eq.1 and made detailed revisions throughout the entire manuscript.