Should Conservation Cut-In Wind Speed Be Tailored to Site-Specific Conditions? Insights from Bat Activity Patterns at Wind Farms in Northern Portugal

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This article addresses a very timely and important topic – the effectiveness of standardized curtailment thresholds (5 m/s) in the context of local bat activity at rotor height.

The greatest value of the work:

• monitoring at 55 m AGL (the actual risk zone),
• translation of the ground-based threshold to the turbine's operating height,
• comparison of three farms,
• application of a GLMM with random effects (month, location).

This is a good research idea.

However, as a reviewer, I would point out several important issues requiring further clarification.

The STRENGTHS of the article include:

1. Relevance of the research problem

The debate on constant vs. site-specific cut-in speed is current (see Adams 2021; Barré 2022). The authors contribute well to this discussion.

2. Monitoring at an altitude of 55 m

This is a key strength of the paper.

The authors rightly emphasize that 90% of ground-based activity does not translate into the collision zone.

3. Differences between farms

Clear differences are visible:

• Gevancas → 90% activity up to 8.87 m/s
• Azinheira → 90% to 6.02 m/s
• Dom João → 90% to 6.19 m/s

This indeed challenges the uniform threshold of 5 m/s. 4. GLMM

The model shows a strong negative effect of wind:

β = −0.48 ± 0.01 SE, p < 0.001

The random effects of month are more pronounced than those of location (σ² = 1.66 vs. 0.23), which makes biological sense.

 

KEY NOTES

1. Logical inconsistency between results and interpretation

The authors show that:

activity decreases with increasing wind speed (GLMM)

but at the same time argue that:

90% of activity occurs above 5 m/s

This is not a statistical contradiction, but requires a better explanation.

Currently, it looks like this:

• model → activity decreases with wind speed
• cumulative plot → a large portion of activity at higher speeds

This results from the wind frequency distribution, but the authors do not explain it.

It is necessary to: explain the difference between:

• relative intensity,
• and cumulative time under given meteorological conditions.

2. Using 90% as a Threshold

The authors assume:

• 80% and 90% activity as potential curtailment thresholds.

Why 90%?

There is no biological or energetic justification.

It is arbitrary.

In management practice:

• cost-effectiveness optimization is used,
• protection is not maximized at the expense of production.

The following should be added: a trade-off analysis (production loss vs. activity reduction) should be added, or at least discussed.

3. Lack of mortality data

The authors:

• analyze acoustic activity,
• but do not have collision data.

Activity ≠ mortality.

The literature (e.g., Arnett 2010) has shown that:

• activity and mortality do not always correlate linearly.

This limitation must be clearly stated.

4. Wind speed extrapolation (Hellmann power law)

The authors assume α = 0.18 (p. 2).

However:

• they do not show the model's sensitivity to changes in α,
• they do not justify whether the terrain actually corresponds to this parameter throughout the season.

This may generate a bias.

At least one sentence should be included in the constraints.

5. Timescale

Monitoring:

• 7 nights/month,
• March–October,
• 56 nights in total.

No:

• seasonal differentiation,
• analysis of the migration period (August–September).

This data is key to mortality.

NOTES FOR THE DISCUSSION

The discussion is valid, but:

1. Too general.
2. It lacks a stronger reference to:

o algorithmic curtailment (Barré 2023),

o smart curtailment (multicriteria approach),

o real-time detection.

3. It lacks reference to species differences.

Does the activity concern:

• Pipistrellus?
• Nyctalus?
• migratory species?

This significantly changes the interpretation.

 

The article requires further refinement in the following areas:

• Clarification of the results' logic
• Stronger methodological constraints
• Lack of reference to mortality
• Lack of analysis of the energy trade-off

 

The manuscript addresses an important and timely conservation issue and provides valuable empirical data from turbine-relevant heights. However, several methodological clarifications and conceptual refinements are needed before publication, particularly concerning the interpretation of cumulative thresholds, the absence of fatality data, and the ecological justification of proposed cut-in adjustments. Major revision recommended.

Comments on the Quality of English Language

I am not able to 100% confirm the correctness of the English language.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This brief report addresses a critical practice-oriented question in wind energy biodiversity conservation: whether standardized conservation-focused cut-in wind speeds can effectively mitigate bat collision risks across different sites. The study has several notable strengths. First, it conducted long-term systematic acoustic monitoring of bat activity at turbine rotor-relevant height (55 m) across three wind farms in northern Portugal, filling a key gap in current mitigation guidelines that mostly rely on near-ground monitoring data. Second, it empirically revealed significant site-specific variation in wind speed thresholds corresponding to peak bat activity, providing solid evidence for the limitation of universal 5 m/s curtailment rules. Finally, its findings directly support adaptive, site-specific wind farm management, with clear practical value for bat conservation amid global renewable energy expansion. However, there are still some issues that need to be further resolved before the manuscript can be published.

 

Major comments:

1. It is recommended that the authors further supplement the core baseline information for the three studied wind farms, including the number of wind turbines at each site, as well as landscape characteristics in the vicinity of the wind farms (e.g., forest coverage, topographic features, and key bat habitats in the surrounding areas). These contextual data could be critical to support the interpretation of the significant inter-site differences in bat activity across the three wind farms presented in the results section.
2. It is recommend that the authors optimize the font display of the figures in the manuscript, with the use of Arial font strongly advised. Such as, the decimal points of the numerical values in the figures appear as commas rather than standard decimal points, which needs to be corrected. 
3. When estimating the wind speed at 55 m above ground level, the authors uniformly applied a roughness coefficient α = 0.18, which corresponds to moderately rough terrain. However, the authors did not specify the cited source of this value, and critically, failed to justify whether a single identical coefficient is applicable to all three wind farms. Do these 3 wind farms have the same landscape heterogeneity? This oversight may lead to significant systematic bias in the wind speed estimation.
4. The authors constructed the GLMM with Poisson distribution. Bat activity data inherently exhibit ecological aggregation characteristics (for example, bat activity is far lower under inclement weather conditions than in warm, windless conditions), which may cause the dispersion of bat activity data to exceed the theoretical range of the Poisson distribution. In particular, the authors’ GLMM results show that the random effect variance of month (σ²=1.66) is much larger than that of wind farm (σ²=0.23), which demonstrates an extremely high degree of seasonal variation in bat activity.
5. The authors only quantified bat activity levels, yet did not collect data on bat collision fatalities caused by wind turbines, resulting in a lack of direct evidence linking bat activity to collision mortality risk. Accordingly, equating "activity level" with "collision mortality risk" in the Discussion section may lead to over-extrapolation of the study conclusions. It is recommended that the authors supplement additional supporting evidence for this association or revise the relevant statements to align with the actual empirical data obtained in this study.

 

Minor comments:

1. The wind speed in Figure 3 lacks units.
2. why the wind speed values corresponding to 80% and 90% of cumulative bat activity are identical in Figure 1.
3. It is recommended to use images to display the GLMM fitting results.
4. It is recommended authors should check the citation carefully. Citation 5 [Toman et al. 2008], whose content focuses on energy expenditure and greenhouse gas emissions, is placed right after the statement that wind turbines should maintain minimum distances from raptor nests and bat roosts.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript evaluates bat activity relative to wind speed at 55 m at three wind farms in northern Portugal, extrapolating wind speeds to rotor-relevant height and using acoustic monitoring and GLMMs to assess relationships and identify wind‑speed thresholds encompassing 80–90% of activity. The authors argue for site‑specific cut‑in speeds rather than a uniform threshold. It is an interesting study but need carefully revision. Detailed comments are as follows:

1) The Hellmann power-law with fixed α = 0.18 is applied but spatial/temporal variability in α is not addressed. This can introduce systematic bias in estimated wind speeds at 55 m.

No uncertainty propagation or sensitivity analysis is presented to show how choice of α or reference-height measurements (45 vs. 85 m) affect the 80/90% thresholds.
Recommendation: quantify uncertainty in extrapolated wind speeds (e.g., sensitivity to α range, bootstrapping) and report confidence intervals for threshold estimates.

2) Acoustic detectors mounted at 55 m may underrepresent activity within the full rotor sweep (which can extend well above/below 55 m depending on turbine). The manuscript acknowledges this but does not quantify how representative 55 m is of collision risk.

Discuss more explicitly limitations of single-height sampling and, if possible, compare with literature on vertical activity profiles or include caveats in conclusions.

3) Analyses aggregate bat passes across species. Collision risk varies by species flight height, maneuverability, and call detectability; aggregating may mask species at greater risk.

If species IDs exist (even coarse guilds), present species- or guild-level patterns or at least report the species composition and discuss how that composition may bias thresholds.

4) While sampling from March–October is appropriate, results should clarify how thresholds vary seasonally (migration vs. breeding) given management implications for temporal curtailment.

Present seasonal (or monthly) cumulative curves or at minimum comment on seasonal differences using model outputs (random effect for month exists—leverage it).

5) The GLMM description lacks detail on overdispersion checks, model fit, zero inflation, and choice of Poisson vs. negative binomial; given large z and significant effect, model adequacy must be shown.

Report overdispersion statistics, conditional and marginal R2, and whether a negative‑binomial or zero‑inflated model was considered. Provide diagnostic plots in supplementary material.

Comments on the Quality of English Language

Several grammatical and formatting errors should be corrected. It's better to have a native speaker revise the manuscript.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript addresses a highly relevant conservation issue by questioning the applicability of uniform cut-in wind speeds across different wind farms. The use of acoustic monitoring at rotor-swept height (55 m) is a significant strength.

However, I recommend clarifying that bat activity is used as a proxy for collision risk, and does not directly reflect mortality rates. Additionally, uncertainty associated with wind speed extrapolation using the Hellmann power law in complex terrain should be acknowledged.

The manuscript would benefit from a more explicit discussion of limitations related to single-height sampling and potential species-specific differences in flight behavior.

Overall, the study provides valuable insights supporting the need for site-specific mitigation strategies.

Comments on the Quality of English Language

I am not able to 100% confirm the correctness of the English language.

Author Response

Please see attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

I have checked the revised version. The authors have adequately responded to and resolved all my previous concerns. I have no further comments.

Author Response

Please see attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Since the author revised the manuscript carefully, I have no more comments.

Author Response

Please see attachment.

Author Response File: Author Response.pdf