Responses of GPS-Tagged Territorial Golden Eagles Aquila chrysaetos to Wind Turbines in Scotland

Round 1

Reviewer 1 Report

A fine paper; no specific comments

Comments for author File: Comments.pdf

Author Response

We thank reviewer 1 for the overwhelmingly positive comments on our original submission. No changes were suggested by reviewer 1 and so in our revised submission we have not made any changes due to this review.

Reviewer 2 Report

This was an interesting and well-written paper focused on the important question of how golden eagles respond to windfarm and wind-turbine locations. I liked how the paper addressed the two interrelated consequences of wind-energy development to this species: habitat loss vs. collision fatality. Another key strength of the paper was the ability to evaluate the response to windfarms by territorial vs. non-territorial birds separately, which I believe is an underrecognized yet highly relevant factor in evaluating wind-energy related impacts. 

I did have several comments/questions regarding the analyses used to address the study objective and reach the conclusion that territorial birds were generally avoiding turbine locations. I elaborate on these points below, followed by a few line-by-line editorial suggestions. 

General comments: 

1.      There was a minor disconnect between the primary study objective stated in the introduction (lines 153 – 157) and the analyses, results, and conclusions. The objective statement is specific to how territorial birds respond to wind farms (so is the title of the ms). But the majority of analyses, results, and discussion were focused on comparisons between territorial and nonterritorial (dispersing) birds is their use of wind turbine locations. This comparison is interesting, but I was expecting to see a direct analysis of the use/avoidance of wind turbine locations by territorial birds. 

2.      It wasn’t clear to me why the author’s concluded that territorial birds were “substantially avoiding” wind turbines (lines 514 – 519), rather than exhibiting indifference to turbine locations (i.e., they select the underlying habitat conditions regardless of whether a turbine is there or not). The conclusion of avoidance implies that use of turbine locations is independent of underlying habitat conditions around the turbines (i.e., high quality habitat conditions are used less than expected by chance if a wind turbine is present). I did not see this clearly demonstrated or analyzed in this study. Rather, the summary statistics for habitat conditions at turbine locations vs. used locations (Table 4) indicated that GET values around turbines were lower than at used locations. 

3.      Did the home range or territory area of territorial birds overlap with the windfarms? I understand that a bird had to have a minimum of 50 records within 1km of a wind farm’s turbine to be included, but this could still happen in cases where the closest turbine is beyond the periphery of a territory or home range area. This is a critically important detail, because if a bird’s territory or home range does not overlap with a windfarm/turbine, the windfarm/turbine may not even be available for use by that bird (for unknown reasons). This highlights the importance of quantifying availability of wind turbines to the birds. It is clear the authors were aware of the availability issue when they used random points to characterize “expected use” of the windfarm. But there was no apparent tie to an individual bird’s home range or territory location. 

4.      This is just my opinion, but it seems like more powerful/simpler approach to address the study objective, given the information available, would have been to use a single analysis of habitat use vs. availability (e.g., methods in Manly et al. 2010 Resource Selection by Animals). Such an analysis would permit one to directly estimate selection coefficients for windfarms or turbine locations, which would address the question, “Do territorial birds use wind turbine locations more or less than expected by chance?”  I raise this issue because I believe many other readers would have the same questions about the lack of a use/availability analytical design to quantify use/availability of windfarms/turbines while controlling for underlying habitat conditions. 

5.      A relatively minor comment, but the terminology used to describe the territorial status of individual birds is inconsistent throughout the text, tables, and figure legends (e.g., “territorial vs. nonterritorial” birds used in abstract and portions of the Introduction as opposed to “range holding” and “dispersing” birds used in other areas of the text and in table headers). I suggest using “territorial” and “non-territorial” consistently throughout, as these are descriptors are the most accurate. 

Line-by-line comments:

Line 114: I would use Hunt et al. 2017 here instead of Hunt 2002; the latter is an update to the former and provides more detailed information about age-specific fatality rates. 

Lines 139 – 152: here you describe reasons why territorial birds may respond differently from non-territorial birds to turbines. All good reasons mentioned, but I would add that territorial birds are also engaged in territorial flight displays/undulations, pair interactions, or territory defense behaviors that may distract from their ability to detect and respond to spinning turbine blades.

 Table 2. I would delete the “current” field….this will be outdated quickly.

Author Response

This was an interesting and well-written paper focused on the important question of how golden eagles respond to windfarm and wind-turbine locations. I liked how the paper addressed the two interrelated consequences of wind-energy development to this species: habitat loss vs. collision fatality. Another key strength of the paper was the ability to evaluate the response to windfarms by territorial vs. non-territorial birds separately, which I believe is an underrecognized yet highly relevant factor in evaluating wind-energy related impacts. 

I did have several comments/questions regarding the analyses used to address the study objective and reach the conclusion that territorial birds were generally avoiding turbine locations. I elaborate on these points below, followed by a few line-by-line editorial suggestions. 

General comments: 

1. There was a minor disconnect between the primary study objective stated in the introduction (lines 153 – 157) and the analyses, results, and conclusions. The objective statement is specific to how territorial birds respond to wind farms (so is the title of the ms). But the majority of analyses, results, and discussion were focused on comparisons between territorial and nonterritorial (dispersing) birds is their use of wind turbine locations. This comparison is interesting, but I was expecting to see a direct analysis of the use/avoidance of wind turbine locations by territorial birds.

This is a fair point. However, in the Introduction (lines140 – 153) we emphasise there were expectations that territorial birds could react differently to turbines than non-territorial dispersing birds. Our previous publications on eagles and wind farms involved only non-territorial birds, hence the repeated comparisons and commentary. As the reviewer perceptively notes in his/her opening statement, this separation has rarely been considered before.

We had the opportunity to compare reactions to the same turbines by both classes, which also allowed us to address the hypothesis that territorial birds could react differently. In essence we found that the two classes did not differ significantly, and territorial birds were not prone to be less wary (as could have been expected).

Though not raised by the reviewer explicitly, we appreciate that this isn’t reflected in the title. To do so would mean a very lengthy title (our choice of title is appropriately focussed on territorial birds, we feel, given previous publications from Scotland). This issue is covered in the Abstract too. We don’t feel that considering reactions of non-territorial birds to the same turbines as territorial birds detracts from the examination of the latter class; rather it is an enhancement. On this specific comment, we don’t see that there should be any revision, but as per responses to later comments from reviewer 2, we have emphasised territorial birds’ responses in a revised text.

This is also picked up in reviewer 2’s line-by-line comments, for other reasons on why territorial birds may react differently (on which we have included the suggestions).     

 

2. It wasn’t clear to me why the author’s concluded that territorial birds were “substantially avoiding” wind turbines (lines 514 – 519), rather than exhibiting indifference to turbine locations (i.e., they select the underlying habitat conditions regardless of whether a turbine is there or not). The conclusion of avoidance implies that use of turbine locations is independent of underlying habitat conditions around the turbines (i.e., high quality habitat conditions are used less than expected by chance if a wind turbine is present). I did not see this clearly demonstrated or analyzed in this study. Rather, the summary statistics for habitat conditions at turbine locations vs. used locations (Table 4) indicated that GET values around turbines were lower than at used locations.

Another good point. We note this at the beginning of section 2.5. Please see our response to comment 4, where we expand on this issue further and note the necessary changes which are in the revised submission. We have tempered our language in some places additionally.  

3. Did the home range or territory area of territorial birds overlap with the windfarms? I understand that a bird had to have a minimum of 50 records within 1km of a wind farm’s turbine to be included, but this could still happen in cases where the closest turbine is beyond the periphery of a territory or home range area. This is a critically important detail, because if a bird’s territory or home range does not overlap with a windfarm/turbine, the windfarm/turbine may not even be available for use by that bird (for unknown reasons). This highlights the importance of quantifying availability of wind turbines to the birds. It is clear the authors were aware of the availability issue when they used random points to characterize “expected use” of the windfarm. But there was no apparent tie to an individual bird’s home range or territory location.

 Yes, the home range of all territorial birds did overlap with the wind farms. A further useful point which has been clarified in the revised text.

4. This is just my opinion, but it seems like more powerful/simpler approach to address the study objective, given the information available, would have been to use a single analysis of habitat use vs. availability (e.g., methods in Manly et al. 2010 Resource Selection by Animals). Such an analysis would permit one to directly estimate selection coefficients for windfarms or turbine locations, which would address the question, “Do territorial birds use wind turbine locations more or less than expected by chance?”  I raise this issue because I believe many other readers would have the same questions about the lack of a use/availability analytical design to quantify use/availability of windfarms/turbines while controlling for underlying habitat conditions.

This partially relates to reviewer 2’s second comment. We are familiar with the Manley et al approach and other ‘use vs availability’ methods and used this approach in a previous paper (Fielding et al. 2020 Ibis: cited in the present submission). We did not use this approach in the present submission because what is considered available habitat is more complicated with additional consideration of what is also available away but close to the points of use re: turbine locations. The big problem with selection coefficients is the definition of available habitat and the spatial context which is a particular issue for the current analyses.

Nevertheless, in effect this use v availability notion is encapsulated by our approach on random points (v use) even though we don’t explicitly frame it as such. We believe from other comments from reviewer 2 that he/she has seen this equivalency. We think that other readers will see this equivalency too, especially after the additional text in the revision.

However, reviewer 2 also makes or infers the additional further insightful point (we believe) that this still doesn’t quite get at the notion of whether the disparity in random point v use in foundational descriptive statistics is an illustration of avoidance. Many random points (in wind farms, around turbines as deemed by the limits of a wind farm) could be in unsuitable habitat and their lack of ‘use’ could be because they were not in suitable habitat, rather than eagles avoiding the turbines.

Hence, in the revised submission we have additionally pruned the random point data to include only those with a GET score of 6 or higher (i.e. preferred eagle habitat: Fielding et al 2020) and similarly for tag data. The premise being that if there was still a proportional difference in use records inside v outside wind farms compared only with random points which were in suitable habitat then this removes the prospect that habitat suitability could be involved. Any difference in lower use implicates the presence of a turbine (i.e. avoidance) not related to (beyond) habitat suitability of turbine locations. This additional material illustrates that the difference in use was probably more due to avoidance of turbine locations, through the presence of turbines, rather than their situation or otherwise in suitable habitat.

We believe this addresses reviewer 2’s insightful comment and we have added several new passages of text on this in the revised submission, across several sections. This additional method on pruning random & use points to only suitable habitat locations (GET 6+) is in the Descriptive Statistics section in Methods. We don’t see that further formal statistical testing is needed (in Results) on this aspect when for territorial birds the difference between random and use records (only in suitable habitat) inside and outside of wind farms is an order of magnitude. Confirming avoidance, we suggest.

 

5. A relatively minor comment, but the terminology used to describe the territorial status of individual birds is inconsistent throughout the text, tables, and figure legends (e.g., “territorial vs. nonterritorial” birds used in abstract and portions of the Introduction as opposed to “range holding” and “dispersing” birds used in other areas of the text and in table headers). I suggest using “territorial” and “non-territorial” consistently throughout, as these are descriptors are the most accurate. 

Yet another good point. Thank you. In the revised submission, we have revised the terminology throughout for consistency, and have emphasised “territorial” and “non-territorial”.

 

Line-by-line comments:

Line 114: I would use Hunt et al. 2017 here instead of Hunt 2002; the latter is an update to the former and provides more detailed information about age-specific fatality rates. Done. 

Lines 139 – 152: here you describe reasons why territorial birds may respond differently from non-territorial birds to turbines. All good reasons mentioned, but I would add that territorial birds are also engaged in territorial flight displays/undulations, pair interactions, or territory defense behaviors that may distract from their ability to detect and respond to spinning turbine blades. In revision, as well as in the Introduction we have also referred to this possibility in the Discussion (opening paragraph).

 Table 2. I would delete the “current” field….this will be outdated quickly. We have changed this field to ‘Status’ rather than delete it because it is relevant to the date when analyses were made on available data.

 

Thanks to the reviewer again. In the revised text we have made the suggested changes.

 

Author Response File: Author Response.docx

Reviewer 3 Report

Manuscript ID: diversity-2492908

Title: Responses of GPS-tagged Territorial Golden Eagles Aquila chrysaetos to Wind Turbines in Scotland

Authors: Alan H. Fielding, David Anderson, Stuart Benn, John Taylor, Ruth Tingay, Ewan D. Westonand D. Philip Whitfield

 

In this manuscript the authors examined the responses of GPS-tagged territorial  eagles at 11 operational Scottish wind farms, contrasting the responses of GPS-tagged non-territorial  (intruding) and territorial eagles to the same turbines within territories. Their results highlighted that territorial eagles substantially avoided the turbines and that the spatial extent of avoidance depended on the habitat suitability of both turbine locations and their wider surroundings. The authors highlight in this study the absence of differences, in terms of mistrust towards the same turbines, between territorial and non-territorial eagles and that, regardless of age or territorial status, the avoidance of the turbines could be conditioned by the location itself of the turbines and the spatial suitability of the areas in which they are located.

In my opinion this research is approached with correct methods, and the results are convincing and very clearly expressed and appropriately discussed. Furthermore, the reading of the text is smooth and well organized, and this makes the text easily understandable. For these reasons, in my opinion, the manuscript is eligible for publication in Diversity after a short review.

 

Monor comments:

L.147-149: “Moreover, despite no evidence suggesting habituation in non-territorial eagles (Fielding et al. 2022), repeated exposure to the same turbines may favour habituation in territorial birds”

Comment: this statement seems to be in contrast with the previous assertion according to which there is no correlation between collisions and implant age (L. 123-126). Explain more about this apparent contradiction, please.

L.307.309: “Turbine locations may be avoided not because of the presence of a turbine but because the turbine is not in habitat (including air space) preferred by golden eagles”

Comment: This could also indicate that the presence of the turbine makes the habitat unsuitable and therefore not preferred. I advise the authors to better analyze this apparent "circular discourse", i.e. the eagles avoid the areas where the turbines are located because these are located in unsuitable habitats or, conversely, the presence of the trurbines makes an area unsuitable which, in their absence , would it be preferred by eagles instead?

 

L. 315.319: “GET provides a topographically 314 based surrogate for the availability of orographic winds, which have  repeatedly been found as influential in habitat selection studies of golden eagle……..”

Comment: In my opinion, topography is only a part of the concept of habitat, which is undoubtedly complex and made up of many ineractive parts, including topography. The authors must consider this factor, clearly inserting in the text that they are considering only a "portion" of the preferred habitat, i.e. the one related to the topography,

justifying and commenting in an appropriate way their choice to analyze the GET rather than other "portions" of the concept of habitat prefeence (for example land use, bioclimatic data, exposure, etc.).

 

L. 569-578: While the authors indicate that new or old sites are being occupied even if near wind farms, in the absence of alternatives, however there is an effect of the presence of these facilities on site selection. Furthermore, it would be useful to compare the productivity of sites with and without wind farms within the occupied territories to analyze possible interferences of the presence offarms on the biology and ecology of the species.

 

L.587-591: “Intriguingly, it appears that while Scottish territorial golden eagles are  wary of most turbines and so avoid them, this does not necessarily  transfer to their choice of nest sites as regards ‘disturbance distance’ or  apparently ‘accepting’ them as (typically) largely unsuitable locations  within their home range.”

Comment: see also previous commenta t L. 307-309, and also, furthermore, more simply, the authors also consider the hypothesis that  the eagles have no alternative, even given the population density of Golden Eagle in the Scottish territories.

Author Response

In this manuscript the authors examined the responses of GPS-tagged territorial  eagles at 11 operational Scottish wind farms, contrasting the responses of GPS-tagged non-territorial  (intruding) and territorial eagles to the same turbines within territories. Their results highlighted that territorial eagles substantially avoided the turbines and that the spatial extent of avoidance depended on the habitat suitability of both turbine locations and their wider surroundings. The authors highlight in this study the absence of differences, in terms of mistrust towards the same turbines, between territorial and non-territorial eagles and that, regardless of age or territorial status, the avoidance of the turbines could be conditioned by the location itself of the turbines and the spatial suitability of the areas in which they are located.

In my opinion this research is approached with correct methods, and the results are convincing and very clearly expressed and appropriately discussed. Furthermore, the reading of the text is smooth and well organized, and this makes the text easily understandable. For these reasons, in my opinion, the manuscript is eligible for publication in Diversity after a short review.

 

Monor comments:

L.147-149: “Moreover, despite no evidence suggesting habituation in non-territorial eagles (Fielding et al. 2022), repeated exposure to the same turbines may favour habituation in territorial birds”

Comment: this statement seems to be in contrast with the previous assertion according to which there is no correlation between collisions and implant age (L. 123-126). Explain more about this apparent contradiction, please.

There is no contradiction in our re-reading of our text. We did not use “collisions” in the previous ‘assertation’ statement or any comment on any correlation. The previous text involved non-territorial birds and didn’t include any data on collisions. What we have said is that territorial birds may more likely habituate to turbines in their territory because they are spatially constrained and so can have repeated exposure to them. Non-territorial birds do not have the same level of repeated exposure to these same turbines and so may be more likely to avoid. We have explained this in the text, and more. Aside from not fully understanding what reviewer 3 is claiming on assertion about “no correlation between collisions and implant age” we can’t see that we should make any more changes, when we have made several on more pertinent points by reviewer 2 and reviewer 3 (see later on coincidence).      

L.307.309: “Turbine locations may be avoided not because of the presence of a turbine but because the turbine is not in habitat (including air space) preferred by golden eagles”

Comment: This could also indicate that the presence of the turbine makes the habitat unsuitable and therefore not preferred. I advise the authors to better analyze this apparent "circular discourse", i.e. the eagles avoid the areas where the turbines are located because these are located in unsuitable habitats or, conversely, the presence of the trurbines makes an area unsuitable which, in their absence , would it be preferred by eagles instead?

This is similar to comments made by reviewer 2, which we have addressed extensively in the revised submission. 

315. 315.319: “GET provides a topographically 314 based surrogate for the availability of orographic winds, which have  repeatedly been found as influential in habitat selection studies of golden eagle……..”

Comment: In my opinion, topography is only a part of the concept of habitat, which is undoubtedly complex and made up of many ineractive parts, including topography. The authors must consider this factor, clearly inserting in the text that they are considering only a "portion" of the preferred habitat, i.e. the one related to the topography,

justifying and commenting in an appropriate way their choice to analyze the GET rather than other "portions" of the concept of habitat prefeence (for example land use, bioclimatic data, exposure, etc.).

We agree that the GET model is not the be all and end all to predict perfectly the use of habitat by golden eagles, at least in Scotland. What we have shown, however, is that it consistently predicts (robustly, with repeated high statistical confidence) the spatial use of land by golden eagles. It does this across a wide range of different ‘vegetative’ habitats and ‘land use’ characteristics across Scotland, which as we note is extremely variable in these features and, also, climatic influences. As has been referenced before, topography seems a universal basis for much habitat selection in golden eagles, across many ecosystems. It provides for a simple but powerful proxy for habitat use/preference/suitability which can extend beyond local vegetative features. This is described in Fielding et al (2020) and we re-affirm this robust proxy in Appendix A for territorial birds, as well as a further sample of non-territorial eagle data beyond those considered by Fielding et al (2020).

We have not made any changes to the submission based on this comment.         

1. 569-578: While the authors indicate that new or old sites are being occupied even if near wind farms, in the absence of alternatives, however there is an effect of the presence of these facilities on site selection. Furthermore, it would be useful to compare the productivity of sites with and without wind farms within the occupied territories to analyze possible interferences of the presence offarms on the biology and ecology of the species.

This is a useful suggestion but beyond the limits of the current submission. 

L.587-591: “Intriguingly, it appears that while Scottish territorial golden eagles are  wary of most turbines and so avoid them, this does not necessarily  transfer to their choice of nest sites as regards ‘disturbance distance’ or  apparently ‘accepting’ them as (typically) largely unsuitable locations  within their home range.”

Comment: see also previous commenta t L. 307-309, and also, furthermore, more simply, the authors also consider the hypothesis that  the eagles have no alternative, even given the population density of Golden Eagle in the Scottish territories.

See previous response to l. 307 – 309, and that this has been dealt with in response to reviewer 2’s comments.

 

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

I have read the revised manuscript carefully and appreciated that the author’s directly addressed my previous review comments in their revisions. I think the authors did a nice job of addressing my comments – I found the additional analysis informative in that it addressed the question of how wind turbines in “good” habitat were used.

There was one lingering issue regarding how the summary statistics were calculated – something I should have caught in my first review. Specifically, the analyses of use/avoidance of wind turbines (described in sections 2.5 – 2.6) appeared to be based on total numbers of “bird records” (GPS-point locations) combined across birds within territorial vs. non-territorial groups (or by windfarm groups, as shown in Table 3). The issue with this approach is that it doesn’t account for among-bird variation in use/avoidance descriptive statistics (even when restricted to preferred habitat conditions). Did some individual birds contribute disproportionately more or less to the percentages reported in section 3.1? This is ultimately a psuedoreplication issue.

Alternatively, the authors could have calculated percentages reported in section 3.1 for each individual bird first, then calculated mean % per bird per group, along with SD or SE, which then provides the reader with important information on the precision of these descriptive measures (i.e., how much percentages varied among birds). I suggest mean % per bird, rather than % across all birds combined, would provide a much more robust measure for comparisons with random point %.

Note that the glmm analysis correctly handled individual bird variation because the authors used bird ID as a random effect term in the models.

Table 6: unclear why you are reporting both AIC and BIC here? This wasn’t explained in the methods. I suggest just using AIC you didn't use a Bayesian approach.

It was a pleasure to read and review this interesting and informative paper.

Author Response

Diversity 2492908 Rev 2 Round 2

 

Open Review

( ) I would not like to sign my review report
(x) I would like to sign my review report

Quality of English Language

( ) I am not qualified to assess the quality of English in this paper
( ) English very difficult to understand/incomprehensible
( ) Extensive editing of English language required
( ) Moderate editing of English language required
( ) Minor editing of English language required
(x) English language fine. No issues detected

 

	Yes	Can be improved	Must be improved	Not applicable
Does the introduction provide sufficient background and include all relevant references?	(x)	( )	( )	( )
Are all the cited references relevant to the research?	(x)	( )	( )	( )
Is the research design appropriate?	( )	(x)	( )	( )
Are the methods adequately described?	( )	(x)	( )	( )
Are the results clearly presented?	(x)	( )	( )	( )
Are the conclusions supported by the results?	(x)	( )	( )	( )

Comments and Suggestions for Authors

I have read the revised manuscript carefully and appreciated that the author’s directly addressed my previous review comments in their revisions. I think the authors did a nice job of addressing my comments – I found the additional analysis informative in that it addressed the question of how wind turbines in “good” habitat were used.

There was one lingering issue regarding how the summary statistics were calculated – something I should have caught in my first review. Specifically, the analyses of use/avoidance of wind turbines (described in sections 2.5 – 2.6) appeared to be based on total numbers of “bird records” (GPS-point locations) combined across birds within territorial vs. non-territorial groups (or by windfarm groups, as shown in Table 3). The issue with this approach is that it doesn’t account for among-bird variation in use/avoidance descriptive statistics (even when restricted to preferred habitat conditions). Did some individual birds contribute disproportionately more or less to the percentages reported in section 3.1? This is ultimately a psuedoreplication issue.

Alternatively, the authors could have calculated percentages reported in section 3.1 for each individual bird first, then calculated mean % per bird per group, along with SD or SE, which then provides the reader with important information on the precision of these descriptive measures (i.e., how much percentages varied among birds). I suggest mean % per bird, rather than % across all birds combined, would provide a much more robust measure for comparisons with random point %.

 

Another very helpful point from this astute and insightful reviewer. We have revised the text in section 3.1 along the lines of the ‘alternative’ approach suggested. This revision is useful in providing the reader with additional information on the precision of the descriptive measures and how this varied among birds. This prefaces the later GLMM analyses in highlighting the influence of individual birds’ behaviour which is seldom considered in studies of birds’ responses to wind farms.

 

Note that the glmm analysis correctly handled individual bird variation because the authors used bird ID as a random effect term in the models.

 

Yes, thank you.

 

Table 6: unclear why you are reporting both AIC and BIC here? This wasn’t explained in the methods. I suggest just using AIC you didn't use a Bayesian approach.

 

This has been revised in Table 6 to include only AIC and any reference to BIC and BIC scores have been removed elsewhere in the text.

 

It was a pleasure to read and review this interesting and informative paper.

 

Thanks again to this reviewer who has substantially and repeatedly improved this submission, with insight, good grace, and appreciation for our efforts. If possible we would like to acknowledge this contribution personally should our submission be published.

 

Submission Date

22 June 2023

Date of this review

31 Jul 2023 22:43:39

 

Author Response File: Author Response.docx