Substantial Variation in Prospecting Behaviour of Young Golden Eagles Aquila chrysaetos Defies Expectations from Potential Predictors

Round 1

Reviewer 1 Report

This is an excellent piece of work which draws on an outstanding dataset, which is clearly growing as the authors amass more data.  As the authors summarise (pasted below) this is a novel study of a key question regarding natal dispersal.  They very openly present their findings pointing to substantial variation in prospecting behaviour, which we still do not understand - but which I believe will emerge from this study in later years.  The methods, analysis and over all depiction of the work is very clear, and  the Discussion is very clear. In lines 295-300, discussing differences between males and female dispersive tendencies, I suggest this is more clearly explained.  The authors should explain in more detail the importance of so many vacant territories in their study, and why this should be so important to the findings.

'We expected a positive association between number of prospecting visits and natal dispersal duration, since with more time to gather information there should be more visits. We also expected fewer prospecting visits in prior vacant territories. Neither of these expectations were supported. There was a non-significant tendency for more prospection visits by males. Our study provides novel information on a seldom-studied behaviour in a large raptor. It illustrates substantial variation in prospecting behaviour, but expectations of potential drivers behind this variation were not confirmed, urging further study.'

 

Author Response

This is an excellent piece of work which draws on an outstanding dataset, which is clearly growing as the authors amass more data. 

We thank the reviewer for the complimentary comment. And indeed, with further research we hope to exploit our large dataset to address several unanswered questions in a seldom-studied field.

In lines 295-300, discussing differences between males and female dispersive tendencies, I suggest this is more clearly explained.  The authors should explain in more detail the importance of so many vacant territories in their study, and why this should be so important to the findings.

On the first point raised - sexual differences in dispersive behaviour and the text at lines 295-300. There are some published differences or expected differences to which we refer. Unfortunately, we cannot yet refer thoroughly to other manuscripts which are still under review in submissions to other journals which highlight further differences between the sexes – other than those we can cite, mostly from our study system.

From these other submissions, even though we can’t cite them – yet - we would nevertheless hope to reassure the reviewer that we have found a significant difference by sex in natal dispersal distance (males settle closer to their natal site – as expected by theory: Whitfield et al 2023). We have also found sexual differences – documented in another submission to another journal – in when males and females first settle on a territory, seasonally and according to the prior occupancy status of the territory. We cannot cite these studies as they are still under review. Hence in this paper we do have to be a bit circumspect. But there are other published differences between the sexes to which we refer. Mostly involving our study system.

On the second point. It is suggested that we explain in more detail the importance of the occurrence of many vacant territories in our study system. We have already outlined this in detail, we feel, in several passages of text and with many citations. In the Introduction, Methods, as well as the Discussion. In trying to balance responses to different reviewers, another reviewer has conversely suggested that we have gone too overboard in how many papers and reports we have cited on this issue. We consider that, given conflicting reviewers’ opinions, that what we have presented on this is – on balance – appropriate. There is a large body of information on this point and without repeating this material again. We feel it is better to refer to previous material. Whitfield et al (2022a) and Whitfield & Fielding (2017) are key publications on this. These are cited for reference (reading).  

Reviewer 2 Report

see attached

Comments for author File: Comments.pdf

Author Response

General comments

This is a very interesting manuscript, well written, concerning a remarkable argument. Thank you. Nonetheless, I found some issues to be addressed, regarding the methods used, listed below.

 

Specific comments

Methods

• Lines 200-203: but occupied by who? A territorial pair or a single bird? Because I expect that in first case the chance to settle are very low but in the latter maybe the presence of an unpaired bird is less problematic, if not an incentive to pair if the bird is of the opposite sex. Please clarify this aspect too.

We refer to several papers on this issue, which include our field checks that the algorithm for determining settlement of a tagged bird encapsulated wider definitions for an ‘occupied territory’. In Golden Eagles this includes the presence of a putative pair of birds. Please refer to Whitfield et al (2022a) and the other papers we have cited which include definitions of ‘occupancy’ from field records. These are many but necessary to document thoroughness. For our study system we have been through this issue previously, and it is better to refer to past material on this rather than recycle and repeat with more repetitive text in another paper. Even though another referee takes issue with how many papers we cite. We prefer to cite all the relevant studies, in expectation that the reader can access these, rather than add the same text in paper after paper.    

• You need to explain here the statistical analyses that you ran. As I can see from the results, you computed simple GLMs, with Gaussian error distribution (i.e. linear models), using Wald test (the summary() command from R) to see the significance. Explain what method you used, how you calculated significance, how you run analyses (R?), and so on.

Thank you. This is a fair and useful comment since the statistical approach should be outlined in the Methods. A passage of text has been added. This includes the suggestions from the referee.

• Wald test is not the best one to test significance: did you try to use LR test? It is more powerful (in car package, Anova() function). For the full model (that is always better because keep together the effects of all variables, as it is in nature), you can use an AIC-approach, considering all the models with these variables (MuMIn package, dredge function) and then see which is the best model or the best set of models explaining you response (you can use delta AIC <=2 or cumulative sum of Akaike’s weight <=0.95). This method is widely used, is reliable and powerful (Burnham and Anderson 2002).

We have added passages of text on this in the revised MS, in the Methods (see above) and the Results. We are aware of AIC methods advocated by the referee, having often used them in past studies. Here our approach was to undertake simple GLMs as a first examination. There was nothing significant in these analyses (very far from it in relation to most of the predictors tested), suggesting need for more ‘powerful’ analyses would be redundant. None should be expected in the use of AICs as advocated by the referee, even if these are more ‘powerful’, given the results of the individual GLMs. We have nevertheless undertaken further analyses as suggested by the referee and have included these in an additional passage of text on multi-model inference. As expected, use of AICs and multi-model analyses have made no difference to our Results. We have also referred to the warning issued by Burnham and Anderson (2002) that the best model as determined by AIC may be spurious and parsimony should be used.    

Discussion

• Lines 318-320: I would remove this sentence.
• If it is important, I would discuss the presence of occupied territories by single birds, and how this fact could have affected the eagles in their choices.

Line 318-320. We have removed this sentence. Not sure why this removal was suggested when reconnaissance/prospecting behaviour is practically easier to record when those breeding opportunities are easier to study in seabirds, because seabirds often nest in colonies (we are not alone in making this point- hence the reference we cite). In large raptors the diametrically opposite breeding distribution is typical. Fewer birds spread out over a wider landscape, with widely distributed breeding/territorial opportunities. But we have removed as suggested, despite our better judgement and providing a conspicuous contrast; and since this paper primarily concerns material towards a special issue on raptor ecology and conservation.

The second point is not important, in our view, since as we described, the presence of a pair of birds – in field confirmations of occupancy – was part of our study (see Whitfield et al. 2022a).

Reviewer 3 Report

Comments to Authors – Substantial Variation in Prospecting Behaviour of Young Golden Eagles (Aquila chrysaetos) Defies Expectations from Potential Predictors

 

Overall, this manuscript includes some very interesting information and was quite interesting.  The manuscript is difficult to read at times, likely due to numerous citations using author names. Please see instructions to authors and format references as numbers rather than author names as this will make reading the manuscript easier.  

 

Are all of the self-citations necessary or can that list be shortened down a bit?  I would like to see it narrowed if possible.  

 

Mentioned in my comments about the Discussion section, is there a way to analyze behaviors of Eagles in the East vs the West?  There is emphasis in methods about the illegal persecution of Golden Eagles in the East and its inherent impacts on Golden Eagle ecology in Scotland, yet the analysis does not address an east/west component.  Instead, one is left to imaging that vacant territories are more common in the east, and occupied territories are more common in the west and therefore vacant/occupied is a stand-in for east/west or visa versa.  The manuscript would be stronger if this could be clarified.    

 

 

1. Introduction

Line 38 – formatting of Clobert et al.  (larger font than the rest)

Line 81 – remove “then”

 

2. Methods

The second paragraph, lines 132-145, should be re-worded and reworked for better clarification.  I believe that the point of the paragraph is to highlight that in the east, grouse managers illegally kill eagles in order to manage for high grouse densities necessary for successful driven grouse hunts and therefore, empty territories exist but represent “ecological black holes”.  As someone familiar with hunting and the hunter/resource interface I arrived at this conclusion only after reading the paragraph more than once.  I think you can condense the information.  I.e. There is a broad west/east divide on the strong influence of illegal persecution of Golden Eagles. In the east, driven shoots of Red Grouse (Lagopus lagopus scotica) are practiced and require management for extremely high densities of grouse coveys (Whitfield et al. 2003, 2004a, b, 2006, 2007, 2008). To maintain such high densities, some land managers illegally kill predators of grouse such as Golden Eagles. Illegal persecution of Golden Eagles has been evident for several decades. These illegal activities substantially contribute to a national environment of numerous vacant eagle territories and a relatively low density in the east, compared to higher densities further west (Watson 2010, Whitfield et al. 2003, 2004a, b, 2006, 2007, 2008, Hayhow et al. 2017, Whitfield & Fielding 2017). Such eastern areas can be attractive to eagles but as they can be lethal this provides a classic example of an “ecological black hole” due to anthropogenic activities (Whitfield et al. 2004a, b, 2008).   

 

Line 148 – extra space before Despite…

 

Line 157 – remove “then”.  Also remove “then” in lines 198 and 201.

 

Consider adding/moving language about using GLM models…on the five predictors after discussion of the 5 predictors and before the Results. 

 

3. Results

 

Figure 1 is partially missing on my proof.  The ‘V’ box is missing for the prior range status boxplot of male and female visits. 

 

Figure 1. Consider adding both boxplots for each sex with occupied and vacant territories onto one graph such as you have with the comparison of F and M, and Prior Range Status.  

 

4. Discussion

Line 302 – extra space before ‘An’

 

Mention of “territorial landscape which was heavily influenced by illegal persecution” i.e. the east.  There should be an analysis of behaviors in the east and west if this is the case.  Or, if this is the case, state that most vacancies were in the east and most occupied were in the west.  

 

 

 

Author Response

Overall, this manuscript includes some very interesting information and was quite interesting.  The manuscript is difficult to read at times, likely due to numerous citations using author names. Please see instructions to authors and format references as numbers rather than author names as this will make reading the manuscript easier.

 

Yes, we are aware of the referencing system & instructions for Diversity papers. If the present manuscript progresses to publication, then these changes in the citations will be enacted, which should alleviate this concern.  

 

Are all of the self-citations necessary or can that list be shortened down a bit?  I would like to see it narrowed if possible.

We think they are necessary (conversion to numerical referencing notations should help if a version of the manuscript is accepted), when we try to summarise critical aspects of our study by referring to many previous studies using the same methods. We are conspicuously aware that we cite many previous studies from our Scottish studies, and other relevant research, as highlighted by the reviewer.

The alternative is to continue to reiterate what several previous papers have already described for our study system. That would take up much more text, which would be redundant repetition, when they have been described, advanced and accepted in several previous publications. We are also very much of the opinion that too many papers in this field of research are minimal in describing the essential basics of critical underlying methods. Hence we spend an appropriate respective amount of text on such, but without too much repetition, if possible. We also credit the interested reader with the capacity to seek out the references to which we refer if he/she doesn’t know them already. There may be many.

Mentioned in my comments about the Discussion section, is there a way to analyze behaviors of Eagles in the East vs the West?  There is emphasis in methods about the illegal persecution of Golden Eagles in the East and its inherent impacts on Golden Eagle ecology in Scotland, yet the analysis does not address an east/west component.  Instead, one is left to imaging that vacant territories are more common in the east, and occupied territories are more common in the west and therefore vacant/occupied is a stand-in for east/west or visa versa.  The manuscript would be stronger if this could be clarified.

We do not believe such an additional layer of analysis would add much. Vacant territories are more common in the east, as we describe and cite several outputs which show this. Hence, we are not of the same opinion that such is left to “imaging” [sic] (we presume this typo should be ‘imagination’). For an earlier sample of the (now, in present study, enhanced) database on settled territories this is explicitly illustrated in Whitfield et al. (2022a, Front Ecol Evol 10: 743598. https://doi.org/10.3389/fevo.2022.743598, Fig. 5), which we cite.

We cite many other studies which have shown that vacant territories are more common in the east, largely down to some areas where driven grouse moor management occurs and there is evidence of illegal persecution. As background (and hence requiring several citations) this shows a major reason why some territories are more likely to be vacant. However, as this illegality is not blanket across the east, it is not so simple as being a simple east/west divide, which we note in appropriate language (see Fig. 5).

As we hope we have already made clear in the submission, vacant/occupied is not a stand-in for east/west, or vice versa. It’s that many vacancies are created by the illegal activities of some land managers pursuing a particular land management objective and these tend to be more common in the east. There are other broad differences between the east and the west (live prey availability, which we also refer to: hence more citations on this aspect…as well as vegetative differences).

The primary expectations on how dispersal behaviours (including reconnaissance/prospecting) may be affected should rest primarily on the prior occupancy status of the settled territory – on which we describe. Why or how there are differences are secondary in this context, even if in the background (Methods and Discussion) it is worth referencing – as we have done. However, if we do not find any relationship with the primary explanatory/predictor, then it is not worth considering any cruder (weaker) metric such as east/west.       

1. Introduction

Line 38 – formatting of Clobert et al.  (larger font than the rest)

Line 81 – remove “then”

 

 

Corrected on line 38, but this will disappear when the referencing system is revised to match Diversity format.

Line 81 – thank you, ‘then’ removed.

2. Methods

The second paragraph, lines 132-145, should be re-worded and reworked for better clarification.  I believe that the point of the paragraph is to highlight that in the east, grouse managers illegally kill eagles in order to manage for high grouse densities necessary for successful driven grouse hunts and therefore, empty territories exist but represent “ecological black holes”.  As someone familiar with hunting and the hunter/resource interface I arrived at this conclusion only after reading the paragraph more than once.  I think you can condense the information.  I.e. There is a broad west/east divide on the strong influence of illegal persecution of Golden Eagles. In the east, driven shoots of Red Grouse (Lagopus lagopus scotica) are practiced and require management for extremely high densities of grouse coveys (Whitfield et al. 2003, 2004a, b, 2006, 2007, 2008). To maintain such high densities, some land managers illegally kill predators of grouse such as Golden Eagles. Illegal persecution of Golden Eagles has been evident for several decades. These illegal activities substantially contribute to a national environment of numerous vacant eagle territories and a relatively low density in the east, compared to higher densities further west (Watson 2010, Whitfield et al. 2003, 2004a, b, 2006, 2007, 2008, Hayhow et al. 2017, Whitfield & Fielding 2017). Such eastern areas can be attractive to eagles but as they can be lethal this provides a classic example of an “ecological black hole” due to anthropogenic activities (Whitfield et al. 2004a, b, 2008).

Many thanks. We have transcribed this suggested condensed text into the revised manuscript.

Line 148 – extra space before Despite…

 

Line 157 – remove “then”.  Also remove “then” in lines 198 and 201.

 

All suggestions have been acted on in the revised MS. Along with a couple of other places where an extra space should also be required.

Consider adding/moving language about using GLM models…on the five predictors after discussion of the 5 predictors and before the Results. 

Thanks for the suggestion which we’ve considered. We feel that it’s better to keep the ordering of text as it is, not least as in a response to reviewer 2’s suggestions on statistical analyses these have been incorporated in the revised MS. This has added a large passage of text in the Methods. We hope that the response to, and text revisions from, reviewer 2’s suggestions will comply with reviewer 3.  

Figure 1 is partially missing on my proof.  The ‘V’ box is missing for the prior range status boxplot of male and female visits. 

 

Apologies – we can see this from our proof too. This shouldn’t be the case and is down to editorial formatting rather than what we supplied. There’s also some odd formatting on Tables too, from our perspective. We leave it to the editors/proofing/formatting of the journal to rectify this should progress to publication be realised.

 

Figure 1. Consider adding both boxplots for each sex with occupied and vacant territories onto one graph such as you have with the comparison of F and M, and Prior Range Status.  

 

Thanks for this useful suggestion, which we have considered in depth. As we understand this comment, we think that the original graphical layout, as submitted, better illustrates the relevant differences according to the four-way presentation.   

Line 302 – extra space before ‘An’

 

Rectified in revision.

Mention of “territorial landscape which was heavily influenced by illegal persecution” i.e. the east.  There should be an analysis of behaviors in the east and west if this is the case.  Or, if this is the case, state that most vacancies were in the east and most occupied were in the west.

As in response to a previous comment on this issue, we feel that the submitted text has already covered this matter thoroughly. We have stated that most vacancies were in the east and most occupied territories were in the west. We have cited several previous publications showing this, too.

Round 2

Reviewer 2 Report

General comments

In this version Authors improved the manuscript, also taking in account my suggestions or explaining the reasons why they were not followed. I have only a minor revision on the clarity of the analysis (listed below), but after it has been followed, I’ll recommend the acceptance.

 

Specific comments

Methods

·       Lines 219-221: ok, you used multimodel inference that is a good choice. However you need to specify how you selected models to retain: using the best one or the best set of models (e.g. with delta AIC <=2)? That’s important. I suggest to modify the sentence as the following “We selected the model with the lowest AIC as the best model but we acknowledged a warning (Burnham and Anderson 2002) about selecting a spurious best model, therefore we included parsimony as an important criterion.” or something similar.

Author Response

Methods

Lines 219-221: ok, you used multimodel inference that is a good choice. However you need to specify how you selected models to retain: using the best one or the best set of models (e.g. with delta AIC <=2)? That’s important. I suggest to modify the sentence as the following “We selected the model with the lowest AIC as the best model but we acknowledged a warning (Burnham and Anderson 2002) about selecting a spurious best model, therefore we included parsimony as an important criterion.” or something similar.

We have included the suggested sentence in R2 revision.