Quick-quick-slow: the foxtrot migration and dynamic non-breeding range

Reviewer #2 (Public Review):

This preprint by Pokrovsky and coworkers is a descriptive study reporting on non-breeding itinerant behaviour of an intrapalearctic migratory raptor, the rough-legged buzzard, and relating such non-breeding movements to snow cover across the European non-breeding range. The article is based on long-term GPS tracking data from a relatively large (n=43) sample of individuals that were equipped with state-of-the-art tracking devices in the Russian Arctic during 2013-2019. The results show that, upon breeding, buzzards migrated rapidly to southern non-breeding areas, located in open areas north of the Black and Caspian seas, where they perform continuous directional movements at a slower pace, initially moving SW (Oct to Jan) and then progressively moving NE (Feb to Apr) before embarking on rapid spring migration. It is suggested that such itinerant behaviour follows variation (expansion and retreat) of snow cover across the non-breeding range.

The results are potentially useful for researchers investigating the ecological drivers of bird movement patterns. The analytical framework appears solid, although some details on the analyses (requested during the previous review round) are still unclear and have not been modified despite explicit requests. Significant weaknesses in the theoretical framework persist in the revised version, including unwarranted claiming of novelty, overselling of importance of the study, and overinterpretation of the data. Below are key points that the authors did not consider when revising their manuscript.

(1) The authors underemphasize the fact that what they term 'fox-trot' migration is actually a well-known patterns for many other migratory species, both in the Nearctic and in the Afro-Palearctic migration systems. Such behaviour has previously been identified as 'itinerant' or 'non-breeding itinerancy', involving an alternation of stopovers and movements between different short-term non-breeding residency areas, or even slow continuous movements, and it seems that the pattern the authors report for this particular species is perfectly in line with such previous evidence. For instance, this is well-documented among migratory raptors, such as the Montagu's harrier, lesser kestrels or black kites, that exploit Sahelian savannahs, where large spatio-temporal variation in greenness and hence resource availability occurs. And, besides the mentioned cuckoos and nightingales, there are studies of red-backed shrikes suggesting the same, as well as of tree swallows in the Nearctic. Therefore, the authors should avoid claiming novelty for this study and introducing unnecessary and confusing new terms in the literature (i.e. the 'fox-trot' migration patterns) when these are definitely not strictly needed as they have been previously observed and defined otherwise. Sentences such as 'We used the rough-legged buzzard as a model..." are also similarly unwarranted. This is simply a descriptive studies reporting on such behaviour in yet another migratory species. The whole introduction is pervaded by a faulty logic. The authors first introduce a new (unwarranted) term based on previous evidence from other studies (none of which felt any need to introduce and describe it); then they assume, for unclear reasons, that the species they are studying should behave in that way; even more worryingly, based on these assumptions, they make specific predictions on how this species should behave, without any sound biological reason for these predictions. I admit I hardly see any scientific logic in this approach.

(2) The species has a very standard migration for a short-distance migrant, by all means. It moves to non-breeding areas, and once there it slowly moves towards the south in autumn and back again in spring, until it departs for pre-breeding migration. This is no different from other trans-Saharan migratory raptor species that spend the non-breeding period in the Sahel. Whether the species perform any short/medium term stopover (frequenting the same are for some days) during the non-breeding stage (between end of autumn migration and onset of spring migration), as is the case of most species showing non-breeding itinerancy, is not reported. The authors only show a slower pace of movement during the non-breeding period compared to autumn and spring migration, without providing any further details. This hinders comparisons with other previous studies.

(3) The current title is unnecessarily general (it may recall rather a review or meta-analysis) and not adequately describing the content of the manuscript. In order to be informative, the title should more tightly reflect the content of the article. A valid alternative would be 'Itinerant non-breeding behaviour of an intra-Palaearctic migratory raptor', as it would be far more adequate and informative.

(4) The text, particularly the Introduction and the Discussion, would greatly benefit from profound reframing in light of the above comments. Upon reading the first sentence of the introduction, it looks surprising that the authors based their suggestion for 'fox-trot' migration based on a very outdated article on the migration of Montagu's harrier based on sparse ring recovery data which merely suggests the existence of 'movements' within the non-breeding areas (i.e. non-breeding itinerancy), while subsequent large scale satellite tracking studies of this species provided compelling evidence for non-breeding area itinerancy (and again, no mention of 'fox-trot' whatsoever). The discussion is entirely framed around potential issues related to accurate monitoring of population size and trends, which the author surprisingly refer to 'conservation implications'. As I already mentioned in my previous review, the 'conservation implications' of this study are nearly negligible. At best, it suggests that caution should be applied when interpreting population trends of migratory species based on non-breeding area counts only, a pattern that is already well known for decades (consider the long-running IWC coordinated by Wetlands International!). In addition, Christmas Bird Count, a long-term monitoring program of AOS, is mentioned without any accurate reference to what it actually is, assuming that any reader would be familiar with a very peculiar monitoring scheme of the Nearctic region.

The final paragraph epitomizes how authors are overstating the importance of this study, claiming for non-existent novelty and even 'discovery': "Our study has identified and characterized a new pattern of migratory behavior, the 'foxtrot migration', along with the associated concept of 'dynamic range'. This discovery has significant implications for conservation strategies and adequate representation of non-breeding habitats".

Author response:

The following is the authors’ response to the current reviews.

We sincerely appreciate the reviewer’s dedication to evaluating our manuscript and raising essential considerations regarding the classification of the migration behavior we described. While the reviewer suggests that this behavior aligns with the concept of itinerancy, we contend that it represents a distinct phenomenon, albeit with similarities, as both involve the non-breeding movements of birds. We acknowledge that our manuscript did not adequately address this distinction and have considered the reviewer’s feedback. In our response, we clarify the difference between the described phenomenon and itinerancy. Our revised manuscript will include a new section in the Discussion to address this issue comprehensively.

In the first part of the review, the reviewer emphasizes that the pattern we are describing is consistent with itinerancy. Regardless of the terminology used, we want to highlight the existence of two different types of migratory behavior, both of which involve movement in non-breeding areas.

The first type, called itinerancy, was first described by Moreau in 1972 in “The Palaearctic-African Bird Migration Systems.” As noted by the reviewer, this behavior involves an alternation of stopovers and movements between different short-term non-breeding residency areas. They usually occur in response to food scarcity in one part of the non-breeding range, causing birds to move to another part of the same range. These movements typically cover distances of 10 to 100 kilometers but are neither continuous nor directional. Moreau (1972) defined itinerancy as prolonged stopovers, normally lasting several months, primarily in tropical regions. He noted observations of certain species disappearing from his study areas in sub-Saharan Africa in December and others appearing, suggesting they may have multiple home ranges during the non-breeding season. Subsequent research, as mentioned by the reviewer, has confirmed itinerancy in many species, particularly among Palaearctic-African migrants in sub-Saharan Africa. In particular, the Montagu’s Harrier has been extensively studied in this regard. The reviewer rightly points out that our study does not include recent findings on this species. In our revised version, we will include references to recent studies, such as those by Trierweiler et al. (2013, Journal of Animal Ecology, 82:107-120) and Schlaich et al. (2023, Ardea, 111:321-342), which show that Montagu’s Harrier has an average of 3-4 home ranges separated by approximately 200 kilometers. These studies suggest that the species spends approximately 1.5 months at each site, with the most extended period typically observed at the last site before migrating to the breeding grounds.

In the second type, birds undertake a post-breeding migration, arrive in their non-breeding range, and then gradually move in a particular direction throughout the season. This continuous directional movement covers considerable distances and continues throughout the non-breeding period. In our study, this movement covered about 1000 km, comparable to the total migration distance of Rough-legged Buzzards of about 1500 km. As observed in our research, these movements are influenced by external factors such as snow cover. In such cases, the progression of snow cover in a south-westerly direction during winter can prevent birds from finding food, forcing them to continue migrating in the same direction. In essence, this movement represents a prolonged phase of the migration process but at a slower pace. Similar behavior has been documented in buzzards, as reported by Strandberg et al. (2009, Ibis 151:200-206). Although several transmitters in their study stopped working in mid-winter, the authors observed a phenomenon they termed ‘prolonged autumn migration.’

In the second part of the review, the reviewer questions the need to distinguish between the two behaviors we have discussed. However, we believe these behaviors differ in their structure (with the first being intermittent and often non-directional, whereas the second is continuous and directional) and in their causes (with the first being driven by seasonal food resource cycles and the second by advancing snow cover). We therefore argue that it is worth distinguishing between them. To differentiate these forms of non-breeding movement, we propose to use ‘itinerancy’ for the first type, as described initially by Moreau in 1972, and introduce a separate term for the second behavior. Although ‘slow directional itinerancy’ could be considered, we find it too cumbersome.

Moreover, ‘itinerancy’ in the literature refers not only to non-breeding movements but also to the use of different nesting sites, e.g., Lislevand et al. (2020, Journal of Avian Biology: e02595), reinforcing its association with movements between multiple sites within habitats. We, therefore, propose that the second behavior be given a distinct name. We acknowledge the reviewer’s point that we did not adequately address this distinction in the Discussion and plan to include a separate section in our paper’s revised version. In the third part of his review, the reviewer suggests an alternative title. Another reviewer, Dr Theunis Piersma, suggested the current title during the first round of reviewing, and we have chosen his version.

In the fourth part of the review, the reviewer questions whether it is appropriate to discuss the conservation aspect of this study. This type of non-breeding movement raises concerns about accurately determining non-breeding ranges and population dynamics for species that exhibit this behavior. We believe that accurate determination of range and population dynamics is critical to conservation efforts. While this may be less important for species breeding in Europe and migrating to Africa, for which monitoring breeding territories is more feasible, it’s essential for Arctic and sub-Arctic breeding species. Large-scale surveys in these regions have historically been challenging and have become even more so with the end of Arctic cooperation following Russia’s war with Ukraine (Koivurova, Shibata, 2023). For North America and Europe, non-breeding abundance is typically estimated once per season in mid-winter. In North America, these are the so-called Christmas counts (which take place once at the end of December), and in Europe, they are the IWC counts mentioned by the reviewer (as follows from their official website - “The IWC requires a single count at each site, which should be repeated each year. The exact dates vary slightly from region to region, but take place in January or February”). Because of such a single count in mid-winter, non-breeding habitats occupied in autumn and spring will be listed as ‘uncommon’ at best, while south-western habitats where birds are only present in mid-winter will be listed as ‘common.’ However, the situation will be reversed if we consider the time birds spend in these habitats.

The reviewer also highlights the introduction’s unconventional structure and information redundancy at the beginning. We have chosen this structure and provided basic explanations to improve readability for a wider audience, given eLife’s readership. At the same time, we will certainly take the reviewers’ feedback into account in the revised version. We plan to include the references to modern itinerancy research mentioned above and to add a section on itinerancy to the Discussion.

We appreciate the reviewer’s input and sincerely thank them for their time and effort in reviewing our paper. While we may not fully agree on the classification of the behavior we describe, we value the opportunity to engage in discussion and believe that presenting arguments and counterarguments to the reader is beneficial to scientific progress.

The following is the authors’ response to the original reviews.

Reviewer #1 (Recommendations For The Authors):

I much enjoyed reading this manuscript, that is, once I understood what it is about. Titles like "Conserving bird populations in the Anthropocene: the significance of non-breeding movements" are a claim to so-called relevance, they have NOTHING to do with the content of the paper, so once I understood that this paper was about the "Quick quick slow: the foxtrot migration of rough-legged buzzards is a response to habitat and snow" (an alternative title), it was becoming very interesting. So the start of the abstract as well as the introduction is very tedious, as clearly much trouble is taken here to establish reputability. In my eyes this is unnecessary: eLife should be interested in publishing such a wonderful description of such a wonderful migrant in a study that comes to grips with limiting factors on a continental scale!

We sincerely appreciate your time and effort in reviewing our manuscript. Thank you for your appreciation of our study.

We agree that the focus of the article should be changed from conservation to migration patterns. We have rewritten the Introduction and Discussion as suggested. We have added the application of this pattern including conservation at the end of the Discussion by completely changing Figure 5. We have also changed the title to the suggested one.

Not sure that the first paragraph statements that seek to downplay what we know about wintering vs breeding areas are valid (although I see what purpose they serve). Migratory shorebirds have extensively been studied in the nonbreeding areas, for example, including movement aspects (see, as just one example, Verhoeven, M.A., Loonstra, A.H.J., McBride, A.D., Both, C., Senner, N.R. & Piersma, T. (2020) Migration route, stopping sites, and non breeding destinations of adult Black tailed Godwits breeding in southwest Fryslân, The Netherlands. Journal of Ornithology 162, 61-76) and there are very impressive studies on the winter biology of migrants across large scale (for example in Zwarts' Living on the Edge book on the Sahel wetlands). Think also about geese and swans and about seabirds!

We have rewritten the first paragraph and it now talks about patterns of migratory behavior. We have also rewritten the second paragraph, now it is devoted to studies of movements in the non-breeding period. We explain how our pattern differs from those already studied and give references to the papers you mentioned.

Directional movements in nonbreeding areas as a function of food (in this case locusts) have really beautifully been described by Almut Schlaich et al in JAnimEcol for Montagu's harriers.

We have added Montagu's harrier example in the second paragraph of the Introduction and the Discussion. We have added a reference to Schlaich and to Garcia and Arroyo, who suggested that Montagu's harriers have long directional migrations during the non-breeding period.

Once the paper starts talking buzzards, and the analyses of the wonderful data, all is fine. It is a very competent analysis with a description of a cool pattern.

Thank you for your appreciation of our study. We hope the revised version is better and clearer.

However, i would say that it is all a question of spatial scale. The buzzards here respond to changes in food availability, but there is not an animal that doesn't. The question is how far they have to move for an adequate response: in some birds movements of 100s of meters may be enough, and then anything to the scale of rough-legged buzzards.

In the new version of the manuscript, we emphasize that this is a large distance (about 1000 km), comparable to the distance of the fall and spring migrations (about 1400 km) in lines 70-72 of the Introduction and 379-383 of the Discussion.

And actually, several of the shorebirds I know best also do a foxtrot, such as red knots and bar-tailed godwits moulting in the Wadden Sea, then spending a few months in the UK estuaries, before returning to the Wadden Sea before the long migrations to Arctic breeding grounds. The publication of the rough-legged buzzard story may help researchers to summarize patterns such as this too. Mu problem with this paper is the framing. A story on the how and why of these continental movements in response to snow and other habitat features would be a grand contribution. Drop Anthropocene, and rethink whether foxtrot should be introduced as a hypothesis or a summary of cool descriptions. I prefer the latter, and recommend eLife to go with that too, rather than encourage "disconnected frames that seek 'respectability'" Good luck, theunis piersma

We thank the reviewer again for his valuable comments and suggestions. We have changed the framing to the suggested one and removed the Anthropocene from the article.

Reviewer #2 (Recommendations For The Authors):

We sincerely appreciate the time and effort you have taken to review our manuscript. We have carefully considered all of your comments, including both public and author comments, and provided detailed responses to each of them below. In addition, we would like to address the most important public comments.

We agree with the suggestion to shift the focus of the article from conservation to migration patterns. Accordingly, we have rewritten both the Introduction and Discussion sections to focus on migration behavior rather than conservation.

However, we respectfully disagree with the suggestion that the migration patterns we describe are synonymous with itinerancy. We acknowledge that our original presentation may have been unclear and may have hindered full understanding. In the revised version, we provide a detailed analysis of migratory behavior in the Introduction that describes how our pattern differs from itinerancy. We also revisit this distinction in the Discussion section. We have also carefully revised Figure 1 to improve clarity and avoid potential misunderstandings.

Regarding the applicability of the described migration pattern, we acknowledge that the Rough-legged Buzzard is not listed as an endangered species. However, we believe that our findings have practical implications. We have moved our discussion of this issue to the end of the Discussion section and have completely revised Figure 5. While the overall population of Rough-legged Buzzards is not declining, certain regions within its range are experiencing declines. We show that this decline does not warrant listing the species as endangered. Instead, it may represent a redistribution within the non-breeding range - a shift in range dynamics. We use the example of the Rough-legged Buzzard to illustrate this concept and emphasize the importance of considering such dynamics when assessing the conservation status of species in the future.

We also acknowledge that the hypothesis of this form of behavior has been proposed previously for Montagu's Harrier, and we have included this information in the revised manuscript. In addition, we agree that the focus on the Anthropocene is unnecessary in this context and have therefore removed it.

We believe that these revisions significantly improve the clarity and robustness of the manuscript, and we are grateful for your insightful comments and suggestions.

As a general comment, please note that including line numbers (as it is the standard in any manuscript submission) would facilitate reviewers providing more detailed comments on the text.

We apologize for this oversight and have added line numbers to our revised manuscript.

Dataset: unclear what is the frequency of GPS transmissions. Furthermore, information on relative tag mass for the tracked individuals should be reported.

We have included this information in our manuscript (L 157-163). We also refer to the study in which this dataset was first used and described in detail (L 164).

Data pre-processing: more details are needed here. What data have been removed if the bird died? The entire track of the individual? Only the data classified in the last section of the track? The section also reports on an 'iterative procedure' for annotating tracks, which is only vaguely described. A piecewise regression is mentioned, but no details are provided, not even on what is the dependent variable (I assume it should be latitude?).

Regarding the deaths. We only removed the data when the bird was already dead. We have corrected the text to make this clear (L 170).

Regarding the iterative procedure. We have added a detailed description on lines 175-188.

Data analysis: several potential issues here:

(1) Unclear why sex was not included in all mixed models. I think it should be included.

Our dataset contains 35 females and eight males. This ratio does not allow us to include sex in all models and adequately assess the influence of this factor. At the same time, because adult females disperse farther than males in some raptor species, we conducted a separate analysis of the dependence of migration distance on sex (Table S8) and found no evidence for this in our species. We have written a separate paragraph about this. This paragraph can be found on lines 356-360 of the new manuscript.

(2) Unclear what is the rationale of describing habitat use during migration; is it only to show that it is a largely unsuitable habitat for the species? But is a formal analysis required then? Wouldn't be enough to simply describe this?

Habitat use and snow cover determine the two main phases (quick and slow) of the pattern we describe. We believe that habitat analysis is appropriate in this case and that a simple description would be uninformative and would not support our conclusions.

(3) Analysis of snow cover: such a 'what if' analysis is fine but it seems to be a rather indirect assessment of the effect of snow cover on movement patterns. Can a more direct test be envisaged relating e.g. daily movement patterns to concomitant snow cover? This should be rather straightforward. The effectiveness of this method rests on among-year differences in snow cover and timing of snowfall. A further possibility would be to demonstrate habitat selection within the entire non-breeding home range of an individual in relation snow cover. Such an analysis would imply associating presence-absence of snow to every location within the non-breeding range and testing whether the proportion of locations with snow is lower than the proportion of snow of random locations within the entire non-breeding home range (95% KDE) for every individual (e.g. by setting a 1/10 ratio presence to random locations).

The proposed analysis will provide an opportunity to assess whether the Rough-legged Buzzard selects areas with the lowest snow cover, but will not provide an opportunity to follow the dynamics and will therefore give a misleading overall picture. This is especially true in the spring months. In March-April, Rough-legged Buzzards move northeast and are in an area that is not the most open to snow. At this time, areas to the southwest are more open to snow (this can be seen in Figure 4b). If we perform the proposed analysis, the control points for this period would be both to the north (where there is more snow) and to the south (where there is less snow) from the real locations, and the result would be that there is no difference in snow cover.

A step-selection analysis could be used, as we did in our previous work (Curk et al 2020 Sci Rep) with the same Rough-legged Buzzard (but during migration, not winter). But this would only give us a qualitative idea, not a quantitative one - that Rough-legged Buzzards move from snow (in the fall) and follow snowmelt progression (in the spring).

At the same time, our analysis gives a complete picture of snow cover dynamics in different parts of the non-breeding range. This allows us to see that if Rough-legged Buzzards remained at their fall migration endpoint without moving southwest, they would encounter 14.4% more snow cover (99.5% vs. 85.1%). Although this difference may seem small (14.4%), it holds significance for rodent-hunting birds, distinguishing between complete and patchy snow cover. Simultaneously, if Rough-legged Buzzards immediately flew to the southwest and stayed there throughout winter, they would experience 25.7% less snow cover (57.3% vs. 31.6%). Despite a greater difference than in the first case, it doesn't compel them to adopt this strategy, as it represents the difference between various degrees of landscape openness from snow cover.

We write about this in the new manuscript on lines 385-394.

Results: it is unclear whether the reported dispersion measures are SDs or SEs. Please provide details.

For the date and coordinates of the start and end of the different phases of migration, we specified the mean, sd, and sample size. We wrote this in line 277. For the values of the parameters of the different phases of the migration (duration, distance, speed, and direction), we used the mean, the standard error of the mean, and the confidence interval (obtained using the ‘emmeans’ package). We have indicated this in lines 302-303 and the caption of Table 1 (L 315) and Figure 2 (L 293-294). For the values of habitat and snow cover experienced by the Rough-legged Buzzards, we used the mean and the error of the mean. We reported this on lines 322 and 337 and in Figures 3 (L 332-333) and 4 (L 355-356).

Discussion: in general, it should be reshaped taking into account the comments. It is overlong, speculative and quite naive in several passages. Entire sections can be safely removed (I think it can be reduced by half without any loss of information). I provide some examples of the issues I have spotted below. For instance, the entire paragraph starting with 'Understanding....' is not clear to me. What do you mean by 'prohibited management' options? Without examples, this seems a rather general text, based on unclear premises when related to the specific of this study. Some statements are vague, derive from unsubstantiated claims, and unclear. E.g. "Despite their scarcity in these habitats, forests appear to hold significant importance for Rough-legged buzzards for nocturnal safety". I could not find any day-night analysis showing that they actually roost in forests during nighttime. Being a tundra species, it may well be possible that rough-legged buzzards perceive forests as very dangerous habitats and that they prefer instead to roost in open habitats. Analysing habitat use during day and night during the non-breeding period may be of help to clarify this. Furthermore, considering the fast migration periods, what is the flight speed during day and night above forests? Do these birds also migrate at night or do they roost during the night? Perhaps a figure visualizing day and night track segments could be of help (or an analysis of day vs. night flight speed) (there are several R packages to annotate tracks in relation to day and night). This is an example of another problematic statement: "The progression of snow cover in the wintering range of Rough-legged buzzards plays a significant role in their winter migration pattern." The manuscript does not contain any clear demonstration of this, as I wrote in my previous comments. Without such evidence, you must considerably tone down such assertions. But since providing a direct link is certainly possible, I think that additional analyses would clearly strengthen your take-home message.

The paragraph starting with "The quantification of environmental changes that could prove fatal to bird species presents yet another challenge for conservation efforts in an era of rapid global change." is quite odd. Take the following statement "For instance, the presence of small patches of woodland in the winter range might appear crucial to the survival of the Rough-legged buzzard. Elimination of these seemingly minor elements of vegetation cover through management actions could have dire consequences for the species.". It is based on the assumption that minor vegetation elements play a key role in the ecology of the species, without any evidence supporting this. Does it have any sense? I could safely say exactly the opposite and I would believe it might even be more substantiated.

We agree with these comments.

We have completely rewritten this section. As suggested, we have shortened it by removing statements that were not supported by the research. We have completely removed the statements about "prohibited management". We have also removed the statement that "forests appear to be of significant importance to Rough-legged buzzards for nocturnal safety" and everything associated with that statement, e.g. the statement about "small elements of vegetation cover", etc. We do believe that this statement is true in substance, but we also agree that it is not supported by the results and requires separate analysis. At the same time, we believe that this is a topic for a separate study and would be redundant here. Therefore, we leave it for a separate publication.

Conclusion paragraph: I believe this severely overstates the conservation importance of this study. That the results have "crucial implications for conservation efforts in the Anthropocene, where rapidly changing environmental factors can severely impact bird migration" seems completely untenable to me. What is the evidence for such crucial implications? For instance, these results may suggest that climate change, because global warming is predicted to reduce snow cover in the non-breeding areas, might well be beneficial for populations of this species, by reducing non-breeding energy expenditure and improving non-breeding survival. I think statements like these are simply not necessary, and that the study should be more focused on the actual results and evidence provided.

We have completely rewritten this section. We removed the reference to the Anthropocene and focused on migratory behavior and migration patterns.