Mixed-Species Flock Diversity and Habitat Density Are Associated with Antipredator Behavior in Songbirds

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The aim of the study was to investigate the anti-predatory behaviour in mixed species foraging groups and the effect of traffic noise, vegetation and social composition. The author used an established experimental paradigm of presenting a predator model in close proximity of a feeder used by foraging mixed species groups. The authors found that the anti-predatory behaviour of some species were influenced by social factors, whereas in other species vegetation density was the most influential factor.

I have some concerns about the study. First, I assumed that one of the main focus, perhaps even the major one, of the study was to investigate how human noise impact anti-predatory strategies in mixed-species groups. This is stated in the very first sentences of the abstract. This is actually an interesting question, whereas other dynamics of mixed-species groups are quite well understood already (Sridhar et al 2009 Anim Behav; Sridhar and Shanker 2013 Behav Ecol Sociobiol). Yet, the impact of human noise (or lack of) turned out to be only a minor focus of the study and the authors focused on already well established dynamics in mixed-species group foraging startegies. Secondly, there are several points regarding the methods and statistical analysis that are unclear and need a much greater clarification. Please find below my detailed comments.  

 

Lines 140-156: why such a time gap between measuring noise level, collecting Lidar data for the vegetation and performing the behavioural trials? For the noise levels you compare three years of data, which is ok and I understand you used the average value of the three measurements at each site. However, the Lidar data was collected is different periods, so can you also confirm that there was no significant time-related difference?

Lines 209-214: sounds like the observer had to place the predator model on the mount while birds were foraging on the feeder. This most likely scare the birds away disrupting their foraging, so how can you exclude that this disturbance has not biased your results?

Line 240-244: what do you mean “For both experiments”? I only see one, i.e. the presentation of the hawk model. Why did you not include the feeding station in your model? In my view this parameter should be included at least as a random term in a GLMM. Did you check the fit of your linear model? I also assume that the birds were not banded, so it is not clear to me how you counted the numbers of conspecifics and flock size.

Line 244-247: there is no P in the formula, please correct

Line 250-251: I do not understand which variables you used in your PCA. In general I also do not understand the need to run a PCA at all on your dataset. Please clarify.

Line 260-270: I do not understand what are these PC components supposed to indicate and which variables were used in the PCA analysis. Also, it makes no sense that calling and seed-taking latencies and rates load onto your PC axes because, as you wrote in the previous paragraph, these are your dependent variables. Please, clarify.

Line 273: but how did you calculated the PCA? Where is the interpretation of your PC axes? Also, here and throughout, results should be presented either in the text or in the table, there is no need to repeat these.

Table 1: I do not understand this table. Looks like the variables are the ones you used in your linear models, which is ok. But what does the grouping of PC scores mean? How did you calculate these PC scores and what do they mean?

Figure 2, 3 and 4: what does “positive” or “negative” relationship or “tendency for an association” even mean? You should include regression lines (in Figure 2 seems a logarithmic is a better fit than linear) with R values and statistical testing. The value in Figure seems heavily biased on 2 nuthatches per feeder and generally looks there is no real difference.

Lines 258-334: the entire result section should be condensed in a single (and short) paragraph reporting all results, no need to make so many paragraphs

Line 336-339: this belongs to introduction or methods

Lines 346-422: in your abstract you stated that “we know little about how such human-induced changes impact these behaviors in the context of mixed-species groups. To address this gap we examined the effects of mixed-species group composition, traffic noise, and vegetation density on antipredator and foraging behavior.” Yet, this aspect is barely mentioned in the discussion, with the attention focused on the leader-follower role of species in mixed groups and the effect of vegetation. You should definitely expand on this as it is the only interesting aspect of your study, otherwise you should completely change the tone of the main focus of your investigation. 

Author Response

We sincerely thank the editors and the two reviewers for their helpful comments and criticisms of our manuscript. We have thoroughly revised the manuscript in light of these comments. Below, we have copied-and-pasted the entire set of reviewer comments and, in text begun with -->, indicate how we responded to the queries and comments. Additionally, in our responses one can find the line numbers that related to the revised text in the manuscript that aimed to answer the reviewers.

If we can provide any additional information, please let me know, and thank you again!

-Todd Freeberg

 

Reviewer 1

REV DIVERSITY Very interesting manuscript (ms) focused on associations among anti-predator behavioural patterns in mixed-species flock of birds in relation to environmental constarints (e.g., habitat density).

Authors are Mother Tongue in English: therefore style and language is, obviously, good. I read it with interest. Sampling design is very representative and data reliable. I think that this ms could be largely read and citable, therefore I think that it deserves to be published on Diversity after MINOR REVISIONS. I have only some minor points to suggest. First, references are limited for a so large topic: please add some further references. Second, some words should be added about the type of forest management affecting the studied habitats: the local history could be important to explain habitat density and, therefore, the behavioural patterns. Third, see about the possible role of intermediate disturbance effects on birds (about vegetation cover). Fourth, what about individual personality in single individual in the flocks? However, a good ms. Here below I have added monir comments and suggestion about these points. Have a nice work and compliments.

-->Thank you for the helpful comments on the manuscript! To address the points raised above, we have added nine additional references (see also comments below). Secondly, we added some text to discuss forestry practices at our study site (lines 158-165). For the third point, we are not sure what the reviewer is asking about beyond the forestry practices just mentioned or the alpha volume habitat density measurement that is in the analyses in the manuscript already. Our apologies if we are missing something obvious! For the fourth point, we are unable to state anything about individuals in this study as we were working almost exclusively with non-color-banded individuals. We raise this as a potential question to address in future work in lines 395-400.

 

POINTS Row 80-81. Interesting the terms ‘core’ and ‘satellite’ applied to species. You should exmplain a bit these concepts.

-->We have added text and new references in lines 82-84 to define these roles in a bit more detail.  

Row 133 and everywhere. ‘spp.’ not in italic.

-->Thank you for catching this! Corrected in lines 136 of the revised manuscript.

Study area and rows 155-166 (forest structure). You should specify if these forests are managed (for example, with coppice management). In this case the change in forest vegetation pattern could affect your data? Density and richness of tit species in managed forests is strongly affected by patterns in forest management: see Ornis Fennica, 93(2), 100-110 (2016). Perhaps some words about the local history of these woods should be added.

-->The reviewer raises an important issue lacking in our original submission. We have added some more detail on the forest areas where the work was conducted, and include this reference, in lines 158-165.

 

 ‘Tufted titmice showed higher feeder avoidance (higher latencies to take seeds and lower rates of seed taking) at sites with higher vegetation density‘. Regarding vegetation density in the surrounding of your seed plots, is it possible an intermediate disturbance effects (review in Annales Zoologici Fennici (Vol. 57, No. 1-6, pp. 67-78, 2020)? For example, changing the vegetation density, both completely exposed situations (low vegetation density) and completely 'closed' situations (high vegetation density) may be less suitable than situations of intermediate vegetation density? See the Springer book about disturbance ecology at https:// link.springer.com/book/10.1007/978-3-319-32476-0.

-->We suspect almost all of our feeder sites occupy a relatively intermediate level of disturbance. As mentioned above regarding the additional detail we have added about the forest areas where the work was done, all of our feeders are located in areas of at least secondary succession – none of the feeders exists in dense primary succession or completely open grasslands outside of forest. We have added these references in some text we added to the discussion as potential future directions to take this kind of research (lines 429-434).

Fig. 1. Add the author of the photo. In conclusions, I suggest to add at least a sentence about the possible role of individual animal personality inside flocks as factor that should be investigated in the future. See, for example: Behaviour, 142: 1149-1157 (2005) and Current Zoology, 57(4), 491-498 (2011) and the seminal handbook of Carere, C., & Maestripieri, D. (Eds.). (2013). Animal personalities: behavior, physiology, and evolution. University of Chicago Press.

-->As mentioned in response to the initial paragraph of the reviewer, we added some text on the personality question in lines 395-400 in the discussion. We also added the photograph information to the Figure 1 caption (line 220).

Add the role of anonymous reviewers and Editors in the Acknowledgments.

-->Done – line 485! Many thanks again for all these helpful suggestions.

Reviewer 2 Report

Comments and Suggestions for Authors

REV DIVERSITY
Very interesting manuscript (ms) focused on associations among anti-predator behavioural patterns in mixed-species flock of birds in relation to environmental constarints (e.g., habitat density). Authors are Mother Tongue in English: therefore style and language is, obviously, good. I read it with interest. Sampling design is very representative and data reliable. I think that this ms could be largely read and citable, therefore I think that it deserves to be published on Diversity after MINOR REVISIONS. I have only some minor points to suggest. First, references are limited for a so large topic: please add some further references. Second, some words should be added about the type of forest management affecting the studied habitats: the local history could be important to explain habitat density and, therefore, the behavioural patterns. Third, see about the possible role of intermediate disturbance effects on birds (about vegetation cover). Fourth, what about individual personality in single individual in the flocks? However, a good ms. Here below I have added monir comments and suggestion about these points. Have a nice work and compliments.
POINTS
Row 80-81. Interesting the terms ‘core’ and ‘satellite’ applied to species. You should exmplain a bit these concepts.
Row 133 and everywhere. ‘spp.’ not in italic.
Study area and rows 155-166 (forest structure). You should specify if these forests are managed (for example, with coppice management). In this case the change in forest vegetation pattern could affect your data? Density and richness of tit species in managed forests is strongly affected by patterns in forest management: see Ornis Fennica, 93(2), 100-110 (2016). Perhaps some words about the local history of these woods should be added.
‘Tufted titmice showed higher feeder avoidance (higher latencies to take seeds and lower rates of seed taking) at sites with higher vegetation density‘. Regarding vegetation density in the surrounding of your seed plots, is it possible an intermediate disturbance effects (review in Annales Zoologici Fennici (Vol. 57, No. 1-6, pp. 67-78, 2020)? For example, changing the vegetation density, both completely exposed situations (low vegetation density) and completely 'closed' situations (high vegetation density) may be less suitable than situations of intermediate vegetation density? See the Springer book about disturbance ecology at https:// link.springer.com/book/10.1007/978-3-319-32476-0.
Fig. 1. Add the author of the photo.
In conclusions, I suggest to add at least a sentence about the possible role of individual animal personality inside flocks as factor that should be investigated in the future. See, for example:  Behaviour, 142: 1149-1157 (2005) and Current Zoology, 57(4), 491-498 (2011) and the seminal handbook of Carere, C., & Maestripieri, D. (Eds.). (2013). Animal personalities: behavior, physiology, and evolution. University of Chicago Press.
Add the role of anonymous reviewers and Editors in the Acknowledgments.
Have a nice work.

 

Author Response

The aim of the study was to investigate the anti-predatory behaviour in mixed species foraging groups and the effect of traffic noise, vegetation and social composition. The author used an established experimental paradigm of presenting a predator model in close proximity of a feeder used by foraging mixed species groups. The authors found that the anti-predatory behaviour of some species were influenced by social factors, whereas in other species vegetation density was the most influential factor.

I have some concerns about the study. First, I assumed that one of the main focus, perhaps even the major one, of the study was to investigate how human noise impact anti-predatory strategies in mixed-species groups. This is stated in the very first sentences of the abstract. This is actually an interesting question, whereas other dynamics of mixed-species groups are quite well understood already (Sridhar et al 2009 Anim Behav; Sridhar and Shanker 2013 Behav Ecol Sociobiol). Yet, the impact of human noise (or lack of) turned out to be only a minor focus of the study and the authors focused on already well established dynamics in mixed-species group foraging startegies. Secondly, there are several points regarding the methods and statistical analysis that are unclear and need a much greater clarification. Please find below my detailed comments.

-->Traffic noise at the forest areas where we work has been – and will continue to be – of interest to us at it varies so much across the 36 feeders we have. We are actually surprised that traffic noise seems to be of so little effect on foraging and anti-predatory behavior of the birds at our feeders. We added some text in the Discussion (lines 435-442) to address this point. We thank the reviewer also for pointing out these references – we have added the Sridhar & Shanker reference to the text in lines 46 and 454-457 and the Sridhar et al. 2009 reference to the text in lines 82-84.

 

Lines 140-156: why such a time gap between measuring noise level, collecting Lidar data for the vegetation and performing the behavioural trials? For the noise levels you compare three years of data, which is ok and I understand you used the average value of the three measurements at each site. However, the Lidar data was collected is different periods, so can you also confirm that there was no significant time-related difference?

-->There has been no substantial change to the forest areas where our feeders are since the original Lidar data were collected – no forestry practices or logging, no forest road alterations, and no major storms that produced tree fall. We therefore have no reason to believe that the Lidar metrics for these 36 sites will have changed much, perhaps especially because our studies were done in the overwintering period when at least the deciduous trees were leafless.

Lines 209-214: sounds like the observer had to place the predator model on the mount while birds were foraging on the feeder. This most likely scare the birds away disrupting their foraging, so how can you exclude that this disturbance has not biased your results?

-->For practical reasons, we have used this method of walking up to a feeder to place a visual stimulus in a series of field experiments over the decades. We have now cited some of these studies in lines 227-230 to make the point that the simple act of walking up to a feeder and then returning to an observation point 10-15 m away has minimal influence on birds’ foraging or calling behavior. Additionally, all the feeders and flocks in this study were presented with the same method, so we don’t have a reason to believe that we biased the birds’ behavior.

Line 240-244: what do you mean “For both experiments”? I only see one, i.e. the presentation of the hawk model. Why did you not include the feeding station in your model? In my view this parameter should be included at least as a random term in a GLMM. Did you check the fit of your linear model? I also assume that the birds were not banded, so it is not clear to me how you counted the numbers of conspecifics and flock size.

-->We are grateful to the reviewer for catching the “both experiments” mistake – we did indeed only do one experiment! We did not include the feeder as a random effect as we had only one data collection period for each feeder. We assessed fit of the linear models by looking at the P-P plots of residuals and also with one sample Kolmogorov-Smirnov tests of residuals against the normal – normality was confirmed in all cases (lines 285-287). We expanded on a point we raised in the original manuscript to make clearer how we determined the number of birds given the birds were unbanded – see lines 211-214.

Line 244-247: there is no P in the formula, please correct

-->This also was an error in the original manuscript – we have corrected the equation in the revised manuscript in lines 261-262.

Line 250-251: I do not understand which variables you used in your PCA. In general I also do not understand the need to run a PCA at all on your dataset. Please clarify.

-->We carried out PCA to reduce the dimensionality of the data set. Instead of carrying out 12 regression models (four DVs [seed-taking latencies and rates and calling latencies and rates] * 3 species), the PCA reduced the data set down to 5 comparisons (one for chickadees and two each for titmice and nuthatches). The predictor variables that went into the PCA analyses were the three social variables and two physical environment variables. We added text in lines 266-267 to make these points clearer.

Line 260-270: I do not understand what are these PC components supposed to indicate and which variables were used in the PCA analysis. Also, it makes no sense that calling and seed-taking latencies and rates load onto your PC axes because, as you wrote in the previous paragraph, these are your dependent variables. Please, clarify.

-->We hope the added text mentioned above addresses this point as well.

Line 273: but how did you calculated the PCA? Where is the interpretation of your PC axes? Also, here and throughout, results should be presented either in the text or in the table, there is no need to repeat these.

-->edited the text in lines 268-281to try to make it clearer what these PC scores were and to reduce the redundancy in the manuscript.

Table 1: I do not understand this table. Looks like the variables are the ones you used in your linear models, which is ok. But what does the grouping of PC scores mean? How did you calculate these PC scores and what do they mean?

-->As in our earlier studies, we sought to assess how these five physical and social environmental variables were associated with our behavioral measures (in this study, the PC scores). This table shows the extent to which the environmental variables ‘predicted’ the behavioral responses to the predator model. We hope the changes to the text that we raised above address this concern.

Figure 2, 3 and 4: what does “positive” or “negative” relationship or “tendency for an association” even mean? You should include regression lines (in Figure 2 seems a logarithmic is a better fit than linear) with R values and statistical testing. The value in Figure seems heavily biased on 2 nuthatches per feeder and generally looks there is no real difference.

-->By “positive” association we mean that as one variable increases (e.g., Forest Alpha Volume), the other variable increases (in this case, the PC1 score for titmice). We did not test for non-linear relationships in these data as we wanted to use the same statistical approaches we used in our earlier published studies on mixed-species flocks at these 36 feeders. Finally, nuthatch group size variation is much smaller than that for chickadees or titmice – nuthatches rarely form groups beyond the female-male pair in the overwintering months. We occasionally see groups of 3 (3 instances in this study) and in one study have seen a group of 4, but the majority of cases are just the pair or, less commonly, one member of the pair. This might explain the reviewer’s comment related to nuthatches.

Lines 258-334: the entire result section should be condensed in a single (and short) paragraph reporting all results, no need to make so many paragraphs

-->believe the text in our Results section is already quite condensed and concise, and so are not sure what the reviewer is advocating here. If the Editor prefers, we could get rid of the three species subsections and put the text all into a single paragraph.

Line 336-339: this belongs to introduction or methods

-->We revised the opening paragraph of the Discussion section to address this comment – see lines 353-363.

Lines 346-422: in your abstract you stated that “we know little about how such human-induced changes impact these behaviors in the context of mixed-species groups. To address this gap we examined the effects of mixed-species group composition, traffic noise, and vegetation density on antipredator and foraging behavior.” Yet, this aspect is barely mentioned in the discussion, with the attention focused on the leader-follower role of species in mixed groups and the effect of vegetation. You should definitely expand on this as it is the only interesting aspect of your study, otherwise you should completely change the tone of the main focus of your investigation.

-->We have altered the text in this section to minimize the discussion of variation in ‘leader-follower’ roles and to circle back more obviously to the tone of the abstract. See especially lines 450-457 and 458-463.