Density-Dependent and Predator-Specific Nest Defense Strategies in Colonially Breeding Saunders’s Gulls

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

In the manuscript “Density-Dependent and Predator-Specific Nest Defense Strategies in a Colonially Breeding Seabirds” the authors describe the nest defence behaviour of the Saunders’s Gulls Saundersilarus saundersi in Southeast Korea. Using taxidermy dummies of the Oriental Magpie Pica serica as a diurnal avian nest predator and the Common Raccoon Dog Nyctereutes procyonoides as a nocturnal mammalian predator,  the authors demonstrated the responses of Saunders's Gulls to the presence of predators in two colonies of different sizes.

The research is important and valuable as the Saunders’s Gull is an endangered species whose population is decrease due to anthropogenic impacts. The hypothesis is well presented and addressed in the Introduction, while the Methodology and the Discussion are clear.

However, in section 2.2. Behavioral Data Collection the authors describe the size of two colonies and the number of nests. However, However, it is unclear whether these were mixed colonies with the Little Tern which were mentioned as a non-predatory control species. If that was the case, please state this explicitly in the text, along with the number of breeding Little Tern pairs..

It is also unclear over how many years the research was conducted. If the study was based on only one year of data, this would be insufficient as the statistical  significance would be low. Additional years of study could reveal different patterns, such as variations in the predation rates of the examined species. Furthermore, it should be clarified whether the birds left the colony after breeding or whether they returned to breed again in subsequent years.

Overall, however, I recommend the paper for publication because it provides valuable evidence that the behaviour of Saunders's Gulls in defending their nests is influenced by both social and ecological factors. However, longer-term studies are essential to confirm these patterns and inform more effective conservation strategies for this endangered species. Please find the attached file to see more comments.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Birds-3954231- Yoon et al. explored the nest defense behavior variation in Saunders’s gulls with nest density and predator type using decoy experiments and mixed-model analysis. The research is conceptually sound and ecologically meaningful and is interesting to the readers of the journal Birds. This manuscript directly tests two biologically meaningful factors (density and predator type) influencing nest defense behavior. It is generally well written and understandable. However, there needs clarifications in the research design and statistical treatment to make it more readable and replicable.

The major issues in the manuscript are-
The use of term ‘Seabirds’ is vague in the title. Since the manuscript deals with one species, it should be reflected in the title itself. So, the title could be- Density-Dependent and Predator-Specific Nest Defense Strategies in a Colonially Breeding Saunder’s Gulls. 
Only two plots (one high-density and one low-density) were used, which confounds “density” with site effects. Any difference attributed to density could instead reflect unmeasured site differences (e.g., microhabitat, predator history, human disturbance). I recommend the authors treat density as a continuous variable across multiple colonies or plots in future work or at least include multiple replicates per density level.
At what time of the day were the experiments done? Since you have one diurnal and another nocturnal predator being tested, time of experimentation also matters. Please mention with justification for the same. 
The total sample size is very small (n = 33 trials; you have written 36 trials in methods, but results are shown for 17+16=33 trials only), with repeated measures from the same colony. Individual gulls likely participated in multiple trials, leading to non-independence of observations. Although “site” was modeled as a random effect, this does not adequately capture within-group autocorrelation. I think it is better to use generalized linear mixed models (GLMMs) with trial identity or observation point as a nested random factor to properly account for repeated measures and spatial autocorrelation.
Besides tested variables, other environmental variables such as weather, time of day, visibility, and breeding stage can strongly influence antipredator responses but were not controlled or recorded. What could be the effects of those variables? Need to discuss, if not included in the experimentation. 
The PCA integrates three behavioral variables (latency, attack rate, mobbing individuals), but the dataset is tiny (n=33). PCA with such a small sample is statistically unstable. I recommend the authors report eigenvalues, % variance explained, and loadings explicitly. Validate PCA using correlation matrix adequacy (KMO, Bartlett test) or avoid PCA altogether given small n.
Latency and attack rates are time/count data, which are often non-normal and heteroscedastic. There is no mention of normality tests, transformations (e.g., log), or use of appropriate distributions (e.g., Poisson, negative binomial). Reanalyze using GLMMs with appropriate error distributions and log or inverse-link functions.
The PCA composite score is interpreted as an overall “defense behavior index,” but its biological meaning depends on variable scaling and correlation structure. The authors should treat PCA-derived results as exploratory; avoid causal inference without validating loadings and cross-validation.  
I have provided some other minor issues in the annotated PDF. I hope they will help the authors revise the manuscript. All the best!  

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This study investigates the effects of (breeding) density and predator type on the (collective) nest defence behaviour of a vulnerable colonial gull species breeding on reclaimed tidal flats in the Songdo reclamation complexes, Incheon, South-Korea. The results are in itself interesting and discussed with regard to (adaptive collective) anti-predator responses and conservation policies. The manuscript is concise and overall well-written. I do have some comments that will hopefully improve the manuscript.

General comments

First and foremost, higher collective nest defence (i.e. higher absolute number of mobbing individuals) at the higher density plot could be, due to the correlative nature of the results, simply be a byproduct of the higher number of individuals present, rather than a spatial (nest/breeding) density effect per se. To establish whether it actually results in higher collective nest defence, I would argue that you should express the number of mobbing individuals relative to (i.e. controlled for) number of nests within a certain predefined radius/area around the point of stimulus presentation. Hence, it would be worthwhile and informative to express the number of mobbing individuals proportional, and assess relative differences in collective nest defence between low- and high-density plots. In other words, these results would give insight into whether, relative to the number of nests (and hence potential total individuals present), collective mobbing differs between plots. This should at least be explored.  

Second, the racoon dog is a nocturnal predator, but if I understood correctly, it was presented during daytime. What is the biologically relevance of this predator in the colony during daylight? In other words, how often are racoon dogs observed in the colony during daytime and do they also prey upon eggs/chicks during the day? If they do not, what is the rationale behind its presentation during daytime and why did you not opt for one of the diurnal mammalian predators? This requires more explicit explaining in my opinion.  

Third, you did not use replication of density-plots (i.e. not multiple high- and multiple low-density plots), which limits the generalizability of results. Especially in the discussion, care should therefore be taken when discussing the ramifications of these results. I notice that the authors repeatedly discuss results in terms of “colonies/sites” while they only considered one high- and one low-density plot within the same colony. Several sentences should be re-phrased as to not overstate results of the current study (see e.g. L223-226, L250-252, L302-306).  

Fourth,  breeding dispersal (after high nest predation) is postulated as one of the research goals of the current study (L68), yet these results are not presented. Instead, this was examined by a previous study (Yoon et al 2018). Although this is appropriate work to cite and discuss, it should not be postulated as one of the aims of this study, given it is not the focus.

More specific comments

Title: change "seabirds" to "seabird"

L44: typo; change to “many eyes effect”,

L83: remove the “and” to improve readability, i.e. “lay 2-4 eggs, the sexes incubate…”.  

L89: consider re-phrasing, as currently it is difficult to read. Suggestion: “… allowing incursions by multiple predators, including both avian (Oriental Magpies, Pica serica; Large-billed Crows, Corvus macrorhynchos) and mammalian predators (Common Raccoon Dogs, Nyctereutes procyonoides; Siberian Weasels, Mustela sibirica; feral dogs and cats, Canis and Felis spp.).

L95: please re-phrase as you do not consider a density-“gradient” but instead use two single plots within the same colony differing in local breeding/nest density.

L95-99: sentence reads difficult, consider splitting it up.

L105: did you count the number of individuals or the number of nests? Also, I guess Drone imagery was done during the incubation period, which might be worth to explicitly mention.

L106-108: what is the actual distance between the high- and low-density plot? Please provide this information.

L148-149: Please provide more specific information (e.g. mean, SD) on the time-interval between different predator type presentations. Also, please provide results for the (absence of an) effect of presentation order (i.e. sequence) of predator types on your behavioural metrics.

L156: it is not clear what the random effect “site” entails given it is the first time this is mentioned. See also L30 in the abstract, what does “sites” mean?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

Thank you for revising the manuscript following the suggestions and clarifying the issues raised. I believe the manuscript is much improved and can be considered for publication in Birds. 

All the best!