Life History and Sociality Predict Variation in Eye Size across Birds
Round 1
Reviewer 1 Report
This manuscript uses a comparative approach to correlate eye size variation in a large sample of birds with sociality and gradients in life history. The author hypothesizes that interspecific variation in predation risk contributes to eye size variation, with the assumption that sociality and life history (as interpreted by clutch size and adult survival) are good proxies for predation risk. I previously reviewed this paper for a different journal and have gone through my previous review and highlighted areas that remain in need of improvement.
General Comments
The author needs to control for foraging height when considering sociality. Flocking species are much more likely to be found in the mid to upper canopy, meaning they are likely exposed to much more light than understory solitary species. Foraging stratum has been shown to be a strong predictor of eye size variation (both locally and across the avian tree of life), as mediated by light intensity gradients. In order to separate the assumption that sociality predicts predation risk from light availability, the author will need to include some metric for foraging stratum (I recommend Elton Traits which provides continuous numeric estimates). Even then, for me the assertion that sociality is a good proxy for predation risk remains dubious, since there many other factors that could explain sociality.
New Comment: The author has not incorporated foraging stratum into the analysis. I strongly recommend including it. If not then the author needs to justify why.
Discussion: Given that the author is using proxy variables for predation risk, the Discussion needs expanding to provide more interpretation and caveats. The correlations found between eye size, life history, and sociality could be explained by many other hypotheses beyond predation risk, and the author needs to explain why he thinks that predation risk is the best one. In general, I found the Discussion to be limited in content and in need of more interpretation.
New Comment: While the Discussion has been expanded since the prior draft, it still lacks a robust discussion of why predation risk is the best explanation for the patterns observed. Furthermore, the strong correlation with the life history gradient is the most interesting contribution of this paper, and the author dedicates almost no space towards its interpretation in the Discussion. Please improve these two deficiencies.
Specific Comments:
Lines 107-115: Foraging stratum has also been shown to correlate with eye size for terrestrial birds. Please justify why you didn’t include it. You should also include it in Table S1. In particular, solitary birds may be more likely to be terrestrial and found in darker understory habitats, which could in part explain your results.
New Comment: Stratum needs to be addressed. See above.
Lines 119-125: You correctly reference body size allometry for eye size. However, you do not correctly account for it in your model (as you have written). The second PCA axis does load highly for body mass, but still incompletely, meaning that you are not fully or correctly accounting for body mass in your regression. I suggest that you either (1) use residual eye size accounting for body mass as your dependent variable or (2) you remove body mass from the PCA and include log(mass) as a dependent variable. Furthermore, you use a log10 transformation. Why not the natural log, which is used in most research on this topic? Either use the natural log or provide a justification for log10. This seems like a detail, but it’s important to maintain consistency among studies.
New Comment: The author did not address this comment. Please do so in this revision.
Line 136: Do you mean “Pagel’s Lambda”?
New Comment: This was not addressed. Please clarify.
Line 133: Do you mean to cite the Bird Tree by Jetz et al. (2012)? The Hackett tree is not complete.
New Comment: This was not addressed. Please clarify.
Line 186: I think you mean to say that relative eye size is larger for species in more closed habitats – this is what your stats say and is consistent with past literature.
New Comment: This error was not changed. Your statistics clearly show that open habitat species have spaller eyes. Please change.
Line 128: Moller 2010 looks at this topic.
New Comment: This was not addressed. Please clarify.
Figure 2: Please indicate with asterisks or letters the statistical relationships among categories so that the reader can quickly interpret pairwise significance among groups.
New Comment: This was not addressed. Please correct.
ESM: Please include a table of correlation coefficients among all variables. This reflects my point made for Line 121. I realize you included the VIF for your analysis, but a correlation table will be more transparent as to the exact relationships among variables.
New Comment: This was not done. Please provide this table.
Lines 292-230: This new paragraph with caveats is helpful, but please provide some citations.
Author Response
This manuscript uses a comparative approach to correlate eye size variation in a large sample of birds with sociality and gradients in life history. The author hypothesizes that interspecific variation in predation risk contributes to eye size variation, with the assumption that sociality and life history (as interpreted by clutch size and adult survival) are good proxies for predation risk. I previously reviewed this paper for a different journal and have gone through my previous review and highlighted areas that remain in need of improvement.
Reply: Thank you for taking the time to review this paper once more. In the following, I explain how I addressed the remaining issues.
The author needs to control for foraging height when considering sociality. Flocking species are much more likely to be found in the mid to upper canopy, meaning they are likely exposed to much more light than understory solitary species. Foraging stratum has been shown to be a strong predictor of eye size variation (both locally and across the avian tree of life), as mediated by light intensity gradients. In order to separate the assumption that sociality predicts predation risk from light availability, the author will need to include some metric for foraging stratum (I recommend Elton Traits which provides continuous numeric estimates). Even then, for me the assertion that sociality is a good proxy for predation risk remains dubious, since there many other factors that could explain sociality.
New Comment: The author has not incorporated foraging stratum into the analysis. I strongly recommend including it. If not then the author needs to justify why.
Reply: I understand the need to control for foraging height. Foraging height was partly included in the lifestyle variable as it contained a category for species foraging on the ground. However, this is not as informative as foraging stratum. To address this issue, I used Elton’s traits, as suggested, and now have a new variable called foraging substrate, which distinguishes between terrestrial species foraging mainly on the ground and those foraging in the under-story, the mid-story, and the canopy. As there was a large overlap between foraging substrate, as defined here, and lifestyle, I dropped the lifestyle variable. I also changed the trophic niche variable to include fewer categories and decrease the overlap with other variables. These changes did not alter any of the conclusions.
Discussion: Given that the author is using proxy variables for predation risk, the Discussion needs expanding to provide more interpretation and caveats. The correlations found between eye size, life history, and sociality could be explained by many other hypotheses beyond predation risk, and the author needs to explain why he thinks that predation risk is the best one. In general, I found the Discussion to be limited in content and in need of more interpretation.
New Comment: While the Discussion has been expanded since the prior draft, it still lacks a robust discussion of why predation risk is the best explanation for the patterns observed. Furthermore, the strong correlation with the life history gradient is the most interesting contribution of this paper, and the author dedicates almost no space towards its interpretation in the Discussion. Please improve these two deficiencies.
Reply: I now provide a lengthier discussion of the life history finding, which I agree is one main finding of this research. With respect to life history, I added the following: The association between eye size and life history fits with the asset protection principle whereby animals invest more in anti-predator defences when they have more to lose in terms of future reproductive potential {Clark, 1994 #988}. In this study, species at the slow end of the life history spectrum, which emphasize adult survival over reproduction, evolved larger eyes, which would be helpful to reduce predation risk by allowing earlier detection. Consistent with the concept of risk aversion, empirical studies have documented more caution with respect to predators for species at the slow end of the life history continuum {Blumstein, 2006 #6469; Møller, 2012 #14955; Ghalambor, 2001 #1987}. These studies have focused on behavioural responses to risk. This study shows that bird species with different life histories can also differ in morphological traits with an anti-predator function like the eye. This is similar to the idea that predation can lead to the evolution of morphological traits aimed at reducing predation risk {Langerhans, 2004 #14956}.
As for the other deficiency, I added the following, which should I hope address the issue: The association between eye size and sociality fits with the idea that high predation risk associated with solitary living can select for the evolution of relatively larger eyes. While species that forage alone rather than in flocks often experience higher risk {Shultz, 2004 #5203; Beauchamp, 2023 #14789}, foraging alone or in flocks is also associated with particular ecological niches that on their own might affect eye size independently of risk. For instance, flocking in birds is associated with body mass, diet, timing of activity, and foraging substrate {Beauchamp, 2022 #14583} and many of these variables are also correlates of eye size {Ausprey, 2021 #14905; Brooke, 1999 #14915; Garamszegi, 2002 #11215; Liu, 2023 #14910}. As I included these potential confounding effects in the phylogenetic model, the association between sociality and relative eye size was not an indirect effect of the above co-factors. Life history is also associated with ecological niches in general {Roff, 2002 #13673} and with flocking in birds as well {Beauchamp, 2022 #14583}. Accounting for confounding variables, life history and flocking were both independently associated with relative eye size.
Species foraging on the ground have relatively larger eyes to capture more elusive light in this forest stratum {Ausprey, 2021 #14905}. My model recovered this trend, but the effect was not significant perhaps reflecting the lower power in my study and additional variables included in the model. If solitary species forage on the ground more often {Thiollay, 1998 #487}, the relatively larger eyes of solitary species could be explained by the darker habitat they inhabit rather than sociality. However, the effect of sociality persisted after controlling for foraging stratum.
If you have other ideas as to why predation risk is not the only explanation for the results, I will be happy to include these ideas in the paper.
Lines 107-115: Foraging stratum has also been shown to correlate with eye size for terrestrial birds. Please justify why you didn’t include it. You should also include it in Table S1. In particular, solitary birds may be more likely to be terrestrial and found in darker understory habitats, which could in part explain your results.
New Comment: Stratum needs to be addressed. See above.
Reply: This has been changed as described earlier. The old Table S1 was just a list of the families included in the analysis. The dataset with full information is available from the author.
Lines 119-125: You correctly reference body size allometry for eye size. However, you do not correctly account for it in your model (as you have written). The second PCA axis does load highly for body mass, but still incompletely, meaning that you are not fully or correctly accounting for body mass in your regression. I suggest that you either (1) use residual eye size accounting for body mass as your dependent variable or (2) you remove body mass from the PCA and include log(mass) as a dependent variable. Furthermore, you use a log10 transformation. Why not the natural log, which is used in most research on this topic? Either use the natural log or provide a justification for log10. This seems like a detail, but it’s important to maintain consistency among studies.
New Comment: The author did not address this comment. Please do so in this revision.
Reply: As far as I know, it is not correct to use residuals as a dependent variable in a regression model or at least not accurate (Freckleton 2002). I point out that an advantage of using PC1 (the principal component related to body mass) is that it is independent (orthogonal) by design from PC2 (the life history component). PC1 was also strongly correlated with body mass as pointed out in the paper. It would seem strange to do a PCA with body mass, survival, and clutch size but then leave aside one of the principal components. PC2 was derived using body mass, so it makes sense to use PC1 in the model as they were scaled together. It is not that I am unwilling to make changes, but that in this particular case, it seems wrong. If this is not acceptable, an option would be to drop life history and only include body mass in the model. For information, the effect of sociality persists in this simpler model, but obvioulsly there is less to discuss than before.
As for using log10, this has been the typical way of dealing with allometry for eye size in animals starting with Ritland and including Brooke (1999), Liu (2012), Huang (2019) and Liu (2023) among others. Would there be an advantage in using loge that I do not know?
Line 136: Do you mean “Pagel’s Lambda”?
New Comment: This was not addressed. Please clarify.
Reply: Yes this is that lambda. This has been corrected.
Line 133: Do you mean to cite the Bird Tree by Jetz et al. (2012)? The Hackett tree is not complete.
New Comment: This was not addressed. Please clarify.
Reply: I made a reference to Hackett in the paper. The website where I got the trees is http://vertlife.org/phylosubsets/ and the option for downloading trees is called Hackett. So I am assuming it is using the information in Hackett’s paper, no?
Line 186: I think you mean to say that relative eye size is larger for species in more closed habitats – this is what your stats say and is consistent with past literature.
New Comment: This error was not changed. Your statistics clearly show that open habitat species have spaller eyes. Please change.
Reply: My mistake. This has been corrected.
Line 128: Moller 2010 looks at this topic.
New Comment: This was not addressed. Please clarify.
Reply: I left out specific references here since they were listed in the introduction. Nevertheless, to address this issue, I added the following: In birds, species exposed to more predation risk tend to flee at larger distances from potential threats and have evolved relatively larger eyes {Møller, 2010 #13341}.
Figure 2: Please indicate with asterisks or letters the statistical relationships among categories so that the reader can quickly interpret pairwise significance among groups.
New Comment: This was not addressed. Please correct.
Reply: The trophic niche variable is no longer in the paper. Diet was not associated with relative eye volume or axial length.
ESM: Please include a table of correlation coefficients among all variables. This reflects my point made for Line 121. I realize you included the VIF for your analysis, but a correlation table will be more transparent as to the exact relationships among variables.
New Comment: This was not done. Please provide this table.
Reply: As far as I understand, it is multicollinearity that matters in a multiple regression and not pairwise correlations. I can provide those correlations if it is justified. Table S1 now provides the VIFs as requested by the other reviewer.
Lines 292-230: This new paragraph with caveats is helpful, but please provide some citations.
Reply: I added 6 references in this section as suggested.
Reviewer 2 Report
The author presents a comprehensive and significant study that correlates eye size with the life history and sociality of 660 avian species. The author analyzes eye volume and axial length and concludes that eye size is associated with life history and flocking, but not with absolute latitude, migration, or timing of activity. The author provides interesting discussions on these findings. I believe a minor revision is necessary to ensure that the resulting paper meets the publication standards required by the international journal Birds.
Recommended Revisions:
1. Introduction: In the fourth paragraph of the introduction, it would be beneficial to provide more detailed information on the potential relationship between life history and eye size.
2. Data collection: The author mistakenly used the transverse diameter of eye specimens to directly calculate eye volume. I recommend carefully reviewing references [10, 11, 13] and adding appropriate referencing for the formula used.
3. Data analysis: It is crucial to clearly explain how the most suitable model was determined through AIC. Additionally, it is important to display the calculated VIF values to enhance the transparency of the analysis.
4. Results: A discrepancy is observed in the statement "eye volume was larger in species foraging in more open habitats." According to Table 1, the data actually show that eye volume is larger in species foraging in more closed habitats instead of open habitats.
5. Discussion: The third paragraph of the discussion mentions that "trophic niche and lifestyle, among others, influenced relative eye size." However, this study lacks strong evidence to support this result. Further examination and clarification regarding this point would be valuable.
6. Conclusions: The conclusion section could benefit from further elaboration to summarize the main findings of the study.
7. Ensure that references are formatted uniformly throughout the paper.
8. Lastly, I noticed a few errors that I will point out line by line below.
Line by line edits:
Line 88: The title of Table 1 should be made more specific, as it currently includes variables such as body size and life history, which do not fall under ecological traits.
Line 116: The title of Table 2 should also be made more accurate and clearly defined. Additionally, the definition of different levels can be described in more detail to avoid confusion, for example, clarifying the distinction between "Invertivore", "Vertivore" and "Aquatic prey".
Line 187-189: The interpretation of results from Figure 3 is difficult. It would be helpful to rephrase these results more accurately, taking into account the information provided in Table 1.
Author Response
The author presents a comprehensive and significant study that correlates eye size with the life history and sociality of 660 avian species. The author analyzes eye volume and axial length and concludes that eye size is associated with life history and flocking, but not with absolute latitude, migration, or timing of activity. The author provides interesting discussions on these findings. I believe a minor revision is necessary to ensure that the resulting paper meets the publication standards required by the international journal Birds.
Reply: Thank you for your thoughtful comments. Below, I explain how I addressed these important issues.
1. Introduction: In the fourth paragraph of the introduction, it would be beneficial to provide more detailed information on the potential relationship between life history and eye size.
Reply: I agree and now provide more information. This now reads as: Due to the emphasis on adult survival, species at the slow end of the continuum are considered less likely to accept risk on themselves to ensure their own survival and future reproduction. By contrast, species at the fast end of the continuum might accept more risk on themselves as their probability of breeding again is low. In general, species that emphasize adult survival and future reproduction are expected to be more cautious {Clark, 1994 #988}, which would favour the evolution of traits that reduce predation risk. The evidence thus far supports an association between life history and risk taking. For instance, species at the slow end of the continuum tend to flee sooner when predators approach {Blumstein, 2006 #6469; Møller, 2012 #14955; Gotanda, 2009 #14957}, have higher levels of hormones involved in reducing risk {Hau, 2010 #11984}, and take fewer risk when raising young {Ghalambor, 2001 #1987}.
2. Data collection: The author mistakenly used the transverse diameter of eye specimens to directly calculate eye volume. I recommend carefully reviewing references [10, 11, 13] and adding appropriate referencing for the formula used.
Reply: In a spheroid, as far as I know, the horizontal transverse diameter is assumed to be equal to axial length since both diameters are measured on the same plane. In other words, the middle of a spheroid where these two diameters are measured is a perfect circle. The formula for a spheroid uses the equatorial and polar diameters, which in Ritland are given by H and V, respectively, as mentioned in the paper. Since the formula is provided, I am perfectly transparent about how eye volume was calculated in the paper. Of course, the assumption that the eye is a spheroid is open for discussion as I acknowledged in the discussion. The spheroid model has been used frequently in the past to measure eye volume. I added a reference to the use of a spheroid to model eye volume. Did I get this right?
3. Data analysis: It is crucial to clearly explain how the most suitable model was determined through AIC. Additionally, it is important to display the calculated VIF values to enhance the transparency of the analysis.
Reply: I now mention that I sought the model with the smallest AIC to determine the best model. I now provide the VIFs in Table S1.
4. Results: A discrepancy is observed in the statement "eye volume was larger in species foraging in more open habitats." According to Table 1, the data actually show that eye volume is larger in species foraging in more closed habitats instead of open habitats.
Reply: Thank you for pointing this out. This was a mistake and was corrected.
5. Discussion: The third paragraph of the discussion mentions that "trophic niche and lifestyle, among others, influenced relative eye size." However, this study lacks strong evidence to support this result. Further examination and clarification regarding this point would be valuable.
Reply: These two variables are no longer in the paper.
6. Conclusions: The conclusion section could benefit from further elaboration to summarize the main findings of the study.
Reply: I agree and expanded the conclusion as followed: In a phylogenetic framework and controlling for known correlates of eye size across birds, the results of this study show that the risk of predation can predict variation in eye size across species of birds. In particular, species at the slow end of the life history continuum, which are considered risk averse, and those that live alone, which cannot rely on others in their groups to reduce predation risk, have evolved relatively larger eyes. This was true when eye size was estimated as a volume or using axial length.
7. Ensure that references are formatted uniformly throughout the paper.
Reply: The references were formatted using EndNote and a template appropriate for Birds. Can you give examples of non-uniform formatting?
Line 88: The title of Table 1 should be made more specific, as it currently includes variables such as body size and life history, which do not fall under ecological traits.
Reply: This table no longer exists.
Line 116: The title of Table 2 should also be made more accurate and clearly defined. Additionally, the definition of different levels can be described in more detail to avoid confusion, for example, clarifying the distinction between "Invertivore", "Vertivore" and "Aquatic prey".
Reply: The title of this table was changed to: Definition of variables potentially associated with eye size across birds. Notice that the table now contains new variables (Diet and Substrate) to replace trophic niche and lifestyle.
Line 187-189: The interpretation of results from Figure 3 is difficult. It would be helpful to rephrase these results more accurately, taking into account the information provided in Table 1.
Reply: It is difficult to represent results from a multiple regression model. This is why I opted for showing residual eye volumes. I added the following in the figure legend to help understand this figure better: Larger residual eye volumes indicate that eyes are larger than predicted given the size of a species and its ecology.
In the results, when I say, for instance, that eye size is larger for solitary species the figure shows that residual eye volume is also larger. The results in this figure and those in the table match exactly.
Round 2
Reviewer 2 Report
The authors have done an excellent job addressing all of my feedback on the previous version of the manuscript. I only have a few minor points left, which I will describe below.
Line 13: Revise “lifestyle”.
Line 113-116: The eye volume, calculated using the eye transverse radius, should be explicitly stated.
Line 153: Revise “trophic niche, lifestyle”.
Line 213-215: The results for the “habitat” variable are not displayed in Figure 3.
Line 310: A discrepancy is observed in the statement “Accounting for confounding variables, life history and flocking were both independently associated with relative eye size”. According to Table 2, foraging alone species had no relatively larger eye volume than foraging occasionally in flocks or alone species (p-value = 0.12).
Line 313: Revise “lifestyle”.
Author Response
Line 13: Revise “lifestyle”.
Reply: This was a holdover from a previous version. Here, I replaced this term with life histories.
Line 113-116: The eye volume, calculated using the eye transverse radius, should be explicitly stated.
Reply: Agreed. This now reads as: … I calculated mean eye size as a volume (cc) for each species using the above two transverse diameters ...
Line 153: Revise “trophic niche, lifestyle”.
Reply: This was a holdover from the previous version. I replaced the two terms with diet and foraging substrate.
Line 213-215: The results for the “habitat” variable are not displayed in Figure 3.
Reply: Yes, I know but adding this variable made the figure really large and cumbersome. I also did not show results for timing of activity and other variables as well. I changed the legend to say that selected variables are displayed.
Line 310: A discrepancy is observed in the statement “Accounting for confounding variables, life history and flocking were both independently associated with relative eye size”. According to Table 2, foraging alone species had no relatively larger eye volume than foraging occasionally in flocks or alone species (p-value = 0.12).
Reply: To avoid confusion, I now use sociality for the name of the factor related to flocking. So sociality has three different levels in Table 1: Flocking, occasional flocking or solitary.
Line 313: Revise “lifestyle”.
Reply: This was changed to foraging substrate.
