Review Reports - New Evidence on the Linkage of Population Trends and Species Traits to Long-Term Niche Changes

Round 1

Reviewer 1 Report

The authors quantified the realized niche of 71 bird species in Italy. Then, they examined its relationship with long-term population terms and species traits.

This study tackled an important issue in avian ecology. Research is well designed, analyzed and discussed.

Overall, the manuscript provides important insights into avian ecology and therefore I propose it for publication in Birds.

Author Response

Reviewer’s comment

The authors quantified the realized niche of 71 bird species in Italy. Then, they examined its relationship with long-term population terms and species traits.

This study tackled an important issue in avian ecology. Research is well designed, analyzed and discussed.

Overall, the manuscript provides important insights into avian ecology and therefore I propose it for publication in Birds.

Authors’ answer

We thank the Reviewer to have appreciated our work.

Reviewer’s comment

English language and style are fine/minor spell check required.

Authors’ answer

We checked English language and style to correct mistakes and to improve the quality of language throughout the paper. Below there are some examples:

Line 2: we changed “vary” with “varies”.

Line 145: we changed “occurrence frequency” with “frequency of occurrence”.

Line 263: we changed “In the E-space definition” with “In defining the E-space”.

Line 379: we changed “with” with “to”.

Line 427: We change “differs” with “differed”

Author Response File: Author Response.docx

Reviewer 2 Report

In this paper the authors examine temporal changes in occupied niche for 71 species of breeding birds in Northern Italy and how these are related to population trends. The findings support previous studies from other geographic regions that demonstrate niche expansion for species with increasing population trends. Importantly, this paper also expands on previous studies by examining additional niche components (e.g. habitat), and attempting to correlate patterns of niche change with species traits (migratory strategy, diet, life history, etc). I found the paper to be well done, informative, and interesting. I believe this is an important contribution to the literature and will be of interest to ornithologists studying climate responses, as well as general ecologists interested in ecology niche theory.

There are a few places throughout the manuscript where minor clarifications are needed. The only somewhat "major" revision I suggest would be to confirm that the observed relationship between niche expansion and population trend (Figure 5, panel 2) is not biased by a single outlier species. More on these minor and major suggestions in the line by line comments below. Additionally, there are some grammatical and sentence structure issues throughout the manuscript. I did comment on each of these, but the paper would benefit from another round of careful edits with this in mind.

L84- "temporal changes in niche overlap" - this reads like you are interested in testing for changes in niche overlap between species. In the context on the rest of your paper, niche overlap is used within individual species over time to simply test for changes in niche occupancy. I think "...how temporal changes in niche occupancy are related to population trends." works as an edit for what you want.

Figure 1- I appreciate all of the information presented in this figure. A couple of small changes might help the presentation slightly. I suggest moving the legend for sampling site dates over the right panel where those symbols are instead of over the landcover map. Secondly I suggest removing or perhaps lightening in color the grid that covers the sampling site map. It is currently too dark and obscures the map.

L158-159. It is unclear what "continue urban matrix" and "discontinue urban matrix" are. (This perhaps should be continuous and discontinuous?, but either way should be further explained.

L202-203. What is the prupose of the Wilcoxon signed rank test? To test for a general trend in niche shift across all species? I would state the purpose of the test in this sentence.

L241. Did setting non-significant trends to 0 have an impact on your results. I wonder if allowing for small non-zero (albeit "nonsignificant") trends would influence the patterns observed.

L281- I'm not sure describing variation in niche breadth across species is valuable. We expect species to differ in niche breadth and centroid. It IS relevant and valuable to describe that there was a general shift in niche observed across species, which you do in the second half of the paragraph, but that gets a bit buried in the list of numbers preceding it.

Figure 3. Several of these figures (a,b,c) show niche expansion into E-space not observed in T1 (red shading extending beyond the green line). Your methods say that your analyses were limited to species observed in both T1 and T2 (L215-217). This is true, correct? These figures should reflect that too.

Figure 4. I really like this figure! It's a good way to show data for all species, and look for whether lack of overlap is due to expansion and/or unfilling for individual species that the reader may be interested in. One suggestion: it might be interesting to use colors indicate population trend?

Figure 5. I wonder how much the observed pattern in niche expansion and population trend (Figure 5, panel 2), and even Schoeners' D (panel 1) is driven by a single species with a very high population trend compared to the other species- which I'm guessing is the Song Thrush?
Just by looking at the figure it appears that the Schoeners' D correlation would become far less significant by removing this species, and the Expansion correlation may even flip direction (from positive to negative). I think you should confirm whether your results are robust to the inclusion/exclusion of this single species, and then further discuss in the Discussion section what this might mean- or at least include that caveat if it does in fact influence your results.

L379. unclear what "species specialization" means here. Specialization in what?

Author Response

Reviewer’s comment

Authors’ answer

We thank the Reviewer for her/his comment and for the overall appreciation of our work.

As regards the major issue of the correlation analysis between population trend and niche metrics, we confirm that results were not driven by a single species (see the answer to this specific issue for evidence and details).

We also checked and improved English language and style as suggested by the Reviewer (see the last answer for some examples).

Reviewer’s comment

Authors’ answer

We thank the Reviewer for the comment and suggestion. Actually, we investigated niche changes within individual species. We modified the sentence as follow: “Another important question is how temporal changes in niche occupancy (i.e., variations of the density of occurrence corrected by the environmental availability in the multidimensional niche space) are related to population trends”. Additionally, in the appropriate cases below, we added the term “niche occupancy” or replaced “density of occurrence” with “niche occupancy”.

Line 108: we added “occupancy” after “niche”.

Line 207: after “corrected by the prevalence of the environments in their range”, we added “i.e., niche occupancy”.

At lines 426-427 we changed “This means that their density of occurrence differs within the available E-space between T1 and T2” with “This means that their niche occupancy in the E-space differed between T₁ and T₂”.

At line 547 we changed “density of occurrence with niche occupancy”.

Reviewer’s comment

Authors’ answer

We accepted the Reviewer suggestions and modified the Figure 1. We moved the legend for sampling site dates over the right panel, and lightened the grid.

Reviewer’s comment

L158-159. It is unclear what "continue urban matrix" and "discontinue urban matrix" are. (This perhaps should be continuous and discontinuous?, but either way should be further explained.

Authors’ answer

We thank the Reviewer to highlight this mistake. We replaced “continue” and “discontinue” with “continuous” and “discontinuous”, respectively. The use of these two terms derived from the terminology used in the official land-cover map DUSAF (https://www.cartografia.regione.lombardia.it/metadata/Dusaf/doc/legenda_DUSAF5.pdf, only in Italian language). At line 158-159, we added some examples in parenthesis by modifying the sentence as follow: “Land-cover variables included continuous urban matrix and infrastructures (e.g., dense urban areas, industrial areas; C110), discontinuous urban matrix (e.g., farmsteads; C112)…..”. Further details on variables were provided in Table S1 in SM, where we also changed the terms “continue” and “discontinue” with “continuous” and “discontinuous”, respectively, and we added some examples in parenthesis.

Reviewer’s comment

L202-203. What is the prupose of the Wilcoxon signed rank test? To test for a general trend in niche shift across all species? I would state the purpose of the test in this sentence.

Authors’ answer

To be clearer about the purpose of the Wilcoxon rank sum test (we did not use the signed rank test), we changed the sentence at line 202-203 as follow: “To test significant variations in the niche breadth and centroid between the two periods across all species, we performed the Wilcoxon rank sum test“.

Reviewer’s comment

L241. Did setting non-significant trends to 0 have an impact on your results. I wonder if allowing for small non-zero (albeit "nonsignificant") trends would influence the patterns observed.

Authors’ answer

We chose to set non-significant trends to zero because in this case the magnitude of the observed trend might be considerably different from the actual trend due to uncertainty around estimates. Setting them to zero, we opted for a more conservative choice.

In order to check the effect of this choice on results, we performed the same analysis (Kendall correlation between each niche metric and population trends), in which:

We did not set any non-significant trend to zero (i.e., each species retained its trend regardless of the significance);
We only set trends with p-value >0.10 to zero (only two species, the Blackbird and the Eurasian Nuthatch, showed 0.05 < p-value < 0.10).

Results were very similar in both i) and ii) compared to results presented in the manuscripts. Below the details:

Schoener’s D and trend: z = -1.961, p-value = 0.050, tau = -0.159. Expansion index and trend: z = 3.341, p-value = 0.0008, tau = 0.271. Unfilling index: z = -0.765, p-value = 0.444, tau = -0.062.
Schoener’s D and trend: z = -1.9675, p-value = 0.049, tau = -0.164. Expansion index and trend: z = 2.975, p-value = 0.0029, tau = 0.249. Unfilling index: z = -0.836, p-value = 0.403, tau = -0.070.

In order to underline the consistency of results, we decided to add the following sentence at line 369: “The results were consistent even when trends close to the threshold of significance (p-value ≤ 0.10) were not set to zero value (Schoener’s D: tau = – 0.164, p-value = 0.049; expansion index: tau = 0.249, p-value = 0.003; unfilling index: tau = – 0.070, p-value = 0.403)”.

Reviewer’s comment

Authors’ answer

We agree with the Reviewer. We cut the first part of the paragraph and moved it in the caption of Figure S2 in SM. This way, results are not weighted down by description of variation in niche breadth and centroid, but this information is still available in SM along with the corresponding box plots.

The paragraph in main text has been changed as follow: “Niche breadths and centroids showed a large variability across species along each of the two PCA-axes (Table S4). Despite within-species differences, we detected a general change only for the niche centroid along PC1 between T₁ and T₂, representing a shift towards warmer climatic conditions (W =1,955, p-value = 0.021; Figure S2)”.

Reviewer’s comment

Authors’s comment

We confirm that analyses on expansion and unfilling indices were restricted to the shared background E-space between the two periods. The function to obtain this plot in R package ecospat (ecospat.plot.niche.dyn) does not allow representing the occupied niche limited to the shared background E-space. Additionally, we decided to not remove the density of occurrence for the non-shared background areas because, in our opinion, it would result in a loss of information about the occupied niche in each period. This way, we showed the observed occupied niche in relation to the available E-space in each period, but at the same time, the shared background environment can be inferred from the contour lines. Conversely, removing from the plots the density of occurrences for those areas of the E-space that were not included in the intersection, it would prevent the visualization of some important data that contributed to define the realized species niches.

Reviewer’s comment

Authors’ answer

We thank the Reviewer for her/his appreciation of this figure. We accepted the suggestion and we changed colors of the figure to display population trends.

Reviewer’s comment

Authors’ answer

We performed the Kendall non-parametric correlation analysis in order to overcome the issues about the violation of the assumptions for the Pearson correlation analysis (presence of outliers, absence of normality and homoscedasticity) and to manage better ties derived from assigning zero values to non-significant trends. Kendall tau is not affected by the presence of outliers. To confirm our results, we re-run the correlation analysis after excluding the Song Thrush. Below the results:

Schoener’s D: z = -1.622, p-value = 0.105, tau = -0.137. Expansion index: z = 2.542, p-value = 0.011, tau = 0.216. Unfilling index: z = -0.243, p-value = 0.808, tau = -0.021. Tau coefficients are very similar and significance does not change. We noticed that the p-value for Schoener’s D changed from 0.053 to 0.105, but it is expected because a species with a positive trend and low Schoener’s D was removed from the sample.

Reviewer’s comment

L379. unclear what "species specialization" means here. Specialization in what?

Authors’ answer

We added “most of” before “species specialization”, and “indices” after “species specialization” to make the sentence clearer.

Reviewer’s comment

Moderate English changes required.

Authors’ answer

We checked English language and style for mistakes and improving the quality of English language throughout the paper. Some examples:

Simple summary: line 2, we changed “is” with “are”; line 3, “it” with “they”.

Line 2: we changed “vary” with “varies”.

Line 145: we changed “occurrence frequency” with “frequency of occurrence”.

Line 263: we changed “In the E-space definition” with “In defining the E-space”.

Line 379: we changed “with” with “to”.

Line 427: we changed “differs” with “differed”.

Author Response File: Author Response.docx

Reviewer 3 Report

Dear Author,

it is very! interesting article. One of my main comments concerns the description of data analysis methods. I am not able to understand how the PCA was developing (see file). I suspect you "threw" everything into one "bag" (analysis). I am not convinced that this is a good strategy (see how much you lose explained variance). Perhaps with this approach there is one possible way, if yes, then explain it in the discussion.

Best wishes!

Comments for author File: Comments.pdf

Author Response

Reviewer’s comment

Dear Author,

Best wishes!

Authors’ answer

Thanks for the appreciation of our work. We decided to adopt this modelling approach because we were interested in quantifying a multidimensional niche space, which took into account both habitat and climatic components. For further details see the answer to the comment on line 262.

Reviewer’s comment

Line 249-251: please, explain, what type of data was used, how this analysis was developed.

Authors’ answer

In agreement with the Reviewer, we explained better how this analysis was developed. First, we moved the sentence explaining which traits were used in the analysis at the beginning of the paragraph to improve the general flow of reading. Then, we modified line 249-253 as follow: “For continuous traits, we performed a principal component analysis (PCA) using as input variables all 18 traits. This way, we reduced the dimensionality of traits into new uncorrelated components representing species life-history and ecological characteristics. Then, we analyzed the correlation between the main axes of the PCA—picked out through eigenvalues, scree plots and percentage of variance explained—and the niche metrics, which were inputted as supplementary quantitative variables”.

Reviewer’s comment

Line 262: you lost a lot of the explained variability because you put all the data into one PCA. I suggest that separate PCAs be performed for: a) climatic data, c) habitat data, etc.

Authors’ answer

We acknowledge that the overall explained variance in PCA, performed to define the E-space, was not very high, especially for PC2. We inputted all habitat and climatic variables together because the goal of PCA was to define a single E-space that took into account both habitat and climatic characteristics, which were a representative set of all scenopoetic variables. We believe that one of the strong points of this study was to try to define and analyze a multidimensional Grinnellian niche without omitting any component. If we had run three distinct PCAs, one for land cover data, one for topography, and one for climatic data, we would have obtained three distinct niches, which we would have to analyze individually. Moreover, since some correlations among habitat, topographic and climatic variables occurred, we think that performing distinct PCAs could have led to overlook such relationships and to an increase of redundancy in results.

Reviewer’s comment

Line 263: should be definition.

Authors’ answer

We thank the Reviewer for the comment. We changed “In the E-space definition” with “In defining the E-space”.

Author Response File: Author Response.docx