Habitat and Body Condition of Small Mammals in a Country at Mid-Latitude

Round 1

Reviewer 1 Report (Previous Reviewer 2)

Comments and Suggestions for Authors

My criticism of earlier drafts of this manuscript centred on two main points, neither of which is adequately addressed in this revised version.

Point 1: Body condition index (BCI), on which this work is based, is a metric akin to density, which is not known whether, and if so, how it affects survival, especially when used across many different species with different ecological, biological and taxonomic characteristics. The revised manuscript remains the same with respect to this point.

Point 2: Should one accept that analysis of BCI could reveal patterns that may not be interpretable but may nonetheless be interesting to stimulate further research, then the effect of habitat on BCI should be tested using a single model in which other factors known to affect BCI (species, age, sex, season etc.) are also included. In this way, one would see whether habitat has any effect over and above the combined effect of all the other factors. Instead of this, the authors have added a small section (“3.2. Analysis of Factors Affecting Body Condition in Relation to Habitat”) that shows (log) BCI is indeed affected by several other factors, as well as habitat, but the single model analysis stops there and the rest of the results are based on the same inappropriate and less powerful non-parametric comparisons of habitat or species and habitat, ignoring the simultaneous effects of important factors such as age and season.

 

Author Response

Rev #1 comments and answers

My criticism of earlier drafts of this manuscript centred on two main points, neither of which is adequately addressed in this revised version.

 

Point 1: Body condition index (BCI), on which this work is based, is a metric akin to density, which is not known whether, and if so, how it affects survival, especially when used across many different species with different ecological, biological and taxonomic characteristics. The revised manuscript remains the same with respect to this point.

Answer: we do not analyze survival, and in small mammal snap-trapping this is not possible to analyze it. Furthermore, we also do not analyze densities here.

 

Point 2: Should one accept that analysis of BCI could reveal patterns that may not be interpretable but may nonetheless be interesting to stimulate further research, then the effect of habitat on BCI should be tested using a single model in which other factors known to affect BCI (species, age, sex, season etc.) are also included. In this way, one would see whether habitat has any effect over and above the combined effect of all the other factors. Instead of this, the authors have added a small section (“3.2. Analysis of Factors Affecting Body Condition in Relation to Habitat”) that shows (log) BCI is indeed affected by several other factors, as well as habitat, but the single model analysis stops there and the rest of the results are based on the same inappropriate and less powerful non-parametric comparisons of habitat or species and habitat, ignoring the simultaneous effects of important factors such as age and season.

Answer: To clarify, we introduced partial eta-square as a measure the effect size for a specific factor, representing the proportion of total variability attributable to that factor, after controlling for other variables in the model. Partial eta-squared shows relative importance of different factors in a multivariate context, measuring the effect size for a specific factor, representing the proportion of total variability attributable to that factor, after controlling for other variables in the model.

We add text “After controlling for other variables in the model, two of these factors, species (?² = 0.095) and habitat group (?² = 0.045) had strongest influence on BCI variation. Season (?² = 0.021) and animal age (?² = 0.013) were weaker. Therefore, we further analyzed the effects of habitat and species on individual body condition indices.”

We tested ANOVA using logBCI for influences of the habitat in all small mammal species, and obtained similar result. Therefore, to stay with real BCI values, we would like to continue with non-parametric test, and have easily understandable figures 4 and 5.

Reviewer 2 Report (Previous Reviewer 3)

Comments and Suggestions for Authors

In my opinion the response and explanations of the authors are satisfactory. Therefore I do not have any further comments and suggest an acceptance of this text.

Author Response

Rev#2 comments and answers

In my opinion the response and explanations of the authors are satisfactory. Therefore I do not have any further comments and suggest an acceptance of this text.

Answer: thank you

Reviewer 3 Report (New Reviewer)

Comments and Suggestions for Authors

In this study you undertook an interesting subject with an analysis of a substantial dataset.  There is a lot of promise in your effort.  However your manuscript needs revision and can be improved.  See the specific comments for ideas of how it may tell a better story.

 

Specific Comments

Line 11:  This is the only place you mention the time frame (1980–2023) of you study.  It is important to also expand on this in the methods section.

Lines 47-49:  “Sometimes these studies misinterpret habitat use and preference, as species preferences require knowledge of resources and conditions [39], which are not investigated.”  This is a pretty significant statement that should be followed up with additional justification and explanation.

Lines 63-64:  “…this has not been investigated in mid-latitudes.”  Where has it been investigated and how was it accomplished?  Did those studies provide some basis and background for this study?

Lines 71-74:  This is confusing since that has not been a basis established for making these statements.  This should be introduced and better explained either in Methods or Results.

Lines 77-78:  This is results and should be introduced and better explained there.

Lines 85-94:  This Figure is not cited in the text and is not discussed.  How is the information presented here used in this study and what is its relevance?

Line 96:  What trapping protocols did you analyze, where were they used, and how did this analysis allow you to categorize habitats?

Line 99:  The forest land cover classes grouped all ages, including clearcuts; this would seem to combine very different habitats which would support very different small mammal communities (e.g,. mature forest vs. clearcuts).

Line 101:  Inland marshes, peat bogs, and inland waters were grouped together.  These seem to provide very different habitats to small mammal communities.

Line 106:  What is the basis of the 50-m buffer?  “…on an island” is confusing.  Does this mean the whole island is considered riparian habitat?

Lines 108-109:  “…a catch line covered several different habitats” this is confusing.  Even if the line crossed several different habitats, wouldn’t each trap station be recorded as in the habitat where it occurred?  It seems that all trap stations were recorded as in mixed habitat even though they were actually in another specific habitat.  I am not sure how fragmentation was defined and determined.  Just because the catch line covered several different habitats does not necessarily mean that they were fragmented.  Each of those several different habitats could actually have been included in large patches (i.e., not fragmented).

Lines 112-113:  “We treated human-caused and biological disturbance as having similar effect on small mammals.  This seems like a bad assumption.  For example, a site being actively used as a landfill should not be included with a site that has recently been burned.

Line 114:  Again, this group contains some very different land cover (and associated different habitats) (i.e., fruit trees, berry plantations, pastures, annual crops).  Certainly the habitats provided by these for small mammals would be very different.

Line 116:  I am really not sure what commensal mean in this context.  I am not sure, from a small mammal’s point of view, that all places where small mammals can use human-related food should be considered together in this analysis.

Line 120-122:  These statements, “ensures compatibility with habitat categories used by other researchers” and “sufficient sample size” really need explanation and justification.

Line 131:  Additional information is needed to describe how trapping was done.  For example, times of the year and day that samples were collected, how trap lines were laid out, length of trap line in each habitat, and more.

Lines 132-133:  Perhaps the results from live trapping and pitfall traps should have removed from the analysis since they were a very small number of individual and since the differences involved could potentially affect the results.

Line 137:  “We processed 28567 individuals of 18 small mammal species…” implies that the authors actually handled these individuals.  Did they actually do this to determine BCI?  If so, there needs to be some discussion on how specimens were handled and preserved after capture.

Lines 137-145:  It would seem appropriate to remove those species with few captures (e.g., <20).

Lines 146-147:  “…even the smallest sample size is sufficient for further analysis.”  Really, how do you know this?

Lines 155-159:  I do not know what this means or what its significance is.

Line 161-162:  “We used the body condition index (BCI) as a proxy for individual fitness.”  What was your rationale for doing this?  And, why did you use Moors’ method?

Lines 165-167:  “Bias in the BCI due to differences in measurement methods [66] was minimal, as over 80% of the individuals sampled were measured by the same person throughout the study period.”  It would be helpful to have an actual assessment of how the methods were implemented to ensure bias did not occur.  Just saying it did not occur is not convincing.

Line 171:  How were seasons defined and how was age determined?

Lines 172-175:  “…part of country where animals were trapped was used as a continuous predictor. We do not use trapping location as a continuous predictor because there are too many of them and a significant portion of them were used only once.”  So, explain a bit more how where the animal was trapped entered into the analysis.

Lines 232-233:  “…as the influence of county part (eastern, northern, western, central, southern, northeastern, 232 northwestern, southeastern, and southwestern)…” this should have been described in the Methods section.

Line 235:  “Based on these results…” explain why these results supported additional analysis.

Line 252:  How small?  And what is your interpretation of the effect of this on the results?

Line 261:  It is not clear why and how these habitats were grouped together.

Lines 310-311:  Where did you confirm that sample sizes were “…sufficiently robust…”?

Line 321:  What are “…optimal habitats…”?

Lines 321-328:  It is not clear what all this means in the context of this study.

Lines 339-346:  You never described how fragmentation was defined for this study and what the characteristics of habitat described as fragmented was.  As a result it is hard to associate the results of this study to what you described as fragmented habitats (e.g., how do your fragmented habitats compare with other studies that described fragmented habitats?).  As a result, this whole discussion about fragmented habitats is currently not supported by your results.

Lines 373-382:  Is it reasonable and meaningful to describe BCI across species?  It may not be, considering the differences in physiology among species.

Lines 405-413:  It is not helpful to describe aspects you will be doing in future analyses.

Line 422:  Your Conclusions section is basically a summary, not what you learned through this analysis.

Author Response

Rev#3 comments and answers

 

In this study you undertook an interesting subject with an analysis of a substantial dataset.  There is a lot of promise in your effort.  However your manuscript needs revision and can be improved.  See the specific comments for ideas of how it may tell a better story.

Answer: we acknowledge your comments and thank you for suggestions to improve manuscript.

 

Specific Comments

 

Line 11:  This is the only place you mention the time frame (1980–2023) of you study.  It is important to also expand on this in the methods section.

Answer: the time period was also shown in beginning of Material and Methods, Line 77.

 

Lines 47-49:  “Sometimes these studies misinterpret habitat use and preference, as species preferences require knowledge of resources and conditions [39], which are not investigated.”  This is a pretty significant statement that should be followed up with additional justification and explanation.

Answer: we clarified this statement

“However, if these studies say “preferred habitat”, they misinterpret habitat use and preference. Knowledge of species preferences require knowledge of resources and conditions [39], which in most mentioned studies were not investigated, except of [29].”

 

Lines 63-64:  “…this has not been investigated in mid-latitudes.”  Where has it been investigated and how was it accomplished?  Did those studies provide some basis and background for this study?

Answer: we added text here, pointing to Discussion (Lines 349-355), where importance of habitat characteristics to body condition was  shown in tropical forest of Brazilia.

 

Lines 71-74:  This is confusing since that has not been a basis established for making these statements.  This should be introduced and better explained either in Methods or Results.

Answer: text transferred to 2.1. Chapter. Indeed, it is better fit after description of used habitat classes.

 

Lines 77-78:  This is results and should be introduced and better explained there.

Answer: We would like to emphasize that years of trapping and area coverage as well as number of trapping sites are not Results, but descriptions of Material.

 

Lines 85-94:  This Figure is not cited in the text and is not discussed.  How is the information presented here used in this study and what is its relevance?

Answer: we apologize about missing citation, now included. Here we illustrate, what categories of CORINE habitats were grouped, and how these habitats are spatially distributed in the country, to have a general picture of the study site. As this was intended af our first publication concerning habitats and body condition, map of the study site is desirable.

Also, we used classification which is not directly related to CORINE categories. Therefore, listing of these categories and shoving how CORINE categories fit to the habitat groups we used, we think was necessary.

 

Line 96:  What trapping protocols did you analyze, where were they used, and how did this analysis allow you to categorize habitats?

Answer: we accept your criticism, which is mainly due to language problems. In Lithuania, a "trapping protocol" is a description of the trapping, which contains details of the trapping - time, location, species and number of individuals caught, trapping effort and habitat. As the data are retrospective and there is probably no standardised methodology for describing habitats, the habitat is given in very different detail. We have therefore changed the expression to "available descriptions of trapping sessions"

 

Line 99:  The forest land cover classes grouped all ages, including clearcuts; this would seem to combine very different habitats which would support very different small mammal communities (e.g,. mature forest vs. clearcuts).

Answer: yes, in the descriptions of trapping sessions there are over 150 different forest characteristics mentioned. Therefore, if we just take coniferous, deciduous and mixed forests, plus young, middle-aged and mature ones, there will be 9 categories, but then we will lose over 12% of trappings where description only says “forest”, and so on. Not speaking that sample size of some categories will be small.

Therefore, in the initial analysis we wild like to generalize habitats into their groups. More detailed classification of habitats might be used for specific analyses of data.

Finally, if we care about compatibility of results, we are afraid that detailed categories, e.g., “young deciduous forest” in Lithuania and Slovakia are not directly comparable.

 

Line 101:  Inland marshes, peat bogs, and inland waters were grouped together.  These seem to provide very different habitats to small mammal communities.

Answer: apologies, but you are not right here, inland waters are not included into “wetland” group.

 

Line 106:  What is the basis of the 50-m buffer?  “…on an island” is confusing.  Does this mean the whole island is considered riparian habitat?

Answer: thank you, we understood where we were not careful – riparian zone on the island was defined also as 50 m buffer, Text was changed as “within 50 m of the shore of a river, lake, or island”.

 

Lines 108-109:  “…a catch line covered several different habitats” this is confusing.  Even if the line crossed several different habitats, wouldn’t each trap station be recorded as in the habitat where it occurred?  It seems that all trap stations were recorded as in mixed habitat even though they were actually in another specific habitat.  I am not sure how fragmentation was defined and determined.  Just because the catch line covered several different habitats does not necessarily mean that they were fragmented.  Each of those several different habitats could actually have been included in large patches (i.e., not fragmented).

Answer: Unfortunately. Registering every trap was not used in the most of investigations, please remember we use retrospective data. To our best knowledge, situation is the same in other countries.

On the other hand, 25*5 m is 125 m of the single trap line, and if it covers several habitats, we understand this as “fragmented habitat”. To clarify, we add the word “single line”.

E.g., there is note “forest, meadows, ecotone, wetland” in the description of the trapping, with no even number of the traps per every of these habitats, we have no more choices as define this habitat as mixed (and fragmented).

 

Lines 112-113:  “We treated human-caused and biological disturbance as having similar effect on small mammals.  This seems like a bad assumption.  For example, a site being actively used as a landfill should not be included with a site that has recently been burned.

Answer: we have no data from recent burn areas and never trapped small mammals in the active landfills. Accepting your comment we added text “Active landfills and recently burned areas were not used as trapping sites”.

 

Line 114:  Again, this group contains some very different land cover (and associated different habitats) (i.e., fruit trees, berry plantations, pastures, annual crops).  Certainly the habitats provided by these for small mammals would be very different.

Answer: Yes we agree. As a scientist, I am not happy to compare trappings of small mammals in alfalfa fields in Czech Republic with trappings in barley and clover in Poland (barley and clover not as separate result). So here is so many different results, which are only comparable on the general level if the scale is broad.

 

Line 116:  I am really not sure what commensal mean in this context.  I am not sure, from a small mammal’s point of view, that all places where small mammals can use human-related food should be considered together in this analysis.

Answer: most popular description is “A "commensal habitat" for small mammals refers to environments where these animals live in close association with human settlements and derive benefits from human activities without significantly harming humans. The term "commensal" comes from the Latin "com-" (together) and "mensa" (table), indicating a relationship where one species benefits while the other is neither helped nor harmed.”

We do not put emphasis on harm, and therefore clarified description by adding text “and similar places providing small mammals with food and shelter.”

 

Line 120-122:  These statements, “ensures compatibility with habitat categories used by other researchers” and “sufficient sample size” really need explanation and justification.

Answer: We acknowledge your comment as for “sufficient” sample size, and rewrote this text as “Grouping habitats into a limited number of categories facilitates compatibility with data from other researchers, as the number of smaller habitat categories is large and difficult to compare on a large scale. In addition, grouping in our case ensured a larger sample size for a wide range of small mammal species in all habitat groups.”

 

Line 131:  Additional information is needed to describe how trapping was done.  For example, times of the year and day that samples were collected, how trap lines were laid out, length of trap line in each habitat, and more.

Answer: text on trapping added “In the absolute majority of cases, snap trapping was done by the standard method [60,61], traps were placed in lines of 25 traps, 5 m apart, set for three days and checked once or twice a day, i.e. in the morning or morning and evening. The traps were baited with brown bread and crude sunflower oil and replaced after rain or when eaten. One to four lines were set per habitat. Most of the small mammals (76.1%) were caught in the fall season, 13.3% in the summer, 6.3% in the spring and 4.3% in the winter.”

 

Lines 132-133:  Perhaps the results from live trapping and pitfall traps should have removed from the analysis since they were a very small number of individual and since the differences involved could potentially affect the results.

Answer: we checked this when writing first version of manuscript, removing of data did not impact results.

 

Line 137:  “We processed 28567 individuals of 18 small mammal species…” implies that the authors actually handled these individuals.  Did they actually do this to determine BCI?  If so, there needs to be some discussion on how specimens were handled and preserved after capture.

Answer: yes, and actually over 80% of these individuals was dissected by the first author (Line 155), therefore bias due to differences in measuring was minimal. Though measures used to calculate BCI are standard, there can be systematic differences if material is handled by several persons. We had only two persons doing this.

As said in Krebs and Singleton, 1993 [66]: “Indices of condition for house mice have a relatively low repeatability because of variation from day to day in body mass and because of variation in length measurements taken by different observers. Bias in measurements among observers must be eliminated to make indices of condition from live animals useful.” – so, in our case this bias was eliminated.

We also added short text on the specimen handling after capture: “Small mammals were kept cold without freezing if measured and dissected the same day after capture, or kept frozen in plastic bags until transfer to the laboratory. Identification of small mammal species was based on external characteristics, that of Microtus voles on differences in their teeth, and that of M. rossiaemeridionalis by genetic methods.”

 

Lines 137-145:  It would seem appropriate to remove those species with few captures (e.g., <20).

Answer: We apologize not saying clear about not analyzing habitat influence for the species with small sample size, it was written only in the caption of Figure 5, Lines 305–306.

We intended to say, that species with sample size not exceeding 10, namely N. milleri, M. avellanarius, A. sylvaticus, and A. amphibius, were not analyzed for habitat and are not included into Figures 2–5. Text about sample size limitation was added to all captions.

 

Lines 146-147:  “…even the smallest sample size is sufficient for further analysis.”  Really, how do you know this?

Answer: the program we use, Statistica for Windows, is not working when sample is below allowable thresholds, but such explanation is not usable. So we accept your comment and deleted part of the sentence.

 

Lines 155-159:  I do not know what this means or what its significance is.

Answer: deleted

 

Line 161-162:  “We used the body condition index (BCI) as a proxy for individual fitness.”  What was your rationale for doing this?  And, why did you use Moors’ method?

Answer: Moor's method is as good as the other body condition indices based on body mass and length. We have used it before, without any questions or critics "why", so we keep using it to keep compatibility of results. On the other hand, these two measures, body mass and body length, are the most common, thus ensuring retrospective use for the other countries and time periods.

Our rationale was based on (1) BCI being a proxy of fitness – based on [54–58}, as we cited in Lines 60–64; then (2) habitat and BCI can be linked by food resources, therefore (3) we do have long-term, multi-species and multi-habitat data available for BCI calculation, and (4) inter-species differences in BCI exist, but in general, ranges of BCI values are overlapping – therefore, we would like to know what is the habitat component in this.

Some aspects of the rationale also are pointed out in discussion, Lines 425–428.

 

Lines 165-167:  “Bias in the BCI due to differences in measurement methods [66] was minimal, as over 80% of the individuals sampled were measured by the same person throughout the study period.”  It would be helpful to have an actual assessment of how the methods were implemented to ensure bias did not occur.  Just saying it did not occur is not convincing.

Answer:

As said in Krebs and Singleton, 1993 [66]: “Indices of condition for house mice have a relatively low repeatability because of variation from day to day in body mass and because of variation in length measurements taken by different observers. Bias in measurements among observers must be eliminated to make indices of condition from live animals useful.” – so, in our case this bias was eliminated.

We inserted text “Body length was measured from the beginning of snout to the rear side of anus opening when individual is laid on the table, flat and straight, on its back (supine).” Pins were not used.

 

Line 171:  How were seasons defined and how was age determined?

Answer: We used four seasons: winter (December, January, and February), Spring (March, April, and May), summer (June, July, and August), and autumn (September, October, and November).

We defined three age groups of small mammals under dissection (adults, subadults, and juveniles), based on the status of sex organs and on the presence and atrophy of the thymus gland. More details are given in [60].

Text added to Material and Methods.

 

Lines 172-175:  “…part of country where animals were trapped was used as a continuous predictor. We do not use trapping location as a continuous predictor because there are too many of them and a significant portion of them were used only once.”  So, explain a bit more how where the animal was trapped entered into the analysis.

Lines 232-233:  “…as the influence of county part (eastern, northern, western, central, southern, northeastern, northwestern, southeastern, and southwestern)…” this should have been described in the Methods section.

Answer: done, inserted in Methods, sentence about locations deleted, as we explained how trapping sites were attributed to the part of the country.

“Once the coordinates of the survey sites were established (in some cases quite roughly, if the location was not precisely specified), they were assigned to the eastern, northern, western, central, southern, northeastern, northwestern, southeastern, or southwestern part of the country, which we used as continuous predictor.”

 

Line 235:  “Based on these results…” explain why these results supported additional analysis.

Answer: we agree, saying “based on these results” was not the best one. To clarify, we introduced partial eta-square as a measure the effect size for a specific factor, representing the proportion of total variability attributable to that factor, after controlling for other variables in the model. Two factors, species and habitat, had the most significant effect on BCIs.

Thus we changed text as:

We used partial eta-square as a measure the effect size for a specific factor, representing the proportion of total variability attributable to that factor, after controlling for other variables in the model.

After that, we selected two factors, species and habitat group, for the further analysis. To maintain compatibility with previous publications, non-transformed BCI and non-parametric Kruskal-Wallis ANOVA were used. For all small mammal species, we tested the influence of habitat as a categorical factor on BCI as a dependent parameter.

 

Line 252:  How small?  And what is your interpretation of the effect of this on the results?

Answer: 1–18 individuals. Therefore these species had limited distribution across habitats and BCI variability was also limited. We changed text explaining this. “It is important to note that, with the exception of M. minutus (N = 337) and N. fodiens (N = 99), sample sizes for species with similar BCIs across habitats were small (N = 1-18), limiting distribution across habitats and BCI variability.”

 

Line 261:  It is not clear why and how these habitats were grouped together.

Answer: post-hoc also revealed, that all members of the first group differ significantly from all members of the second group. This indicates distinct differences in shrew fitness between these habitat groups. We extended text to say this. ‘’ BCIs of S. araneus in all habitats included into the first group significantly differed from BCIs of S. araneus in all habitats of the second group.”

 

Lines 310-311:  Where did you confirm that sample sizes were “…sufficiently robust…”?

Answer: apologies, robust is not synonym for sufficient. We had in mind, that even in the worst represented habitat number of trapped small mammals (N = 392 in shrub habitat) allow further statistical analysis. However, this is self-evident, so we have deleted the sentence

 

Line 321:  What are “…optimal habitats…”?

Apologies and thanks – term was used after the other author we cite and expected these two are synonyms, but while "optimal" can often mean "best" within a certain context, "best" is a more general term that signifies the highest quality or standard overall. "Optimal" takes into account specific conditions or constraints that define what is best in that particular scenario.

We tried to avoid to repeat “best”, but now changed “optimal” back to “best”.

 

Lines 321-328:  It is not clear what all this means in the context of this study.

Answer: just to show, that habitat and specuies relation is not simple and straightforward, but complex and multilayered,

 

Lines 339-346:  You never described how fragmentation was defined for this study and what the characteristics of habitat described as fragmented was.  As a result it is hard to associate the results of this study to what you described as fragmented habitats (e.g., how do your fragmented habitats compare with other studies that described fragmented habitats?).  As a result, this whole discussion about fragmented habitats is currently not supported by your results.

Answer: as we wrote, if one trap line (125 m) covers several habitat types, we define such habitat as fragmented. So far we have no comparisons with other studies, therefore, Lines 341–344 explain, why fragmentation is negative to small mammals.

 

Lines 373-382:  Is it reasonable and meaningful to describe BCI across species?  It may not be, considering the differences in physiology among species.

Answer: In several of the cited research studies, comparisons of various aspects of small mammals are conducted across species with different physiology and trophic group. Therefore, we expect that comparing BCI (Body Condition Index) across species and habitats is valid and worthwhile.

 

Lines 405-413:  It is not helpful to describe aspects you will be doing in future analyses.

Answer: we rewrote this part of text, excluding future analyses.

 

Line 422:  Your Conclusions section is basically a summary, not what you learned through this analysis.

Answer: we significantly shortened text, removing all details.

1. We found that the representation of small mammal species was habitat dependent, with certain species dominating certain habitats. The most suitable habitat was meadows, where 17 small mammal species were found, with 9 species being the most abundant.
2. BCI variation across habitats was species-specific, indicating habitats and species with the highest and lowest average BCIs.
3. No correlation was found between the proportion of species in a habitat and its BCI. Higher BCI values were found to be characteristic of non-dominant species.

Round 2

Reviewer 3 Report (New Reviewer)

Comments and Suggestions for Authors

I have received the revised review request and I am satisfied with the revision.

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

I found discrepancies in the data between the abstract and the conclusion. Please unite.

Abstract
The highest average BCI for most of these species was found in disturbed habitats, with S. minutus and M. arvalis showing the highest BCI in agricultural habitats. The lowest average BCI for most species was found in mixed habitats, while C. glareolus and M. arvalis exhibited the lowest BCI in shrub habitats.
Conclusion
The highest average BCI of S. araneus, A. agrarius, A. flavicollis, C. glareolus, A. oeconomus and M. agrestis was observed in disturbed habitats, while the lowest was observed of S. minutus and M. arvalis in agricultural habitats.

Author Response

Reviewer #1 comments and answers

 

Comment:

I found discrepancies in the data between the abstract and the conclusion. Please unite.

Abstract: The highest average BCI for most of these species was found in disturbed habitats, with S. minutus and M. arvalis showing the highest BCI in agricultural habitats. The lowest average BCI for most species was found in mixed habitats, while C. glareolus and M. arvalis exhibited the lowest BCI in shrub habitats.
Conclusion: The highest average BCI of S. araneus, A. agrarius, A. flavicollis, C. glareolus, A. oeconomus and M. agrestis was observed in disturbed habitats, while the lowest was observed of S. minutus and M. arvalis in agricultural habitats.

 

Answer:

Corrected, the error was in the conclusions. Thank you and please accept our apologies.

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript is based on an impressive data set, covering a very long time period and the full range of habitats of a country. This makes it instantly attractive. However, it is based on the notion of Body Condition Index (BCI) which is problematic. A significant number of published papers, including some cited by the authors (e.g., Wilder at al. [54] and Wishart et al. [56]), point out that BCI is NOT a very useful measure of body condition and it is not necessarily a good index of fitness or survival ability. There are too many factors that may affect BCI. The authors themselves write “We did not include animal age, sex and reproductive parameters in this analysis, as they have already been analysed [60]” (Lines 70-72). Seasonal variation is also likely to be very important. I find that taking these factors into account would provide a more convincing result. Of course, a different statistical method would be required in such a case. Biometrical data, such as body mass or body length, are usually normally distributed or, if not, can be transformed to a normally distributed variable (often a simple log transformation is sufficient). This would allow using a General Linear Model or other parametric statistical method. Such methods tend to be much more powerful than non-parametric ones and would allow looking at the effect of habitat after having accounted for the effect of all of the other above-mentioned factors known to affect BCI.
Even so, I am not sure that the problems associated with the use of BCI would be eliminated.

As a further suggestion for improvement, I would also consider concentrating on a smaller number of species, such as shrews or herbivorous microtines, to minimise the effect of different body shape and function. This might allow better comparisons between species, with more convincing conclusions.

 

Comments on the Quality of English Language

Overall, I found the manuscript readable but several points in all sections of the text need to be checked because they are ambiguous or hard to undestand.

Author Response

Reviewer #2 comments and answers

Comment:

This manuscript is based on an impressive data set, covering a very long time period and the full range of habitats of a country. This makes it instantly attractive. However, it is based on the notion of Body Condition Index (BCI) which is problematic. A significant number of published papers, including some cited by the authors (e.g., Wilder at al. [54] and Wishart et al. [56]), point out that BCI is NOT a very useful measure of body condition and it is not necessarily a good index of fitness or survival ability. There are too many factors that may affect BCI. The authors themselves write “We did not include animal age, sex and reproductive parameters in this analysis, as they have already been analysed [60]” (Lines 70-72). Seasonal variation is also likely to be very important. I find that taking these factors into account would provide a more convincing result. Of course, a different statistical method would be required in such a case. Biometrical data, such as body mass or body length, are usually normally distributed or, if not, can be transformed to a normally distributed variable (often a simple log transformation is sufficient). This would allow using a General Linear Model or other parametric statistical method. Such methods tend to be much more powerful than non-parametric ones and would allow looking at the effect of habitat after having accounted for the effect of all of the other above-mentioned factors known to affect BCI.
Even so, I am not sure that the problems associated with the use of BCI would be eliminated.

 

Answer:

As you mentioned in the comment, our dataset is really one of the largest ever used for BCI analysis. We believe that no one has ever analyzed BCI in a comprehensive way, comparing ALL species and ALL habitats from a single area. There may be several reasons for this, but the most important is that they have not used retrospective data and have been limited to short-term data and comparisons of a few small mammal species. The other studies compared the BCIs of taxonomically distant species with different body sizes. Therefore, the method may have seemed limited to the authors.

We agree with what you say citing Wilder et al. 2015, but this author uses different interpretation of the usefulness of BCI. And we hardly expect there are retrospective data for such evaluations. To date, we have only short-term isotopic studies of small mammal hair (as a proxy for their dietary space) and are still collecting data from the rarely trapped species.

As for the Wishart et al., 2024 – they compared limited number of contrasting species. Therefore, results are very different from our study. Again, for their method there are no retrospective data.

In our understanding, retrospective data (in our case, small mammal body mass and body length, both used to calculate BCI) are those collected in the past and covering a long period of time. These data can come from various sources and collected for different purposes. They have the advantage of ready availability, a large sample size, and longitudinal analysis. However, they may be subject to different data quality, bias, and limited control over variables.

In our case, the sample is heterogeneous, with some species being less numerous despite the total sample size being large. However, there is no bias in the measurements, as they were taken by the same person. We cover all small mammal species that can be trapped by snap traps in the country, as well as all the main habitats of small mammals in our latitude. Furthermore, species are sympatric and, in most cases, even syntopic.

As for the second part of your comment, we agree with it, but we already did GLM and found that “Among categorical factors, species had the most influence (F_18,25961 = 1496.6), followed by animal age (F_2,25961 = 1161.3), and gender (F_2,25961 = 34.3). The influence of trapping decade (F_4,25961 = 19.4), season (F_3,25961 = 13.6), and habitat (F_8,25961 = 16.5) was less expressed, but all listed categorical factors were significant at p < 0.0001.” this text is published in [60], but was not available at a moment of submission. So now we can use it, citing ourselves, but in limited form avoiding excessive self-citation.

Given the availability of extensive data, we opted for a mixed publishing model. In the initial publication [60], we present a list of significant factors and analyze those related to species (BCI correlation with species, age, gender, and reproductive status across all species). The manuscript under review deals into the analysis of habitat influence. Subsequently, we will examine the Chitty effect and minimum BCI values. Following this, we will analyze time factors, such as long-term and seasonal variability of BCI. Finally, we anticipate presenting an overview of the results in comparison with other body condition evaluation methods. It is possible that a BCI based on body size could be rehabilitated if the samples are large and the species being compared are of similar size. Small mammals are a suitable group for this purpose. So far we did not analyze rats.

 

Comment:

As a further suggestion for improvement, I would also consider concentrating on a smaller number of species, such as shrews or herbivorous microtines, to minimise the effect of different body shape and function. This might allow better comparisons between species, with more convincing conclusions.

Answer:

In the future, we plan to analyze BCIs for individual species. However, we do not exclude your suggestion, which would be to analyze variations in BCIs for different trophic groups. According to our data analysis strategy, the current paper focuses on the influence of habitat and species. Therefore, we are comparing the maximum possible species-habitat sample.

The comparison of BCI variation across a large number of species and habitats represents a novel approach. The use of BCI based on the two simplest measures of body size suggests that similar data may be available from other countries.

Although there are differences in BCI related to species and habitat, the ranges of BCI are similar and overlapping. Therefore, we believe that comparing BCIs of all available small mammal species has some advantages.

The majority of other investigations focused on a few small mammal species, or compared BCIs of taxonomically disparate species that differ in body size and life histories. In contrast, our study compares sympatric species, the majority of which are even syntopic.

Reviewer 3 Report

Comments and Suggestions for Authors

My main doubts are about an influence of seasonal conditions upon BCI which can be more pronounced than habitat conditions. Findings of the paper could be more convincing if data would be divided at least for the beginning and the second half of the vegetative season. Reader is not sure whether compared data were collected from the same season or whether data from various habitats come from different parts of the year. This issue is briefly mentioned in lines 341-346 and 365-369 but it deserves broader explanation.

Trophic conditions vary considerably in coniferous and deciduous stands but here all forests are in one category.

I also have doubts whether Corine classes 131, 132, 133 should be analysed together with colonies of cormorans

Comments for author File: Comments.pdf

Comments on the Quality of English Language

The text would benefit from a consultation with a native speaker

Author Response

Reviewer #3 comments and answers

Comment:

My main doubts are about an influence of seasonal conditions upon BCI which can be more pronounced than habitat conditions. Findings of the paper could be more convincing if data would be divided at least for the beginning and the second half of the vegetative season. Reader is not sure whether compared data were collected from the same season or whether data from various habitats come from different parts of the year. This issue is briefly mentioned in lines 341-346 and 365-369 but it deserves broader explanation.

 

Answer:

First of all, let us answer your comment presenting GLM results from the paper, published in Animals after the submission of paper you reviewed. “Among categorical factors, species had the most influence (F_18,25961 = 1496.6), followed by animal age (F_2,25961 = 1161.3), and gender (F_2,25961 = 34.3). The influence of trapping decade (F_4,25961 = 19.4), season (F_3,25961 = 13.6), and habitat (F_8,25961 = 16.5) was less expressed, but all listed categorical factors were significant at p < 0.0001.” This text was not available in printed form at a moment of submission.

Your assertion is indeed correct; the season exerted a considerable influence, although it was less pronounced than that of the habitat. However, in the current analysis, we utilize data pooled from all seasons to examine BCI variations between species in different habitats. As the pooling process included all seasons, winter, spring, summer, and autumn (though not all seasons were present in every decade of the 1980–2023 time frame), we limit our analysis to habitat and species. The time factor (long-term and seasonal variation of BCI) will be analyzed in greater detail at a later stage. We add text into revised manuscript to explain this.

It can be reasonably assumed that the seasonal influence on body mass depression in voles will be less pronounced in the middle latitudes than in the north, given the species distribution limits. However, there is currently no data available to support this hypothesis, nor any publications that can be used for comparison. Text added to Line 358 in revised manuscript:

In our previous study, we observed growth depression in M. arvalis and C. glareolus. It occurred in January and February in juveniles and in January, February and March in subadults [Balčiauskienė et al., 2009]. However, BCI was not used in this study. We will re-evaluate the seasonal effect on BCI in further analyses.

 

Comment: Trophic conditions vary considerably in coniferous and deciduous stands but here all forests are in one category.

Answer: however, there are numerous additional categories of forest, beyond those of deciduous and coniferous trees. The trophic conditions of forests vary with their age, as evidenced by our published papers on the succession of meadow to forest. Other factors influencing trophic conditions include the presence or absence of shrubs, grass cover, and the composition of trees. Therefore, a detailed analysis of these factors can be conducted, but it is limited to one group of habitats and, in most cases, to one or two forest specialist species of small mammals. The quantity of our data is considerable, and it is not feasible to present it in a single paper. We agree with your observation and have included additional text to explain why the topic of "forests" was not analyzed in greater depth.

 

Comment: I also have doubts whether Corine classes 131, 132, 133 should be analysed together with colonies of cormorans

Answer: We acknowledge the validity of your concerns, but we believe it is necessary to provide an explanation regarding the rationale behind the grouping in question.

The 131, 132, and 133 Corine classes represent specific areas that have been subjected to disturbance, and in the case of landfill sites, they have also been enriched with various chemicals. Cormorant colonies cannot be considered "forest" due to the high degree of disturbance and enrichment by C and N. Our previous research (not cited except for [75]) indicates that small mammal communities in active cormorant colonies are affected in numerous ways. Therefore, we treated "disturbance," whether biological or non-biological in origin, as similar.

 

Comments from PDF file

 

Line 104: what is "close"? 10 m, 100m? 1 km?

Answer: 50 m, added to the text

 

Line 146: corrected as advised

Line 211: corrected as advised

Line 264: corrected as advised

Line 293: what does this mean?

Answer: This should have led to a discussion about whether the species is most common in the habitats that provide the best conditions for it. We have rewritten the text for clarity. The discussion continues in the next paragraph, as there are trade-off elements in the choice of habitat.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

In their reply to my comments the authors give a partly satisfactory explanation of why they carried out their analysis the way they did. They, nonetheless, consider Body Condition Index (BCI) as meaningful and useful, something that is not obvious to me at all.

Different body tissues have different densities with bone tissue tending to have higher values, muscle and other soft tissues being close to 1 g/cm³ and fat probably having values below 1 g/cm³. The question is does BCI reflect different proportions of the various tissues in the body? Perhaps more bone and less fat for a high BCI? If so, is this beneficial to survival? Instead, does BCI mean more soft tissues, fat included, with respect to the animal’s skeleton? If so, is this beneficial to survival? In modern humans a (very?) high BCI of this type would generally be synonymous with obesity, extremely low BCI would be emaciation. What about small mammals of such a wide taxonomic and life history variation? Of course natural selection would not allow such extremes to survive for long. Yet there are differences between species, age classes, habitats and so on. What do they mean? How (by what means) do they come about? How can they be interpreted and what is their significance? This is the core of my scepticism and these questions are ot addressed in the manuscript (the revised manuscript has only very minor corrections) nor in the authors’ response.

A secondary problem, if for the sake of argument BCI was accepted as meaningful, is that the raw values of BCI are used in this analysis, when it is already known that so many other factors affect it. I can still not why this analysis uses BCI and not the residuals after the effect of the other factors (species, age, gender, season etc.) has been accounted for.

Comments on the Quality of English Language

A native English speaker would be able to iron out many of the small language problems that I think need improvement.

Author Response

Comments of Academic Editor

I would like to thank you for reviewing the manuscript. However, in my opinion, the review presented does not address the criticisms made by reviewer 2, which I believe to be pertinent, and as a result I cannot accept this version of the manuscript. I would encourage you to submit a new version of this manuscript in which this point is adequately revised.
The central point of this criticism is that BCI is influenced by other variables that were not statistically controlled in the analysis, such as the sex and age of each specimen collected and the season in which they were caught (among others). Without controlling for these effects, assessing the association of BCI with habitat is subject to sources of bias, jeopardising the validity of the results and their implications. The right thing to do is to change the analysis so that the effect of habitat (a fixed factor) can be assessed by controlling for these other sources of variation. There are different alternatives for doing this.  One of them has already been mentioned by the aforementioned reviewer. Transforming the response variable (log or square root for example) can be tried so that variation in this parameter can be assessed using a more statistically rigorous analytical procedure than Kruskal Wallis, such as the GLM suggested by reviewer 2 or even a mixed effect GLM (GLMM), the latter being indicated to also control for the effect of the collection site or trapping site, which could be treated as a random variable if more than one trapping site has been sampled in each habitat (which is likely). The GLMM would further increase the analytical rigour of the manuscript. This GLM (or GLMM) should test for the effect of habitat, sex, age and season (fixed effects) and trapping site (random effect). This can be done in a single analysis (log BCI ~ habitat+sex+age+season) or, as suggested by reviewer 2, in two stages, the first assessing the relationship between BCI and sex, age and season (log BCI ~ sex + age + season) and the second, using the residuals from the first analysis as the response variable to investigate the effect of habitat (residuals glm1 ~ habitat). This second alternative is perhaps easier, since the influence of these variables (sex, age and season) has already been evaluated in another article recently published by your group.   Other corrections, clarifications and more specific improvements were also made by me directly in the manuscript in case you decide  to resubmit e new manuscript.

 

Answer to Academic Editor

Thank you for giving us a possibility to re-submit modified manuscript. To acknowledge your comments we:

1. Transformed BCI to log BCI, and checked normality
2. Performed GLM, using habitat, species, age, sex and season as categorical factors, while part of the country as random effect. 321 sampling sites are too much to include into the model as random effect. Given that in some sites trapping was done only once. Therefpre we think, that nine parts of the country are acceptable in thos respect.

We decided to make GLM as single run. Necessary changes done in Material and Methods, and Results, where we included one subchapter to present model results. However, we seen not appropriate presenting log BCI in further text.

 

We also acknowledge all your comments done in the text of manuscript PDF file.

Comment Line 69: Please clarify the ecological reasoning here. To me this hypothesis imply that interspecific (not intraspecific)  competition makes more sense here since the higher the species richness the higher the number of different species competing for the same resources.

Answer: apologies for mistype, text was intended to say interspecific concurrence.

 

Comment Line 87: The legend on the right of the figure only shows Level 3 land use classes. Please inform which one of these (if any) are Level 2.

Answer: apologies, Level 2 is not presented. We deleted sentence.

 

Comment Line 186: corrected

Line 225: corrected as proposed

Line 363: parentheses corrected as proposed

 

Comment Line 367: This sentence is confusing since this paragraph starts stating the below average BCI of S. minutus. So why you are sayng "higher" instead of "lower" BCI in this sentence? Please correct or clarify your reasoning here

Answer: thank you for noticing this, sentence really was unclear. We noted that not only was the BCI of S. minutus in agricultural habitats higher than that of any other species, but it was also slightly higher than the BCI of S. araneus and N. fodiens in both agricultural and commensal habitats. Text was corrected.

 

Comment Line 388: It is not clear which findings from your study you are implying here. Further it is not clear what correlation you are referring to here: correlation between fragmentation and individual fitness? between fragmentation and population demographics? Please clarify.

Answer: we clarified with text “Our results confirm the negative influence of habitat fragmentation on the body condition of small mammals: for seven of the eight most abundant species, the BCI was lowest in mixed (fragmented) habitats. The only exception was M. arvalis.”

 

Comment Line 413: I suggest changing this sentence to:

Overall, the relationship between BCI and habitat was species-specific.

Answer: changed as proposed