Testing the Poleotolerance Lichen Response Trait as an Indicator of Anthropic Disturbance in an Urban Environment
Round 1
Reviewer 1 Report
@page { margin: 50px } p { margin-bottom: 6.25px; line-height: 120% }
I carefully went through the manuscript by Bernardo Rocha and co-workers. The article deals with the effectiveness of poleotolerance index, proposed by Nimis for Italy, in other regions of Europe.
This is a relevant and interesting topic, and the presented work seems to me rigorous and methodologically correct, however, the use of “only one” index, although sounds as very useful, is critical and some revisions should be necessary to improve the robustness of the results.
In particular, I think that the authors should be discuss the effectiveness of poleotolerance against other funcional traits in order to better clarify the advantage of the use of poleotolerance of lichens to distinguish areas with different anthropic disturbance levels.
In general terms, the paper is well written. Some paragraphs may be enriched with more references. The methodology used to collect data in the field and for statistical and spatial analysis is thoroughly explained.
A list of comments is subsequently reported to detail those points of the text which should/may be re-checked to improve the clarity of the contents.
1. Introduction
Line 16: “noise”. Anthropic activities causes noise, but lichens are affected by noise? Authors can clarify this point?
Lines 63-64: “Lastly, the results provided by ecological indicators are easily interpreted and can be communicated to politicians, other decision-makers and citizens in a simpler way.” The statement must be better supported: why an "ecological indicator" is simpler to interpret than a measure (i.e. of concentration of pollutants)?
Lines 70-72: “Many studies have already successfully used their response to pollution [14,15] or to the heat island effect [16-18] in urban environments. When using lichen diversity as an ecological indicator, several metrics can be applied, depending on the type and intensity of the environmental driver [19].” The discussion here is a bit lacking, references should be added to recent studies and standards (e.g. European standard BS EN 16413:2014).
Lines 82-84: “The notion of poleotolerance or poleophoby as a response trait to measure human disturbance levels was first introduced by Barkman [24]. The author proposed a poleophoby scale for lichens and algae from northern Europe, later used in several air pollution studies in this region [25].” Also Trass, H. (1973) proposed an index of poleotolerance (IP), perhaps this index would be worth discussing, along with the fact that the airborne pollutants have changed over time.
Lines 100-101: ”This disturbance gradient should be potentially reflected on epiphytic lichen communities in a way that allows us to test this trait efficiency.” authors can add reference to clarify why are the epiphytic communities the most suitable for this type of research?
2. Materials and Methods
Lines 138-141: “Although the European methodology [35] recommends using only one species of phorophyte, this was impossible due to the great diversity of phorophytes in Lisbon. To address this limitation, only phorophytes with medium bark roughness were used for sampling.” This is a critical point: I think authors should discuss deeply this topic: how many different phorophytes have been used? Are there any lichen species found only on a specific phorophyte? Only the roughness have been take in accounts to select phorophytes, not other characteristic of bark (i.e. pH, exfoliation, water retention)? The results obtained on different tree species have the results been compared?
Figure 2. The graphs are poor legibile due to overlapping labels, authors must improve their legibility
3. Results
Line 208: “In total, we identified 38 species (Table S3)”. Could the Table S3 be moved to the main text? This would make the graphs with the species code names more readable. In the Table S3 species names should be reported in the complete form (with the autors).
If this is not possible move the table S3 to the main text, the species must be reported in the complete form the first time they were cited (e.g. at the line 235: “Enterographa crassa (DC.) Fée”).
4. Discussion
Lines 330-331: “This study presents a novelty regarding the application of the poleotolerance response trait as an ecological indicator of human disturbance in urban environments.” Is the poleotolerance index more effective than others funcional traits to detect anthropic disturbances? Authors should be compare his effectiveness against other funcional traits, e.g. eutrophication, aridity, growth forms, reproductive strategy (provided by Nimis in ITALIC website, as well) or the presence of pruina and secondary metabolites (as recently proposed by the same group).
Some minor points:
Line 106: “2,8 million people” should be “2.8 million people” and data from “Instituto Nacional de Estatística, 2012” why is not listed as a reference?
Line 107: “1484/km2” should be “1,484/km2”. “The climate is Mediterranean” maybe adding a reference?
Line 112: “European Urban Atlas” here also, can authors add a reference?
Line 118: “equitable”perhaps better use the term “balanced”?
Line 166: “(m2)” should be “(m2)”
Figure 1: two categories have very similar colors (DLDUF and RAL), would it be possible to change the color of RAL to increase clarity?
Line 194: “(r2)” should be “(r2)
Line337-338 species’ names should be formatted in italic
Line 338: “lichens’ distribution” should be “lichen distribution” ?
Author Response
Reviewer 1
I carefully went through the manuscript by Bernardo Rocha and co-workers. The article deals with the effectiveness of poleotolerance index, proposed by Nimis for Italy, in other regions of Europe.
This is a relevant and interesting topic, and the presented work seems to me rigorous and methodologically correct, however, the use of “only one” index, although sounds as very useful, is critical and some revisions should be necessary to improve the robustness of the results.
In particular, I think that the authors should be discuss the effectiveness of poleotolerance against other functional traits in order to better clarify the advantage of the use of poleotolerance of lichens to distinguish areas with different anthropic disturbance levels.
Answer: This comment is highly pertinent. We didn’t use other functional traits, as we were focusing on the poleotolerance trait. But we agree that it would be useful to add some discussion on what the advantages of using this trait over others could be. A recent study by “Matos, P.; Vieira, J.; Rocha, B.; Branquinho, C.; Pinho, P.: Modeling the provision of air-quality regulation ecosystem service provided by urban green spaces using lichens as ecological indicators. Science of the Total Environment 2019, 705-713” performed in the same study area and with the same sampling methodology analysed three response traits: humidity requirements, type of growth form and eutrophication tolerance in lichens. Their results showed that there were no overall clear shifts in the communities mediated these traits. We added a paragraph in our manuscript discussion about it: “In this work we intentionally focused on a single trait, poleotolerance, with the purpose of understanding if this trait could be useful as an indicator of anthropic disturbances in an urban context. Our results showed a shift in lichen communities from low tolerance species to high tolerance species as we move to more disturbed areas. A recent work [14] using the same methodology in the same study area explored other three different traits, growth form, eutrophication tolerance and humidity requirements, and did not observe any shifts in lichen communities mediated by these three traits. This is an indication that in urban environments, such as this encountered in Lisbon, the use of the poleotolerance trait may be more advantageous then using other more commonly used traits, giving a clearer indication about the disturbance gradient. Humidity requirements, growth form and eutrophication tolerance are frequently used to indicate disturbance related to the urban heat island effect (microclimatic variations) and nitrogen pollution. The fact that poleotolerance trait performed better here the advantage of using it when we aren’t able to clearly disentangle those two environmental drivers, highlighting its potential as an integrated indicator of overall anthropic disturbance in urban settlements. The presence of pruina or secondary metabolites showed also promising results in South American cities as indicators of the effects of disturbances associated to urban environments [28,57]. In our work we did not test these traits. The poleotolerance is, for now, restricted to southern European cities that share some of the lichen flora of Italy. Pruina or secondary metabolites have the advantage of being independent of this classification, having therefore more potential to be applied outside Europe, so future works should try to explore it in Europe, to understand if they have also the potential to be used here”.
In general terms, the paper is well written. Some paragraphs may be enriched with more references. The methodology used to collect data in the field and for statistical and spatial analysis is thoroughly explained.
A list of comments is subsequently reported to detail those points of the text which should/may be re-checked to improve the clarity of the contents.
1. Introduction
Line 16: “noise”. Anthropic activities causes noise, but lichens are affected by noise? Authors can clarify this point?
Answer: We listed some overall effects caused by human disturbance, without looking in particular to which may, or may not, affect lichens. But lichens aren´t, in fact, affected by noise. Thus, we recognize that it may raise some doubt and misunderstanding in readers and so the term “noise” was removed from the sentence. The sentence now reads “Urban environments are densely populated areas buzzing with a wide range of anthropic activities that cause disturbances like air pollution or the heat island effect, threatening both human and environmental health”.
Lines 63-64: “Lastly, the results provided by ecological indicators are easily interpreted and can be communicated to politicians, other decision-makers and citizens in a simpler way.” The statement must be better supported: why an "ecological indicator" is simpler to interpret than a measure (i.e. of concentration of pollutants)?
Answer: We did not intend to say that the results provided by ecological indicators were better that those given by measurements from analytical methods. We meant to say “simple way”. The sentence was corrected. In any case, the fact that it provides a simple way to communicate results is intrinsically linked to the definition of ecological indicators. As defined in the manuscript, ecological indicators are “measurable characteristics of the structure (e.g., genetic, population, habitat, and landscape pattern), composition (e.g., genes, species, populations, communities, and landscape types), or function (e.g., genetic, demographic/life history, ecosystem, and landscape disturbance processes) of ecological systems”. This definition is by Niemi, G.J. & McDonald, M.E. (2007). Application of ecological indicators. Annual Review of Ecology and Systematics. 35. 89-111. As these authors also say, an intrinsic aspect of a good ecological indicator is that it is capable of integrating the information of a complex system (an ecosystem for e.g.) into a simple measure that can be easily communicated to the public. By simple it is meant simpler that all the complexity of effects in an ecosystem. And the fundamental advantage for using an ecological indicator is that while the measuring the concentration of a pollutant gives us an indication how the atmosphere is changing, but only ecological indicators can give us an insight on the effects of those atmospheric changes on ecosystems. To make this clearer, the previous sentence was changed to try to clarify this advantage: (starting in line 63) “Besides this, they offer an integrated response of the effects of pollutants in the ecosystem while analytical methods give us the concentration of pollutants on a given time without integrating the effects it may have in an ecosystem. Lastly, the results provided by ecological indicators are easily interpreted and can be communicated to politicians, other decision-makers and citizens in a simple way”.
Lines 70-72: “Many studies have already successfully used their response to pollution [14,15] or to the heat island effect [16-18] in urban environments. When using lichen diversity as an ecological indicator, several metrics can be applied, depending on the type and intensity of the environmental driver [19].” The discussion here is a bit lacking, references should be added to recent studies and standards (e.g. European standard BS EN 16413:2014).
Answer: To accommodate the reviewer suggestion, we have altered this paragraph. It now states: “Lichens are life forms consisting of a symbiosis between a fungus and algae and/or cyanobacteria, in the case of epiphytes, growing over a plant substrate. They balance their water and nutrient content with the surrounding environment, due to their inability to regulate its content, being this the reason for their excellence as ecological indicators. Epiphytic lichens are one of the most versatile groups of organisms that can be used as indicators of atmospheric pollution [14,15], climate change [16,17] or the urban heat island [18]. In the last 20 years, standard methodologies have been developed to use them in air pollution and climate change monitoring studies [19,20], and in Europe the method is now the European standard norm (European Standard EN 16413:2014).Many studies have already successfully used their response to pollution [21,22] or to the heat island effect [18,23,24] in urban environments. When using lichen diversity as an ecological indicator, several metrics can be applied. A recent work has shown that under an ecological indicator perspective, several metrics can be used, and what determines the choice of the metrics is the type and intensity of the environmental driver [25]”.
Lines 82-84: “The notion of poleotolerance or poleophoby as a response trait to measure human disturbance levels was first introduced by Barkman [24]. The author proposed a poleophoby scale for lichens and algae from northern Europe, later used in several air pollution studies in this region [25].” Also Trass, H. (1973) proposed an index of poleotolerance (IP), perhaps this index would be worth discussing, along with the fact that the airborne pollutants have changed over time.
Answer: We did come across Trass poleotolerance index during our research. We decided to only add Barkman’s work due to the fact that he was the first and the main reason why Nimis decided to create its own poleotolerance trait (the one we use on our work) for Southern Europe. Nevertheless, we added the sentence “Trass [32] also proposed a poleotolerance index for lichens, taking into account the frequency, bark cover degree and vitality of lichens species present in different landscapes of Estonia. However, like Trass [32] stated in is work, the index is limited to the lichen flora found in Estonia and the scale of bark cover degree and vitality is rather subjective.” regarding Trass’ index as suggested.
Regarding airborne pollutants, although we agree that the fact that airborne pollutants have changed over time and could affect the accuracy of these trait developed by Barkman, Trass and others, this is not the case for the poleotolerance trait. This trait was developed as a response trait to integrate the effects of overall human disturbance, of which air pollution is only of them. Other traits exist more devoted to specific disturbance factors, like for instance nitrogen pollution (eutrophication tolerance trait. Regarding change over time, we have also changed the text to accommodate this request. We added a sentence explaining this: It now states “The major potential of this trait is that it doesn’t focus solely on the effects of pollutants in lichens communities but rather it looks at all anthropic disturbances in an integrated way and also how those disturbances, their type and intensity have changed over the years”.
Lines 100-101: “This disturbance gradient should be potentially reflected on epiphytic lichen communities in a way that allows us to test this trait efficiency.” authors can add reference to clarify why are the epiphytic communities the most suitable for this type of research?
Answer: As requested, the following references were added:
[40] “Branquinho, C.; Matos, P.; Pinho, P. Lichens as ecological indicators to track atmospheric changes: future challenges. Indicators and surrogates of biodiversity and environmental change. CSIRO Publishing, Melbourne, CRC Press, London 2015, 77-87” and
[41] “Nowak, D.J.; Jovan, S.; Branquinho, C.; Augusto, S.; Ribeiro, M.C.; Kretsch, C.E. Biodiversity, air quality and human health. In: Connecting global priorities: biodiversity and human health: a state of knowledge review. Geneva, Switzerland: World Health Organization: 63-74. Chapter 4. 2015, 63-74”.
2. Materials and Methods
Lines 138-141: “Although the European methodology [35] recommends using only one species of phorophyte, this was impossible due to the great diversity of phorophytes in Lisbon. To address this limitation, only phorophytes with medium bark roughness were used for sampling.” This is a critical point: I think authors should discuss deeply this topic: how many different phorophytes have been used? Are there any lichen species found only on a specific phorophyte? Only the roughness have been take in accounts to select phorophytes, not other characteristic of bark (i.e. pH, exfoliation, water retention)? The results obtained on different tree species have the results been compared?
Answer: We understand the referee’s concern. We will try explaining better how this was done. In fact, it wasn’t possible to find always the same tree species in the 41 sampling sites. Most of them were done using Jacaranda mimosifolia, which was the most widely spread species in Lisbon green spaces. Around 60% of the sampling sites had these species. 30% were done in Tipuana tipu, which has similar bark characteristics to jacaranda. The remaining phorophytes belonged to other species that had a similar bark texture to this, most of them impossible to identify as the work was performed during winter and were all deciduous. Unfortunately we have not recorded this information for each site. We will do that in future studies so we cannot say whether some species were specific to some phorophytes. Concerning bark exfoliation, we did select the phorophytes taking into consideration that all of them did not lose bark from exfoliation. Regarding bark pH, we used lichen traits related to preference for substrate pH and found that lichens preferring more basic substrate conditions dominated in the smaller green spaces (also with lower NDVI), and the community shifted towards lichens preferring more shadow and acid substrates in bigger and more densely vegetated substrates. This is expectable as smaller green spaces are less protected from dust deposition from traffic, which changes bark pH for more basic pH, and being dominated by more basophilous species. We have a previous work where we demonstrate it (reference 28 in the manuscript). This data is not shown in the paper. A recent work (reference 14 in the manuscript) using the same methodology and in the same study area performed by us also analysed lichen functional groups related to water requirements. The only pattern found related more hygrophytic species (need more water) to bigger and more densely vegetated green spaces. These traits are responding to the main environmental variables underlying the disturbance gradient (smaller green spaces more disturbed than bigger ones) indicating that these are the predominant factors, and not the type of phorophyte, and suggesting that any factor linked to phorophyte characteristics did not have a major influence in the results.
Figure 2. The graphs are poor legible due to overlapping labels, authors must improve their legibility
Answer: Figure 2 was changed in order to improve legibility, as requested by the reviewer.
3. Results
Line 208: “In total, we identified 38 species (Table S3)”. Could the Table S3 be moved to the main text? This would make the graphs with the species code names more readable. In the Table S3 species names should be reported in the complete form (with the authors).
If this is not possible move the table S3 to the main text, the species must be reported in the complete form the first time they were cited (e.g. at the line 235: “Enterographa crassa (DC.) Fée”).
Answer: Species names were rewritten to their complete form as requested, both in table S3 and in the main text. However, it was not possible to move Table S3 to the main text.
4. Discussion
Lines 330-331: “This study presents a novelty regarding the application of the poleotolerance response trait as an ecological indicator of human disturbance in urban environments.” Is the poleotolerance index more effective than others funcional traits to detect anthropic disturbances? Authors should be compare his effectiveness against other funcional traits, e.g. eutrophication, aridity, growth forms, reproductive strategy (provided by Nimis in ITALIC website, as well) or the presence of pruina and secondary metabolites (as recently proposed by the same group).
Answer: As explained above, we have added in the discussion a comparison of their effectiveness against other traits: “In this work we intentionally focused on a single trait, poleotolerance, with the purpose of understanding if this trait could be useful as an indicator of anthropic disturbances in an urban context. Our results showed a shift in lichen communities from low tolerance species to high tolerance species as we move to more disturbed areas. A recent work [14] using the same methodology in the same study area explored other three different traits, growth form, eutrophication tolerance and humidity requirements, and did not observe any shifts in lichen communities mediated by these three traits. This is an indication that in urban environments, such as this encountered in Lisbon, the use of the poleotolerance trait may be more advantageous then using other more commonly used traits, giving a clearer indication about the disturbance gradient. Humidity requirements, growth form and eutrophication tolerance are frequently used to indicate disturbance related to the urban heat island effect (microclimatic variations) and nitrogen pollution. The fact that poleotolerance trait performed better here the advantage of using it when we aren’t able to clearly disentangle those two environmental drivers, highlighting its potential as an integrated indicator of overall anthropic disturbance in urban settlements. The presence of pruina or secondary metabolites showed also promising results in South American cities as indicators of the effects of disturbances associated to urban environments [28,57]. In our work we did not test these traits. The poleotolerance is, for now, restricted to southern European cities that share some of the lichen flora of Italy. Pruina or secondary metabolites have the advantage of being independent of this classification, having therefore more potential to be applied outside Europe, so future works should try to explore it in Europe, to understand if they have also the potential to be used here”.
Some minor points:
Line 106: “2,8 million people” should be “2.8 million people” and data from “Instituto Nacional de Estatística, 2012” why is not listed as a reference?
Answer: “2,8 million people” was replaced by “2.8 million people” as suggested. Reference to “Instituto Nacional de Estatística, 2012” wasn´t with the appropriate reference styling. It was corrected to “[42] - Ine, I. Censos 2011 resultados definitivos-Portugal. Lisboa-Portugal: Instituto Nacional de Estatística, IP 2012.”
Line 107: “1484/km2” should be “1,484/km2”. “The climate is Mediterranean” maybe adding a reference?
Answer: “1484/km2” was replaced by “1,484/km2” as suggested. A new reference “[43]” from “World Map of the Köppen-Geiger climate classification updated” by Kottek et al. 2006 was also added to the manuscript.
Line 112: “European Urban Atlas” here also, can authors add a reference?
Answer: A reference “[46]” was added after “European Urban Atlas”.
Line 118: “equitable” perhaps better use the term “balanced”?
Answer: The term “equitable” was replaced by the term “balanced” as suggested.
Line 166: “(m2)” should be “(m2)”
Answer: “(m2)” was replaced by “(m2)” as suggested.
Figure 1: two categories have very similar colors (DLDUF and RAL), would it be possible to change the color of RAL to increase clarity?
Answer: A new map was added were the colour of “RAL” land cover was shifted from light grey to light pink, thus improving clarity like suggested.
Line 194: “(r2)” should be “(r2)
Answer: “(r2)” was replaced by “(r2)” as suggested.
Line337-338 species’ names should be formatted in italic
Answer: All species names were corrected and formatted in italic as suggested.
Line 338: “lichens’ distribution” should be “lichen distribution”
Answer: “lichens’ distribution” was replaced by “lichen distribution” as suggested.
Author Response File: 
Author Response.docx
Reviewer 2 Report
L 54 valuable instead of great
L 59-64 OK I agree that ecological indicators are very useful. However, from these sentences it seems that the traditional chemical-physical monitoring is completely unsuseful since all the concerns may be addressed using ecologicsl indicators. I think this is noty the case and should be stated here
L 65-66 start with Epiphytic lichens are….and state groups of organisms
L 82 evaluate instead of measure
L 121 eliminate sampling
L 355 eliminatye lichens ad total
L 404 elucidate instead of clear
Author Response
Reviewer 2
L 54 valuable instead of great
Answer: The term “great” was replaced by the term “valuable” as suggested.
L 59-64 OK I agree that ecological indicators are very useful. However, from these sentences it seems that the traditional chemical-physical monitoring is completely unsuseful since all the concerns may be addressed using ecological indicators. I think this is not the case and should be stated here
Answer: We did not intend to say that the results provided by ecological indicators were better that those given by measurements from analytical methods. However we do realise that the way we have written the sentence could lead to that conclusion. To address this issue we altered that sentence and the previous one. It now reads (starting in line 63) “Besides this, they offer an integrated response of the effects of pollutants in the ecosystem while analytical methods give us the concentration of pollutants on a given time without integrating the effects it may have in an ecosystem. Lastly, the results provided by ecological indicators are easily interpreted and can be communicated to politicians, other decision-makers and citizens in a simple way.”
L 65-66 start with Epiphytic lichens are….and state groups of organisms
Answer: Sentence was reformulated to “Epiphytic lichens are one of the most versatile groups of organisms that can be used as indicators of atmospheric pollution.” as suggested.
L 82 evaluate instead of measure
Answer: The term “measure” was replaced by the term “evaluate” as suggested.
L 121 eliminate sampling
Answer: The term “sampling” was eliminated as suggested.
L 355 eliminate lichens ad total
Answer: Sentence was reformulated. It now stands as “(…) accounted for 51% of total species (38% and 13%, respectively), (…)” as suggested.
L 404 elucidate instead of clear
Answer: The term “clear” was replaced by the term “elucidate” as suggested.
Author Response File: Author Response.docx
Reviewer 3 Report
- I suggest to revisit the text of the abstract. Half of the abstract do not represent the results of the study. The abstract should state briefly the purpose of the research, the principal results and major conclusions.
- The methodology recommends using only one species of phorophyte but in the study were used several unknown tree species. In this case tree have to be with similar bark properties and this means not bark roughness but in for example acidity and nutrient content. This is an important parameter which also influenced the results of the study. I suggest to show the influence of tree bark on species diversity or use only one tree species.
- Data analysis: is it suitable for the data to use Spearman correlation?
- Figures: names of species are overlapped and therefore they are uninformative.
- Check spaces between words and numbers (ex. 113 row: 1ha), superscripts (ex. 107: km2, 194: r2).
Author Response
Reviewer 3
- I suggest to revisit the text of the abstract. Half of the abstract do not represent the results of the study. The abstract should state briefly the purpose of the research, the principal results and major conclusions.
Answer: Diversity Journal sets the guidelines for the Abstract structure. This is the information in their website: (…)should follow the style of structured abstracts, but without headings: 1) Background: Place the question addressed in a broad context and highlight the purpose of the study; 2) Methods: Describe briefly the main methods or treatments applied. Include any relevant preregistration numbers, and species and strains of any animals used. 3) Results: Summarize the article's main findings; and 4) Conclusion: Indicate the main conclusions or interpretations.”
In our abstract we have rewritten a part of the abstract to better follow this structure and guidelines.
1) Background: “Urban environments are densely populated areas buzzing with a wide range of anthropic activities that cause disturbances like air pollution or the heat island effect, threatening both human and environmental health. Mitigating its impacts implies understanding the integrated effects that those disturbances exert on urban environments. Lichen biodiversity is frequently used as an ecological indicator, being able to integrate their effects in a quantifiable way. The poleotolerance response trait classifies lichens according to their tolerance to human disturbance, but it was developed for Italy’s flora, and has seldom been applied outside Italy or in urban context studies. The aim of this work is to assess this trait suitability as an indicator of urban anthropic disturbance and test it outside Italy”.
2) Methods: “For that, we sampled lichen diversity in 41 green spaces in Lisbon. Lichens were classified into the respective poleotolerance trait functional groups and their community weighted mean related with three type of environmental variables used as surrogates of urban disturbance.”
3&4) Results + conclusions: “We showed that disturbance-tolerant functional groups could be used as an ecological indicator of the integrated effects of environmental disturbances. Some species were clearly misclassified so we propose reclassification for those. Natural and semi-natural functional groups didn’t behave like expected, nevertheless, disturbance-tolerant functional groups have potential to be used in in others southern European cities”.
- The methodology recommends using only one species of phorophyte but in the study were used several unknown tree species. In this case tree have to be with similar bark properties and this means not bark roughness but in for example acidity and nutrient content. This is an important parameter which also influenced the results of the study. I suggest to show the influence of tree bark on species diversity or use only one tree species.
Answer: We understand the referee’s concern. We will try explaining better how this was done. In fact, it wasn’t possible to find always the same tree species in the 41 sampling sites. Most of them were done using Jacaranda mimosifolia, which was the most widely spread species in Lisbon green spaces. Around 60% of the sampling sites had these species. 30% were done in Tipuana tipu, which has similar bark characteristics to jacaranda. The remaining phorophytes belonged to other species that had a similar bark texture to this, most of them impossible to identify as the work was performed during winter and were all deciduous. Unfortunately, we have not recorded this information for each site. We will do that in future studies so we can say whether some species were specific to some phorophytes. Concerning bark exfoliation, we did select the phorophytes taking into consideration that all of them did not lose bark from exfoliation. Regarding bark pH, we used lichen traits related to preference for substrate pH and found that lichens preferring more basic substrate conditions dominated in the smaller green spaces (also with lower NDVI), and the community shifted towards lichens preferring more shadow and acid substrates in bigger and more densely vegetated substrates. This is expectable as smaller green spaces are less protected from dust deposition from traffic, which changes bark pH for more basic pH, and being dominated by more basophilous species. We have a previous work where we demonstrate it (reference 28 in the manuscript). This data is not shown in the paper. A recent work using the same methodology and in the same study area performed by us also analysed lichen functional groups related to water requirements (reference 14 in the manuscript). The only pattern found related more hygrophytic species (need more water) to bigger and more densely vegetated green spaces. These traits are responding to the main environmental variables underlying the disturbance gradient (smaller green spaces more disturbed than bigger ones) indicating that these are the predominant factors, and not the type of phorophyte, and suggesting that any factor linked to phorophyte characteristics did not have a major influence in the results.
- Data analysis: is it suitable for the data to use Spearman correlation?
Answer: While Pearson is used for linear relationships, Spearman is used for monotonic relationships (linear or not). Community data, such as ours, varies usually monotonically with environmental variables. Pearson correlations are more appropriate for linear relationships, while Spearmen perform better when the relationships tend to be monotonic. For that reason, we selected Spearman correlation coefficients for this work. We added information regarding why we selected this type of correlation coefficient in Materials and Methods. It now reads “Spearman correlations (to account for possible nonlinearity in the relationships) were performed between NMS scores, the CWM of each functional group and the environmental variables.”
- Figures: names of species are overlapped and therefore they are uninformative.
Answer: Figure 2 was changed to improve legibility, as requested by the reviewer.
- Check spaces between words and numbers (ex. 113 row: 1ha), superscripts (ex. 107: km2, 194: r2).
Answer: In line 142 “1ha” was replaced by “1 ha”. Superscript had already been corrected following suggestion by Reviewer number 1.
Author Response File: Author Response.docx
Reviewer 4 Report
This work investigates the association between epiphytic lichen species traits and land cover differences in the City of Lisbon. The aim was to use lichens as an integrative measure of spatial differences in human environmental disturbance. Land cover differences were mapped but environmental disturbances were not specifically identified, mapped or quantified and the relative contribution of one or more antagonistic/synergistic factors that may influence lichen establishment and survival distribution are not offered. As such, this was an attempt to compare lichen community traits that exist at the present time with present-day land use without providing insight into historical conditions or describing key regulator variables such as ambient air quality, smog days or known regulator gradients that might be defined using SO2, NOx and/or other constituents of ambient air.
This study inventoried lichen abundance on 4 phorophytes “near” the centroid of 41 urban green areas in the City of Lisbon. In total, 164 phorophytes and 656 quadrats were used. Key but unspecified assumptions were that the sampled trees (an indeterminate mix of species) had similar regulator variables (e.g., pH, light, humidity levels) regardless of micro-site conditions (solitary grown? overlapping canopies? different bark chemistry?) and that whatever mix of unmeasured variables existed was in balance with current stressors. These are arguably unreasonable expectations.
The study design and analysis used the ITALIC lichen trait classification system to calculate a mean weight (CWM) for lichens at each site and sites were placed into one of three site categories (green area, NDVI, surrounding land cover). Land cover was characterized and mapped using ArcGIS with land use subdivided into classes that are named but are not uniquely defined (e.g., herbaceous vegetation associations, green urban areas, land without current use, public, isolated structures, etc.). A land use weight was assigned to each sampling location based on seven radii (50m to 2000m) extending out from the centroid of the sampling location. The land use classes were presented as surrogates for human environmental disturbance that were statistically compared with the traits (CWM) of lichens on four trees at each site. This exercise produced functional groups of lichens.
After considerable discussion and analysis of correlations, the study reached conclusions that we should have guessed at the outset. We learned for example that a high CWM is found at the high end of a disturbance gradient. Is this not a circular argument?
I wanted to like this paper and I had hoped to find a useful approach that involves the use of GIS and lichen bioindication. However, I am left to wonder about the appropriateness of the statistical treatments.
Overall concerns:
1. Double-counting, renaming and pseudoreplication:
Size and radius: It is improper to reason that the size of the sampled green area is statistically independent from the length of a radius that includes that same green area. This is autocorrelation and pseudoreplication.
Land use and radius: It is also wrong to statistically contrast the quality of land use along the radius with values such as NDVI which are reported for the green area itself (which includes all or part of that radius). The authors mask this double-counting by using the word radii and “buffer” interchangeably. A radius is not a statistically independent “buffer” but in this work it is treated as such even when part of that radius is also part of the green area. This is autocorrelation, renaming and pseudoreplication.
Sampling location, area size, shape and centroid: The size and shape of each green area was used to identify the centroid of that area and the centroid defined where trees were sampled. Lichen values (CWM) at the centroid represent the sampled area. If the “buffer” is to remain statistically independent of the green area then each radius must begin outside of the green area itself. As written, it seems that the authors are correlating the attributes of each green area (e.g., area) with itself (e.g., the proportion of a radius or buffer that is green).
2. Trajectory of radii: Perhaps I missed it, but it is unclear if each radius follows a single trajectory that gets extended in length or if each radius of different length is an independent radius running at a different angle out from the centroid. Repeated use of the same trajectory would introduce autocorrelation and pseudoreplication.
3. Double counting: The size and shape of each green area was used to identify the centroid of that area and that centroid defined the sample location. Then, the green area attributes were described by area, NDVI, etc. Those NDVI and other values (including the rules used to pick the sampled trees) are not statistically independent of each other. Apparently the authors are correlating the attributes of each green area (e.g., area) with itself (the proportion of a radius that is green). This is pseudoreplication.
4. Sledgehammer: This study “uses a sledgehammer to crack a nut.” Just because GIS is able to generate 30+ categories for land classification doesn’t mean that those classes are useful/meaningful or should be used. A quick estimate of the green space surrounding a sampling site would provide a simple, fast and perhaps more effective measure of the amount of “buffer” to human disturbance exists at a site. The authors never tell us what advantage is provided by the approach used in this study. Why should anyone use this approach when simple approaches exist?
5. Stating the obvious: Consider for example Lines 349-354, where a link is made between big green spaces, high NDVI and low CWM. Readers should know that green space MUST have a high NDVI so there is no need to state the obvious.
6. Ignoring the obvious:
Nitrophytes: Unfortunately no distinction was made and no documentation or subanalysis was offered to investigate spatial differences in abundance between lichens that are known nitrophytes, acidophytes and neutrophytes. Why?
Water: The authors acknowledge that many other studies have found that waterbodies and micro-climate influence lichen abundance and distribution. Why is this not explored or controlled for in this study?
Human population: The introduction calls attention to urbanization and the growing human population, but the authors do not make a direct link between the human population distribution in the City of Lisbon and human environmental disturbance. Why? Where are the humans in this urban environment? Do high CWM areas and highest “anthropic disturbance” match the highest human population densities in this city?
Maps and raw data: Waterbodies are not labelled/shaded on the maps. Monsanto is mentioned several times but is not shown on any map. Is the raw data available in a data repository?
Author Response
Reviewer 4
This work investigates the association between epiphytic lichen species traits and land cover differences in the City of Lisbon. The aim was to use lichens as an integrative measure of spatial differences in human environmental disturbance. Land cover differences were mapped but environmental disturbances were not specifically identified, mapped or quantified and the relative contribution of one or more antagonistic/synergistic factors that may influence lichen establishment and survival distribution are not offered. As such, this was an attempt to compare lichen community traits that exist at the present time with present-day land use without providing insight into historical conditions or describing key regulator variables such as ambient air quality, smog days or known regulator gradients that might be defined using SO2, NOx and/or other constituents of ambient air.
Answer: We understand your concern, but the objective of this work is not to use lichens as an integrative measure of spatial differences in human disturbance nor to disentangle each of the multiple environmental drivers that are acting in Lisbon. Our objective is, as you can read in the introduction and abstract, to test the suitability of the poleotolerance response trait as an indicator of anthropic disturbance in a context of urban environments, particularly outside Italy, for where this classification was designed for. As such, we don’t believe such information should be present in the paper, as this was not its objective. The poleotolerance trait was not designed to disentangle different types of environmental disturbance, nor was it designed to distinguish between the impacts of different pollutants. It was designed as a trait that integrates the overall human disturbances, regardless of what they are. To clarify this, we added a sentence in the introduction saying this “The major potential of this trait is that it doesn’t focus solely on the effects of pollutants in lichens communities but rather it looks at all anthropic disturbances in an integrated way and also how those disturbances, their type and intensity have changed over the years”.
This study inventoried lichen abundance on 4 phorophytes “near” the centroid of 41 urban green areas in the City of Lisbon. In total, 164 phorophytes and 656 quadrats were used. Key but unspecified assumptions were that the sampled trees (an indeterminate mix of species) had similar regulator variables (e.g., pH, light, humidity levels) regardless of micro-site conditions (solitary grown? overlapping canopies? different bark chemistry?) and that whatever mix of unmeasured variables existed was in balance with current stressors. These are arguably unreasonable expectations.
Answer: We understand the referee’s concern. We will try explaining better how this was done. In fact, it wasn’t possible to find always the same tree species in the 41 sampling sites. Most of them were done using Jacaranda mimosifolia, which was the most widely spread species in Lisbon green spaces. Around 60% of the sampling sites had these species. 30% were done in Tipuana tipu, which has similar characteristics to jacaranda. The remaining phorophytes belonged to other species that had a similar bark texture to this, most of them impossible to identify as the work was performed during winter and were all deciduous. Unfortunately we have not recorded this information for each site. We will do that in future studies. Regarding micro-site conditions, in fact they varied. Small green spaces tend to have less canopy overlap, which results in higher deposition of dust and other pollutants (see Matos et al. 2019 for a discussion on this – reference 14 in the manuscript) and potentially higher light intensity. This is part of what we are looking for. This is why NDVI is used as a proxy of vegetation density and green space area to account for the fact that the bigger a green space area, the less disturbed the site is. In fact, we did find that using lichen traits related to preference for substrate pH and tolerance to light conditions lichens preferring more light and tolerating more basic substrate conditions dominated in the smaller green spaces (also with lower NDVI), and the community shifted towards lichens preferring more shadow and acid substrates in bigger and more densely vegetated substrates. This data is not shown in the paper. This results indicate that the disturbance gradient in the sampling site follows our expectations (smaller gardens more disturbed, bigger one less disturbed). In addition, this indicates that these are the predominant factors, and not the type of phorophyte, and that any factor linked to phorophyte characteristics did not have a major influence in the results.
The study design and analysis used the ITALIC lichen trait classification system to calculate a mean weight (CWM) for lichens at each site and sites were placed into one of three site categories (green area, NDVI, surrounding land cover).
Answer: I’m afraid this is a misunderstanding. Sites were not placed in one of three categories as suggested by the reviewer. For each sampling (the centroid of each green space), as we mention in M&M, we calculated three types of variables: area, NDVI (in the centroid and in a buffer of 100 m around it) and the surrounding land cover are. As so, each sampling site has its own area value, NDVI values (NDVIgs; NDVIb100) and characterization of surrounding land cover (area of each).
Land cover was characterized and mapped using ArcGIS with land use subdivided into classes that are named but are not uniquely defined (e.g., herbaceous vegetation associations, green urban areas, land without current use, public, isolated structures, etc.).
Answer: As pertinently identified by the reviewer, land covers were grouped into four different classes (HDLC; LDLCa; LDLCb; LDLCc). Those groups are not mentioned in the M&M because they are a result, where we defined them in the results (lines 291-300). To make it clear we will mention them in the Materials & Methods. To correct our lapse we have added, starting in line 217, “From the 25 land cover types, five were discarded (Airports; Construction sites; Mineral extraction and dump sites; Open spaces with little or no vegetation and Wetlands), as they weren’t significantly correlated with the CWM nor to the ordination scores, at any buffer distance (Table S1). For the remaining land covers, the best buffer distance was selected. After that, land cover types with the same buffer distance and similar correlation trend (correlation signals) with the first and second axes from the NMS and the CWMs of all three functional groups were grouped. Four groups were created (HDLC; LDLCa; LDLCb; LDLCc)”. Also, in the results, the paragraph regarding explanation and description of the grouped land covers was altered. Starting in line 282, it now states “Ordination scores and the CWM of all functional groups were correlated with the environmental variables to determine the size of the buffer to use for each land cover (Table S1). As previously stated, from the 25 land cover types, five were immediately discarded. Remaining land cover types were grouped based in their buffer distances and correlation trends. From there four groups were created. The first group was the HDLC (High disturbance land covers) comprising 5 land covers (all at a 500 m buffer) associated with high human disturbance (Discontinuous dense urban fabric; Industrial, commercial, public, military and private units; Port areas; Other roads and associated land; Water). The remaining three groups comprise land-covers associated with low human disturbance, the Low disturbance land covers (LDLC), with three different buffer distances: LDLCa, LDLCb and LDLCc, with 1000, 1500 and 2000 meters buffers, respectively. LDLCa comprises the Arable land, Herbaceous vegetation associations and Sports and leisure facilities land covers. LDLCb comprises the Fast transit roads and associated land, Land without current use, Pastures and Permanent crops. LDLCc comprises the Discontinuous very low-density urban fabric, Forests and Isolated structures”.
A land use weight was assigned to each sampling location based on seven radii (50m to 2000m) extending out from the centroid of the sampling location. The land use classes were presented as surrogates for human environmental disturbance that were statistically compared with the traits (CWM) of lichens on four trees at each site. This exercise produced functional groups of lichens.
Answer: As correctly stated by the reviewer, land covers area surrounding each sampling point were used as surrogates of human disturbance and were correlated with the CWM values of each functional group. However, the reviewer conclusion that this method produced the functional groups is incorrect. Functional groups are the result of grouping species based on their identical response/tolerance to a specific environmental variable. They are the way that within a single categorical trait (in this case human disturbance), different species are classified, or categorized. This information was not produced by us, but, as mentioned in M&M, was instead retrieved from the ITALIC site (http://dryades.units.it/italic/), (High disturbance, Low disturbance and Semi-natural and natural).
After considerable discussion and analysis of correlations, the study reached conclusions that we should have guessed at the outset. We learned for example that a high CWM is found at the high end of a disturbance gradient. Is this not a circular argument?
Answer: Although results can be more or less expected in some cases, based on common sense or previous data, such initial assumptions would be, somehow, reckless if unproved. Our study focused on an underexplored trait in an underexplored scenario (urban environment). Because of this, there was no guarantee that there would be a positive correlation between high CWM and areas with higher human disturbance. In fact, though a high CWM value for high disturbance species was found in the high end of the disturbance gradient, the same was not true for low disturbance species that, in this case, had the highest CWM values on the exact extreme, or to the other functional groups belonging to this trait. Overall, this is, fortunately, what should be expected, meaning that the trait classification is working as it was primarily designed for in the case of two functional groups (high and low disturbance). But our results also showed that this is not the case for semi-natural and natural functional groups.
I wanted to like this paper and I had hoped to find a useful approach that involves the use of GIS and lichen bioindication. However, I am left to wonder about the appropriateness of the statistical treatments.
Overall concerns:
1. Double-counting, renaming and pseudoreplication:
Size and radius: It is improper to reason that the size of the sampled green area is statistically independent from the length of a radius that includes that same green area. This is autocorrelation and pseudoreplication.
Answer: I’m afraid there was a misunderstanding regarding the buffers and how the three types of environmental variables characterizing each site were used. Although we agree that the radius of the buffer and the size of the sampled green areas are not independent, we have no reason to believe that that would affect our results. Environmental variables were not correlated between themselves. They were only correlated with CWM and ordination axes. So this case does not apply to our results. Secondly, just to clarify, radii were not used for any statistical purposes in our manuscript. We simply used ArcGis software to create buffers (circles) with different sizes (therefore with different radius) around each sampling site and then used those buffers (circles) to identify which land cover types were surrounding those sites at different distances and to calculate the area they occupy. So again, we believe this concern does not apply to our results.
Land use and radius: It is also wrong to statistically contrast the quality of land use along the radius with values such as NDVI which are reported for the green area itself (which includes all or part of that radius). The authors mask this double-counting by using the word radii and “buffer” interchangeably. A radius is not a statistically independent “buffer” but in this work it is treated as such even when part of that radius is also part of the green area. This is autocorrelation, renaming and pseudoreplication.
Answer: Again, we did not use the radius to identify the land covers surrounding the sampling sites. The radius of each buffer was solely used and mentioned in our manuscript in order to give the reader the perception of the dimension and size of the drawn buffers. In order to clarify this, we added the sentence “Buffers should be interpreted as drawn imaginary circles, with different radius, created with the sole intention to delimit distinct areas around the sampling sites. These buffers were used to identify all land covers surrounding all the sampling sites, at different distances” in line 206-209. Because, unlike the poleotolerance trait, it wasn’t possible for us to find a single environmental variable capable of depicting the integrated driver called human disturbance, three different environmental variables were used as surrogates of three different drivers of urban disturbance. These though obviously correlated between them (e.g. if we have more green are around a sampling site, this implies we have of course less urban fabric), they were related statistically with each other in the analysis performed in this paper.
Sampling location, area size, shape and centroid: The size and shape of each green area was used to identify the centroid of that area and the centroid defined where trees were sampled. Lichen values (CWM) at the centroid represent the sampled area. If the “buffer” is to remain statistically independent of the green area then each radius must begin outside of the green area itself. As written, it seems that the authors are correlating the attributes of each green area (e.g., area) with itself (e.g., the proportion of a radius or buffer that is green).
Answer: As correctly observed by the reviewer, in order for the buffer to remain statistically independent of the green area then each radius must begin outside of the green area itself. However, it is not our intention in this manuscript to correlate the areas of the green spaces with the buffers and associated land covers. Both of them are environmental variables and in this study, environmental variables were not correlated between them but rather correlated with the axis of the NMS and the CWM of the functional groups.
2. Trajectory of radii: Perhaps I missed it, but it is unclear if each radius follows a single trajectory that gets extended in length or if each radius of different length is an independent radius running at a different angle out from the centroid. Repeated use of the same trajectory would introduce autocorrelation and pseudoreplication.
Answer: We believe this is in fact a misunderstanding. Again to clarify this matter, radii were not used for any statistical method in our manuscript. We simply used ArcGis software to create circles (buffers) with different sizes (therefore with different radius) around each sampling site and then used those buffers (circles) to identify what land covers types were surrounding those sites at different distances and to calculate the area they occupy. So again, we believe this concern does not apply to our results.
3. Double counting: The size and shape of each green area was used to identify the centroid of that area and that centroid defined the sample location. Then, the green area attributes were described by area, NDVI, etc. Those NDVI and other values (including the rules used to pick the sampled trees) are not statistically independent of each other. Apparently the authors are correlating the attributes of each green area (e.g., area) with itself (the proportion of a radius that is green). This is pseudoreplication.
Answer: As we mentioned previously, we did not correlate the characteristics of each green space (our environmental variables) between them. They were only correlated with the CWM of the functional groups and ordination axes of the NMS.
4. Sledgehammer: This study “uses a sledgehammer to crack a nut.” Just because GIS is able to generate 30+ categories for land classification doesn’t mean that those classes are useful/meaningful or should be used. A quick estimate of the green space surrounding a sampling site would provide a simple, fast and perhaps more effective measure of the amount of “buffer” to human disturbance exists at a site. The authors never tell us what advantage is provided by the approach used in this study. Why should anyone use this approach when simple approaches exist?
Answer: We believe that also here, the concept of buffer was misinterpreted. Regarding the suggestion made by the reviewer that a “quick estimate of the green space surrounding a sampling site would provide a simple, fast and perhaps more effective measure of the amount of “buffer” to human disturbance exists at a site”, we want to clarify again that the buffer around each sampling site is not a buffer created to mitigate the urban disturbances surrounding the sampling sites but rather an imaginary circle created with ArcGis software to identify an area around it and enable us to calculate the land covers types and its areas surrounding the sampling sites. Again, to clarify this, we added the sentence “Buffers should be interpreted as drawn imaginary circles, with different radius, created with the sole intention to delimit distinct areas around the sampling sites. These buffers were used to identify all land covers surrounding all the sampling sites, at different distances” The use of land covers rather than an estimate of the green space surrounding a sampling site are important because we use them as a proxy for the human disturbance. For example, a green space surrounded by urban fabric is substantially more disturbed than another one nearby water from a lake or river. Therefore it’s important to understand the land covers around the sampling sites and not only focus on the green spaces. However, like stated by the reviewer if taken and treated individually, the amount of land covers in Lisbon would be too many. Because of that, like described in the materials and methods, from the 25 initial land covers, 5 were immediately taken away. The remaining 20 were the grouped into 4 groups. This was done precisely to address the concern stated by the reviewer in this point.
5. Stating the obvious: Consider for example Lines 349-354, where a link is made between big green spaces, high NDVI and low CWM. Readers should know that green space MUST have a high NDVI so there is no need to state the obvious.
Answer: Although this may be truth, bigger green areas do not necessarily translate into higher NDVI values. NDVI values are higher or lower depending on the amount of surface covered by green vegetation. For example, a big green area with trees but with bare soil will likely have a lower NDVI value than a smaller green area also with trees but with grass instead of bare soil. The link between higher NDVI, bigger green spaces and low CWM may also not be true and falls also on another doubt the reviewer stated previously. Like answered in that point, “Although results can be more or less expected in some cases, based on common sense our previous data, in our case such initial assumption would be, somehow, reckless. Our study focusses on an underexplored trait in an underexplored scenario (urban environment). Because of this, there was no guarantee that there would be a positive correlation between high CWM and areas with higher human disturbance”. And though one CWM is low (Semi-natural functional group), other are high. So it is important to underline these relationships.
6. Ignoring the obvious:
Nitrophytes: Unfortunately no distinction was made and no documentation or subanalysis was offered to investigate spatial differences in abundance between lichens that are known nitrophytes, acidophytes and neutrophytes. Why?
Water: The authors acknowledge that many other studies have found that waterbodies and micro-climate influence lichen abundance and distribution. Why is this not explored or controlled for in this study?
Answer: This comment is highly pertinent. We didn’t use other functional traits, as we were focusing on the poleotolerance trait. However, this relationship was investigated by us earlier and published in another paper cited in the manuscript (Matos et al. 2019 – reference 14 in the manuscript). Nonetheless, because we understand that this could be an interesting topic to help discuss the advantages of this trait over those, we have added a paragraph to the discussion focused on this: “In this work we intentionally focused on a single trait, poleotolerance, with the purpose of understanding if this trait could be useful as an indicator of anthropic disturbances in an urban context. Our results showed a shift in lichen communities from low tolerance species to high tolerance species as we move to more disturbed areas. A recent work [14] using the same methodology in the same study area explored other three different traits, growth form, eutrophication tolerance and humidity requirements, and did not observe any shifts in lichen communities mediated by these three traits. This is an indication that in urban environments, such as this encountered in Lisbon, the use of the poleotolerance trait may be more advantageous then using other more commonly used traits, giving a clearer indication about the disturbance gradient. Humidity requirements, growth form and eutrophication tolerance are frequently used to indicate disturbance related to the urban heat island effect (microclimatic variations) and nitrogen pollution. The fact that poleotolerance trait performed better here the advantage of using it when we aren’t able to clearly disentangle those two environmental drivers, highlighting its potential as an integrated indicator of overall anthropic disturbance in urban settlements. The presence of pruina or secondary metabolites showed also promising results in South American cities as indicators of the effects of disturbances associated to urban environments [28,57]. In our work we did not test these traits. The poleotolerance is, for now, restricted to southern European cities that share some of the lichen flora of Italy. Pruina or secondary metabolites have the advantage of being independent of this classification, having therefore more potential to be applied outside Europe, so future works should try to explore it in Europe, to understand if they have also the potential to be used here”.
Human population: The introduction calls attention to urbanization and the growing human population, but the authors do not make a direct link between the human population distribution in the City of Lisbon and human environmental disturbance. Why? Where are the humans in this urban environment? Do high CWM areas and highest “anthropic disturbance” match the highest human population densities in this city?
Answer: Although the idea expressed by the reviewer is highly pertinent and could be of great interest for this work as complementary information that would allow for better contextualization of our results, the true is that we don’t have the human population densities data at the spatial scale of this work. However, the “Continuous urban fabric (S.L.: >80%); Discontinuous dense urban fabric (S.L.: 50%-80%); Discontinuous medium density urban fabric (S.L.: 30%-50%); Discontinuous low density urban fabric (S.L.: 10%-30%); Discontinuous very low density urban fabric (S.L.: <10%)” land covers were used as proxy for those densities, as more or less urban fabric is connected to more or less human densities.
Maps and raw data: Waterbodies are not labelled/shaded on the maps. Monsanto is mentioned several times but is not shown on any map. Is the raw data available in a data repository?
Answer: We have shaded Tagus river with a light blue colour, in Figure 1 and Figure 4(A/B/C) to improve visualization of those maps. Regarding Monsanto, we changed the colour of all green spaces sampled in our work and added the label “Monsanto Park” on top of the respective green space. Besides, to follow the reviewer suggestion to better highlight it, we added information in the legend of Figure 1. It now reads: “Figure 1. Study area showing the distribution of the sampling sites (red circles; n=41) within the green spaces and the city’s land covers (adapted from Urban Atlas 2012). Green Spaces; CUF - Continuous urban fabric (S.L.: >80%); DLDUF - Discontinuous low density urban fabric (S.L.: 10%-30%); DMDUF - Discontinuous medium density urban fabric (S.L.: 30%-50%); GUA - Green urban areas; HDLC - High disturbance land covers; LDCLa - Low disturbance land covers (1000 m buffer); LDCLb - Low disturbance land covers (1500 m buffer); LDCLc - Low disturbance land covers (2000 m buffer); Other land covers and RAL - Railways and associated land.”Monsato Park” is the biggest green space present in our study area”.
Regarding the raw data, we did not make it available in a data repository as that is not mandatory for this journal.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
Dear Editor,
The revised version of the paper has been deeply improved, and the authors have clarified the “missing points” of the previous version.
I think that this new version is suitable for publication with a last minor revision: authors should add some more information about phorophytes (at lines Lines 154-155: “To address this limitation, only phorophytes with medium bark roughness were used for sampling.”): in particular tree species and the discussion about factors linked to bark characteristics vs. environmental characteristics as they reported in the “Author's Notes” document.
Best regards
Author Response
Reviewer 1:
The revised version of the paper has been deeply improved, and the authors have clarified the “missing points” of the previous version.
I think that this new version is suitable for publication with a last minor revision: authors should add some more information about phorophytes (at lines Lines 154-155: “To address this limitation, only phorophytes with medium bark roughness were used for sampling.”): in particular tree species and the discussion about factors linked to bark characteristics vs. environmental characteristics as they reported in the “Author's Notes” document.
Answer: We feel your concerns are highly pertinent. Therefore, we added the following sentence in line 159: “We estimate that 60% of phorophytes were Jacaranda mimosifolia D. Don, as this species is one of the most widely spread species in Lisbon green spaces, with the adequate characteristics for this work. Tipuana tipu (Bentham) Kuntze accounted for around 30% and the rest were from unidentified species as sampling was conducted during winter and these remaining phorophytes were all deciduous.”
Author Response File: Author Response.docx
Reviewer 4 Report
I thank the authors for their detailed response and I am now in favor of publication. Their comments have greatly added to my understanding of their work. However, I will attempt to make a few constructive suggestions for the authors/editor to consider. The aim after all is to make the paper stronger and more useful - easier to follow.
I must admit that I do not use the poleotolerance approach and was struggling to follow what was being done. I now understand "buffers" and "radius" ... the newly added sentence is very helpful.
I now appreciate that the authors are not attempting to elucidate the causal factors that drive lichen distribution in urban areas. But, other readers (who seldom see the word "poleotolerance" and do not use the poleotolerance approach) might need to be better educated about what the approach involves and how it differs from what others are doing. Readers may be familiar with other approaches e.g., lichen zones are mapped using the IAP or other indices or by identifying zones based on species richness. Those studies may also use 4 trees per area as the sample sites but they do not use GIS to identify buffers and radii (imaginary circles etc.). Some of those papers are also concerned about the diffusion of pollutants from a point source and the location of sampling sites up or downwind from the point source .. etc. In that sense it can be argued that one site is not independent of another.. hence spatial and temporal auto-correlation may be a concern to some readers. Readers will immediately question why this study does not consider upwind/downwind or any source at all. Much of this is well explained in the new paragraph you have added (in response to my comment #6) but it would help if the basics were better explained at the beginning of the paper .. so as to educate and help set the reader's expectations.
This could involve a clear comparison of the poleotolerance approach with other approaches.. followed by an explanation of why the poleotolerance approach is useful.. then... ending the paper by suggesting possible next steps in this research (e.g., subsample to identify abiotic differences such as the amount of paved area or concrete or identify water, human population distribution, wind direction, or whatever etc.). You might want to briefly mention that one of the limitations of the paleotolerance approach is that it creates/uses boundaries that cannot readily be matched to some other data sets (e.g., the spatial distribution of the human population).
I suggest these things not as a criticism but as a way to strengthen the paper and increase its value to an undergraduate/graduate student who may be missing the logic/uniqueness of the poleotolerance approach.
The extra sentence that was added helps us to better understand the GIS process. It would also be helpful to briefly explain how NDVI may/may not increase as the size of the green area increases (as was done in your response to my comment).
Good to see the changes to the map and caption.
Overall I'm impressed by the detail and the organization of the paper. However, the paper is densely written and this revised version could use a few more words added near the beginning and in the discussion to lighten up the paper and make key points clearer. I strongly suspect that an advanced undergraduate student (or a tired academic) who seldom sees the word "poleotolerance" would soon have questions right near the start of the paper. This could be addressed by reminding us of key concepts. Failing to do this could turn readers off right at the start. The goal (as I see it) should be to educate the reader and guide them through the methodology and the stats.
Author Response
Reviewer 4:
I thank the authors for their detailed response and I am now in favor of publication. Their comments have greatly added to my understanding of their work. However, I will attempt to make a few constructive suggestions for the authors/editor to consider. The aim after all is to make the paper stronger and more useful - easier to follow. I must admit that I do not use the poleotolerance approach and was struggling to follow what was being done. I now understand "buffers" and "radius" ... the newly added sentence is very helpful.
Answer: We are glad that your main concerns were addressed. Thank you for all your previous suggestions/request as they have improved the overall quality of the manuscript.
I now appreciate that the authors are not attempting to elucidate the causal factors that drive lichen distribution in urban areas. But, other readers (who seldom see the word "poleotolerance" and do not use the poleotolerance approach) might need to be better educated about what the approach involves and how it differs from what others are doing.
Answer: We do understand your concerns and fully agree with your point. To address this issue we added, in line 106, the following sentence. “This trait is not intended to distinguish between different human disturbance sources, like other traits (e.g. eutrophication, humidity requirements or bark pH), but as a simple integrating trait that indicates disturbance as a whole. In this sense, it could be a useful quick tool to help signal overall human disturbance, for instance, signalling areas that deserve attention and where future studies should focus to identify the individual environmental factors driving the impacts.”
Readers may be familiar with other approaches e.g., lichen zones are mapped using the IAP or other indices or by identifying zones based on species richness. Those studies may also use 4 trees per area as the sample sites but they do not use GIS to identify buffers and radii (imaginary circles etc.). Some of those papers are also concerned about the diffusion of pollutants from a point source and the location of sampling sites up or downwind from the point source .. etc. In that sense it can be argued that one site is not independent of another, hence spatial and temporal auto-correlation may be a concern to some readers. Readers will immediately question why this study does not consider upwind/downwind or any source at all. Much of this is well explained in the new paragraph you have added (in response to my comment #6) but it would help if the basics were better explained at the beginning of the paper .. so as to educate and help set the reader's expectations. This could involve a clear comparison of the poleotolerance approach with other approaches.. followed by an explanation of why the poleotolerance approach is useful..
Answer: Thank you for expressing your concerns. However, like said in the last answer, this trait advantage compared to other traits is that it works as a proxy of the various human disturbances as a whole, thus the integrated view he talk in the introduction “The major potential of this trait is that it doesn’t focus solely on the effects of pollutants in lichens communities but rather it looks at all anthropic disturbances in an integrated way and also how those disturbances, their type and intensity have changed over the years. This trait is not intended to distinguish between different human disturbance sources, like other traits (e.g. eutrophication, humidity requirements or bark pH), but as a simple integrating trait that indicates disturbance as a whole. In this sense, it could be a useful quick tool to help signal overall human disturbance, for instance, signalling areas that deserve attention and where future studies should focus to identify the individual environmental factors driving the impacts”, (line 103). It’s not our intention in this manuscript to subdivide the human disturbance into the different possible disturbance sources usually existent in urban environments as if that was the objective other traits would be more suitable and would have been used (nitrogen pollution -> eutrophication trait; Dust -> pH substrate trait; etc). Our goal here was to test this particular trait because it has the advantage (compared to other mentioned above) of being quick tool to see disturbances in urban environments as a whole without the need to look at each of its components individually.
then... ending the paper by suggesting possible next steps in this research (e.g., subsample to identify abiotic differences such as the amount of paved area or concrete or identify water, human population distribution, wind direction, or whatever etc.).
Answer: We have added some sentences in discussion (line 452) and conclusions (line 463) highlighting your concerns:
· Discussion sentence: “We also suggest, for future works centered in this trait that, if possible, datasets of population densities and distributions should be used as these types of data can be a very suitable proxy for human disturbance levels inside the studied areas.”
· Conclusion sentence: “Because this is an integrative trait, it has the potential to be used even when we do not know the underlying environmental factors driving changes, signalling disturbed areas that should deserve further attention in future studies to individually asses the drivers of change, a very important tool under an urban management perspective.
You might want to briefly mention that one of the limitations of the poleotolerance approach is that it creates/uses boundaries that cannot readily be matched to some other data sets (e.g., the spatial distribution of the human population).
Answer: This is a very good point, because like you said, human population densities and distribution in urban environments can be a very good proxy for disturbance levels. We would like to clarify that the fact that we weren´t able to cross our data with this kind of population datasets is not a disadvantage of the trait per se but is rather a limitation that arises from the fact that this kind of data is not available in our country with the required spatial resolution needed for our work. We did however added a sentence in the discussion (line 452) clarifying this: “We also suggest, for future works focusing on this trait that, if possible, datasets of population densities and distributions should be used as these types of data can be a very suitable proxy for human disturbance levels inside the studied areas”.
I suggest these things not as a criticism but as a way to strengthen the paper and increase its value to an undergraduate/graduate student who may be missing the logic/uniqueness of the poleotolerance approach.
The extra sentence that was added helps us to better understand the GIS process. It would also be helpful to briefly explain how NDVI may/may not increase as the size of the green area increases (as was done in your response to my comment).
Answer: We added the following sentence in the Materials&Methods (line 196): “NDVI is used as a proxy of vegetation density and green space area. However, that doesn’t necessarily mean that bigger green spaces have higher NDVI values as the NDVI values are higher or lower depending on the amount of surface covered by green vegetation and not on the area of the green space (e.g. a bigger green space with trees and bare soil will have lower NDVI values than a smaller green space with trees and grass instead of bear soil).”
Good to see the changes to the map and caption. Overall I'm impressed by the detail and the organization of the paper. However, the paper is densely written and this revised version could use a few more words added near the beginning and in the discussion to lighten up the paper and make key points clearer. I strongly suspect that an advanced undergraduate student (or a tired academic) who seldom sees the word "poleotolerance" would soon have questions right near the start of the paper. This could be addressed by reminding us of key concepts. Failing to do this could turn readers off right at the start. The goal (as I see it) should be to educate the reader and guide them through the methodology and the stats.
Answer: Thank you so much. We believe that we the extra sentences and concepts we’ve added following your suggestions, the paper is now written in a clear and easier to follow.
Author Response File: Author Response.docx
