What Insight Does the Alien Plant Species Richness in Greece Offer for the Different Invasion Biology Hypotheses?

Round 1

Reviewer 1 Report

In this manuscript, Kallimanis and colleagues present an interesting study of alien plant species richness in Greece, analyzing patterns of different invasion biology hypotheses. I think that this type of study is very important to understand ongoing processes connected with alien plant invasions. Overall, the authors do a pretty good job of explaining a complicated system related to invasive alien plant species. The analyses are appropriately conducted, and the authors cover interesting points in the discussion. The manuscript thus significantly increases our knowledge regarding problems with alien plant species related to a very specific Mediterranean eco-system. I am looking forward to seeing this work published.

Author Response

Thank you for your supportive comments

Reviewer 2 Report

Review of “What insight does the alien plant species richness in Greece offer for the different invasion biology hypotheses?”

 

In this article, the authors attempt to understand the causes of alien plant species in Greece in relation to current invasion biology hypotheses. The authors present a correlative study of alien species abundance with abiotic and biotic variables. Strong significant correlations were found between native and alien species richness and different variables. However, the authors failed to produce any insights into these patterns as they relate to invasive biology hypotheses. This may be attributed to misinterpretation of the data, lack of multivariate data analysis and contradicting statements. I, unfortunately, cannot approve the manuscript for publication in Diversity in its current state. Below, I provide detailed comments:

Introduction:

L87-90: More empirical information is needed to describe the number of native and alien species and their distribution.

L97: A hypothesis cannot shape patterns!

L100-101: Explain how your results are expected to change or add to the conservation policy or at all.

L103: What does MAES stand for?

Materials and Methods

L111: “more than 1.2 M occurrences” of how many species? How many are native and how many alien?

L112: What does EEA stand for?

Selection of variables:

The authors need to explain why they only selected mean annual temperature, temperature seasonality, total annual precipitation, and precipitation seasonality and not all 19 bioclim variables and after that only selected variables that were not strongly correlated. This is a major caveat in the study and needs to be reconsidered.

L121: How fine is fine resolution?

L124: Does “cell” refer to the 10X10 grid?

L125: How does landscape data from land cover classes?

L126: “To quantify the impact of the transport network…” on what?

L127-130: How did you examine whether using Corine land cover data from a different period?

If you found quantitative differences, does that not mean the number of patches will be different? That would mean differences in fragmentation.

Why have you selected the 2000 CLC and not later years such as 2018? Surely the most recent data would be the most accurate.

L140: What other landcover types would be less than 70%, after you have excluded areas with no record?

L149-149: You did not test the hypotheses, instead you determined which hypothesis best explains your results. Please rephrase.

L150: “To test the propagule pressure hypothesis”: what would the assumptions be about these areas and alien species abundance?

L156: The empty niche hypothesis: I do not see that you have quantified the “climatic” niche or environmental space of native and alien species. For this, you needed to have done some multivariate study where you can determine if these groups of species actually have some differences in climate distribution.

It is also important the authors describe the major vegetation types in Greece. As it is a Mediterranean climate there might be some common clades that have evolved there, which would be different or the same from alien clades. This may influence the competition or environmental susceptibility between species.

 

Results

L171: How do you define a hotspot?

L172: Describe the distribution in terms of the CLC.

Table 1:

Add a column to Table 1 that groups the predictors into different hypotheses (as you did in Table 2). For example, if you combine the predictor variables for climate you will find that the sum of 28.7 %, is greater than the native species richness (27.9%). This disproves your suggestion that the biotic acceptance or biotic resistance hypothesis is at play here. Even more apparent is that if you landscape diversity together you have a value of 35.5%. This indicates that the resource availability cluster of hypotheses probably best explains the abundance of alien species in Greece.

The study could also benefit from multivariate results of native and alien species in relation to different predictors. This may provide a combination of factors/hypotheses at play.

L210: “The effect of mean annual temperature is almost linear with higher 210 alien species richness observed in the hottest areas”. This is not true! But where is the correlation coefficient to support your statement? They should be presented for all the predictors tested, e.g., for Vegetation displayed the opposite trend, with lower alien species richness in areas 217 dominated by natural and semi-natural vegetation. Did you divide the classes further into natural and semi-natural vegetation? This was not described or mentioned in your materials and methods.

L234: Do not sound too subjective in your results section by using words such as “interestingly” and “surprisingly”

L244: How ecologically significant is a difference in 0.6 degrees Celsius? How high or low is species rich under different climatic variables?

L286-287: This statement is contradictory to your results that showed that alien species richness increase with native species rich.

L299-300: This is a contradictory statement to L234.

L335: Contradictory!

The section on Solanum elaeagnifolium is perhaps the most interesting part of the study. I would encourage that the authors rather focus the study on this one species and examine which hypothesis best explains its abundance in Greece.

References

 

I found numerous references that are not in the format of Diversity. Please revise. 

Overall use of English is fine.

Author Response

Thank you for your comments. Following each comment our response is in italics. 

 

In this article, the authors attempt to understand the causes of alien plant species in Greece in relation to current invasion biology hypotheses. The authors present a correlative study of alien species abundance with abiotic and biotic variables. Strong significant correlations were found between native and alien species richness and different variables. However, the authors failed to produce any insights into these patterns as they relate to invasive biology hypotheses. This may be attributed to misinterpretation of the data, lack of multivariate data analysis and contradicting statements. I, unfortunately, cannot approve the manuscript for publication in Diversity in its current state. Below, I provide detailed comments:

 

Introduction:

L87-90: More empirical information is needed to describe the number of native and alien species and their distribution.

We have added information the number of plant taxa recorded in Greece and are endemic to Greece (please see lines 88-90 of our revised manuscript)

 

L97: A hypothesis cannot shape patterns!

Yes, the reviewer is right it was an unfortunate expression, we revised the text to improve clarity (please see lines 97-99)

 

L100-101: Explain how your results are expected to change or add to the conservation policy or at all.

We have added a sentence on the potential application of our findings on conservation policy (please see lines 106-108). Furthermore, we have discussed this issue more extensively in the end of our manuscript.

 

 

L103: What does MAES stand for?

MAES stands for Mapping and Assessment of Ecosystems and their Services (as we clarify in line 104-105)

 

Materials and Methods

L111: “more than 1.2 M occurrences” of how many species? How many are native and how many alien?

In the revised text we report that those occurrences are from about 7450 native species and subspecies and 457 alien species and subspecies (please see lines 116-117).  

 

L112: What does EEA stand for?

EEA stands for European Environment Agency (as we clarify in line 119)

 

Selection of variables:

The authors need to explain why they only selected mean annual temperature, temperature seasonality, total annual precipitation, and precipitation seasonality and not all 19 bioclim variables and after that only selected variables that were not strongly correlated. This is a major caveat in the study and needs to be reconsidered.

We revised this part to clarify that the selection of only three of the 19 variables was due to the collinearity among the entire dataset and necessary to avoid the issues of collinearity in our analysis (please see line 122-125).

 

L121: How fine is fine resolution?

The finest Corine thematic resolution is the third level of their thematic classification scheme and it includes 44 classes (please see line 131-132).

 

L124: Does “cell” refer to the 10X10 grid?

Yes (please see line 133).

 

L125: How does landscape data from land cover classes?

CLC1990 is the dataset of Corine Land Cover for 1990 which uses the classification of the different land cover classes.

 

L126: “To quantify the impact of the transport network…” on what?

On diversity patterns (please see line 136).

 

L127-130: How did you examine whether using Corine land cover data from a different period?

If you found quantitative differences, does that not mean the number of patches will be different? That would mean differences in fragmentation.

Why have you selected the 2000 CLC and not later years such as 2018? Surely the most recent data would be the most accurate.

We chose the 1990 dataset because that was the period when most of the plant diversity data were collected. To examine whether using Corine land cover data from a different period would affect the observed results we repeated the analysis keeping the other data and using landscape data from different time periods (with a more detailed focus on the 2000 dataset). There were differences in the numbers observed per cell, but those differences were random and did not display a bias, so the overall statistical inference holds. 

 

L140: What other landcover types would be less than 70%, after you have excluded areas with no record?

The Corine land cover dataset includes also water bodies such as marine areas.

 

L149-149: You did not test the hypotheses, instead you determined which hypothesis best explains your results. Please rephrase.

The reviewer is correct. We rephrased this paragraph to improve clarity (please see lines 157-178)

 

L150: “To test the propagule pressure hypothesis”: what would the assumptions be about these areas and alien species abundance?

Following the previous point we rephrased the entire paragraph. We explain that the propagule hypothesis assumes that alien species abundance and richness should be higher closer to the points of introduction of the alien species in the territory (i.e. close to ports, airports, train and road networks) (please see lines 162-166).

 

L156: The empty niche hypothesis: I do not see that you have quantified the “climatic” niche or environmental space of native and alien species. For this, you needed to have done some multivariate study where you can determine if these groups of species actually have some differences in climate distribution.

As the reviewer correctly pointed out, in this study we did not test the invasion biology hypotheses directly, but indirectly. We examine if and to what extent the patterns and drivers of alien and native species richness are in accordance with the predictions of the different hypotheses. Thus, we did not carry out distribution modelling but we did build multivariate models for the patterns of diversity and their drivers. The main assumption of the empty niche hypothesis is that the availability of unoccupied niches is a key factor driving alien and native diversity. According to this hypothesis, successful invaders can exploit empty niches that are not being used by native species. If alien species utilize different niche space than native species, the effect of predictors on the diversity patterns of aliens and natives will be different. This is our indirect way to assess the importance of the empty niche hypothesis (please see lines 171-175).    

 

It is also important the authors describe the major vegetation types in Greece. As it is a Mediterranean climate there might be some common clades that have evolved there, which would be different or the same from alien clades. This may influence the competition or environmental susceptibility between species.

Unfortunately, information of the clade that a specific population originates from and possible differences between clades of the same species are not available for the scale of our analysis. Also vegetation in the present study was quantified using the Corine land cover classification scheme that identifies the same vegetation classes throughout the European Union.

 

Results

L171: How do you define a hotspot?

We define hotspots as areas of highest species richness. To improve clarity, we rephrased the sentence (please see line 186).

 

L172: Describe the distribution in terms of the CLC.

Done (please see lines 187-188).

 

Table 1:

Add a column to Table 1 that groups the predictors into different hypotheses (as you did in Table 2). For example, if you combine the predictor variables for climate you will find that the sum of 28.7 %, is greater than the native species richness (27.9%). This disproves your suggestion that the biotic acceptance or biotic resistance hypothesis is at play here. Even more apparent is that if you landscape diversity together you have a value of 35.5%. This indicates that the resource availability cluster of hypotheses probably best explains the abundance of alien species in Greece.

To estimate the combined effect of climatic variables on the distribution of species richness we performed a multivariable GAMM, the results presented in the next table of our results, where it is apparent that the climatic model explains 15.9% of the variance (the differences is due to the fact that part of the effect of one variable is also explained by another variable and thus the combined model accounts for less than the sum of the individual models). It is true that the combined model for the landscape factors accounts for slightly more variance than the native species richness model alone, but it also includes several parameters and the combined effect could not be associated with a single hypothesis but several different hypotheses like propagule pressure, empty niche, resource availability as discussed in more detail. 

 

The study could also benefit from multivariate results of native and alien species in relation to different predictors. This may provide a combination of factors/hypotheses at play.

The results of the multivariate models are presented in Table 2, Figures 2 and 3 and in the results lines 241-277.

 

L210: “The effect of mean annual temperature is almost linear with higher 210 alien species richness observed in the hottest areas”. This is not true! But where is the correlation coefficient to support your statement? They should be presented for all the predictors tested, e.g., for Vegetation displayed the opposite trend, with lower alien species richness in areas 217 dominated by natural and semi-natural vegetation. Did you divide the classes further into natural and semi-natural vegetation? This was not described or mentioned in your materials and methods.

The analysis this refers to is the GAMM model that includes all variables, but these specific patterns (for mean annual temperature, and vegetation) were also observed in the single variable mode. This models also correct for the spatial autocorrelation observed in both our dependent and independent variables. The GAMM models typically report some goodness of fit statistics such as the coefficient of determination (R²) or Akaike information criterion to judge the strength of the model. Here we used R² for this purpose. Table 1 presents the statistics for the single predictor models and table 2 and extensive part of the results focus on the statistics of the combined model. The relationships examined are non-linear and complex and thus correlation coefficients are not typically estimated. To assess the importance of different predictors in the overall fit of the GAMM model, we compared the fit of the complete model with the fit of the equivalent model with all other predictors except the one we are interested in, and the greater the loss of fitness the greater the importance of this predictor for the complete model.

The reviewer has a point on the difficulty of visually representing the effects of predictors in GAM models. In contrast with linear regression models, in nonlinear regression models one cannot interpret the shape of the regression line from the summary. Therefore, visualization is an important tool for interpretating nonlinear regression models. Here we plotted the partial effects of the predictor variables using the plot.gam function of mgcv package. In essence these plots keep constant the value of all other predictors, and show how the predictor of interest affects the dependent variable. These plots for all predictors are included in figure 2. There the second panel in the first line displays the partial effect of mean annual temperature in the overall GAMM model predictions, and could be seen that the relationship is not linear, initially it increases rapidly and then it increases slowly to not at all. The overall pattern is that areas of high alien species richness are areas of above average mean annual temperature, similarly the panel in the third line shows the partial effect of vegetation on the GAMM model which is linear and negative.

In our study for quantifying landscape metrics we used the Corine land cover classification. The official title of the third group of land cover classes is “forests and semi natural vegetation” for ease of comprehension we referred to this class as vegetation and includes both natural and semi-natural vegetation (e.g. pastures). We did not analyze the difference between the two and therefore did not mention it in our methods section.

 

 

L234: Do not sound too subjective in your results section by using words such as “interestingly” and “surprisingly”

Done we removed the words interestingly and surprisingly from our manuscript.

 

L244: How ecologically significant is a difference in 0.6 degrees Celsius? How high or low is species rich under different climatic variables?

This is a very interesting question. In our study when we refer to significance, we refer to statistical significance. But indeed, the ecological significance is a far more important question for which we cannot provide a formal answer. A back of the envelope quick and dirty answer may be that if mean annual temperature increases by 0.6 degrees and all else remains constant, we may see an average increase of alien species richness by 0.5 species and an average decrease of native species by 3.2 species in every cell of our analysis. But these rough estimates are subject to many assumptions, and we could not argue that this is a scientifically robust answer, therefore we have not added anything in our manuscript.

 

L286-287: This statement is contradictory to your results that showed that alien species richness increase with native species rich.

This sentence further strengthens the argument that alien plant species richness increases with native plant species richness in Greece. It underlines that this is not observed only in the scale of our analysis (with grain size of 100 km²) but has also been observed in a paper reporting a fine scale study (with grain size orders of magnitude smaller of about 100 m²). We revised the sentence to improve clarity (please see line 303-306).

 

L299-300: This is a contradictory statement to L234.

Perhaps our initial text was not clear. The mechanism in the lines mentioned is a mechanism put forward in the literature to explain the association between native and alien diversity. Our findings (as the reviewer correctly points out, and as we explained later on the paragraph) do not support this mechanism. We revised the text to make clear that this proposal is from the literature and not supported by our findings (please see line 319-320 and 323).

 

L335: Contradictory!

Sory for the lack of clarity that gave the impression of contradiction. We revised our text to reflect that in this study we used landscape as a proxy for resource diversity. But plants use many resources that may not be estimated by the landscape metrics used (eg soil nutrients, soil moisture, light). The point of this sentence is that it is possible that some other dimension of resource availability (not reflected by the landscape metrics we used as a proxy) may play an important role and we did not have the empirical data to quantify it (please see line 350-355).

 

 

The section on Solanum elaeagnifolium is perhaps the most interesting part of the study. I would encourage that the authors rather focus the study on this one species and examine which hypothesis best explains its abundance in Greece.

We agree that the story of Solanum elaeagnifolium is a fascinating story and has received a lot of research attention both for Greece and globally. For the interested reader we provide a couple of examples of recent studies focusing on this issue in Greece, the more relevant of which has been included in our manuscript. Therefore, we feel that the further study of S. elaeagnifolium is out of the scope of the present study.

Krigas, N., Tsiafouli, M. A., Katsoulis, G., Votsi, N. E., & van Kleunen, M. (2021). Investigating the invasion pattern of the alien plant Solanum elaeagnifolium Cav.(silverleaf nightshade): Environmental and human-induced drivers. Plants, 10(4), 805.

Krigas, N., Votsi, N. E., Samartza, I., Katsoulis, G., & Tsiafouli, M. A. (2023). Solanum elaeagnifolium (Solanaceae) invading one in five Natura 2000 protected areas of Greece and one in four habitat types: What is next?. Diversity, 15(2), 143.

 

References

 

I found numerous references that are not in the format of Diversity. Please revise. 

We revised references format.

 

Reviewer 3 Report

This article deals with theoretically important issues of plant invasions. The particular value of the article lies in the fact that it raises some extremely important questions, to which the answers are often biased.

The paper is well written and the text is coherent and logical. Without going into the merits of the article, I would like to draw the attention of the authors to weaknesses or points of discussion. 

1. In the introductory section, I suggest citing the references immediately after using the information taken from the source, rather than at the end of the paragraph (e.g. lines 53-59). 

2. In my opinion, it would be very useful to provide, if possible, a general map of the species richness in Greece, based on the same grid system as for the distribution of alien species.

3. The term 'Seasonality' should be lowercased (line 202).

4. I suggest that the discussion paragraph (lines 299-333) be divided into two or three paragraphs, covering the separate issues under consideration. Chapter 4.2 should also be divided into paragraphs.

 

Minor revisions of English are required.

Author Response

Thank you for your comments, our response is given after each comment

 

This article deals with theoretically important issues of plant invasions. The particular value of the article lies in the fact that it raises some extremely important questions, to which the answers are often biased.

 

The paper is well written and the text is coherent and logical. Without going into the merits of the article, I would like to draw the attention of the authors to weaknesses or points of discussion.

 

1. In the introductory section, I suggest citing the references immediately after using the information taken from the source, rather than at the end of the paragraph (e.g. lines 53-59).

We revised the introduction to associate citations with content more closely.

 

2. In my opinion, it would be very useful to provide, if possible, a general map of the species richness in Greece, based on the same grid system as for the distribution of alien species.

In the revised figure 1, we have added a map of plant species richness of Greece (please see page 5 of revised manuscript)

 

3. The term 'Seasonality' should be lowercased (line 202).

Done

 

4. I suggest that the discussion paragraph (lines 299-333) be divided into two or three paragraphs, covering the separate issues under consideration. Chapter 4.2 should also be divided into paragraphs.

The section on the environmental preferences of alien and native species has been divided in two paragraphs: one focusing on the differences in environmental preferences of aliens and natives and one focusing on the empty niche and limiting similarity hypotheses. (please see page 10)

The concluding remarks of section 4.2 have been divided in three paragraphs: one focusing on the association between native and alien species richness patterns, one focusing on the potential impacts of climate change and one focusing on the wider conservation implications of our finding especially regarding ecosystem services. (please see page 11)

Round 2

Reviewer 2 Report

We used only three of the nineteen WorldClim bioclimatic variables due to the strong collinearity among the bioclimatic variables. For example, in Greece, the driest quarter of the year is also the hottest quarter of the year.

Response to authors

This is not the way one should test for collinearity. The authors first include all 19 bioclimatic variables and STATISTICALLY test for collinearity. They might find that the mean annual temperature, temperature seasonality, and precipitation seasonality show collinearity.

 

We chose the 1990 dataset because that was the period when most of the plant diversity data were collected. To examine whether using Corine land cover data from a different period would affect the observed results we repeated the analysis keeping the other data and using landscape data from different time periods (with a more detailed focus on the 2000 dataset). There were differences in the numbers observed per cell, but those differences were random and did not display a bias, so the overall statistical inference holds. 

Response to authors

The purpose of this study was to look at land cover and how it influences species distribution, and not the other way around. Based on this statement above you are assuming that the species richness did not change since 1990. Surely your species data is much more outdated than your landcover data. I strongly suggest that you use the most recent species and land cover datasets.  

 

The Corine land cover dataset includes also water bodies such as marine areas.

Response to authors

Then you should mention it in the text.

 

As the reviewer correctly pointed out, in this study we did not test the invasion biology hypotheses directly, but indirectly. We examine if and to what extent the patterns and drivers of alien and native species richness are in accordance with the predictions of the different hypotheses. Thus, we did not carry out distribution modelling but we did build multivariate models for the patterns of diversity and their drivers. The main assumption of the empty niche hypothesis is that the availability of unoccupied niches is a key factor driving alien and native diversity. According to this hypothesis, successful invaders can exploit empty niches that are not being used by native species. If alien species utilize different niche space than native species, the effect of predictors on the diversity patterns of aliens and natives will be different. This is our indirect way to assess the importance of the empty niche hypothesis (please see lines 171-175).    

Response to authors

I do not see the results of the multivariate analysis (MVA). What you are presenting is each result from the MVA. But it should be presented in multivariate space. You should produce a plot that shows the association between native and invasive species with the three environmental variables. Only if the native and invasive species show differences in environmental space then you can assume the invasion biology hypotheses is at play here.

 

Unfortunately, information of the clade that a specific population originates from and possible differences between clades of the same species are not available for the scale of our analysis. Also vegetation in the present study was quantified using the Corine land cover classification scheme that identifies the same vegetation classes throughout the European Union.

Response to authors

You can group your species into different taxonomic ranks up to the level of clade. Please describe the vegetation classes. Since Greece has a Mediterranean climate, the vegetation should be described as such. Based on the statement that the same vegetation class is found throughout the EU suggests that Greece is not unique in its vegetation and species composition. Please provide an accurate description of the vegetation.

 

We define hotspots as areas of highest species richness. To improve clarity, we rephrased the sentence (please see line 186).

Response to authors

Highest and lowest are relative terms, please provide actual values of this high species richness for western Greece, Peloponnisos, the urban and peri-urban area.

(Corine land cover classes 187 111, 112, 121, 122, 123, 124, 131, and 132)

Response to authors

These land cover types mean nothing to readers who are not extremely familiar with CLCC.  Identify each land cover class.

(Figure 1 top).

Response to authors

Change top to A and Bottom to B. Label them as such in Figure 1.

 

Table 1:

Add a column to Table 1 that groups the predictors into different hypotheses (as you did in Table 2). For example, if you combine the predictor variables for climate you will find that the sum of 28.7 %, is greater than the native species richness (27.9%). This disproves your suggestion that the biotic acceptance or biotic resistance hypothesis is at play here. Even more apparent is that if you landscape diversity together you have a value of 35.5%. This indicates that the resource availability cluster of hypotheses probably best explains the abundance of alien species in Greece.

T Response to authors

he authors have failed to address this.

 

In our study for quantifying landscape metrics we used the Corine land cover classification. The official title of the third group of land cover classes is “forests and semi natural vegetation” for ease of comprehension we referred to this class as vegetation and includes both natural and semi-natural vegetation (e.g. pastures). We did not analyze the difference between the two and therefore did not mention it in our methods section.

Response to authors

This might be for ease of comprehension, but it does not make sense ecologically. Surely semi-natural regions will have different species composition may allow for greater invasion. It is my opinion that by reducing the complexity to R2 you have overlooked nuances in your results which may prove to be more interesting.

 

This is a very interesting question. In our study when we refer to significance, we refer to statistical significance. But indeed, the ecological significance is a far more important question for which we cannot provide a formal answer.

Response to authors

Perhaps look at literature that may have found that minor changes in temperature can have a significant impact on species distribution.

A back of the envelope quick and dirty answer may be that if mean annual temperature increases by 0.6 degrees and all else remains constant, we may see an average increase of alien species richness by 0.5 species and an average decrease of native species by 3.2 species in every cell of our analysis. But these rough estimates are subject to many assumptions, and we could not argue that this is a scientifically robust answer, therefore we have not added anything in our manuscript.

Response to authors

 

If you cannot explain the reason then it is best to remove the statement. 

Minor grammatical mistakes are in the manuscript. 

Author Response

We have also uploaded our response as a word document to include the CCA plot we prepared in response to reviewers comment

 

We used only three of the nineteen WorldClim bioclimatic variables due to the strong collinearity among the bioclimatic variables. For example, in Greece, the driest quarter of the year is also the hottest quarter of the year.

Response to authors

This is not the way one should test for collinearity. The authors first include all 19 bioclimatic variables and STATISTICALLY test for collinearity. They might find that the mean annual temperature, temperature seasonality, and precipitation seasonality show collinearity.

Our response

We apologize for not making it clear, that is what we have done using the correlation coefficient as a metric of collinearity. The correlation coefficient between mean annual temperature and temperature seasonality is -0.28, the one for mean annual temperature and precipitation seasonality is 0.68, while the one between temperature seasonality and precipitation seasonality is 0.70.

 

We chose the 1990 dataset because that was the period when most of the plant diversity data were collected. To examine whether using Corine land cover data from a different period would affect the observed results we repeated the analysis keeping the other data and using landscape data from different time periods (with a more detailed focus on the 2000 dataset). There were differences in the numbers observed per cell, but those differences were random and did not display a bias, so the overall statistical inference holds. 

Response to authors

The purpose of this study was to look at land cover and how it influences species distribution, and not the other way around. Based on this statement above you are assuming that the species richness did not change since 1990. Surely your species data is much more outdated than your landcover data. I strongly suggest that you use the most recent species and land cover datasets.  

 Our response

We used the largest available, spatially explicit database (and constantly updated) for plant species diversity in Greece. The majority of the data was collected in the early 1990s, and updated with varying effort, in temporal and spatial terms, until today (ongoing). Moreover, since then (1990s), there are studies on biodiversity with available data, but they offer considerably less volume of information or complete coverage of the country. If more recent data at adequate scale were available, we would also prefer to analyze them and quantify the differences that occurred in the intervening period. By EU standards, Greece is a species rich country, but a data poor one.

 

The Corine land cover dataset includes also water bodies such as marine areas.

Response to authors

Then you should mention it in the text.

  Our response

We explicitly do so in line 159.

As the reviewer correctly pointed out, in this study we did not test the invasion biology hypotheses directly, but indirectly. We examine if and to what extent the patterns and drivers of alien and native species richness are in accordance with the predictions of the different hypotheses. Thus, we did not carry out distribution modelling but we did build multivariate models for the patterns of diversity and their drivers. The main assumption of the empty niche hypothesis is that the availability of unoccupied niches is a key factor driving alien and native diversity. According to this hypothesis, successful invaders can exploit empty niches that are not being used by native species. If alien species utilize different niche space than native species, the effect of predictors on the diversity patterns of aliens and natives will be different. This is our indirect way to assess the importance of the empty niche hypothesis (please see lines 171-175).    

Response to authors

I do not see the results of the multivariate analysis (MVA). What you are presenting is each result from the MVA. But it should be presented in multivariate space. You should produce a plot that shows the association between native and invasive species with the three environmental variables. Only if the native and invasive species show differences in environmental space then you can assume the invasion biology hypotheses is at play here.

   Our response

Multivariate analysis is a term that covers a wide array of statistical analyses that include many variables, such as ordination, MANOVA, multiple regression. We expanded on the later approach using generalized additive models with many independent variables. The reviewer seems to be referring to ordination (eg CCA), which is a tool to reduce the complexity of multidimensional data in fewer dimensions, It allows the visualization of complex data structures to few axes (typically two) that allows a visual inspection of the data. To show that native and invasive species show differences in environmental space (which is expressed by 9 variables in our study reflecting both climate and landscape), we performed a Constrained Correspondence Analysis for the environmental variables and using native and alien species richness as constraints. We attach a word document file that at this point shows the biplot of this CCA model, which explained 13% of the variance in the data and clearly shows that alien species richness differs to native species richness along the first axis of CCA. However, this is a visual interpretation and, as such, different viewers may reach different conclusions. Also, CCA is not supported by robust inferential statistics as the analyses presented in our manuscript.

 

Unfortunately, information of the clade that a specific population originates from and possible differences between clades of the same species are not available for the scale of our analysis. Also vegetation in the present study was quantified using the Corine land cover classification scheme that identifies the same vegetation classes throughout the European Union.

Response to authors

You can group your species into different taxonomic ranks up to the level of clade. Please describe the vegetation classes. Since Greece has a Mediterranean climate, the vegetation should be described as such. Based on the statement that the same vegetation class is found throughout the EU suggests that Greece is not unique in its vegetation and species composition. Please provide an accurate description of the vegetation.

    Our response

We have added a brief description of the main vegetation types in Greece, and provide a citation to a recent study that explicitly and accurately describes the vegetation of Greece. Please see lines 132-140.

We define hotspots as areas of highest species richness. To improve clarity, we rephrased the sentence (please see line 186).

Response to authors

Highest and lowest are relative terms, please provide actual values of this high species richness for western Greece, Peloponnisos, the urban and peri-urban area.

    Our response

The highest recorded alien species richness in our data set is 77 alien taxa per 10x10 km grid cell. Generally areas of high alien species richness had more than 28 alien taxa per cell. Please see lines 195-196.

 

(Corine land cover classes 187 111, 112, 121, 122, 123, 124, 131, and 132)

Response to authors

These land cover types mean nothing to readers who are not extremely familiar with CLCC.  Identify each land cover class.

     Our response

Done. Please see lines 197-201

(Figure 1 top).

Response to authors

Change top to A and Bottom to B. Label them as such in Figure 1.

     Our response

Done (following the journal’s template).

Table 1:

Add a column to Table 1 that groups the predictors into different hypotheses (as you did in Table 2). For example, if you combine the predictor variables for climate you will find that the sum of 28.7 %, is greater than the native species richness (27.9%). This disproves your suggestion that the biotic acceptance or biotic resistance hypothesis is at play here. Even more apparent is that if you landscape diversity together you have a value of 35.5%. This indicates that the resource availability cluster of hypotheses probably best explains the abundance of alien species in Greece.

Response to authors

The authors have failed to address this.

      Our response

We apologize for not making our response clear. We have not added an extra column in Table 1, because R² values from different models could not be added to predict the R² value of the combined model. For example, if we add all R² values for alien species models in Table 1 the sum would be 107%, something that cannot be accurate. Therefore, to analyze the combined effect of different sets of variables we build the combined models presented in Table 2. Although the resource availability cluster of hypotheses may best explain the alien species richness, this is not evident in our analysis. Perhaps some other proxies for resource availability may indeed produce this outcome, our results are limited to the analysis we performed,

 

In our study for quantifying landscape metrics we used the Corine land cover classification. The official title of the third group of land cover classes is “forests and semi natural vegetation” for ease of comprehension we referred to this class as vegetation and includes both natural and semi-natural vegetation (e.g. pastures). We did not analyze the difference between the two and therefore did not mention it in our methods section.

Response to authors

This might be for ease of comprehension, but it does not make sense ecologically. Surely semi-natural regions will have different species composition may allow for greater invasion. It is my opinion that by reducing the complexity to R2 you have overlooked nuances in your results which may prove to be more interesting.

 Our response

We apologize for not making it clear in the previous response. The reason we did not analyze natural and seminatural vegetation separately is that we do not have this information available. Greece has a long history of humans interacting with the natural environment. Even areas, that today appear pristine, are under human pressure (e.g. grazing for animal husbandry) for thousands of years. Thus separating natural from semi-natural vegetation is not a trivial task. And there is no dataset available in Greece to separate a natural forest from a semi-natural forest. Therefore, we could not perform any analyses on the question.  

 

This is a very interesting question. In our study when we refer to significance, we refer to statistical significance. But indeed, the ecological significance is a far more important question for which we cannot provide a formal answer.

Response to authors

Perhaps look at literature that may have found that minor changes in temperature can have a significant impact on species distribution.

Our response

Thank you for this. We have added a sentence in the discussion citing an influential review to highlight the point.

 

A back of the envelope quick and dirty answer may be that if mean annual temperature increases by 0.6 degrees and all else remains constant, we may see an average increase of alien species richness by 0.5 species and an average decrease of native species by 3.2 species in every cell of our analysis. But these rough estimates are subject to many assumptions, and we could not argue that this is a scientifically robust answer, therefore we have not added anything in our manuscript.

Response to authors

 If you cannot explain the reason then it is best to remove the statement. 

Our response

We have not added the statement.

 

 

Author Response File: Author Response.docx