Species Abundance Distributions Patterns between Tiankeng Forests and Nearby Non-Tiankeng Forests in Southwest China

Round 1

Reviewer 1 Report

This paper aim to find the plant diversity difference between outer and inside the Tiankeng, one important nature plant reserve usually in China. Several models have been used in his paper and finnaly two model were used to find their possible mechnism for the pattern. I think it is good enough for publicatio after some revision.

my main concern is that several models have been compared, and most of them are not statsitically significant, and AIC was used to choose the best model for the model selection. Please explain the criterion for your choose the best  model for this paper. If different model selected, what kind of result could be achieved?

In all, please clarify the roubustness of uncertainty of your results and finding.

  The second is please cite more references in recent publication. If not Tiankeng, possibly, Nature Reserve study should be cited for strongthen the story. 

Author Response

Response letter

On behalf of all the contributing authors, I would like to express our sincere appreciations of your letter and reviewers’ constructive comments concerning our article entitled “Species Abundance Distributions Patterns between Tiankeng forests and Nearby Non-Tiankeng forests in Southwest China” (Manuscript ID diversity-1506346). These comments are all valuable and helpful for improving our article. According to the editor and reviewers’ comments, we have made extensive modifications to our manuscript and supplemented extra data to make our results convincing. In this revised version, changes to our manuscript were all highlighted within the document by using colored text. Point-by-point responses to the nice associate editor and two nice reviewers are listed below this letter.  And the attachment is the revised manuscript.

 

Independent Review Report, Reviewer 1#

This paper aim to find the plant diversity difference between outer and inside the Tiankeng, one important nature plant reserve usually in China. Several models have been used in his paper and finnaly two model were used to find their possible mechnism for the pattern. I think it is good enough for publicatio after some revision.

QUESTION 1. My main concern is that several models have been compared, and most of them are not statsitically significant, and AIC was used to choose the best model for the model selection. Please explain the criterion for your choose the best model for this paper. If different model selected, what kind of result could be achieved?

Response: Thanks for your suggestion. We performed K-S test based on 999 times bootstrap simulation to verify the fitting results between the observed and simulation values from six models. When the p>0.05, this test indicates that the model has passed the test; there was not statistically significant between the simulated and observed SAD patterns. The goodness-of-fit of model was determined by the lowest AIC value. If the AIC value of model is more than 2 (△AIC>2) and lower than others, then this model is considered significantly better than other models. Our AIC test of goodness-of-fit results, showed that the neutral models (△_mean AIC=1.3, AIC weights _mean =26.0%) performed better than the niche (△_mean AIC=22.7, AIC weights _mean =0.1%) and statistical models (△_mean AIC=2.7, AIC weights _mean =24.1%) in the Tiankeng forest, while the statistical models (△_mean AIC=2.4, AIC weights _mean =20.0%) performed better than the niche (△_mean AIC=11.2, AIC weights _mean =15.0%) and neutral (△_mean AIC=3.5, AIC weights _mean =15.1%) models in the non-Tiankeng forests. The neutral model has the smallest AIC value in Tiankeng forests, while the statistical model has the smallest AIC value in the non-Tiankeng forests. Therefore, in Tiankeng forests, the main driving force of Tiankeng forests is the neutral process, and in non-Tiankeng forests, it is the niche and neutral processes. The above contents were revised in line 228-234, 264-270, 340-344.

 

QUESTION 2. please clarify the roubustness of uncertainty of your results and finding.

Response: Thanks for your suggestion. All repeat results were removed and the finding was re-summarized the conclusion. And the uncertainty of my results was added in part 4.3, line 408- 416 as follows: Different patterns of SAD may indicate specific environmental conditions. It is doubtful whether there would have other ecological processes that respond to the species co-occurrence. Because each forests species has its own specific role as a result of evolutionary processes [39]. Moreover, multiple-mechanism models may have better results when fitting the same data. The model fitting results based solely on the patterns of species abundance cannot fully verify the ecological mechanism behind it. Therefore, we suggested that the neutral process was important for the SAD pattern in Tiankeng forests, a further analysis needed to be carried out in combination with environmental factors and spatial patterns.

 

QUESTION 3.The second is please cite more references in recent publication. If not Tiankeng, possibly, Nature Reserve study should be cited for strongthen the story.

Response: Thanks for your suggestion. We have tried our best to cite more references in recent publication. The number of articles published in 2018 and above accounts for 70% of the total.

Author Response File: Author Response.pdf

Reviewer 2 Report

This is an interesting article aimed to compare species abundance distribution between two habitat types in southwest China. The manuscript is concise and well written. I have three major comments and a few specific comments outlined below.

Major comment 1.- The manuscript focus is too narrow. I wonder how two sites in southwest China are representative enough and would be from interest for a wider audience, keeping in mind that Diversity is a broad journal.

Major comment 2.- Data analysis sub-section within the Methods section is unclear. From the information provided is very difficult to grasp what was done and why. For instance, a Kruskal Wallis test is mentioned first, but then a K-S test (a Kolmogorov-Smirnov test I presume) is cited. Also, it is not clear how AIC values were estimated for each model.

Major comment 3.- In the Results section, it is unclear how goodness-of-fit was determined to the different SAD models. Particularly, the use of AIC is not properly implemented as it is mixed with K-S tests (two different approaches). Furthermore, AIC ranking was not properly presented, providing only AIC values does not take the full advantage of this approach, as delta AIC and AIC weights were not presented.

Specific comments

Line 19. It is Akaike, not Ahaike, please correct it throughout the manuscript
Lines 19-20. AIC is suited per se to assess goodness-of-fit
Line 88. It should be ‘studies’ not ‘studied’
Lines 244-245. Please cite R packages (not only mention them).
Line 287. What ‘abundants’ mean?
Line 295. There is a missing space between ‘results’ and ‘(Figure 3)’
Figure 2 is little informative and repeats information included in Table 1.

Author Response

Response letter

On behalf of all the contributing authors, I would like to express our sincere appreciations of your letter and reviewers’ constructive comments concerning our article entitled “Species Abundance Distributions Patterns between Tiankeng forests and Nearby Non-Tiankeng forests in Southwest China” (Manuscript ID diversity-1506346). These comments are all valuable and helpful for improving our article. According to the editor and reviewers’ comments, we have made extensive modifications to our manuscript and supplemented extra data to make our results convincing. In this revised version, changes to our manuscript were all highlighted within the document by using colored text. Point-by-point responses to the nice associate editor and two nice reviewers are listed below this letter. And the attachment is the revised manuscript.

Independent Review Report, Reviewer 2#

This is an interesting article aimed to compare species abundance distribution between two habitat types in southwest China. The manuscript is concise and well written. I have three major comments and a few specific comments outlined below.

 

QUESTION 1. The manuscript focus is too narrow. I wonder how two sites in southwest China are representative enough and would be from interest for a wider audience, keeping in mind that Diversity is a broad journal.

Response: Thanks for your suggestion. I would like to make some clarification. First of all, Tiankeng forest is a type of fragmented forest. As of 2020, more than half of the world’s natural habitat have been lost. The problem of habitat fragmentation is still very serious, and habitat loss and degradation caused by chemical transformation is one of the main reasons for biodiversity loss, but the understanding of the maintenance mechanism of fragmented habitat plant communities is still unclear. Second, Tiankeng Forest is a special forest distributed in the negative terrain habitat of Tiankeng. It is an ecologically sensitive area in the karst area to monitor the global climate change. Finally, Tiankeng Forest preserves the most primitive and complete forest in the Karst Stone Mountain area, and it helps us to suggest vegetation restoration and reconstruction in ecologically sensitive areas. And that was added in line 66-67, 84-85, 87-90.

 

QUESTION 2. Data analysis sub-section within the Methods section is unclear. From the information provided is very difficult to grasp what was done and why. For instance, a Kruskal Wallis test is mentioned first, but then a K-S test (a Kolmogorov-Smirnov test I presume) is cited. Also, it is not clear how AIC values were estimated for each model.

Response: Thanks for your suggestion. The Methods section and results section were improved. Such as :We used two curves (species ECDF and species abundance curve) to visually observe the SADs pattern in six sampling sites, and a Kruskal-Wallis test (K-W) was used to compare the SAD’s difference in the ‘dplyr’ packages of R software 4.0.0. (R core team, Vienna, Austria) [35,40]. That was added in line 165-168.

  We performed K-S test based on 999 times bootstrap simulation to verify the fitting results between the observed and simulation values from six models. When the p <0.05, this test indicates that the model has not passed the test; otherwise, the model is accepted [39]. The goodness-of-fit of model was determined by the lowest AIC value. If the AIC value of model is more than 2 (△AIC>2) and lower than others, then this model is considered significantly better than other models [40]. These basic statistical analyses were conducted using the ‘sads’ and ‘Matching’ packages in R software 4.0.0. (R core team, Vienna, Austria) [40-41]. That was added in line 227-234.

 

QUESTION 3. In the Results section, it is unclear how goodness-of-fit was determined to the different SAD models. Particularly, the use of AIC is not properly implemented as it is mixed with K-S tests (two different approaches). Furthermore, AIC ranking was not properly presented, providing only AIC values does not take the full advantage of this approach, as delta AIC and AIC    were not presented.

Response: Thanks for your suggestion. We performed K-S test based on 999 times bootstrap simulation to verify the fitting results between the observed and simulation values from six models. When the p>0.05, this test indicates that the model has passed the test; there was not statistically significant between the simulated and observed SAD patterns. The goodness-of-fit of model was determined by the lowest AIC value. If the AIC value of model is more than 2 (△AIC>2) and lower than others, then this model is considered significantly better than other models. And the  delta AIC and AIC weight were revised as following:

Our AIC test of goodness-of-fit results, showed that the neutral models (△_mean AIC=1.3, AIC weights _mean =26.0%) performed better than the niche (△_mean AIC=22.7, AIC weights _mean =0.1%) and statistical models (△_mean AIC=2.7, AIC weights _mean =24.1%) in the Tiankeng forest, while the statistical models (△_mean AIC=2.4, AIC weights _mean =20.0%) performed better than the niche (△_mean AIC=11.2, AIC weights _mean =15.0%) and neutral (△_mean AIC=3.5, AIC weights _mean =15.1%) models in the non-Tiankeng forests. The neutral model has the smallest AIC value in Tiankeng forests, while the statistical model has the smallest AIC value in the non-Tiankeng forests. Therefore, in Tiankeng forests, the main driving force of Tiankeng forests is the neutral process, and in non-Tiankeng forests, it is the niche and neutral processes. The above contents were added in line 228-234, 264-270, 340-344.

 

QUESTION 4. English language and style are fine/minor spell check required

Response: Thanks for your suggestion. We have tried our best to polish the language in the revised manuscript. The expert with experience in language proofreading & polishing provided the linguistic services.

QUESTION 5

Line 19. It is Akaike, not Ahaike, please correct it throughout the manuscript.

Response: The “Ahaike” was corrected to “Akaike”in Line 19.

Lines 19-20. AIC is suited per se to assess goodness-of-fit.

Response: The (AIC) test was used to assess goodness-of-fit of each models between the simulated and observed SAD patterns.

Line 88. It should be ‘studies’ not ‘.

Response: The ‘studied’ was corrected to “studies”in Line 88.

Lines 244-245. Please cite R packages (not only mention them).

Response: It was corrected in Line 167-168, 246-249, “a Kruskal-Wallis test (K-W) was used to compare the SAD’s difference in the ‘dplyr’ packages of R software 4.0.0. (R core team, Vienna, Austria)”, “These basic statistical analyses were conducted using the ‘sads’ and ‘Matching’ packages in R software 4.0.0. (R core team, Vienna, Austria) [40-41].”

Line 287. What ‘abundants’ mean?

Response: The ‘abundants’ was corrected to “abundance” in line 287.

Line 295. There is a missing space between ‘results’ and ‘(Figure 3)’

Response: The space was added in line 233-34.

Figure 2 is little informative and repeats information included in Table 1.

Response: Thanks for your suggestion. The Figure 2 was removed.

Author Response File: Author Response.docx

Reviewer 3 Report

This manuscript quality is good for publishing in the journal. And the method is conventional, not advanced method. The conclusion is good, and  is in line with expectations. 

Since this area study is just on the early stage in China domestic and study output is little, even in deficiency, this innovative study is made with theoretical meaning and practical value.  And the data is very interesting and forward-looking, howevr, the  scientific questions are not be concised and focused on, adequately, and the questions look too simple, even though I am not the specialist in this field. Otherwise, the quality of this manuscripy will be better.

All in word, this manuscript is good and and fit for the request of Diversity to publish.

Comments for author File: Comments.pdf

Author Response

Response letter

On behalf of all the contributing authors, I would like to express our sincere appreciations of your letter and reviewers’ constructive comments concerning our article entitled “Species Abundance Distributions Patterns between Tiankeng forests and Nearby Non-Tiankeng forests in Southwest China” (Manuscript ID diversity-1506346). These comments are all valuable and helpful for improving our article. According to the editor and reviewers’ comments, we have made extensive modifications to our manuscript and supplemented extra data to make our results convincing. In this revised version, changes to our manuscript were all highlighted within the document by using colored text. Point-by-point responses to the nice associate editor and two nice reviewers are listed below this letter. And the attachment is the revised manuscript.

Independent Review Report, Reviewer 3#

This manuscript quality is good for publishing in the journal. And the method is conventional, not advanced method. The conclusion is good, and is in line with expectations.

Since this area study is just on the early stage in China domestic and study output is little, even in deficiency, this innovative study is made with theoretical meaning and practical value. And the data is very interesting and forward-looking, however,

QUESTION 1. the scientific questions are not be concised and focused on, adequately, and the questions look too simple, even though I am not the specialist in this field. Otherwise, the quality of this manuscripy will be better.

Response: Thanks for your suggestion. I would like to make some clarification. The scientific question is simple but meaningful. First of all, Tiankeng forest is a type of fragmented forest. As of 2020, more than half of the world’s natural habitat have been lost. The problem of habitat fragmentation is still very serious, and habitat loss and degradation caused by chemical transformation is one of the main reasons for biodiversity loss, but the understanding of the maintenance mechanism of fragmented habitat plant communities is still unclear. Second, Tiankeng Forest is a special forest distributed in the negative terrain habitat of Tiankeng. It is an ecologically sensitive area in the karst area to monitor the global climate change. Finally, Tiankeng Forest preserves the most primitive and complete forest in the Karst Stone Mountain area, and it helps us to suggest vegetation restoration and reconstruction in ecologically sensitive areas. And that was added in line 66-67, 84-85, 87-90.

 

QUESTION 2. How to arrange the order of key words?By importance? By alphabetical order? Or by the request of journal's guideline?

Response: By importance. 

 

QUESTION 3. “ is one of a important way in line 51 ”Do authors know the problem for linguistic problem?

Response: The “ is one of a important way” was corrected by “ is one of the important ways”.

 

QUESTION 4. Please mark which city, this area belongs to. How warm and how cool? Please give the detailed data for the average data in the past ten years, at least.Same problem. To give the average data at least ten years. Longer is better. in line 112-118.

Response: Thanks for your suggestion. Line 112-118 has been corrected to “This study was carried out in the Dashiwei Tiankeng Group branch, located at of 24°30′N to 25°03′N and 106°10′E to 106°51′E, in Leye, Guangxi, south-west China. The study area has a big elevation difference, higher in the southwest (Guizhou plateau) and lower in east, north and west (Guangxi basin). The local climate is mid-subtropical monsoon. There was a distinct rainy season from May to October.The annual average rainfall is 1400mm, rainfall accounts for 85% of the mean precipitation, annual relative humidity is 85%, and mean average temperature is 16.6-23℃”.

QUESTION 5. The FOC was completed in many years ago, the systematic positions of many genera and species were changed. If the species and genus names/numbers are being changed, the data analysis and the conclusion will have a bit dubious. How to resolve this problem?

A monograph can help, entitled The Distribution and Conservation Status of Native Plants in Guangxi, China by Wei Yi-Gang (2018). The book offers the latest scientific names and the previous names before the latest revision. The manuscript should refer to it.

Response: Thanks for your suggestion. I have checked and referred the latest edition of online flora of China (http://www.iplant.cn/foc/) and Wei Yi-Gang (2018) in my references, and the results showed the same. 

 

QUESTION 6. I strongly add the namers behind the scientific name of species, especially when the species firstly appear.

Response: Thanks for your suggestion. The namers behind the scientific name of species in this paper was added in line 295-296, 393 and Table 1. Such as Choerospondias axillaris B. L. Burtt & A. W. Hill and Rhaphiolepis indica Lindley.

Author Response File: Author Response.pdf

Reviewer 4 Report

In their manuscript “Species abundance distributions patterns between Tiankeng forests and nearby non-Tiankeng forests in Southwest China”, Huang and Yang provided the different SADs patterns between Tiankeng and non-Tiankeng forests, and suggested that the neutral process is the key point of Tiankeng community construction. The results in this study are meaningful which would enhance the protection of the Tiankeng forests for their unique ecosystem functions and services.

 

Overall, although detailed information was provided on the characteristics of Tiankeng and non-Tiankeng forests including their species compositions, SADs patterns and good-to-fit of 6 models, the manuscript can be greatly improved by resolving issues from several aspects including language and logic especially in Introduction and M&M sections. Importantly, I can’t follow the findings of SADs patterns, which was purely descriptive and deductive, and authors should give more mechanistical explanations for the patterns they observed based on their data. Logical description and deduction should be the core. Moreover, the Results is too lengthy and there is too much overlap with the data in Tables and Figures, and same to the Conclusion. Please pay more attention to the language because it is the weakest part in this manuscript, I think. See following review report for more details.

 

1. L63-68 There is a contradiction in this sentence. Many researchers studied the models for SAD, …., the objects of SAD models are limited. The authors may mean the application of SAD models in the negative terrain is limited. Please rephrase it to avoid the indistinction.
2. L101 This is the first time to mention ECDF curve, please explain it in the parentheses.
3. L138-140 I can’t calculate the accurate area of the study site. Each site was divided into 4 plots at scale 20*20 m², 8 plots at 10*10 m², which totally 2400 m² for each habitat type. Please check or explain it.
4. L141-156 The authors selected 6 Tiankeng for three types habitats (inner Tiankeng, fringe of Tiankeng and non-Tiankeng). The study involved species dispersal and community composition, I think it is better to select two independent Tiankeng terrains and set up transects, which including three habitats described above, respectively. Moreover, the Tiankeng forest is a negative terrain on the slope, light should be a key environmental limitation factor for species in Tiankeng forests. The authors didn’t seem to have mention whether the sample sites in this study located on a sunny or a shady slope.
5. L249-252, L254-261 Please delete these sentences, which have already showed in Figure 1 and Table 1. The same to other paragraphs in Result section. It is too lengthy.
6. L265-268 This sentence is more of a corollary and should be moved to discussion section.
7. L270 There is no significant difference in numbers of family, genus and species among different habitats. If it suggested that species dispersal patterns did not significantly affect community composition in these three habitat types?
8. L282-285 The same to L265-268, please move it to discussion. Additionally, this sentence is purely deductive, the authors should give more mechanistical explanations for the patterns based on the data.
9. L294-304 Delete this sentence or make it concise.
10. Figure 2 & Table 2. I think they show the same thing and recommend just keep Table 2, as Figure 3 doesn’t show the differences clearly.
11. L373-378 This sentence in the Discussion section is overlap with the Results, please check and delete it.
12. I found the conclusion quite long and I don’t think that it is informative to emphasize the result too much detail, because you have already described them in the former. I recommended to abbreviate the conclusion into one paragraph.

 

Author Response

Response letter

On behalf of all the contributing authors, I would like to express our sincere appreciations of your letter and reviewers’ constructive comments concerning our article entitled “Species Abundance Distributions Patterns between Tiankeng forests and Nearby Non-Tiankeng forests in Southwest China” (Manuscript ID diversity-1506346). These comments are all valuable and helpful for improving our article. According to the editor and reviewers’ comments, we have made extensive modifications to our manuscript and supplemented extra data to make our results convincing. In this revised version, changes to our manuscript were all highlighted within the document by using colored text. Point-by-point responses to the nice associate editor and two nice reviewers are listed below this letter. And the attachment is the revised manuscript.

Independent Review Report, Reviewer 4#

In their manuscript “Species abundance distributions patterns between Tiankeng forests and nearby non-Tiankeng forests in Southwest China”, Huang and Yang provided the different SADs patterns between Tiankeng and non-Tiankeng forests, and suggested that the neutral process is the key point of Tiankeng community construction. The results in this study are meaningful which would enhance the protection of the Tiankeng forests for their unique ecosystem functions and services.

QUESTION 1 Overall, although detailed information was provided on the characteristics of Tiankeng and non-Tiankeng forests including their species compositions, SADs patterns and good-to-fit of 6 models, the manuscript can be greatly improved by resolving issues from several aspects including language and logic especially in Introduction and M&M sections. 

Response: Thanks for your suggestion. Some aspects including language and logic was removed and corrected in line 52, 62, 62-71, 95-118. Such as the line 101-104 were removed. Line 64-68 were revised to “Therefore, many scholars mostly adopted a combination of multiple models to analyze the SAD pattern in different objects, including alpine meadows [19], tropical forests [20], subtropical evergreen-deciduous broadleaved mixed forests [21]. However, the current research on forest communities of SAD models in negative terrain habitats is still unclear.”

 

QUESTION 2 Importantly, I can’t follow the findings of SADs patterns, which was purely descriptive and deductive, and authors should give more mechanistical explanations for the patterns they observed based on their data. Logical description and deduction should be the core.

Response: Thanks for your suggestion. The SADs patterns are presented in the part 3.2 and discussed in the part 4.2. In the part 3.2, the SADs patterns is that “the species accumulation rate was the highest in the fringe of Tiankeng forests, middling in the nearby non-Tiankeng forests, and lowest is in the inside of Tiankeng forests (Figure 2b); the Jaccard index showed that the species similarity between the fringe and inside of Tiankeng forests was the highest (0.36), followed by the non-Tiankeng forests and the fringe of Tiankeng forests (0.29), and between the inside of Tiankeng forests and the non-Tiankeng forests was the lowest (0.27) (Figure 2c)”.

And in the part 4.2, we give a mechanistical explanations as following: 

On the one hand, the fringe of Tiankeng forest is surrounded by non-Tiankeng forest and encircles the inside of the Tiankeng forest [44]. The non-Tiankeng forest is usually distributed in the foothills, ridges, and slopes near villages and farmlands, and is greatly affected by human activities which enhance alien species to the Tiankeng forests [22,45]. On the other hand, the inside of the Tiankeng forest is located at the bottom of the negative topographic structure, and this benefits species accumulation.

The Tiankeng formation can be tracked back to the Paleocene [46]. At that time, species in this area would freely migrate. Later, the Tiankeng and the Tiankeng forests gradually evolved and formed along with the geological movement. By now, with global warming, nearby non-Tiankeng forests were suffering soil loss, water stress and degraded, some species could not adapt to their stress environment and took refuge in the Tiankeng negative terrain [29,47]. Sufficient water and heat condition, stable growing environment, and tundra organic-rich soil in Tiankeng provide a favorable place for the heterogeneity and diversity of species [24,29]. Therefore, by comparing the SADs difference between Tiankeng and non-Tiankeng forests, we can determine that the Tiankeng forest is a plant refuge formed by surrounding species that have adapted to the environmental changes. One of the reasons is that the negative topographic structure promotes the accumulation of species, and the other is that the external environment deteriorates, and species are attracted by the hot and humid environment of Tiankeng.

 

QUESTION 3 Moreover, the Results is too lengthy and there is too much overlap with the data in Tables and Figures, and same to the Conclusion. Please pay more attention to the language because it is the weakest part in this manuscript, I think.

Response: Thanks for your suggestion. The repeat parts were moved, the revised part is as following:

3.1. Species composition of Tiankeng forest and nearby Non-Tiankeng forests

A total of 3599 individuals of woody plants were recorded in the study sites and the species composition of woody plant was rich and complex. There were 20 families, 21 genera and 31 species in DC-TK, 24 families, 27 genera and 32 species in LX-TK, 27 families, 27 genera and 48 species in LJ-BY, 26 families, 26 genera and 50 species in SM-BY, 30 families, 30 genera and 47 species in LA-FS, 30 families, 31 genera and 41 species in SW-FS. And if we compare the importance value of dominant species at different sites, the dominant species in the inner area of Tiankeng forests is rarely distribution in the nearby non-Tiankeng forests, while the dominant species nearby non-Tiankeng forests is frequently distributed in the inner and the fringe of tiankeng forest (Table 1).

3.2. SADs pattern in Tiankeng and non-Tiankeng forests

There was a significant difference in SADs patterns (Kruskal-Wallis chisq=16.798, p<0.05) between the Tiankeng and non-Tiankeng forests (Figure 2a). The species accumulation rate was highest in the fringe of Tiankeng forests, and lowest in the inside of Tiankeng forests (Figure 2b). The Jaccard index showed that the species similarity between the fringe and inside of Tiankeng forests was the highest (0.36), followed by the non-Tiankeng forests and the fringe of Tiankeng forests (0.29), and between the inside of Tiankeng forests and the non-Tiankeng forests was the lowest (0.27) (Figure 2c).

3.3. Goodness-of fit of SADs models in Tiankeng and non-Tiankeng forests

We observed differences in the test values at the 40m×40m scale between forest with different plots. K-S tests indicated no significant differences between the simulated and actual SAD values and all models have passed the K-S test (P>0.05) between Tiankeng and non-Tiankeng forests (Figure 3). AIC test of goodness-of-fit results, showed that the neutral models (△mean AIC=1.3, AIC weights mean =26.0%) performed better than the niche (△mean AIC=22.7, AIC weights mean =0.1%) and statistical models (△mean AIC=2.7, AIC weights mean =24.1%) in the Tiankeng forest, while the statistical models (△mean AIC=2.4, AIC weights mean =20.0%) performed better than the niche (△mean AIC=11.2, AIC weights mean =15.0%) and neutral (△mean AIC=3.5, AIC weights mean =15.1%) models in the non-Tiankeng forests. According to the results of neutral theory parameters and the species diversity (Table 2), the fundamental biodiversity number (θ), species immigration rate (m), species richness and Shannon-Wiener index was the highest in the fringe of the tiankeng forest, and the lowest was in the inside the Tiankeng forests.

 

QUESTION 4

L63-68 There is a contradiction in this sentence. Many researchers studied the models for SAD, …., the objects of SAD models are limited. The authors may mean the application of SAD models in the negative terrain is limited. Please rephrase it to avoid the indistinction.

Response: Thanks for your suggestion. It has been corrected to “Different models can only demonstrate a part of community ecological process. Therefore, many scholars mostly adopted a combination of multiple models to analyze the SAD pattern in different objects, including alpine meadows [19], tropical forests [20], subtropical evergreen-deciduous broadleaved mixed forests [21]. However, the current research on forest communities of SAD models in negative terrain habitats is still unclear.”

L101 This is the first time to mention ECDF curve, please explain it in the parentheses.

Response: Thanks for your suggestion. It was corrected to “Empirical cumulative distribution functions (ECDF) curve” in line 94.

L138-140 I can’t calculate the accurate area of the study site. Each site was divided into 4 plots at scale 20*20 m2, 8 plots at 10*10 m2, which totally 2400 m2 for each habitat type. Please check or explain it.

Response: Thanks for your suggestion. Line 138-140 were corrected to “ There are six sampling site for three habitats types, each site were divided into sixteen plots at scale 10 m × 10 m, which totally 3200m² for each habitats type”. 

L141-156 The authors selected 6 Tiankeng for three types habitats (inner Tiankeng, fringe of Tiankeng and non-Tiankeng). The study involved species dispersal and community composition, I think it is better to select two independent Tiankeng terrains and set up transects, which including three habitats described above, respectively. Moreover, the Tiankeng forest is a negative terrain on the slope, light should be a key environmental limitation factor for species in Tiankeng forests. The authors didn’t seem to have mention whether the sample sites in this study located on a sunny or a shady slope.

Response: Thanks for your suggestion. We have also considered this issue before. But there some problems. Firstly, the habitat is fragmented, it is difficult to find the three types of habitats in the same Tiankeng. Secondly, even if they are found, the tiankengs are too steep to involve for humans. Finally, water and light are the key environmental limiting factors for tiankeng forest species. There is no absolute sun or shady slope at Tiankeng.

L249-252, L254-261 Please delete these sentences, which have already showed in Figure 1 and Table 1. The same to other paragraphs in Result section. It is too lengthy.

Response: These sentences in 249-252, L254-261 were deleted.  

L265-268 This sentence is more of a corollary and should be moved to discussion section.

Response: These sentences were moved to the discussion section, part 4.1.

L270 There is no significant difference in numbers of family, genus and species among different habitats. If it suggested that species dispersal patterns did not significantly affect community composition in these three habitat types?

Response: Thanks for your suggestion. In the aspect of numbers of family, genus and species among different habitats is little difference in quantity, but in other aspects are significant difference. There is significant difference in SADs pattern, and the change of importance value of dominant species also proved that species dispersal patterns affect community composition in these three habitat types. The species dispersal from the inside of Tiankeng forests to the nearby non-Tiankeng forests is limited, while species has unlimited dispersal from nearby non-Tiankeng forests to the inside of Tiankeng forests via the fringe of Tiankeng forests. And that leading to the internal enclosed environment of Tiankeng could be conducive to the independent evolution of species in the Tiankeng forests, which provided habitats for more precious and primitive species.

L282-285 The same to L265-268, please move it to discussion. Additionally, this sentence is purely deductive, the authors should give more mechanistical explanations for the patterns based on the data.

Response: Thanks for your suggestion. L282-285 were moved to discussion in 4.2, and it was explained for two reasons in line 314-320, “ On the one hand, the fringe of Tiankeng forest is surrounded by non-Tiankeng forest and encircles the inside of the Tiankeng forest. The non-Tiankeng forest is usually distributed in the foothills, ridges, and slopes near villages and farmlands, and is greatly affected by human activities which enhance alien species to the Tiankeng forests. On the other hand, the inside of the Tiankeng forest is located at the bottom of the negative topographic structure, and this benefits species accumulation.”

L294-304 Delete this sentence or make it concise. L373-378 This sentence in the Discussion section is overlap with the Results, please check and delete it.

Response: Thanks for your suggestion. The L294-304 and L373-378 were deleted, and the L294-304 were corrected to “and most species in inner Tiankeng forests originated from nearby non-Tiankeng forests via fringe area.”

Figure 3 & Table 2. I think they show the same thing and recommend just keep Table 2, as Figure 3 doesn’t show the differences clearly.

Response: Thanks for your suggestion. The table 2 and Figure 2 was removed, and the results of K-S test and AIC value were added to the new Figure 2.

I found the conclusion quite long and I don’t think that it is informative to emphasize the result too much detail, because you have already described them in the former. I recommended to abbreviate the conclusion into one paragraph.

Response: Thanks for your suggestion. The conclusion has been abbreviated into one paragraph as following:

The comparing of SADs between Tiankeng and non-Tiankeng forests is essential for its management, conservation and sustainable utilization in Dashiwei Tiankeng Group. Assessing the role of biodiversity in Tiankeng forests could help in understanding climate change, approaches to the wise use of sustainable resources, preserving biological diversity and exploring the community maintenance mechanism. For example, we found that the species dispersal from the inside of Tiankeng forests to the nearby non-Tiankeng forests is limited, while species has unlimited dispersal from nearby non-Tiankeng forests to the inside of Tiankeng forests via the fringe of Tiankeng forests. Moreover, some species in Tiankeng forest tend to become more rare and precious. Therefore, this study results shows that neutral process incites Tiankeng forests to be a plant refuge throughout the fitting of niche and neutral models with nearby non-Tiankeng forests. Finally, the SAD pattern was related to the environment, climate, topography and human activities. Tiankeng forests will remain its own unique ecosystem functions and ecological position if we enhanced the protection of the nearby non-Tiankeng forest.

Round 2

Reviewer 2 Report

I want to congratulate the authors for the revision. The manuscript has considerably improved now. I am sure that it will make a good contribution. After reading the revised version I do not have further comments. There are a couple of missing spaces but that can be easily fixed during the proof stage.

Reviewer 4 Report

All my comments have been properly addressed.

I think this manscript is ready for publication.