Distinct Community Assembly Mechanisms of Different Growth Stages in a Warm Temperate Forest
Round 1
Reviewer 1 Report
Comments to the manuscript entitled “Changes in community assembly mechanisms with vegetation growth in warm temperate forest”
The manuscript assesses the role of biotic interactions and abiotic filtering on community assembly of saplings and adult trees along an elevational gradient, by looking at the phylogenetic structure and beta-diversity of these communities. To do so, the authors used the elevational gradient of Mount Lao, in northern China, where adult tree and sapling species were recorded in 69 forest plots distributed in six transects from low to high elevation. They show that saplings and adults differed in their phylogenetic structure, the first being more convergent and the latter divergent than expect by chance. Additionally, soil ammonium nitrogen content was the most important predictor for adults, while geographical distance was the most important predictor for saplings.
The study is quite interesting and I appreciate the amount of effort put into it. I appreciate that the study has very clear and concise goals. However, there are many issues that need to be addressed in a thorough revision. The introduction and conclusion should be expanded and backed-up with more solid theoretical explanations and examples from the literature. There are many inconsistencies and imprecise terminology which make the reader confused. I tried to provide many suggestions which I hope will help the authors to improve the manuscript.
Detailed comments and suggestions:
Title
Suggestion: ‘Distinct community assembly mechanisms of different growth stages in warm temperate forest’ ‘Distinct community assembly mechanisms of different growth stages in warm temperate forest’
Abstract
In general, I would recommend to keep the sentences short and direct. This makes the text easier to read and conveys clearer messages.
L8-10: the subject of this sentence is so long it makes it difficult to get the verb right (it should be ‘provide’). I would suggest to either remove ‘as a complement to species-centric approaches in biodiversity studies’ or try to split it into two sentences: ‘Community phylogenetic structure and diversity analysis can be useful complements to species-centric approaches in biodiversity studies, by providing new insights into the processes driving community assembly.
L10-14: this sentence could be split into many small ones. There is too much information packed here, making it difficult to follow. For example: ‘In this study, we aimed to understand the differences in the relative importance of abiotic filtering and biotic interactions on community assembly at different vegetation growth stages. We also looked at the influence of geographical distance, altitude, terrain, and soil factors. To do so, we examined the phylogenetic structures and β-diversities of saplings and adults along different abiotic gradients.’
L12: ‘altitude’. I suggest using ‘elevation’ instead, throughout the whole manuscript. See (McVicar and Körner 2013).
L13: remove ‘respectively’. Here it would mean that you only analyzed the phylogenetic structure of saplings and beta-diversity of adults.
L14-16: this sentence is a bit vague and does not really explain what is clustered, the phylogenetic structure of the saplings? Suggestion: ‘The results show that instead of being random, the phylogenetic structure of saplings tended to be clustered whereas that of adults tended to be divergent. This means that the relative force of abiotic filtering and biotic interactions changes throughout growth.’
L16: ‘GDMs’. Spell it since it is being used for the first time?
L16-20: again, please break this very long sentence into small, easy to follow, pieces.
L17: ‘diffusion restriction’ has not been mentioned or explained in the context of the study, so it is hard to understand this part of the sentence without having read the manuscript.
L19-20: this last part of the sentence is very hard to follow.
Introduction
It is in general well written, and I appreciate that the sentence structure is much more direct and clear than in the abstract. However, the introduction is rather short, and could be expanded, as a few key topics for understanding the background of the study are missing. I would suggest to base the structure of the introduction on the objectives (L76-80), which are quite clear. In order to provide the necessary background to understand these objectives, the authors should explain 1. Possible mechanisms involved in community assembly: abiotic filtering (different abiotic factors), biotic interactions and dispersal limitation. 2. Debate between neutral and niche theory (stochastic vs deterministic processes). 3. How these mechanisms reflect into different assembly/phylogenetic patterns (convergence, divergence and random). Also explain a bit about niche conservatism and the link between phylogenetic and functional diversity. 4. Why is it important to look at the different vegetation growth stages, and what to expect in terms of assembly patterns and mechanisms? (completely missing) 5. Mountains as study systems (L52-61 are ok). 6. Study region hypotheses and objectives (L62-80 are ok).
Please review all references. Some references are not fitting.
L25-27: the first sentence can be considered plagiarism from ref 1 (Gastauer et al. 2020, which is wrongly written as Markus): ‘Environmental heterogeneity, naturally fragmented occurrence, dispersal limitation and isolation permit independent patch dynamics and contribute to elevated diversity in these ecosystems (Korner and Spehn 2019).’ Please remove it from the manuscript!
I would suggest to start with something like ‘Understanding the role of the different processes driving community assembly and diversity patterns remains a central topic in ecology.’
L27-29 I am missing an explanation of how the mechanism of biotic filtering actually works, as you have done for biotic interactions below.
L29-31: I am not sure this whole part about scale is really necessary, since it is not exactly addressed in the rest of the manuscript. It might make sense to keep it if a link is made later in the methods to justify the choice of abiotic variables: ‘Since this is a local scale study, we chose to focus on soil variables.’
L31-32: ‘a primary role in the species assemblages in local communities’ this expression does not make sense to me. Perhaps something like ‘a primary role in the sorting of species into local communities’?
L38-39: explain a bit about niche conservatism and the link between phylogenetic relatedness and functional traits, so that the reader can understand why you perform the analysis of phylogenetic signal in the selected functional traits.
L44: here and throughout the manuscript, keep the terminology consistent and use ‘abiotic filtering’ instead of ‘filtration’.
L43-45: there are two issues with this sentence. First, it contradicts the rest of the paragraph about neutral theory. A random phylogenetic structure should be evidence for stochastic processes, and as said in the next sentences, ‘dispersal limitation and local extinction’, instead of ‘abiotic filtration, in combination with biotic interactions’. Second, and please correct me if I am mistaken, but I did not find anything in the listed references 15 and 16 that supports this statement. I think I found in the 5th paragraph of the introduction of reference 17 the source for this sentence, but the authors (Kembel & Hubbell) offer no reference for it and it appears to be mere speculation. So I would suggest to remove this sentence, or move it to the end of the paragraph.
L49-51: as explained in my general comments, this part should be expanded into its own paragraph.
L52-53: I suggest changing ref 20 for (Körner et al. 2017).
L57: here and throughout the manuscript, keep the terminology consistent and use ‘abiotic filtering’ instead of ‘habitat filtration’ or other variations.
L73-80: In this final part, I am missing any information about phylogenetic structure. What are the expectations?
Methods
L91: ‘Mount Lao is rich in vegetation’ does not read well. I would suggest something like ‘The two main vegetation types in Mount Lao are temperate deciduous broad-leaved forests and temperate coniferous forests’.
L100: I think ‘transects’ is more clear than ‘sample lines’.
L109-120: This is the first time functional traits are mentioned, without any link to the whole background of the study which has been presented so far. As suggested already, the link between phylogenetic and functional diversity should be explained in the introduction. The reader is missing the information that the phylogenetic approach is used as a proxy to functional diversity, based on the idea that closely related species are ecologically/functionally more similar (niche conservatism) and that information on phylogenetic distance might contain unmeasured aspects of plant functioning which are not captured by the selected traits of a particular study.
I would also suggest here to start the paragraph making reference to why traits are being used. For example: ‘In order to confirm that phylogenetic relatedness reflects how ecologically similar species are, we selected the following functional traits:…’. This way, the reader can understand why functional traits are being mentioned now and it also works as a link to section 2.4 of the methods and the phylogenetic signal analysis.
L109-110: this is not a good justification of why these traits should be selected. Maybe try to justify the selection based on the their relation to the major axes of plant form and function (Díaz et al. 2016).
L122-125: this is a bit hard to understand even with the appendix S1, and reference 36 is in Chinese. What is meant with slope position being divided into three levels? Was slope taken at 3 different parts of the plot or did you classify 0-15o as downhill, and so on? If so, why not use the actual slope values? Please clarify.
Remove ‘This’ from ‘This aspect…’
In Table 2, why is position considered a continuous numerical variable here? If I understood correctly it is a categorical variable. Why is aspect varying between 1 and 5? In the main text and in S1, it has only 4 levels.
L149: I would replace ‘proved’ with ‘assessed’.
L155: a short justification of why NRI was chosen out of the wide range of available phylogenetic distance metrics is missing. Why not use MDP or Rao?
L175-182: It is not clear to me why is the variability of environmental factors calculated and how it helps to understand the questions of this study. I would remove it.
L184: Did you use some criteria (such as AIC or R2) to compare the linear and quadratic regressions and choose the ‘best’ one? Keeping both I think does not make sense. And are these mixed models? Shouldn’t the authors account for random factors such as transect (sample line) and elevational band?
L187: nowhere in the manuscript it is mentioned what GDM stands for. Please add.
L185-194: From my limited understanding of GDMs, the response variable should be a dissimilarity matrix among all plots (calculated as Sørensen index, e.g.), indicating compositional turnover. So I don’t understand why and how was the net relatedness index (NRI) fitted here instead of the βSOR or βSIM, unless this is a typo.
In any case, I don’t see how the analysis of the phylogenetic β-diversity helps to understand the questions proposed in this study. This aspect was not clearly presented in the introduction and hypotheses, and is almost not present in the discussion. I would suggest to remove it, and instead use other kind of models to evaluate the influence of the different abiotic variables on NRI. This would make the analysis more straightforward and directly related to the objectives of this study.
Figure1: all figures are very pixelated (not sure if this happened during the preparation of the pdf by the journal or if the original quality was already low). The legend should explain more about the study, so it can be understood on its own. For example: ‘Locations of the 69 (20 m x 20 m) forest plots sampled along the elevational gradient of Mount Lao, northern China. The plots were established on temperate deciduous broad-leaved and temperate coniferous forests, and tree adults as well as saplings were recorded.’
Results
The results are over packed, making it difficult to keep in mind the main objectives of the study. I suggest moving some parts to the supplementary materials.
L196-201: I think the heading of section 3.1 could be removed and this general paragraph can stay without a heading.
L198-202: as I explained in the methods, I think this part is not necessary. The appendices are missing, so I cannot tell how the information about the plots is presented there, but for me it would be very useful to see each of these variables plotted against elevation for each transect (sample line). That would give a more clear overview of how all these abiotic components vary among the plots.
Table2: the legend, as mentioned for Figure 1, has to give some context about the study, i.e., explain key aspects so the reader can understand the table by itself. Meaning of C.V should be mentioned.
L206: typo in ‘By removing the rare species with fewer than five species’. Instead ‘By removing the rare species with fewer than five observations’? This means that 57 species had such low frequencies?
L207-211: the interpretation of the results are a bit weird. The significant phylogenetic signal means that, for those traits, phylogenetic relatedness should contain signatures of the trait differences between species and should, therefore, provide equivalent results in the further investigation of the processes behind community assembly. For the other two traits, a trait-based approach should still be used, since the phylogenetic approach does not contain signatures for those traits. This is also why I think both trait-based and phylogenetic approaches should be done in parallel, as done by (López-Angulo et al. 2018). As concluded by López-Angulo et al, these should be used as complimentary analysis, and not as proxies of the other.
As the authors here already collected all the trait information, it should be relatively simple to implement a trait-based analysis in parallel (using some metric of functional diversity). I think the manuscript would benefit from that.
Figure 3: I suggest moving this to the appendices, and adding in parentheses in L224-229 the Pearson correlation results next to the relevant information.
Figure 4: as mentioned before, I think this should be a mixed model or a generalized mixed model.
L249-250: ‘We chose adults because it was an aggregate macro variable.’ I believe this sentence was misplaced here. Please remove, and if added in another part explain better.
Discussion
L297-300: in these two opening sentences, new terms are used, such as ‘clustering’ and ‘overdispersion’, which have never been introduced. Changing the naming of core concepts throughout the manuscript makes it confusing for the reader who cannot tell these terms are meant as synonyms of previously introduced terms (aggregated, clustered, convergent). It is best practice to define the core concepts and stick to the same terminology. I recommend to define in the introduction the possible dispersion patterns: convergent, divergent, random (in terms of traits, phylogeny and the associated mechanisms), and use them throughout the manuscript.
L300-303: first, there is some word missing after ‘randomly’. Second, I am not convinced that this statement is well supported by the analysis, or that it makes sense to group the NRIs of both adult trees and saplings together. The different growths stages were sampled at different scales and the analysis did not contemplate grouping of the two groups. One of the motivations of the study design was exactly that different processes of community assembly act on different growth stages, and the authors do find results to support this when looking at each group separately.
L319: stay consistent. Change ‘filtration’ to ‘abiotic filtering’.
L342-348: first, this sentence should be split into smaller pieces. Second, the term ‘diffusion restriction’ has only been mentioned in the abstract and now in the conclusion. If I understand correctly, this is equivalent to the previously used term ‘dispersal limitation’. Please standardize.
References
Some references have the first name of the author spelled out instead of the last name (e.g., ref 1, 3).
Please revise that all references are actually appropriate. Ref 12, for example, is clearly misplaced as a reference for L41-42. The whole point ref 12 is to debunk the use of phylogenetic data to evaluate assembly mechanisms, so I do not see how it fits to L41-42.
Appendices
S1: see comments to L122-125.
S4: The values for latitude have a weird range from 0 to over 34000. Which coordinate system is this? Please present the coordinates in the same format as in Figure 1 (degrees, minutes, seconds).
References used in this review
Díaz, S. et al. 2016. The global spectrum of plant form and function. - Nature 529: 167–171.
Körner, C. et al. 2017. A global inventory of mountains for bio-geographical applications. - Alp. Bot. 127: 1–15.
López-Angulo, J. et al. 2018. Interactions between abiotic gradients determine functional and phylogenetic diversity patterns in Mediterranean-type climate mountains in the Andes (I Kühn, Ed.). - J. Veg. Sci. 29: 245–254.
McVicar, T. R. and Körner, C. 2013. On the use of elevation, altitude, and height in the ecological and climatological literature. - Oecologia 171: 335–337.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This study deals with the evaluation of the sources of explanation for the assembly of forest tree and shrub communities. It is well written and of an adequate length. Despite authors looked at well-known concepts and mechanisms, the main interest of the paper stands in the use, in an elegant way, of phylogenetic and Beta diversity metrics to explore how and why understory and canopy communities are distinct in a large geographical area.
Nevertheless, this study is somewhat confirmatory, and, to me, its soundness is thus downsized to a case study. Moreover, I do have important concerns about the design, which will limit the scope and significance of the content.
I do think that these concerns can be addressed by changing some statistical analyses and some inferences in the discussion.
Main concerns:
- The way phylogenetic and B diversity metrics were used is really interesting but, to me, such use has to be associated with a systematic sampling in the studied area to avoid the nested effect of transect sampling. Of course, this is critically difficult to achieve so that transect sampling is ok, but you can not produce a mean NRI or B value for the whole area before checking for transect effect (as random or fixed effect). Indeed, transect may dramatically control for community assembly and you need to confirm that the relation is true in all these subsamples. Since you measured plots at regular altitudes, you did establish a systematic sampling within a transect (i.e. at the transect scale), but not between (i.e. not at the whole area scale). To me, after presenting results for all transects at each vegetation stage, if they all go in the same direction, you will be allowed to merge them and be confident to discuss the results per transect or for the whole area.
- My second concern is about the comparison of the canopy vs sapling stage. Despite they are probably true, I am not presently confident about the strong conclusions you make. Direct comparison is possible if both communities have very similar range of species but (if my calculations are ok) saplings have 45 species not represented in canopy layer and canopy has 17 species not present in saplings. Only 45 species were shared by both groups. First, that mean that biodiversity of saplings is much higher and may contains shrub species that will never reach canopy (neither because of biotic nor abiotic reasons/mechanisms). This has critically mathematical consequences on phylogenetic metrics (such has NRI or B diversity because the potential pools of species (i.e. gamma) are different between both vegetation stage. Furthermore, based on the 20x20 plots you sampled many more individuals for sapling stage than for trees (increasing the chance to find more species in the sapling stage). The use of contrasting sampling area can help to control for density dependent effects. These two biases tend to increase the chance to get higher diversity in the sapling stage and thus possibly higher NRI and B diversity.
o Nevertheless, I am pretty sure that your conclusions are good, but you need to deal, discuss or at least acknowledge this potential biases.
Some minor comments
- Some sentences are really too long (in the summary for example - L10-14 or L16-20)
- L187: you didn’t define GDM in your text. Please write it once before using the acronym.
- Check the edition of table 2 (parentheses in the first column)
- L206: Despite necessary, I do not like the removing of rare species. They are key and important species in the ecosystems and not taking them into account is not totally right. It is an open comment, but we need to find a way to include them or do analyses with and without them.
- L206 “with fewer than five OCCURRENCES/INDIVIDUALS” (not species)
- Figures 5 and 6 are low quality, improve size of the charts and resolution
- Conclusions are probably true, but you need to deal or talk about our limitations before stating them like this.
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Round 2
Reviewer 1 Report
I really appreciate the effort put into the revision of the manuscript. The conclusion has improved substantially. The authors were very thorough and replied to all the comments raised. They implemented many of the suggested changes. Additional clarification was also provided about some unclear points. I am satisfied with the current revised version, but there are still some additional comments to be addressed, and some previous comments which were addressed in the reply letter but not implemented in the manuscript. I know this could be due to time pressure, but I would suggest to ask for an extension and not feel pressured by tight journal deadlines.
Detailed comments and suggestions:
Title
I know it was my suggestion, but I think there is an ‘a’ missing: Distinct community assembly mechanisms of different growth stages in a warm temperate forest
Abstract
L27: Dispersal limitation comes out of nowhere. Either add ‘geographical distance’ in parentheses next to it, or use geographical distance.
Introduction
L57: ‘species pool’ instead of ‘species bank’?
L70: ‘Functional trait studies are important means’?
L73: references about 1. measuring functional traits and 2. intraspecific variation is missing.
L75: ‘proxy for functional diversity’. ‘However’ instead of ‘but’?
L83: I suggest removing ‘(referred to as ‘null models’)’, and in the following lines substitute ‘null models’ with ‘expected by chance’.
L95: add some explanation about dispersal limitation in forest communities. If there are many stochastic events why is ‘dispersal limitation’ used in the context of this study? Later in the discussion this is also not discussed: What does it actually mean for the communities/saplings to be strongly influenced by dispersal limitation?
L103: I still miss more explanation (perhaps a whole paragraph, as I previously suggested) about the different growth stages, why they should be investigated separately and what are the expectations for each (similar to the paragraph about mountain ecosystems). It is very difficult to understand why the authors decided to look at saplings and adults separately, especially when the only reference cited is only available in Chinese.
Methods
L231: now the explanation about ‘position’ is more clear. Consider renaming it to ‘elevational band’.
L326: ‘Nestedness’
L333: ‘multiple-site’ or ‘multiple-plot’? Otherwise I am not sure what is meant as a site, and it should be clarified. Same applies to L338.
L336: remove ‘matrix’
L339: ‘metrics’ or ‘indices’ instead of ‘matrices’?
Also, add which of these dissimilarity indices were used for the GDM analysis. The reader only finds out in the results (and then it is not clear why beta-sor or beta-sim are used for different (?) GDM analysis, Table 5 vs Figures 7 and 8). Having to go back and forth to solve these questions makes it hard to read the manuscript in a flowing way. Finally, if the index used is the Sorensen and/or Simpson, why is nestedness even calculated? It does not add anything to the story/questions, as even in the discussion this part is completely disconnected from the rest. I insist in recommending to remove the components of beta-diversity which are not further used in the GDM analysis and stick to Sorensen and/or Simpson (whichever is the one actually used in the GDM analysis), or integrate the whole idea of nestedness better into the whole manuscript.
L351-352: why was the significance of individual predictors not tested? It is possible to do so with the same R function. This would give actual support to the discussion of which predictors are affecting the phylogenetic dissimilarity of adult trees and saplings.
The two last paragraphs about β-diversity and GDM are still lacking some context in the sense of which question/objective from the introduction they are addressing. I understand there is some explanation about the GDM analysis in the discussion, but that is rather late. It helps the reader to have a clear thread throughout the whole manuscript, explicitly connecting the objectives and key terms presented in the introduction with the choice of methods and following results/discussion. So I would add something at least for the last two paragraphs, explaining why you did these additional analysis. Regarding β-diversity, it is lacking completely, and from what I understood it was only used to feed into the GDM. So say this explicitly. Otherwise the reader is left wondering how are the different components of beta-diversity helping to answer the initial study questions. For the GDM part, as mentioned in the reply letter, something like: ‘In order to assess the effect of dispersal limitation (stochastic processes) on community assembly, we performed a GDM analysis’ or ‘The GDM analysis allowed us to assess the effect of dispersal limitation on community assembly, by taking into account the effect of geographical distance.’
Results
L356-3576: Remove together with Table 2? Nowhere in the manuscript this analysis or results are shown. The tables mentioned show raw data and descriptive statistics, but no test results. An alternative to keep Table 2 would be to actually mention what is shown in the table, e.g., ‘Environmental factors varied greatly between plots’ (without mentioning any test or significance). Again, Position (or elevational band) should be a categorical variable (because position 2 is not necessarily twice as large as position 1, and so on), so presenting an average, min and max makes no sense.
L375: missing a ‘by’ in ‘expected by random differentiation’.
L395: ‘converge’ instead of ‘aggregate’ for keeping the wording consistent.
L401-04: remove the part ‘, and the slopes of the fitting lines were basically the same in 401 different transects (Fig 4)’ and ‘, and the phenomenon was more pronounced at lower elevations (less than 200m) (Fig 6)’, as well as Figures 4 and 6. Transect was used as a random factor, so this means you are accounting for the differences among transects, but you are not interested in the differences, only in the general results of all transects together.
Figures 3 and 5: I would merge them as you had in the previous version.
Table 5, Figures 7 and 8: as I mentioned before, it is not made clear in the methods which dissimilarity index is used in the GDM analysis and here it is confusing to see one table referring to the results regarding Sorensen’s (or is it a typo?) and the following figures regarding Simpson’s index. Table 4 shows they are basically the same, so why not proceed with only one of them?
Also regarding Table 5, the legend is a bit misleading, as it only mentions percent deviance explained, which is the first row of the table. The rest is not mentioned.
Discussion
Add references to the Figures throughout the discussion, so it is easier to know which results are being discussed.
L459-555: as I explained before, this whole paragraph does not fit into the story of the manuscript. Besides, the point about high turnover and different species being restricted to different zones does not fit well with the fact that no evidence was found here for abiotic filtering acting on adult trees (due to mild conditions, as the authors explain).
L556-558: the full model was significant, but the testing of individual predictors is missing as I mentioned before. The additional analysis would make this whole discussion better supported.
L561: Please correct me if I am wrong, but I believe summed coefficient values for a predictor is not the same as explained variance (even if they might have a similar interpretation).
Appendices
S1: Add legend or title like in the other files.
S4: Please add in the legend or footnote the full name of the columns.
S5: The measuring units show weird symbols for me (LDMC£¨%£©, WD£¨g/cm3£©). I tried opening the file with different software, all in English, and the same problem occurred.
Old Figure 3 is still missing from the appendix.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
