Effects of Fire on Diversity and Aboveground Biomass of Understory Communities in Seasonally Dry Tropical Forest in Western Thailand

Round 1

Reviewer 1 Report

1. The pioneer species are almost the light demanding species that grow up first after fire, and then shade tolerant species grow up later. Because the prescribed burning just decreases small percentage of canopy cover of tall tree, the variation of penetrating light through canopy of this type of tree would be small or neglect small. Hence, the parameter affecting aboveground biomass and the growth condition of regenerated species, except for competition among these species, might include the change of soil nutrition before and after prescribed burning. Attaching the relative data of soil nutrition would be better for this study.

2. According to the content of Table 4, “MDFB, DDF, and MDF plots” on Line 187 might be revised as “DDF, MDF, and MDFB”. 

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

General observations

The manuscript analyzes variations in composition, richness, abundance, and aboveground biomass of plants in function of forest types, fire control regimes, and life forms. The manuscript lacks a potential explanation in terms of the theoretical basis that precedes their study and characteristics as species analyzed as study area, i.e., a hypothesis. The theoretical basis that justifies the sources of variation is weak, therefore I suggest that they present effects of fire control regimes on life forms, composition, richness, abundance, and aboveground biomass. The information presented in introduction will allow authors to propose a hypothesis that was tested with the methods that reported on methods section. The methods report a parametric test to compare two samples, Wilcoxon, but it only could be applied on biomass and canopy variables, continuous nonparametric variables. However, authors omit present analyses to compare differences in species composition, abundance, and richness. The grouping of species into life forms can lead to statistical type II error, since grouped species with different heights, such as Asteraceae and Fabaceae which were grouped in forb life form. The justification and analysis of seedlings in deciduous dipterocarp forest was absent from the methods and introduction. The results present graphs with confidence intervals and means, but the analyses used are nonparametric, therefore, is not proper report position and dispersion parametric descriptors. I suggest that authors use whisker plots. The discussion and conclusion should be updated according to the changes in the introduction and analyses that I suggested.

Individual Observations

Introduction

Line 20: Keywords: I suggest that authors use words different from the title

Line 23-38. The authors report in their first paragraph of introduction: “The filtered light can limit growth and survival of certain species and therefore determine the composition and structure of understory communities”, which suggest that authors will be use filtered light as independent variable to explain the composition and structure of understory communities. However, in methods and results they use filtered light as dependent variable. Therefore, I suggest that the authors present and justify in introduction the sources of variation that they analyzed: with and without fire control regimes, forest type, life form and plant species, since are important part of the study.

Line 23. I suggest that author use active voice in their orations. For example, for "While covering only 2.85% of the Earth surface, tropical deciduous forests stand out among the other biomes, for their high biodiversity and unique floral composition", could be change for "Tropical deciduous forests covering only 2.85% of the Earth surface, but contain high biodiversity and unique floral composition.

Line 26-27. Please enlarge description of canopy counterpart, "more vulnerable than their canopy counterpart". Authors refer to perennial species?

Line 49-51. Studies of the effect on forest of APs HKK and HTS are uncommon. However, I consider that authors could present results of other forest that analyzed effects of fire suppression on plant community, which can use to propose a hypothesis. How did they expect plant community respond to fire suppression? e.g.

https://cdnsciencepub.com/doi/10.1139/X07-210

https://link.springer.com/article/10.1007/s10980-013-9924-7

https://fireecology.springeropen.com/articles/10.1186/s42408-019-0062-8

https://www.frontiersin.org/articles/10.3389/ffgc.2022.940302/full

Methods

Figure 1. I suggest that in map of Thailand include the surrounding countries and seas, as well as a regional view of the country's position at continent or semi-continent scale. Also, I suggest that change the unique symbol of sample plot by a symbol for each forest type.

Line 108. The grouping of species into life forms can lead to type II error, since groups include species with different heights, such as Asteraceae and Fabaceae which were grouped in forb life form. I suggest that authors use multivariate analyses to compare structure and biomass between forest types and fire control regimes. The advantage of using multivariate analysis is that the identity of the variables (species) and sampling sites is not lost, unlike univariate analyses. Therefore, I suggest that authors apply the PERMANOVA test to determine if exist differences between sources of variation (forest types or fire control regimes). The analysis can be performed in the free program PAST (https://www.nhm.uio.no/english/research/resources/past/). If significant differences exist, they can apply the SIMPER test to determine which species contribute most to difference in species composition.

Results

Line 127. The section 3.1 is too descriptive and lacks an analysis to identify differences in species composition among the sources of variation. Therefore, I suggest that authors use PERMANOVA and SIMPER analyses (see observation line 108).

Line 155. The analysis of richness was absent of methods. I suggest that richness would have to be estimated and compared with non-parametric extrapolation of rarefaction curves (or intervals), which can be calculated in the program Estimates  https://www.robertkcolwell.org/pages/estimates

Line 173. The Wilcoxon test is for comparing two groups, but Table 1 presents five types of forest and only reports one probability value, also, omitted report calculated value.

Table 4. The test applied is a nonparametric, therefore parametric statistical descriptors (mean and standard deviation) do not represent the samples. For example, in figure 4 confidence intervals of FORB versus AO or SCRUBS versus MDFB are not intersected, which occur when two samples are statistically different, but probabilities do not reflect significant differences.

Lines 180-181. The authors report that “whereas that of forbs and grasses showed a distinct decrease (p < 0.05, Figure 3)”, but here is no difference, the interpretation is erroneous, since the probability is greater than 0.05.

Figure 3. Authors report 10 comparisons, but only two probability values are reported and lack calculated value of Wilcoxon.

Line 203. The results of section 3.5 should be justified in introduction and report the procedure to obtain them in methods.

I suggest applying an association analysis, such as the multiple correspondence analysis (https://rpubs.com/gaston/MCA), given that they analyzed the frequency of individuals in three variations sources: before-after fire, forest type and species.

Discussion

I suggest that the authors discuss the factors that may compromise their results and conclusions, for example sampling design or statistical analysis used. In addition, it would be interesting that authors present future studies that should be carried out to support their results or to improve the understanding of the effect of fire control.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

The paper entitled “Effects of Fire on Diversity and Aboveground Biomass of Understory Communities in Seasonally Dry Tropical Forest in Western Thailand” focused on determining the species composition and aboveground biomass and comparing the changes in aboveground biomass of understory plants before and after prescribed burns in two study sites. The results maybe helpful for managers and foresters to manage forest structure and control fire. The paper prepared good, and results described well and interesting, however I need some minor revision:

1. LN 69: “about 173 km2” 2 must be a small letter above the m

  2. Provide a paragraph in introduction as a literature review that represents the studies same as your study in different climates as well as tropical

3 3. What are the differences between two sites that you want to compare? Write about it in study area?

  4. Table 4: use stars to show significant differences

  5. Write about F values as well as p values in whole manuscript

   6. What are the limitations in your study? Mention all in the text.

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

This paper has revised. I have no more comment.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

General observation.

The authors made significant improvements to their manuscript. However, I believe they could reconsider the use of statistical tests to support their arguments. The authors propose that "Fire suppression can lead to accumulation of fuel loads and alter the composition of the understory and tree communities". The potential explanation is that the fuel load and understory community composition will be altered. Altered, I understand, can include anything from non-significant to significant changes. I assume the authors are looking for non-random trends to propose solid conservation strategies, which can be derived with statistical analyses. The advantage of statistical analysis is that it will allow discerning if results derived from the effect of a variable independent variable or by random chance.

Particular observations

Authors: We added a paragraph to address overall hypotheses regarding what we expected to see in our study site, based on the previous literature as suggested by the reviewers.

 

The sentences presented in lines 49 to 62 suggest that a single fire regime and history of cover change will alter the understory plant composition of dipterocarp forests. The authors propose that "Fire suppression can lead to accumulation of fuel loads and alter the composition of the understory and tree communities". The potential explanation is that the fuel load and understory community composition will be altered. Therefore, there is one source of variation (fire regime) that will modify two variables (fuel load and understory community composition). By deduction, it can be interpreted that fuel load will increase, but such an assertion is implicit, when it should be explicit; and it is not clear for species composition alterations. Thus, I suggest that they can propose a hypothesis that addresses the response that they expect to get, for example, that areas with a controlled fire regime will have a higher fuel load than unmanaged areas. In this regard, the authors present a parametric test to test this hypothesis. However, species composition lacks a direction of the results that authors expected to obtain before conducting the field work and a statistical test to determine if effect of the independent variable is real or due to chance. The authors present a list of species (species composition or life forms) at the sampling sites but omitted to use a statistical test to determine whether the variation between and within groups of the independent variable (fire regime or forest type) is because of these variables or to chance.

The authors' justification for omitting a statistical test to analyze species composition is since it corresponds to another document. However, in the current manuscript it is mentioned that the species composition will alter, so it is necessary to use an analysis that determines that the observed variations are because of the fire regime and not to chance. The document presents a table (2) with the most dominant species by forest type and a supplementary file with the species recorded, information that can be used to determine if the species composition before and after a fire varies statistically. I consider that if authors limit themselves to report only a list of species in the study sites, then the document will have a high regional component, where readers who are in contact with these species may be interested in the manuscript; but if they apply statistical tests, then the results they present may be used by a wider audience, since they will give more weight to the effect of the fire regime or forest type on the species than individual response of species.

Finally, I suggest that the hypothesis could be more explicit and avoid that reader inferring what the authors want to test. For example, that sites with fire management will have lower biomass than unmanaged areas, which can be tested with the non-parametric statistical test that use. In the case of species composition, the word alter implies change, hence differences. Therefore, I suggest their hypothesis expresses what the authors want to test. If test differences, then they should use a statistical analysis.

 

Authors: The current study focuses on the effect of the fire on biomass and forest structure, in particular the life forms of understory plants, because the study was designated as a non-hunting area, and the data on availability of herbivore food source is a priority for the management of this area. Therefore, the comparison of difference in species composition and richness before and after fire is a subject for another study.

 

The authors' response to omit the use of statistical tests to compare composition, richness and abundance is inconsistent with what is presented in the study. In line 58 of version 2 they refer to "alter the composition of the understory and tree communities" and from line 164 to 185 they report a summary of the taxa recorded. Therefore, they should present an analysis that evaluates the alteration of species composition depending on fire management and forest type. The reader will be able to identify more clearly the emerging pattern with results that indicate that there is an effect of the independent variable on species composition, and not by because of chance. With respect to abundance, the authors present a table (2) in which they report the most dominant species, but it is also necessary that these values be supported with statistical tests that allow the authors to give non-random responses. Finally, in lines 180-182 the authors report "The pattern of species richness in these forest types was similar to that before the fire except for the unburned plots, where MDFB showed the most diverse understory plants". In this regard, the similarity in the pattern of richness must be supported with an analysis that proves that the number of species is statistically similar and more diverse.

 

Auhors: The PERMONOVA and/or SIMPER analyses could certainly be done with the current dataset, but we believe that the results will distract the main point of the comparison of life forms and biomass and after the fire.

The objective of PERMANOVA and SIMPER is compared and identify the species that lead to these differences, respectively. The PERMANOVA test allows to identify significant differences between categories of one or two sources of variation, which is the multivariate variant of ANOVA and Kruskal-Wallis test. I believe that the test would not distract the reader; on the contrary, if authors use these analyses, then they would reduce the number of statistical tests that the authors would use. For example, to compare biomass they had to perform 10 statistical tests, and this analysis cannot evaluate the interaction of fire regime and forest type. The two-way PERMANOVA allows them to analyze the two sources of variation, report the radius of variation between and within groups (F value), the probability value and the matrix of comparisons between the categories of sources of variation in the same analysis, which would not distract the reader. On the other hand, Table 2 presents the dominant species of each vegetation type. But these abundances and biomass differ between fire management regimes? The PERMANOVA test will allow you to determine whether it is the more abundant species that are causing the differences or the less abundant species. In addition, if the study is oriented to the vegetation consumed by herbivores, where each species has a differential use of resources and tends to select its food, then it is necessary to identify in statistical terms which are the plant species (dominant or non-dominant) that present differences between fire regimes and forest types.

Authors: The rarefaction is normally used when the sampling efforts are uneven. Since the number of plots before and after the fires are the same, meaning the same sampling efforts. We are not attempting to compare the species composition between the forest types here, and therefore do not see the need for this particular analysis.

The authors argue that a "The pattern of species richness in these forest types was similar to that before the fire except for the unburned plots, where MDFB showed the most diverse understory plants". However, they omit to present an analysis to support this statement.

Authors: The Figure 3 showed the significant decreases of forbs and grasses with p-value less than 0.05 in the burned plots (left panel). The interpretations are already consistent with the reported p-values.

In this regard, I apologize, I was referring to the interpretation of the first part of that sentence (lines 179-180) in which they refer to "When considering each life form separately, we found that AGB of bamboos 179 and tree seedlings showed a slight increase (p > 0.05, Figure 3)". The probability is greater than 0.05, so referring that there is a slight increase is incorrect, since the acceptance threshold of the null hypothesis is that the value must be equal to or greater than 0.05. In this case they report that it is greater than 0.05, so there is no significant relationship between both variables.

Authors: The community analysis (species-by-site multivariate analysis), while it could be done, was beyond the scope of the current manuscript, as we are attempting to estimate the biomasses of the understory for the management purpose. Adding another section on community analysis will involve discussing and dissecting effects of fire on individual understory species, which will add to the considerable length of the manuscript.

The test would synthesize the tables and text presented by the authors, since a graph would display the species that are most frequent depending on the type of forest and the fire control regime, which would summarize Table 2, the paragraph found from line 164 to 185 and the supplementary file.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 3

Reviewer 2 Report

The changes made by the authors improved the quality of their manuscript, which will allow readers of the journal to use similar  theoretical and methodological approaches to those made in this paper. Therefore, I consider that the manuscript can be published.