Decline in Aboveground Biomass Due to Fragmentation in Subtropical Forests of China
Round 1
Reviewer 1 Report
This is an interesting study focusing on the overall impacts of fragmentation on the structure and function of subtropical forests in China. The authors used a large dataset from permanent plots in addition to field data collected in different forest fragments in the study area. The findings, while not surprising, are interesting in that they highlight the negative effects that fragmentation and its associated components (e.g., edge effects) have over the dynamics and structure of these forests. Overall, it is an exciting study that have the merits to be published. However, there are major issues with regards to how the authors analyzed the data and furthermore how they discuss the apparent negative effect that fragmentation have on AGB. In general, the ‘story’ is not well connected. Importantly, in my opinion there seems to me missing critical information on how the data was analyzed: why some other simple, yet more appropriate, analytical techniques (e.g., non-parametric anova). I also found that some of the results presented in the tables from the GLMs seem to contradict what the authors are saying in the discussion. A better explanation of the so called ‘simulations’ is needed, and the figures are really hard to read. In the attached pdf file, I have included a series of comments, questions and suggestions that should be addressed to improve the quality of the paper. I encourage the authors to revise and resubmit as it is an interesting and valuable study. Finally, I am happy and open to revise a new version of this study should the authors decide to do this.
Comments for author File: 
Comments.pdf
Author Response
Point 1: Decline in ,,,
Response 1: Thanks for your suggestion. We have changed the reductions instead of decline in in the title. Please see the Page 1, Line 2.
Point 2: I suggest starting the abstract with this sentence followed by the previous one.
Response 2: We have placed this sentence ahead of the previous one. Please see the Page 1, Lines 16-18.
Point 3: This is a bit confusing. It is clear that DBH decreased with decreasing patch size but the rest of the sentence is not clear. Suggest rewording this.
Response 3: We have revised this sentence. The meaning of this sentence is the average diameter of DBH increased with decreasing patch size in forest fragments. The average diameter of DBH increased with decreasing quadrat size within two non-fragmented forests. Please see the Page 1, Lines 25-27.
Point 4: if quadrats are simply the sampling method is this really neccesary_ why not simply say non-fragmented forests?
Response 4: We have deleted some words in this sentence. Please see the Page 1, Line 28.
Point 5: Consider a different word that is better connected with the findings: carbon? ecosystem services?
Response 5: We have changed this word instead of aboveground biomass carbon density. Please see the Page 1, Line 31.
Point 6: between 6 to 17% of...
Response 6: We have revised this part. Please see the Page 1, line 43.
Point 7: these ecosystems (to avoid repetition)
Response 7: We have revised this part. Thanks for your suggestion. Please see the Page 2, Line 47.
Point 8: Instead of this sentence why not simply list a few of these human-caused disturbances?
Response 8: We have listed a few of human-caused disturbances in this part. Please see the Page 2, Line 48.
Point 9: what is the context here? are you referring to 'ecological' communities? if so, I think a different word is needed to avoid confusion with people's. Example: "The impacts of forest fragmentation on the structure and composition of forest communities can be complex".
Response 9: We have revised this sentence. Please see the Page 2, Lines 51-52.
Point 10: Juts delete and continue both sentences. "Forest fragmentation leads to....and isolation [7], and often can result in new...
Response 10: We have deleted these two words and joint the two sentences. Please see the Page 2, Line 53.
Point 11: Similarly, no need to repeat "Edge effects" and both sentences can be integrated into one.
Response 11: We have deleted the edge effects and integrated into one sentence. Please see the Page 2, Lines 56-58.
Point 12: especially in subtropical forests.
Response 12: We have added these words after “remains unclear”. Please see the Page 2, Line 73-74.
Point 13: This seems to be out of place here.
Response 13: We have deleted this sentence.
Point 14: These hypotheses are essentially one that refers to the effects of fragmentation in AGB and the potential mechanisms. Authors are assessing species composition, so a direct hypothesis about this part should be included too.
Response 14: We have re-write our hypothesis. Please see the Page 2, Lines 77-79.
Point 15: Please add a location figure for readers not familiar with the region.
Response 15: We have added the location figure in the manuscript. Please see the Page 3, Line 112.
Point 16: and the first census...just keep the flow of the sentence.
Response 16: We have revised this part. Please see the Page 3, Line 114.
Point 17: I want to reiterate that it would be useful to have some reference map of these fragments if possible.
Response 17: We have added the location figure in the manuscript. Please see the Page 3, Line 111.
Point 18: I don't neccesary have anything against this, but I wonder why 'forcing' normality if the data is non-normal? A non-parametric ANOVA test (e.g., Kruskal Wallis) could have been used here. It shouldn't take too much time for authors to run this and see if results may change?
Response 18: We have revised this part. Please see the Page 4, Lines 165-175.
Point 19: Does this include the variables that may have been log-transformed too? Some transformation might be neede yes but adding another standardization might affect the analysis.
Response 19: The purpose of data standardize was just to direct compare the estimated coefficients in the results of GLM.
Point 20: See comments later in the results about issues with these models: no AIC, no R2 reported, and no comparison between models.
Response 20: We used the stepwise multiple regression to chose the best model in GLM. This process was based on AIC, etc. However, we did not show these parameters in our results. We just show the final results. We have changed our expression in this part. Please see the Page 4, Line 181.
Point 21: Suggest stating the aspects that were especifically addressed: structure and compoisition of forest fragments and continous forests
Response 21: We have revised this part according to your suggestion. Please see the Page 5, Line 185.
Point 22: I recommend not using numbers or bullet points to report results.
Response 22: We have revised our results in paragraphs. Please see the Results part, started from Page 5, Line 184.
Point 23: This figure is problematic for different reasons: 1) is so small that we can't really appreciate the relationships; 2) I am not sure why AGB is not expressed in Mg per ha if that is the units that are being used in the X axis? 3) Abundance is the same of density? if so, it should simply says trees per ha 4) The difference between the dots and the lines are not clear. Authors say this is the 'mean' but the shapes look very weird, especially for richness and dbh.
Response 23: We have changed a better and larger image with higher resolution ratio. All X axis in the figure are the patch size of 69 forest fragments. The unit of AGB is Mg. Abundance in this figure represent number of stems in the quadrat or patch, actually is different with density. The black dots are the observed data in 69 forest fragments with different patch sizes. The grey dots are the simulated sampling with different quadrat sizes with two non-fragmented forests. The lines are the mean of grey dots within same sampled quadrats size. More details were showed in method part and reference (He, 2011).
Point 24: The size of the figure is too small. Why not including some 95% confidence intervals for each line?
Response 24: We have changed this figure with a larger one with higher resolution ratio. Actually, the figure showed different type of plots (interior, middle, and edge plots) in 69 forest fragments with different patch sizes. There is no repetition within each fragment. Therefore, no 95% confidence intervals for each line.
Point 25: how this figure is showing edge effects? we often tend to express these effects based on distance to the edge not necessarily as a function of size per se.
Response 25: Most of previous studies tend to express edge effects based on distance to the edge. Our manuscript also expressed edge effects based on distance. We established three types of plots (interior, middle, and edge plots) within each forest fragment. The interior plots usually located in the core area (center of each fragment) of each fragment. The edge plot usually located near (almost beside forest edge) forest edge areas. According to our method, the middle plot often located in the middle of the former plots. So, the distance of the middle plot to forest edge depends on the patch size of the fragment. The distance may higher in larger sized fragments. Therefore, the trend of middle plots in the figure showed the edge effects based on distance.
Point 26: But in Table 3, the coefficient for fragmentation is positive?
Response 1: We made a mistake of table 3. The coefficient for fragmentation is negative. Please see the Page 6, Line 229.
Point 27: These numbers don't say much without adding things like the R2c and R2m: the conditional (c) and marginal (m) R2 for all models. R2c is indicative of the variance explained by both fixed and random effects, whereas R2m indicates the variance explained by fixed effects only. When the R2m value is close to R2c, most of the variation explained in the biomass components is caused by the predictors (fixed effects), rather than by plot differences (random effect) (Nakagawa and Schielzeth, 2013)
Response 27: In the present study, GLM was used to estimate the factors determining AGB in forest fragments and non-fragmented forest. Stepwise regression was used to selected best models based on the lowest AIC. However, we just show the final results (estimated coefficients and p values) of lowest AIC in the table.
Point 28: But the figures show the opposite: as patch size decrease there is less biomass? I wonder if there is any effect on the multiple data normalization and standardization.
Response 28: The figures showed that as patch size decrease there is less biomass. So, actually there is positive relationship between patch size and AGB. So, the coefficient of patch size here is correct.
Point 29: Authors only analized a count of species to use as richness. While this statement might be true, this study didn't analyze biodiversity as a whole.
Response 29: We have revised this part. Please see the Page 7, Line 234.
Point 30: This is already said in line 245-246 in this paragraph. Might be good to tight these two sentences together.
Response 30: We have moved this sentence following the former one. Please see the Page 7, Lines 239-241.
Point 31: This in any case should be part more of the introduction or just to open the discussion. This is not discussing authors' results.
Response 31: We have deleted this sentence.
Point 32: Again, this 'biodiversity' effect of fragmentation might be true but there wasn't any result on this. At least make sure this has a better connection with the rest of the paragraph.
Response 32: We have revised this part. Please see the Page 7, Line 270.
Point 33: I'm curious to understand why in most of the discussion there is an agreement that AGB declines with declining patch size and yet in Fig 1 DBH seems to increase fr smaller patches? please clarify.
Response 33: AGB declines with declining patch size is main results of the present study. Although higher mean DBH in smaller patcher, we think these results were reasonable. The mean DBH: (1) In forest fragments, smaller patch sizes were more affected by edge effect, that is few stand stems, larger diameter (higher light intensity), leads to higher mean DBH; (2) In non-fragmented forest, smaller quadrat size contains fewer stems. However, the proportion of larger trees increased as quadrat size decreasing (the smaller the quadrat size is, the higher the probability of larger trees being drawn according to former studies). This also leads to a litter higher mean DBH. In addition, GLM showed that mean DBH had positive effect on AGB, which is agree with previous studies. However, the abundance is the biggest contributor to AGB in the present study. Therefore, despite higher mean DBH in smaller fragment, lower abundance leads to lower AGB.
Point 34: This need some citations
Response 34: We have added citations in the text. Please see the Page 8, Line 291.
Point 35: This is related to my question about DBH and patch size: if DBH was higher in smaller fragments what are the mechanisms that explain this apparent edge effect?
Response 35: Our data showed that the smaller fragments contain fewer stems. Furthermore, proportion of larger trees was higher and proportion of understory trees was lower in smaller fragments according to previous studies. In addition, few stems could receive more sunlight which leads to higher diameter. Disturbances and bad environment near forest edges may increase the mortality of trees, especially smaller trees. Therefore, the community characteristics of fragments were affected by edge effect.
Point 36: This whole paragraph need extra work that connects with the results of this work.
Response 36: We have deleted this paragraph because of we did not study the abiotic factors in these forest fragments. In the future, we could study these factors affecting community characteristics in these fragments.
Reviewer 2 Report
Dear Authors,
Please find comments in the attached file
Comments for author File: Comments.pdf
Author Response
Point 1: This belongs to results or discussion part
Response 1: We have moved this part to the discussion part. Please see the Page 8, Lines 325-326.
Point 2: What was the age, soil, density, tree species
Response 2: We have added this information in the manuscript. Please see the Page 3, Lines 101-1-3, Page 3, Lines 116-117, Page 3, Lines 125-127.
Point 3: The size of the forests viaries a lot. I would expect differenct fragmentation in 3 ha forest compared to 40 ha
Response 3: We have added some information in the manuscript. The differences of community characteristics due to fragmentation were shown in Figure 3.
Point 4: Why not to present formula coefficients here?
Response 4: Actually, we put the formula coefficients in the appendix, and upload when submitting this manuscript. We thought that the formula coefficients are developed by other researcher, so we just cited this in the main text.
Point 5: But you did not used different sizes for other 69 forests.
Response 5: In the present study, patch sizes of 69 forest fragments is the area gradient. We compared AGB of these 69 different-patch-sized fragments with different sized quadrat within two non-fragmented forest. We used the different sizes for these fragments.
Point 6: It is not alowed to present the results like conclusions, write normal paragraph please
Response 6: We have revised this part into normal paragraph. Please see the results parts from Page 5, Line 186.
Point 7: In which forests? 69, or 2 forest reserves
Response 7: We have revised our expression in this part. Please see the Page 5, Line 189.
Point 8: Figures are hard to understand to small letters. Also quality of figures is too low
Response 8: We have replaced the figure with higher quality figures. Please see the Page, Line. All X axis in the figure are the patch size of 69 forest fragments. The unit of AGB is Mg. Abundance in this figure represent number of stems in the quadrat/patch, actually is different with density. The black dots are the observed data in 69 forest fragments with different patch sizes. The grey dots are the simulated sampling with different quadrat sizes with two non-fragmented forests. The lines are the mean of grey dots within same sampled quadrats size.
Round 2
Reviewer 1 Report
Dear authors,
I have revised the second version of this manuscript and found it is in a much better shape. The topic of forest fragmentation is relevant for the readers of this journal and beyond. However, I still think there are some issues mostly linked to the writing style that in my opinion is over using short sentences one after the other lacking a good connection within each paragraph. In some cases, these short sentences still repeat a few topics (e.g. tree mortality increasing in edges is said several times along the manuscript). The figures in this new version are still very low in resolution and is difficult to appreciate the results but I am hoping authors can work with the journal to fix this. Being a non-native english speaker myself I try to avoid making too much emphasis on language ion my reviews but as you will see in the attached pdf there are some issues that require attention from the authors regarding language and writing style. I hope the authors find these suggestions useful. Regards
Comments for author File: Comments.pdf
Author Response
Point 8: our results estimated AGB in species-rich,,, in south China
Response 8: We have revised this part.
Point 9: by an increased tree mortality
Response 9: We have deleted this part due to already in the text.
Point 10: undesirable consequences for long-term carbon storage
Response 10: We have added the consequences of the AGB storage in large trees in the future under the background of climatic change.
Point 11: fragmented forests
Response 11: We have revised this sentence in the Conclusions part.
Point 12: enhance
Response 12: we have change the enhance instead of contribute to better evaluating the role of ,,, in the Conclusion part. Thanks for your suggestions.
Reviewer 2 Report
Thank you for taking my comments into account
Author Response
Thanks for your suggestions.
