Dynamic Top Height Growth Models for Eight Native Tree Species in a Cool-Temperate Region in Northeast China
Round 1
Reviewer 1 Report
Dear Authors,
I really appreciate the effort you have done to improve your work. Thank you very much to take my comments in consideration.
Author Response
Dear Reviewer 1,
thank you very much for reviewing our manuscript and for your positive feedback.
Best regards
Reviewer 2 Report
Dear authors,
You have improved the article by making changes. Chapter 2, Materials and Methods is now better described and clearer for the reading public. Also, by changing some aspects related to the applied methods, you presented the new results in a clear way. I appreciate your effort.
I read the new version of the manuscript carefully and I identified some small editing errors, presented below.
Line 15: Please, delete a dot after "relationship".
Lines 27-28: I would add the unit (m) for AAB and RMSE values. They refer to meter.
Line 156: "Table 1" should be text using bold option and you should delete a space between "4," and "26".
Line 198: I would replace "30" with "27-30" or "27 to 30" as it results from Table 2.
Line 329: A dot is need to be added after "stabilizes".
Line 330: Is the text a new paragraph?
Lines 359-364: The title of Table 3 is not before the table.
Line 367: "3. Results" should be left-aligned text, bold text, no indentation, on a single line.
Line 377: Please, delete a space between "years" and "with".
Line 384: Change "Fig. 2" with "Figure 2".
Line 386: "Table 3" is, in fact, "Table 4".
Lines 389-391: I would add the unit "m" after AAB and RMSE values.
Line 394: "Table 3" is "Table 4", please use bold option.
Line 395: You present the unit for RMSE but not for AAB.
Figure 2: A space is need to split the name of variables from unit on the horizontal and vertical axis titles in each chart.
Line 412: The title should be left-aligned text, no indentation, using Italic option.
Lines 431-432: A comma should be added after "conditions" and after "point)".
Lines 435-437: The paragraph should be edited as normal text, not as subtitle (Tab, no Italic, justify). Please, split the subtitle 4.2 from the paragraph!
Line 453: Please, add a space before "The".
Line 461: Please, add a space after "trees".
Line 488: Please, delete a space between "0.99" and "for".
Lines 563 and 602: These lines are empty.
Author Response
Dear Reviewer 2,
thank you very much for reviewing our manuscript. We have taken into account all of your comments.
Best regards
Reviewer 3 Report
General comments
In the current version of the manuscript, authors have addressed my most important comments in my previous reviews, such as “It seems that you have addressed only a few concerns, and because of this, your manuscript has been improved to some extent. My other concerns were not addressed, such as either applying mixed-effects modeling or autoregressive error-structured modeling in order to reduce the bias due to temporal autocorrelations, as these problems exist in long-term time series data. Applying the nonlinear least square regression, which is also known as a nonlinear OLS regression, does not solve these autocorrelations problems without inclusion of autoregressive error-structures into the site index models. Without applying one of them, resulting models could produce substantial bias. So please consider addressing this issue in the next revision. Also without applying ADA or GADA, base-age specific approach, which is conventional method of site index modeling, does not attract the readerships and does not provide the advancement of new knowledge in the field”. Thus, this manuscript is substantially improved. I do not have major concerns on this version, but have only few minor ones, which should be addressed before acceptance.
Minor issues
Line 19-20: Do these numbers also include measurements of annual shoots? Please make this clear.
Line 21-24: You derived dynamic equations from base equation with GADA, and fitted such dynamic equations to your height-age data using nested regression. This is the main message here. Please formulate the text accordingly.
Line 25-26: Please use the same terms consistently, equation or model, use one consistently throughout the manuscript.
Line 26-29: These selected models or this selected model. Please present the results of the best model only. Delete acronyms, as they are not in use again in the abstract.
Line 29-31: I do not agree with this formulation. What do you mean by homogenous? Distribution of residuals should band-like across the zero line; however, figure 3 does not show this for some tree species. Thus, please formulate the text carefully; using term “very good” does not suit for all species, but for few.
Line 35: other relevant stand attributes…. using the term “parameter” here would be confusing to model parameters.
Line 45: ….dominant trees
Line 63: Please mention other aims also, and what do you mean by it here? Site index model or growth and yield model? Specify.
Line 67-87: It is not necessary to describe the historical background of site index modeling. There exist several studies on this. Please shorten the text, but increase the number of citations, mostly the site index models or dominant height-age models developed in recent years.
Line 88-102: Not today, but in recent years, ADA and most preferably GADA has been used to develop site index models (or dominant height growth models), as this approach is suited to both the short-time series data with no common base age of the height-age series (NFI, PSP data) and long-time series data with common base-age series (SA data). Please cite some examples of ADA or GADA models based on each of these datasets also.
Line 117-118: …..including site index models…..
Line 120-121: Please specify the size of individuals for which you measured the shoot height, for example individuals of juvenile stage
Line 193-205: Did you consider the assumption 100 largest trees per ha should be the main basis for selecting the dominant trees for measurement or SA? Please mention this assumption also, as this is international practice, which you need to follow in your sampling design.
Line 219-221: Do you mean suppressed trees? Dominant trees in the past may not be dominant in the present days, due to influence of disturbance factors.
Table 2: Number of stands in second column. Define SD.
Line 258-262: I think it is also necessary to add 0.5 year as correction factor to age according to Carmen’s rule. Maybe you have done this, but need mentioning here.
Line 303-309: Did you try for other alternatives of AR (x), such as second order, third order, etc.? Because, problems of correlations of stem analysis data may not be properly resolved by AR(1). Why do you choose AR (x), why not other alternatives, such as ARMA, CS, etc.. Please give justification.
Line 322: Please replace reference (40) with better reference, such as that published in Forest Ecology and Management (https://doi.org/10.1016/j.foreco.2011.07.037)
Line 336: check here!
Table 3: It seems perfect model. Do you know why R2 is so high? Because, each of the growth series is adequately described by GADA- or ADA-based model. Thus, GADA is standard and highly precise method of height-age modeling or other stand attributes modeling.
Line 397-400: I was suggesting you to connect the points of observations by line of the residuals of each tree in my previous review, but you did not follow, why? Connection shows clearly whether there would be real trends or not. Please do show the residual series in this figure. As mentioned earlier, I am not fully agreed with your statement homogenous residuals…… as some tree species do not have this. For P. aspecrate, there is largely under-estimation. If you want to keep this figure as it is (dotted plots), please overlay the loess curve for showing data trends or calculate mean residuals by age class of 10 cm interval and overlay such mean using line on the residuals dots.
Line 435-438: why text is in italics?
Author Response
Dear Reviewer 3,
thank you very much for reviewing our manuscript. Regarding your suggestions and comments, we would like to propose the following answers:
Line 19-20 (now 22-23): Do these numbers also include measurements of annual shoots? Please make this clear.
Answer: We have clarified it.
Line 21-24 (now 24-27): You derived dynamic equations from base equation with GADA, and fitted such dynamic equations to your height-age data using nested regression. This is the main message here. Please formulate the text accordingly.
Answer: We have rephrased the sentence.
Line 25-26 (now 28-29): Please use the same terms consistently, equation or model, use one consistently throughout the manuscript.
Answer: We have used the term model throughout the manuscript
Line 26-29 (now 29-32): These selected models or this selected model. Please present the results of the best model only. Delete acronyms, as they are not in use again in the abstract.
Answer: We presented only the results of the best model
Line 29-31 (now 33-35): I do not agree with this formulation. What do you mean by homogenous? Distribution of residuals should band-like across the zero line; however, figure 3 does not show this for some tree species. Thus, please formulate the text carefully; using term “very good” does not suit for all species, but for few.
Answer: We rephrased the sentence.
Line 35 (now 41): other relevant stand attributes…. using the term “parameter” here would be confusing to model parameters.
Answer: We changed „parameter“ into „attributes“
Line 45 (now 51): ….dominant trees
Answer: We have added „dominant“
Line 63 (now 70-71): Please mention other aims also, and what do you mean by it here? Site index model or growth and yield model? Specify.
Answer: We have added this information in the text.
Line 67-87 (now 76-96): It is not necessary to describe the historical background of site index modeling. There exist several studies on this. Please shorten the text, but increase the number of citations, mostly the site index models or dominant height-age models developed in recent years.
Answer: We have shorten the text and increased the cites.
Line 88-102 (now 109-126): Not today, but in recent years, ADA and most preferably GADA has been used to develop site index models (or dominant height growth models), as this approach is suited to both the short-time series data with no common base age of the height-age series (NFI, PSP data) and long-time series data with common base-age series (SA data). Please cite some examples of ADA or GADA models based on each of these datasets also.
Answer: We followed your suggestions.
Line 117-118 (now 142-143): …..including site index models…..
Answer: We have included „site index models“
Line 120-121 (now 147-149): Please specify the size of individuals for which you measured the shoot height, for example individuals of juvenile stage
Answer: The annual shoot length has been measured at all analyzed trees and has been controlled by counting tree-rings of discs taken at specified stem heights.
Line 193-205 (now 223-238): Did you consider the assumption 100 largest trees per ha should be the main basis for selecting the dominant trees for measurement or SA? Please mention this assumption also, as this is international practice, which you need to follow in your sampling design.
Answer: As the stand height curve of dominant trees in the analyzed, old even-aged stands is rather flat the height of dominant trees can be used as measure for the 100 largest trees per ha.
Line 219-221 (now 226-227): Do you mean suppressed trees? Dominant trees in the past may not be dominant in the present days, due to influence of disturbance factors.
Answer: We mean suppressed and damaged trees, where tree-rings could not be clearly identified
Table 2: Number of stands in second column. Define SD.
Answer: SD has been calculated separately for each stand
Line 258-262 (now 292-296): I think it is also necessary to add 0.5 year as correction factor to age according to Carmen’s rule. Maybe you have done this, but need mentioning here.
Answer: We have not done this correction, because such a correction would not have a relevant effect on the results.
Line 303-309 (now 339-349): Did you try for other alternatives of AR (x), such as second order, third order, etc.? Because, problems of correlations of stem analysis data may not be properly resolved by AR(1). Why do you choose AR (x), why not other alternatives, such as ARMA, CS, etc.. Please give justification.
Answer: In fact, we discussed the autocorrelation problem with a mathematician. A look at the residuals showed, that the residuals increased systematically with higher ages. So, there were two possibilities: (1) adjust the nonlinear regression to account for the increase with age to get normally distributed residuals or (2) improve the results with the help of a first-, second-, third- … order correlation. Since AR(x) is a widely used method in forest science, we applied it and were able to correct almost all of the autocorrelations using AR(1).
Only for a small number of trees the autocorrelation could not be corrected completely with AR(1). In this case, however, a correction with AR(2) or with the help of an additional linear regression cannot be justified mathematically correct, because only few individual trees are affected and there is no systematics behind it.
Line 322 (now 357): Please replace reference (40) with better reference, such as that published in Forest Ecology and Management (https://doi.org/10.1016/j.foreco.2011.07.037)
Answer: We have replaced the reference
Line 336 (now 371): check here!
Answer: We have adjusted the numbers.
Table 3 (now Table 4): It seems perfect model. Do you know why R2 is so high? Because, each of the growth series is adequately described by GADA- or ADA-based model. Thus, GADA is standard and highly precise method of height-age modeling or other stand attributes modeling.
Answer: May be the use of AR(1) significantly increased the R2 values.
Line 397-400 (now 448): I was suggesting you to connect the points of observations by line of the residuals of each tree in my previous review, but you did not follow, why? Connection shows clearly whether there would be real trends or not. Please do show the residual series in this figure. As mentioned earlier, I am not fully agreed with your statement homogenous residuals…… as some tree species do not have this. For P. aspecrate, there is largely under-estimation. If you want to keep this figure as it is (dotted plots), please overlay the loess curve for showing data trends or calculate mean residuals by age class of 10 cm interval and overlay such mean using line on the residuals dots.
Answer: We read your previous review and couldn't find any clue there. Of course, the points of the observations can also be connected by the line of the residuals of each tree. We have followed your suggestion and used lines for the trees instead of points
Line 435-438 (now 477-479): why text is in italics?
Answer: This is a formatting error and was changed to normal text
Best regards
This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.
Round 1
Reviewer 1 Report
GENERAL COMMENTS
The aim of this study was to develop the first site index curves for 8 major native species in Northeast China. The proposed tools are very useful to achieve sustainable forest management in China and it would be very convenient if they were improved in near future. I found the manuscript to be overall well written and much of it well described. I enjoy reading and I would like to congratulate the authors for its work, but it needs to be improved.
Minor questions or clarifications:
- Please, review links in references 9 and 15, links are broken.
- Line 137-140 “Nevertheless, environmental conditions in the surrounding forest, e.g. the wind-sheltering effect of adjacent stands, and practical aspects of forest operations, like soil compaction, may influence the height growth potential of the dominant trees”
Comment: The authors assume the transcendence of the environmental conditions to define properly the potential productivity but none of them has been included in the previous classification, being especially interesting soils properties. I guess there is no edaphic information in the forest inventory, was it not possible to have done soil analysis to complete the characterization?,
Major questions:
- Line 59-60 “The site index describes the site productivity and is defined as the top height of a stand at a reference age of usually 50 or 100 years”.
Comment: Reference age is directly related to rotation age (growth speed) of the species to cover as much variation as possible, which allows a good classification of the different forest sites detected. This means an age close to half rotation age or longer. Considering 50 or 100 years as common values reference ages is very ambiguous and exclude a huge list of species, including the fast-growing tree species. This would seem too strong statement taking into account the tendency of the forest management in this area. Please, update this information and justify with the appropriate bibliography.
- Line 78-81 “Since no site maps were available at the time of stem disc collection, a provisional relative classification of the site had to be done based on tree growth data of the most recent forest inventory, taking into account topographical conditions like elevation a.s.l. and exposition”
Line 122-124 “Assuming height growth differences on slopy terrain, a sample of three trees at top, middle and lower slope was selected in each investigated stand. In plain areas three samples of three trees per stand were taken.”
Comment: In order to define the provisional relative classification I don´t understand why the authors don´t take into account for the first moment all physical conditions (elevation, exposition and slope) to characterize from this point of view the forest stands. It would be appropriate to indicate (e.g. table format) the provisional assignment of productivity to different forest stands. I´m sure there is an explanation.
- Line 45-46 “In recent decades forest establishment was emphasized, leading to extensive young forests, while old forests are rare”
Line 117-119: “In order to exclude inter-specific competition, even-aged monocultures and well stocked stands with relatively low-intensity management regimes were preferably selected”
Comment: I would have appreciated a brief ecological description of each species, even some sylvicultural aspects as expected rotation. There are some notes in the text about forest management, but it´s difficult to conclude how the situation is today and the next future for these tree species and this generates confusion.
- Line 190-191 “The reference age of 30 years was chosen to provide the possibility of site assessment for even rather young stands”
Comment: Is there any reference to support this decision?. Different species with different growth speed, very heterogeneous in old ages representation, are being evaluated with the same reference age. Why didn´t you propose a reference age adapted to the theoretical rotation of each specie?
I recommend the authors to include the best reference age to which site index must be referenced. You can use, for example, the method used by Diéguez-Aranda U, Burkhart HE and Amateis RL, 2006. Dynamic site model for loblolly pine (Pinus taeda L.) plantations in the United States. Forest Science 52, 262-272. However, it must be considered that practical use requires reference age to be defined. Moreover, site index is key not only for growth and yield purposes but also as an independent variable for any ecological study. Therefore, I recommend its determination for each site index curves.
- Figure 2. Productivity of forest sites classified by stand heigh (m) at given stand age (years) represented on a site index scale (24 m, 20 m, 16 m,…4 m) for each tree species
Comment: Could you explain how is it possible to model site index curves for 4 and/or 24 m with absence of data?. Why do you project the age for site index curves in several cases 40, 50, even 100 years (P. sylvestris) from the sample data?
- Line 294-295 “However, the rotation age of the respective tree species in China is between 15 and 40 years and therefore currently rather short [37].”
Comment: Taking into account the size of the country and the variations of the ecological conditions, could we assume the generalization include in reference [37] for the tree species of the study (broad leaf species included):
Pp 13. In China, the balance of “natural forests” (currently 66 percent) and plantation forests is shifting gradually in favour of even-aged pure stands, managed mostly with short rotations of between 15 to 40 years
Pp 92. Plantation forests are commonly subjected to age-class management with short rotation periods (ranging from 10 to 30 years), followed by clear-felling, removal and burning of log-ging debris, and replanting with a limited number of economically significant tree species (e.g. Cunninghamia lanceolata, Pinus massoniana, Eucalyptus spec., Casuarina equisetifolia, Acacia mangium etc.).
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
Dear authors,
Your article is interesting by results and also by methods used. Your work has materialized in site index curves of eight tree species from Northest China that are a novelty for this region. Considering that long-term observations from permanent plots are required to produce site index curves, it was an excellent idea to use stem analysis data in order to reconstruct the relationship between height and age. This way, you have managed to reduce the limitations of the fact that long-permanent observations are not common in China.
Some insignificant editing errors can be easily eliminated.
Lines 11-12: Species variety for Larix gmelinii and Pinus sylvestris are not written using Italic fonts as elsewhere in the text.
Line 137: Please, add a space after "site productivity".
Line 166: Please, change "disks" with "discs".
Line 199 and other places: I would choose R2 not R2.
Line 213: Please, add a space between "11" and "m".
Line 247: There is too much space between the figure number and the title.
Lines 309-312: Please, use Italic fonts to write tree species.
Line 325: Please, add a space between "Carr." and "and".
Line 342: Please, add a dot after "Mr".
S.M. Lines 22 and 31: Please, delete a space before "specific".
S.M. Table S1: I'm not sure about "Betulus". Do you refer to "Betula pedula"?
S.M. Line 64: Please, add a space between "Carr." and "and".
S.M. The way of citing is different than one used in the article.
Author Response
Dear Reviewer 2,
thank you very much for reviewing our manuscript. We've got rid of any editing errors you suggested.
Best regards
Reviewer 3 Report
Summary of the study
Authors developed the model describing top height-age relationship (site index model) using both the stem analysis data and shoot length measurement data acquired from eight different tree species in Northeast China. A single growth function (Sloboda function) was used to fit the height-age data using the ordinary least square (OLS) regression. Fitting performance of the model for each species seemed adequate with 93%-98% variations of the height-age relationships described. Model fitting performance for different species is different.
General comments
Authors have written the manuscript generally well regardless of the modeling methods employed, so I enjoyed reading the manuscript. Modeling top height (dominant height) started some decades ago with the use of the traditional approaches, such as guide curve method and base-age specific approach (current study also adapted this), and continued with the much more advanced approach, such as base-age invariant approach (also known as dynamic approach). In recent years, dynamic modeling approaches, such as algebraic difference approach (ADA), more preferably the generalized ADA (GADA) has been used for developing more accurate top height growth models compared to those developed using the traditional approach. The traditional approaches are outdated, as the dynamic approach applied with the mixed-effects modeling is mostly in practice in modeling studies of top height growth (site index modeling) in recent years. It should be also noted that the concept of ADA or GADA is not a new also, as this approach started in 1974, but frequently after 2000. In recent years, ADA or GADA is applied together with the mixed-effects modeling, which can substantially improve the fitting precision. In contrast to this, the traditional approach used with OLS regression (current study also adopted this), which is not appropriate to model the hierarchical data with the observations dependence and significant correlations For this data, ADA and GADA is widely applied to develop not only top height growth models, but also stand basal area growth models, individual tree radial growth models, and stand biomass models with the application of nested regression, mixed effects modeling or autoregressive error-structure modeling. Thus, the current study, which is based solely on the traditional approach, more specifically the base-age specific approach, in general, does not add the naïve knowledge to the forest biometrics. Thus, potential audience of the journal will not be benefitted from the current version of the manuscript that will be published in any journal. Because, there are a bundle of literature available containing the modeling site index with traditional methods. I picked up some major issues, as below.
- Base function: Authors picked only one function, the Sloboda function to fit the top height-age relationship. However, they did not provide the convincing reason to choose such a function. Other several versatile growth functions are also available, such as Chapman-Richard function, Korf function, Hosfeld function, and Weibull function, which need to be evaluated also. The data originated from different site conditions may be fitted differently to the same growth function, and thus, it is necessary to pick some more growth functions of the different forms (fractional and exponential forms) and evaluate their fitting performance, as they may exhibit the best fitting improvement and predicting performances in many other top height growth modeling studies. Authors did not consider deriving ADA or GADA from their function (see below for detail), but fitted the base function with OLS regression, which is not considered appropriate and robust modeling. Choosing only one function can be the unfair decision and inappropriate evaluation.
- GADA model formulation: The base functions mentioned above may be used to derive ADA or GADA models considering one or more parameters of the base functions as site-specific parameters. While making a proper derivation of ADA or GADA from each of those versatile functions, modelers may follow either the step-by-step process (4 steps) using their real data (see Cieszewski and Bailey, 2000; Cieszewski 2001, 2003; Nord-Larsen, 2006) or assume any of the one or more parameters of the base function as site-specific parameters and make a derivations of ADA or GADA based on such assumption. The ADA always assumes only one parameter of a base function to be site-specific parameter. This is also not appropriate way of making the model remarkably highly robust and flexible. Alternatively, authors need to consider the GADA-based models which can describe much larger variations of top height-age relationship, and they are more robust and flexible to produce the polymorphic curves with variable asymptotes, which are the basic properties of height growth. The expansion of any growth functions (base functions) has to be done according to the growth theory, which only GADA follows, i.e., GADA includes any two of the parameters (asymptotic, rate, and shape parameter) of the base function as site-specific parameters, and therefore it should have higher fitting performance than ADA-based model.
- Modelling approach: As mentioned earlier, authors have applied OLS regression modeling, which is not appropriate for data that are nested temporally and spatially, i.e., multiple observations within the same tree, multiple trees within the same plot, and multiple plots within stand, etc. The estimated parameters of the models are substantially biased because of the observations dependency and correlations in the hierarchically-structured data. Thus, it is necessary to deal with this problem by applying either the auto-regressive error structure modeling or mixed-effects modeling. However, authors did consider applying none of them.
- Model estimation: authors did not mention about what data structures were used and how model parameters were estimated in their Sloboda model. After formulating ADA or GADA equations, which contains height (h) and age (t) as variables [e.g. h1 = f (t0, t1, h0)], the same variable (height: h1 or h0) appears in both the left-hand side and the right-hand side of the equations. Therefore, the assumption in the fitting procedure that the response variable includes errors (sampling and measurement errors), while the independent variable should be assumed to be measured without errors. It is because that there would be error-in-variable problem, meaning that height appearing in the right-hand side of the equation are subject to errors. This leads to the significantly highly biased-parameter estimates (e.g. Krumland and Eng, 2005; Cieszewski and Strub, 2007). The authors should therefore need to apply one of the base-age invariant (BAI) approaches while estimating parameters of the ADA or GADA models: iterative nested regression (INR) (e.g., Krumland and Eng, 2005; Cieszewski et al. 2000; Cieszewski et al. 2007), dummy variable (DV) (Nord-Larsen, 2006), integrative evaluation (IE) approaches, to estimate site-specific parameters simultaneously with global parameters. However, among these, INR is most suitable and computationally much efficient in the case for a data set of any size. The BAI approach is theoretically and practically desirable as there is no violation of regression assumptions on error-free independent variables, which are estimated as site-specific parameters, and therefore it is more likely to identify the true height growth trends. Site-specific parameters are equivalent to site index (h0, t0), and estimation technique is unaffected by any arbitrary choice of the base-age. The INR is more efficient and stable than other two, especially in the computation processes, and therefore it is mostly preferred in dominant height growth and site index modelling works. Authors, therefore, are suggested to consider analyzing data by applying INR in their revision works. If fitted the ADA or most preferably GADA model with INR with autoregressive error structure modeling, the problems appeared in Figure 3 (as a result of inappropriate function chosen and fitting method employed) can be substantially reduced.
References
Cieszewski, C. J., Harrison, M., & Martin, S. W. (2000). Practical methods for estimating non-biased parameters in self-referencing growth and yield models. Daniel B. Warnell School of Forest Resources, University of Georgia.
Cieszewski C.J., Strub M. 2007. Parameter estimation of base-age invariant site index models: Which data structure to use? - A discussion. Forest Science, 53, 552-555. <Go to ISI>://000250131500002
Krumland B, Eng H (2005) Site index systems for major young-growth forest and woodland species in northern California. California Department of Forest and Fire Protection, Sacramento, CA. Forestry Report 4: 1-219.
Nord-Larsen, T., 2006. Developing dynamic site index curves for European beech (Fagus sylvatica L.) in Denmark. Forest Science 52, 173-181. <Go to ISI>://000236943800007
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
First of all, I appreciate the efforts of the authors for the responses to the comments of the first review. I understand the difficulties that the study area presents due to the historical absence of forest tradition in China, so the tools provided in its study are a good step to advance towards sustainable forest management.
Taking all this into account, I consider it necessary the following recommendations for improvement in the text.
Line 59-62: “The site index describes the site productivity and is defined as the top height of a stand at a reference age of usually 50 or 100 years”.
Comment Review 1: Reference age is directly related to rotation age (growth speed) of the species to cover as much variation as possible, which allows a good classification of the different forest sites detected. This means an age close to half rotation age or longer. Considering 50 or 100 years as common values reference ages is very ambiguous and exclude a huge list of species, including the fast-growing tree species. This would seem too strong statement taking into account the tendency of the forest management in this area. Please, update this information and justify with the appropriate bibliography.
Answer: We agree that the reference age may reflect to some extent the rotation age. In Germany for instance, we use commonly a reference age of 100 years (see for example Assmann Franz 1962 “Vorläufige Fichtenertragstafel für Bayern”). In Swizerland Badoux in 1966-1969 developed yield tables for Norway spruce, Silver fir, Larix and European beech using a reference age of 50 years. The choice of the reference age is mainly a question of practicality. In the investigatied regions in China very few trees are older than 50 years. Therefore, we have chosen 30 years as reference age.
Comment Review 2: There is no doubt that the lack of data is an inconvenience to carry out a study with the rigor with which it is carried out in countries where sustainable forest management has been a reality for more than 150 years. For that reason, I consider it a success.
This is the case in Germany; however, it cannot be an exclusive reference to illustrate the concept of site productivity worldwide. As the text is written, it seems that the site index at a reference age of 50 or 100 years is an absolute truth throughout the world, disregarding the global diversity in forestry matters.
It is for this reason, and because of the lack of field data that the study area presents, that I think it is essential to enrich the introduction, reviewing the scientific literature worldwide on the standard reference ages to describe the site index.
Well focused, it can be a good argument in the methodology and subsequent discussion, to justify the choice of a single reference age (30 years) for all the species studied.
Reviewer 3 Report
Thanks for attempting to answer my concerns raised in the first review. It seems that you have addressed only a few concerns, and because of this, your manuscript has been improved to some extent. My other concerns were not addressed, such as either applying mixed-effects modeling or autoregressive error-structured modeling in order to reduce the bias due to temporal autocorrelations, as this problems exist in long-term time series data. Applying the nonlinear least square regression, which is also known as a nonlinear OLS regression, does not solve these autocorrelations problems without inclusion of autoregressive error-structures into the site index models. Without applying one of them, resulting models could produce substantial bias. So please consider addressing this issue in the next revision. Also without applying ADA or GADA, base-age specific approach, which is conventional method of site index modeling, does not attract the readerships and does not provide the advancement of new knowledge in the field.
