Insights into the BRT (Boosted Regression Trees) Method in the Study of the Climate-Growth Relationship of Masson Pine in Subtropical China

Round  1

Reviewer 1 Report

The manuscript “Insights into the growth-climate relationship of Masson pine in subtropical China” deals with a recurrent topic in dendroclimatology: the non-linearity of the climate-growth relationships. Within these complex interactions it is very important to provide efficient tools to estimate and possibly predict growth responses to climate variability. The intent of the authors to search for more flexible models is therefore appreciated. However, the manuscript presents numerous and significant problems concerning structure, contents and language. I’d suggest a resubmission after a complete reorganization of the entire structure including an extended language revision by an English editorial expert.

The manuscript is poorly organized and data presented often in a confusing way (the different sections are not suitably filled with the appropriate text), Several information are lacking considering that this is strictly a dendroclimatological analysis and all the different appropriate steps should be fully described. 

The aims of the study are not clearly stated and therefore conclusions appear rather vague. 

Results analysis and discussion are often overlapping creating confusion to the reader (figures should all be discussed within the results section and not in the discussion that should explain the outcomes and compare results obtained with the existing literature.

The data analysis tends to be overwhelming since it covers in excessive detail also climate variables with scarce effects on tree growth. 

The copious literature about nonlinear responses has not been suitably taken into consideration. The model used (not very common in dendroclimatology) is proposed as a suitable method but the outputs could have been better appreciated if compare with more common methods.

It would have been useful (especially for not Chinese readers) to provide a short profile of the ecological requirements of Pinus massoniana also considering its wide distribution area and its altitudinal plasticity. It is also quite surprising that the species is mainly sensitive to the same climatic variables (Sept-Nov precipitation and Feb-May temp.) regardless of the geographic and environmental diversity of its natural or man-induced range.

Two basic basic remarks: 

the common name is in regular character and the latin name in italic and not viceversa: masson pine (Pinus massoniana). Please change it throughout the entire text.

Table and figure legends should be self-explaining and provide all information necessary to its understanding avoiding to refer to the main text.

Specific comments

The title does not clearly specify the main topic covered in the manuscript, therefore it should be modified and include reference to the application of non-linear methods for detecting climate-sensitivity of the species. The same problem applies to the choice of the keywords.

Introduction:

provide reference to avoid speculative statement (e.g. line 39, 74-75, 77);

state clearly the specific aims of the study by formulating hypotheses to be tested

Material and methods

Keep distinct and clear sampling areas (provide short climatic description for the 5 study the areas since the figure 2 is too general) from sampling procedures and from analysis methods. Fig. 1 should be improved.

Tree-ring preparation and measurement procedures as well as chronologies constructions are very poorly described and should be consistent with the broad existing literature

Table 2 is very confusing and shows that autocorrelation at some sites is still very high (as clearly visible in fig. 3) indicating an inadequacy of the standardization methods. Please provide also main chronologies features.

Results

Provide an appropriate Results chapter avoiding what happened in chapter 4 where the first part about the BRT should go to Materials and Methods whereas some other parts are more fitted in a Results chapter.

The different results performed by JLN and TBS (fig.5) are not specifically explained from an ecological perspective (e.g. are artificial stands? at lower altitude at different managemenr regimes?) and should be treated in the Discussion section.

Discussion 

Very long and often including parts more appropriate to Results (e.g. comments about fig. 8 and 9). 

Conclusions

Are vague and should refer to the study aims eventually achieved with the analysis.

References

Literature about the use of non-linear methods in dendroclimatology are scarce and should be implemented.

More specific comments on language aspects are useless now and could be provided only after a complete text reorganization prior to resubmission.

Author Response

Dear reviewer, thank you give me the chance to major revision. And now, I response to your comments. In my new manuscript, based on your suggestion, I have adjust the contents. The detail response comments are upload to the system. Thank you for taking the time to review my paper.

Author Response File: Author Response.doc

Reviewer 2 Report

The paper addresses the fast developing field of dendrochronology -  assessment of climatic signals in tree-ring records of tropical tree species.  The authors assembled a good size data set of Masson pine ring widths from subtropical China (13 sites). The problem is that the study has been presented inadequately.

Some critical aspects of tree-ring chronology development (cross dating validation and site chronology statistics) and site characteristics have been omitted which impedes the evaluation of results.

Another serious problem is that even linguistically the English is ok, many parts of the manuscript are not comprehensive because of incorrectly used terminology or misunderstanding the tree-ring methods.  As an example see lines 37-38 and 141 with very odd or false statements that have been attributed to other authors, who have never made these statements neither have reached these conclusions.

Unfortunately, the manuscript demonstrates that authors’ misunderstanding the main principals of dendrochronology and time series statistics.

I would not comment either believe to the presented results on the nonlinear relationship between climate change and tree rings  because the following flaws

1) the cross-dating of their tropical tree-ring chronologies (inter serial correlation of tree-ring series within a site) and the chronology statistics are unknown, and 2) the differences in growth conditions between the sites and areas are unknown.

Another concerns:

The paper organization and structure is not solid. Methods includes some results (Line 128-133 ) and the results include methods (175-195). Conclusions and Abstract are not matching. 

It appears that quality control of data that are routinely applied prior any modeling are unnecessary included the results. For example, instead of just mentioning the detrending technic applied to tree-ring width series and a statement on the quality of developed tree-ring width index chronology, the authors went to a long critic of various detrending approaches in dendrochronology then concluded with this statement:  Line 154-160 “Because NEXP method is not suitable for removing non-climatic variance from tree ring chronology, we only consider grid climate data and SP chronology data(Fig. 3). In addition, the expressed population signal (EPS), a valuable metric of only indicates the variability of the population signal

 strength was introduced (Briffa and Jones 1990; Buras 2017).

”   Unfortunately this does tell us nothing about the EPS or suitability of developed chronologies for this type of analysis which is necessary to provide.

All collected sites show a different response (sometimes opposite from one site to another) to the temperature and precipitation parameters.  Since there is no clear statement the tree growth conditions the interpretation of these result is doubtful.

Tropical dendrochronology is a very young and challenging field. Exploration of climatic signals in tropical tree-ring variance must be persuaded with extreme cautious following established principals and protocols.

Author Response

Dear reviewer, thank you give me the chance to major revision. And now, I response to your comments. In my new manuscript, based on your suggestion, I have adjust the contents. The detail response comments are upload to the system. Thank you for taking the time to review my paper.

Author Response File: Author Response.doc

Reviewer 3 Report

This is my review for the paper "Insights into the growth-climate relationship of Masson pine in subtropical China" by Hongliand Gu and others.

It should be positively evaluated that the paper deals with the nonlinear response of subtropical pines to climate. However, the manuscript must be drastically improved. See the comments below.

* Reviewers are not grammar checker. The English must be revised throughout the manuscript before submission. There are too many grammatically incorrect sentences and improper wordings.  There are quite many improper and incomprehensible expressions for scientific writing.

* With respect to the standardization of ring-width series, it is now a common sense that the use of negative exponential curve is not "conservative" any more. I recommend the authors once to draw age-aligned curves. I also recommend to have several trials of the climate-growth analysis for different cut of length of spline.

* Since the analytical method, i.e. BRT, is relatively new in dendroclimatology, the authors must more plainly and clearly explain the meaning and merit in using the measures, relative importance , partial dependence, etc. The explanation at L. 165-173 is not enough, and rhetorical expresion like "elbow" should be avoided.

* I do not see any reason to use the irrelevant data set in Section 4.2. What is the necessity to present the Friedman.1 data set with the equation?

* I suggest to present the results using the conventional PCA analysis. I also suggest the authors to show scatter plots of the pronounced relationships between ring-width indices and climate variables, e.g. prior Sep-Nov precip and TBS and so on. Nonlinear response of plants to environmental factors is just a common sense, but it is very informative to readers of the journal if forms of non-linearity are shown.

* One small but important question: What do the authors mean by "dark art" for detrending? As far as I searched the OCR-ready pdf of Hal Frit's book, there are no such words. I request to show the part of the book. Even if there are the words, the expression is quite ill-advised and must be changed to another expression.

Author Response

Dear reviewer, thank you give me the chance to major revision. And now, I response to your comments. In my new manuscript, based on your suggestion, I have adjust the contents. The detail response comments are upload to the system. Thank you for taking the time to review my paper.

Author Response File: Author Response.doc

Round  2

Reviewer 1 Report

The authors have considerably improved the manuscript accepting most of the suggested comments and reorganising and optimizing the structure. It only needs a minor but general language revision by a professional editor.

Author Response

Dear reviewer, thank you give me the chance of minor revision on my manuscript. And now, I will report my revision work to you. Thank you for criticizing and correcting my manuscript in your busy schedule. Best sincerely.

Author Response File: Author Response.doc