Dear editor, Dear Dr. Yuhe Ji,

We would like to thank you for your interest in our work, and also for your evaluation, and most importantly, providing some constructive comments and suggestions, which are indeed very valuable to improve our work. Below is our one-by-one reply to the concerns and suggestion. The author replies were illustrated in ITALIC. Some of our replies included updated figures, and they were attached as supplementary file to our response.

Thanks again.

Best regards

on behalf of all the authors

Zongshan Li and Maierdang Keyimu

The precipitation during non-growth season (coincide with non-monsoon season) in the monsoonal areas are extremely important for the forest ecosystems, because they are the critical moisture source in the early growth season when the monsoon has not arrived. This manuscript provided with opportunity to observe the long-time variability of the NGS precipitation in the southeastern Tibetan Plateau, where historical NGS precipitation was not reconstructed before, and such work provides with important knowledge to evaluate the future of regional forest development. The background of the study and importance of the current investigation were introduced well. The methodology was sound and provided in detail. The results and discussions were convincing and were presented logically in order. However, there are some minor points which needs to be addressed, as well as some suggestions which can be considered to refine the work.

1. It was used as “non-growth season precipitation” on the title of the manuscript, however, it was used as “winter precipitation” in the keywords, such mixed usage also existed at some other places within the main text, please synchronize.

Author reply:

Thanks for the comment. We will use as “non-growth season precipitation” throughout the revised manuscript.

1. Please separate the ombrothermic diagram of the climate variables form Fig. 1, because an important term “saddle shaped rainfall pattern” was mentioned in the discussion, and it was not clear to observe such rainfall pattern in the current status of Fig. 1.

Author reply:

Thanks. We will separate the ombrothermic map in the revised manuscript (it will be appeared as Fig. 2 in the revised manuscript). Please refer to the attached supplement for the separated and refined ombrothermic diagram.

1. Please alternate a higher resolution map of the study area because a lot of information were hard to obtain from the current map. In addition, please provide an image showing the landscape of the tree-ring sampling site.

Author reply:

Thanks for the suggestion. We will replace a re-created map of the investigation area (updated Fig. 1) which included more information about the geographical conditions of the study area (combing the comments by referee 1). Fig. 1 also included the landscape image of tree ring sampling site. Please refer to the attachment for the updated Fig. 1.

1. What is the sample depth In Fig. 2? Is it number of tree-ring cores or number of trees?

Author reply:

The sample depth in Fig. 2 is the number of tree-ring cores. We have sampled one tree-ring core per tree. We will replace the updated Fig. 2 (it will be Fig. 3 in the revised manuscript). Please refer to the attachment for the updated Fig. 2.

1. It is important in dendrochronology (of course not limited to) science to present the complex and abstract results in a clear and easy to understand way for the readers. Please place the Rbar, EPS, and Sample depth on the right side of the Fig. 2, and use similar tick position for them, in this way, it would be easier for readers to have a sense that, which sample depth corresponded with which EPS and Rbar.

Author reply:

Thanks for the careful attitude and the suggestion. We will replace the updated Fig. 2 according to the comment. Please refer to the attachment for the updated Fig. 2.

1. It is suggested to re-check the running EPS value, because it seems that sample depth was already quite high when the EPS reached threshold value.

Author reply:

Thanks. We have re-calculated the EPS value carefully, and the result was the same with the original one. Combining the suggestion of referee 1, we have used EPS criteria of 0.85 to truncate the reliable length of the chronology (updated Fig. 2, please see attachment).

1. Please mention the meaning of Durban Watson test (Table2) in the results section, what does it imply.

Author reply:

We will remove the Durban Watson test (which represents the autocorrelation) from the Table 2, because what we used is the residual chronology (without autocorrelation), and presenting DW does not make sense.

1. Please add the unit of NGS precipitation in the transfer function.

Author reply:

We will add the unit of NGS precipitation in the transfer function.

1. Too much area was involved in the Fig. 7 to show the spatial representativeness of the reconstruction and actual NGS precipitation.

Author reply:

Thanks for the suggestion. We have re-conducted the spatial correlation analysis, and updated the Fig. 7 (it will be Fig. 8 in the revised manuscript). We will replace the updated figure in the revised manuscript. Please refer to the attachment for the updated Fig. 7.

1. Some references were not inserted in the main text, while some citations were not provided in the reference list. Besides, the references should be re-organized according to the journal template. For instance, it was used as “D’Arrigo, R. D” at one place, while it was used as “D’Arrigo, R” at another place. “Clim. Dynam.” in one place, while “Clim. Dyn.” in other place, and so on, please rectify.

Author reply:

Thanks for the careful attitude of Dr. Yuhe Ji. We will check through the main text of the manuscript and also the reference list, and rectify the existing problems within citations and references.

Dear editor, Dear referee 1,

First of all, we would like to thank you for your comments and suggestion, some of them are critical but invaluable to improve our original work. Secondly, we benefited from the discussion process, while we were solving existed problems and answering questions, we gain some insights, found new data resources, and improved our data visualization skills, which surely benefit us in our future scientific work.

Below is our one-by-one reply to the concerns and suggestion of the referee 1. The author replies were illustrated in ITALIC. Some of our replies included updated figures, and they were attached as supplementary file to our response.

Thanks again.

Best regards

on behalf of all the authors

Zongshan Li and Maierdang Keyimu

General comments

Tree rings response to local climate conditions, and the temperature and moisture in the growing season are usually the controlling factors for radial growth. In semi-arid regions on the southern Tibetan Plateau, where the climate is cold and dry, including the upper treeline, tree-ring width of coniferous species respond to growing season climate variables, e.g. early summer, warm season, or the whole year average (Zhang et al., 2015; Liu et al., CR 2012; Liu et al, 2014 QSR). Liu et al (NSR) comprehensively revealed that tree rings in SETP respond to growing season or annual precipitation and only a few tree ring chronology respond to PDSI due to the pre-monsoon moisture deficiency. As a result the physiological dynamics of tree ring/NGS correlation should be carefully and reasonably clarified. Consider the climate background and the tree growth/climate patterns, it is suggested to see the tree ring index responding to pre-monsoon climate variables.

Comparisons with previous results in the nearby area should include the difference and attributions within series other than consistence only.

There were lots of tiny mistakes in the manuscript and some figures needs improving. Acceptance could be done after the second revision.

Specific comments

1. It’s unnecessary to show descriptive statistics of the reconstructions in the abstract. Concentrations on the key results are mainly demanded.

Author reply:

Thanks for the suggestion. We will remove the lines (and the leave-one-out verification parameters indicated the reliability of the reconstruction) which interpreted the descriptive statistic of the chronology or reconstruction.

2. Tree rings are more and more of important in paleoclimatology. It’s interesting to see that ‘tree rings’ are written in different way. Some are ‘tree rings’, and some are ‘tree-rings’. Early dendrochronologists or students need standard of the terms. Would the authors like to say something on this?

Author reply:

Thanks for the simple but not simple question. Honestly, I (first author) have not really paid much attention to the context in which we use the terms of “tree ring” and “tree-ring”, but I was using the latter one in most of the cases. But the above question by the referee 1 really made me think, and I have looked through a few books in tree-ring science seeking for an answer. I have realized some basic differences in the application of “tree ring” and “tree-rings”. Understanding of the usage of “tree-ring” is relatively easy. It is used in compound situations, such as “tree-ring samples”, “tree-ring width/density”, “tree-ring data”, “tree-ring indices”, “tree-ring chronology”, “tree-ring series”, “tree-ring analysis”, and “tree-ring research”. But it is used as “tree ring/s” when it is a separate and independent unit, i.e., “application of tree rings”, “climate signal in tree rings”, “use tree rings to date years”, “wide/narrow tree rings”, and so on. I’d like to thank the referee 1 to make me ponder over the correct usage of above-mentioned terminologies.

Accordingly, we will check through the whole manuscript and run corrections on the existing mistakes.

3. Line 37, ...’of the planet Earth’…, delete planet please.

Author reply:

Thanks. We will delete as suggested.

4. Figure 1, besides the study site and the sites from previous sequence’s reference, we knew quite few from the figures. We couldn’t see where the sites are, and what the key geographical settings are nearby.

Author reply:

Thanks for the comment. We will replace an updated the location map of the study area. Please refer to the attachment for the updated figure.

5. Line 99, 3.26e gridded? Latest version of CRU is 4.04 (2020, Nature scientific data)

Author reply:

Thanks. We have downloaded the scPDSI data using KNMI climate explorer (http://climexp.knmi.nl/select.cgi?id=someone@somewhere&field=scpdsi). On this website, the latest version of PDSI data provided was 3.26e. We will try to download the latest version of scPDSI data from other sources and run tree growth – climate relationship analysis again, and observe if the correlation results between TRW and scPDSI were the same/different.

6. Line 96, 101-102, 27.17 N, 99.28 Eà17° N, 99.28° E; 27.0-27.5 N, 99.0-99.5 Eà27.0-27.5 °N, 99.0-99.5° E.

Author reply:

Thanks for pointing out the mistake. We will modify accordingly.

7. The EPS was below 1475 A.D., and the sample depth was less 7 according to Figure 2. Why didn’t safely choose the confident period since 1600 A.D. for reconstruction?

Author reply:

Thanks for the suggestion. Combining the suggestions of referee 1 and Dr. Ji Yuhe in the public comment, we have re-calculated the running EPS and Rbar values of the chronology, and updated the Fig. 2 (it will be appeared as Fig. 3 in the revised manuscript). As suggested by the referee 1, we have used the EPS criterion value of 0.85 to truncate the most reliable length of the TRW chronology and used it for the reconstruction (updated Fig. 5). We will replace the updated Fig. 2 and updated Fig. 5 (will be Fig. 6) in the revised manuscript. Please refer to the attachment for the updated figures.

8. Table 1, was it number of cores or trees? If it was number of cores, well, 38 cores rather than trees make the reconstruction since 1475 A.D. disputable.

Author reply:

Thanks for the concern. It was the number of trees and the cores. One core per tree was sampled intending to increase the sampling representativity. We will replace the updated figure (the name of the scale of the figure on the right side has been modified). Please refer to the updated Fig. 2 in the attachment.

9. The tree grown/temperature correlation pattern looks rather weird. There wasn’t positive correlation coefficient found at all. If we take the climate factors together, it was found TRW negatively correlated with temperature but positively with precipitation in current May. It is typically the pattern that the hemlock radial growth is limited by the pre-monsoon drought. The non-significant correlation coefficients with PDSI could possible attributed to temperature during the pre-monsoon season. Have the authors ever tested the correlation between TRW and pre-monsoon drought (prior December to current May)?

Author reply:

Thanks for pointing out the mistake. (1) We will rectify the interpretation of the correlation between TRW and temperature. (2) We have checked the relationship between TRW and PDSI of aggregated months (previous year December to current year May). The correlation value between aggregated PDSI and TRW was weaker (R = 0.47) than the correlation value between NGS precipitation and TRW (R = 0.56), and thus we would keep the original variable (NGS precipitation) as reconstruction target.

10. Figure 6 displayed comparison between this study and previous results where the sampling site are close. Visually compared, besides the common wet/dry variations in decadal scales were identified, much difference could be easily found. Obviously, Zhang’s and Li’s series showed increasing trend during the 2000s, but the other three series didn’t. For Zhang’s series, which was a compo-site reconstruction, could the authors adopt only sites that are close to Lijiang? During 1540s-1580s, 1680s-1720s, 1840s-1920s, no common variation patterns were identified, and some were even contrarily varied. By the way, the ‘year’ scale of the figure was shown in the window of 80 years, and it is difficult to read. Why didn’t show it in every 50-year step?

Author reply:

Thanks for the concerns. We would like to answer the above question by separating it into three sections:

1. As mentioned by the referee 1, apart from similarities, dissimilarities were also existed among different reconstructions. The dissimilarities among reconstructions can be attributed to (1) different tree species in chronologies which have different morphological structures and different drought tolerance capacities, (2) different reconstruction target (PDSI/precipitation), (3) seasonal differences in reconstruction target (annual/summer/winter/different aggregations), (4) sample replication, (5) different methods of detrending the tree ring measurement series and different chronology establishment methods (standard chronology/residual chronology), (6) different length of calibration period.

We have summarized the differences among the compared reconstructions as in Table S1 (please see attachment), and because of these, there appeared dissimilarities among the variabilities of different reconstruction series.

1. We have used the average PDSI reconstruction series of Zhang et al (2015) in the southeastern Tibetan Plateau to carry out the comparison with our reconstruction series.
2. We have updated the Fig. 6 (it will be appeared as Fig. 7 in the revised manuscript) according to the suggestion by the reviewer (using 50 years interval to separate), please refer to the attachment for the updated Fig. 6.

11. Extreme dry/wet years were investigated, did they were consistent with other results in Fig 6? What did those extreme years imply? In terms of the spatial correlation analysis, did those extreme years spatially exist? Were there coincident with the Asian Monsoon Atlas (Cook et al, 2010)

Author reply:

Thanks for the questions and suggestions. We have defined the years which had more/less than one times of the standard deviation of NGS precipitation as wet/dry NGS years; two times of the standard deviation of NGS precipitation as extreme wet/dry NGS years. According to the above definition, appearances of wet/dry years were too frequent, therefore, we have selected the extreme wet/dry years to demonstrate variability of wet and dry years.

We have compared the extreme wet/dry NGS years with reconstruction series of Fan et al., 2008, Fang et al., 2010, Li et al., 2017, and Zhang et al., 2015. By comparison, some of the extreme wet/dry years in present reconstruction series were consistent with other reconstructions; some of the extreme wet years in present reconstruction were not extreme wet but wet in other series; some of the extreme dry years in present reconstruction were not extreme dry but dry in other series. We have made clear comparison of the extreme wet/dry years among different reconstructions in Table S2, S3, please refer to the attachment.

We have extracted the drought series of Asian Monsoon Atlas (Cook et al.2010) from the nearest point (http://drought.memphis.edu/MADA/Extract.aspx) and compared it with our NGS precipitation reconstruction (please refer to updated Fig. 6 in the attachment). The extreme wet years in our reconstruction were coincided with the extreme wet years in the MADA; six out of 11 extreme dry years in our reconstruction were matched with the extreme dry years in MADA (please refer to the Table S2, S3 in the attachment).

12. Line 262, was the 1920s-1930s drought called World War I drought in southeastern China? (Kang et al., 2013, QI).

Author reply:

Thanks for the comment. We have checked the reference provided, and found that the drought period (1920s) should be called “China mega-drought”. We will modify as “China mega-drought” in the revised manuscript.

Dear editor, Dear referee 2,

We would like to thank referee 2 for his/her participation in the discussion and evaluating our work, as well as providing important comments and suggestions. Particularly, the suggestion of adding some isotope analysis-based investigations in the discussion section is indeed quite beneficial for us to further illustrate the eco-physiological importance of NGS precipitation on radial tree growth. Such discussion strongly supports our findings about the association between radial tree growth and NGS precipitation.

Below is our one-by-one reply to the concerns and suggestion of the referee 2. The author replies were illustrated in ITALIC. Some of our replies included updated figures, and they were attached as supplementary file to our response.

Thanks again.

Best regards

on behalf of all the authors

Zongshan Li and Maierdang Keyimu

Referee 2 comments:

Using tree-ring width data, Keyimu et al. (2021) presented a non-growing season precipitation reconstruction from 1475 to 2005 on the southeastern Tibetan Plateau. Given that there are lot of summer precipitation or temperature reconstructions in this region, it is very interesting to obtain non-growing seasonal precipitation reconstruction. Overall, this study is well designed with reasonable data analysis, producing a robust result and conclusion. I suggest to accept this manuscript after minor revision. Detailed comments and suggestions are as follows:

1. Line 1, "non-growth" or "non-growing", which is suitable? Please check

Author reply:

Thanks for the comment. After checking many literatures, we have considered to use as “non-growing”. We will replace as “non-growing” through the whole manuscript.

1. Line 1, Add "A" before "531-year"

Author reply:

Thanks. The title of the MS will be changed because according to the EPS value of the TRW chronology, the updated length of the reconstruction will be 406 years (A.D. 1600-2005). Therefore, we will change the title as “A 406-year non-growing season precipitation reconstruction in the southeastern Tibetan Plateau”.

1. Lines 79-83. It is better that only "Figure 1" should be in bold, other text should be normal. Same for other tables and figures

Author reply:

Thanks. We will do so.

1. Lines 192-193, it is a little difficult to see green and yellow bars, maybe it's better to change to other color combinations.

Author reply:

Thanks for the suggestion. We will replace the upgraded the Fig. 6 (it will be Fig. 7 in the revised manuscript) combining the comments of referee 1 and referee 2. Please refer to the attachment for the updated Fig. 6.

1. Line 206-233. More detailed discussions are needed. It appeared the underlying mechanisms about the non-growing season precipitation signals of tree-ring widths were lacking. The non-growing season precipitation signals of tree-ring widths seemed to imply the non-monsoon (e.g., winter) precipitation was used for tree growth. Maybe tree-ring oxygen isotopes could provide some evidence to support non-monsoon precipitation usage of tree growth.

Author reply:

Thanks a lot for the valuable suggestion about the isotope analysis which was indeed important to improve the discussion over the underlying mechanism between NGS precipitation and radial growth of forest hemlock in current study.

In the revised manuscript, we will add detailed discussion about the importance of NGS precipitation on radial tree growth combining some isotope-based findings, below is the content which we are going to add:

“This is because tree growth is often water stressed in the early stages of its growth in each year on the SETP when the monsoon precipitation does not arrive (Bräuning and Mantwill, 2004; Zhang et al., 2015), and the earlywood of tree rings mainly use spring melt water (Zhu et al., 2021). The eco-physiological importance of NGS precipitation on tree growth and tree water usage was also revealed by isotope ratios method-based investigations. Brinkmann et al’s (2018) study showed that nearly 40% of the uptaken water by Fagus sylvatica and Picea abies trees in a temperate forest of middle Europe are sourced from NGS precipitation. Tree-ring oxygen isotope ratios (δ18O) are demonstrated to contain NGS precipitation signals in the Himalayan region (Huang et al., 2019; Zhu et al., 2021). Huang et al’s (2019) study revealed that NGS precipitation (snowfall) increased the snow-depth and the later snowmelt compensated soil moisture in the spring and early summer, which was a crucially important water source for the Juniper growth in the southwestern Tibetan Plateau. Zhu et al’s (2021) investigation in the western Himalaya revealed that formation of earlywood in tree rings of Pinus wallachina depended on the snowmelt originated from NGS precipitation”.