the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 11 Jan 2021
Overcoming model instability in tree-ring-based temperature reconstructions using a multi-species method: A case study from the Changbai Mountains, northeastern China
Abstract. The unstable sensitivity of growth-climate relationships greatly restricts tree-ring-based paleoclimate reconstructions, especially in areas with frequent divergence
problems, such as the temperate zone in northeast China. Here, we propose an original tree-species mixing method to overcome this obstacle and improve the stability and reliability of reconstruction models. We take the tree-ring based growing-season minimum temperature reconstruction for the northern Changbai Mountains in northeast China as an example to illustrate the method. Compared with previous temperature reconstruction models, our reconstruction model is more stable and reliable and explains up to 68 % of the variance. It is also highly consistent with historical records and tree-ring-based temperature reconstructions from the nearby Xiaoxing'an Mountains and from across the Northern Hemisphere. Our reconstruction uses two different tree species and is more accurate than temperature reconstructions developed from a single species. Over the past 259 years (AD 1757–2015), five significant cold periods and five warm periods were identified. The reconstruction indicates rapid warming since the 1980s, which is consistent with other instrumental and reconstructed records. We also found the Atlantic Multidecadal Oscillation plays a crucial role in driving the growing-season minimum temperature in the northern Changbai Mountains.
- Preprint
(2557 KB) - Metadata XML
-
Supplement
(580 KB) - BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on cp-2021-2', Anonymous Referee #1, 02 Feb 2021
The authors presented a 259-year long (from AD 1757 to 2015) April-September minimum temperature record reconstructed from tree-ring widths from six sites in Changbai Mountains of northeastern China. Different from traditional method that mainly use a single tree species for dendroclimatological reconstructions, the authors reconstructed the target temperature using tree rings combined from two species. They demonstrated the appropriateness of the method by comparing the tree growth-climate relationships in two single species and their combinations, and showing that the association of tree rings with regional April-September minimum temperature was stronger using data from combination of the two species than from each of the single species. From the temperature reconstruction, they identified the warm and cold periods in the past and discussed the linkage of the regional temperature variability with Atlantic Multidecadal Oscillation (AMO). I am not expertised in AMO. The following comments are mainly about tree rings.
My major concern is on the biological basis of combining tree-ring data from the two species.
- Line 144: “Significant correlations between P. koraiensis and F. mandshurica chronologies, however, only exist for some sites”. What’s the reason for these un-correlations? And what is the biological rational for combining the tree-ring data from such un-correlated site chronologies?
- Line 153: “However, there is a differential temporal instability in the growth-climate relationship between P. koraiensis and F. mandshurica” and Fig. 3. It seems to me that the differential temporal instability in the growth-climate relationships between the two species was not removed in the combined regional chronology, rather, it was simply mixed. It is still unclear how the correlation coefficients between tree rings and temperature increased from 0.672 and 0.762 for the two single species to 0.824 for mixture of the two species. If this question is not answered, it is hard to be applied for other studies. What I guess is that the combined chronology was smoothed by the mixing which favored correlation coefficients for time series with obvious trends.
- Curiosity: As shown in Fig. 3, the lower correlations in growth-climate relationship appeared for high values of the species PK and for low values of the species FM. I suggest the authors to have a try for removal of these high values of PK and low values of FM (the level of high or low depends on authors’ definition) in development of the combined chronology. The biology for this practice is that, when PK grows wide (and FM grows narrow), factors other than April-September minimum temperature join together to limit tree growth, thus the wide rings in PK (and narrow rings in FM) no longer act as reliable signal for the target temperature. I am not sure about the correctness of this suggestion but the authors could have a try. This means you are taking actions (rather than simple mixing) to deal with the temporal instability.
Minor points:
- Title: please consider if the “overcoming model instability” and “multi-species” are appropriate.
- Line 40: “however”, is there a turning between the two sentences?
- Line 42: “reduction in tree-ring indices and temperature sensitivity”, reduction in tree-ring indices or reduction in sensitivity of ...?
- Line 53: “Pinus” or “P.”.
- Line 58: “frequent”?
- 1, the map at right: the site name of the top line: a typo?
- 2: right bottom: how about draw the trend line from the year 1986?
- Line 125: “greater than (less than) high than”: a typo?
- 3: The level of significance for correlation coefficients should be tested differently for time series with a trend and without a trend.
- Line 222: “consistent” with
- Line 231:”possibility” or reliability?
- In the early period of the reconstruction (close to AD 1757), how many sites are from PK and how many from FM?
Citation: https://doi.org/10.5194/cp-2021-2-RC1 -
AC1: 'Reply on RC1', Liangjun Zhu, 17 Apr 2021
Dear editor, dear referees,
We are so sorry for this late reply. Thank you for your constructive comments on our manuscript. The comments were valuable for helping us to revise and improve our paper, as well as the important guiding significance to our future research. We will do our best to improve the MS according your suggestions.On behalf of the authors
Yours sincerely
LiangjunReply to Anonymous Referee #1:
The authors presented a 259-year long (from AD 1757 to 2015) April-September minimum temperature record reconstructed from tree-ring widths from six sites in Changbai Mountains of northeastern China. Different from traditional method that mainly use a single tree species for dendroclimatological reconstructions, the authors reconstructed the target temperature using tree rings combined from two species. They demonstrated the appropriateness of the method by comparing the tree growth-climate relationships in two single species and their combinations, and showing that the association of tree rings with regional April-September minimum temperature was stronger using data from combination of the two species than from each of the single species. From the temperature reconstruction, they identified the warm and cold periods in the past and discussed the linkage of the regional temperature variability with Atlantic Multidecadal Oscillation (AMO). I am not expertised in AMO. The following comments are mainly about tree rings.
My major concern is on the biological basis of combining tree-ring data from the two species.
1. Line 144: “Significant correlations between P. koraiensis and F. mandshurica chronologies, however, only exist for some sites”. What’s the reason for these un-correlations? And what is the biological rational for combining the tree-ring data from such un-correlated site chronologies?
Reponses: P. koraiensis and F. mandshurica are different species and they have different response to main climate factors. The results of tree growth-climate relationships show a contrasting sensitive pattern of tree growth to minimum temperature (Tmin). We found that the FM is more limited by Tmin in the high Tmin period, while the PK more like limited by Tmin when low Tmin period. The biology for this practice is that, when the Tmin is high (warming period), factors other than April-September Tmin join together to limit tree growth of PK, thus the wide rings in PK no longer act as reliable signal for the target temperature. While the Tmin is low (cold period), other factors may join together with Tmin to limit tree growth of FM, thus the narrow rings in FM no longer act as reliable signal for Tmin. This is the biological rational for we combining the tree-ring data of two species even there do not correlated in some sites.
2. Line 153: “However, there is a differential temporal instability in the growth-climate relationship between P. koraiensis and F. mandshurica” and Fig. 3. It seems to me that the differential temporal instability in the growth-climate relationships between the two species was not removed in the combined regional chronology, rather, it was simply mixed. It is still unclear how the correlation coefficients between tree rings and temperature increased from 0.672 and 0.762 for the two single species to 0.824 for mixture of the two species. If this question is not answered, it is hard to be applied for other studies. What I guess is that the combined chronology was smoothed by the mixing which favored correlation coefficients for time series with obvious trends. Curiosity: As shown in Fig. 3, the lower correlations in growth-climate relationship appeared for high values of the species PK and for low values of the species FM. I suggest the authors to have a try for removal of these high values of PK and low values of FM (the level of high or low depends on authors’ definition) in development of the combined chronology. The biology for this practice is that, when PK grows wide (and FM grows narrow), factors other than April-September minimum temperature join together to limit tree growth, thus the wide rings in PK (and narrow rings in FM) no longer act as reliable signal for the target temperature. I am not sure about the correctness of this suggestion but the authors could have a try. This means you are taking actions (rather than simple mixing) to deal with the temporal instability.
Reponses: we will try your suggestion to test if it is work.
Minor points:
4. Title: please consider if the “overcoming model instability” and “multi-species” are appropriate.
Reponses: we will revise the title.
5. Line 40: “however”, is there a turning between the two sentences?
Reponses: we will do so.
6. Line 42: “reduction in tree-ring indices and temperature sensitivity”, reduction in tree-ring indices or reduction in sensitivity of ...?
Reponses: we will do so.
7. Line 53: “Pinus” or “P.”.
Reponses: we will do so.
8. Line 58: “frequent”?
Reponses: we have deleted it.
9. 1, the map at right: the site name of the top line: a typo?
Reponses: we will replot the figure.
10. 2: right bottom: how about draw the trend line from the year 1986?
Reponses: we will plot new figure.
11. Line 125: “greater than (less than) high than”: a typo?
Reponses: we will do so.
12. 3: The level of significance for correlation coefficients should be tested differently for time series with a trend and without a trend.
Reponses: we will do so.
13. Line 222: “consistent” with
Reponses: we will do so.
14. Line 231: ”possibility” or reliability?
Reponses: we will do so.
15. In the early period of the reconstruction (close to AD 1757), how many sites are from PK and how many from FM?
Reponses: 3 sites for FM and 2 sites for PK can reach the AD 1757.Citation: https://doi.org/10.5194/cp-2021-2-AC1
-
CC1: 'Comment on cp-2021-2', Feng Chen, 05 Feb 2021
General remarks:
The manuscript presented a multiple species method that significantly improves the accuracy of tree-ring-based climate reconstructions in areas with unstable growth-climate relationships. And also the linkage of the regional temperature of Changbai Mts in northeast China to aspects of the AMO index is one of the highlights of the MS. Overall, this paper is well-organized, and the results and conclusions are well supported by data. It provided a new idea for the development of tree-ring-based climate reconstruction. I suggest this MS being accepted by CP if the coauthors can address all reviews’ concerns and taking into account of comments in CPD (if they are reasonable).
Below I present some suggestions that the authors could find useful when revising the manuscript.
Specific comments:
- Figure 2: In the captions, the authors use the abbreviation “Tmin” and “Tmax” to represent the minimum and maximum temperature. However, subscripts are not used in the Figure 2a and Table 3.
- Line 134: “Atlantic Multi-decadal Oscillation (AMO)” should be “Atlantic Multidecadal Oscillation (AMO)”.
- Line 171: The abbreviation of the number of cases sometimes uses N, sometimes uses n (see table 1), please keep the same.
- Lines 176, 186 and 187: Change the “observed” to the “actual” to keep the same in whole text.
- Line 187: Change the “1959-2014” to the “1959-2015”.
- Table 3: Give the UNIT of minimum temperature, and add the notes for “Tmin”.
- In discussion section 4.2, the authors pointed out that their new temperature reconstruction has a higher quality than regional records. It is better to compare the correlations between the existing (TS, CB and LB in figure 6) reconstructed temperature records and other temperature records (XXA and NH). A new table should be added.
- In figure 6: What is the R and Rlow. The same in figure 8.
- Line 268: Change the “The Atlantic Multidecadal Oscillation (AMO)” to the “The AMO”.
- Line 285: “……the actual April-September AMO index……”.
- Line 291: “……and the reconstructed AMO index (annual Sea Surface Temperature anomalies for the North Atlantic) from Gray (2001)……”.
- Line 294: “……the AMO index used in this study (thin line)……”.
- Figure 9: The arrows in Figure 9a are not clear, please replot it.
- Please double check and update the references. For example, Zhu et al., 2020a (CATENA) should be update to 2021.
Citation: https://doi.org/10.5194/cp-2021-2-CC1 -
AC2: 'Reply on CC1', Liangjun Zhu, 17 Apr 2021
Dear editor, dear referees,
We are so sorry for this late reply. Thank you for your constructive comments on our manuscript. The comments were valuable for helping us to revise and improve our paper, as well as the important guiding significance to our future research. We will do our best to improve the MS according your suggestions.
On behalf of the authors
Yours sincerely
Liangjun
Reply to CC:
General remarks:
The manuscript presented a multiple species method that significantly improves the accuracy of tree-ring-based climate reconstructions in areas with unstable growth-climate relationships. And also the linkage of the regional temperature of Changbai Mts in northeast China to aspects of the AMO index is one of the highlights of the MS. Overall, this paper is well-organized, and the results and conclusions are well supported by data. It provided a new idea for the development of tree-ring-based climate reconstruction. I suggest this MS being accepted by CP if the coauthors can address all reviews’ concerns and taking into account of comments in CPD (if they are reasonable).
Below I present some suggestions that the authors could find useful when revising the manuscript.
Specific comments:
- Figure 2: In the captions, the authors use the abbreviation “Tmin” and “Tmax” to represent the minimum and maximum temperature. However, subscripts are not used in the Figure 2a and Table 3.
Reponses: we will keep the same.
- Line 134: “Atlantic Multi-decadal Oscillation (AMO)” should be “Atlantic Multidecadal Oscillation (AMO)”.
Reponses: we will do so.
- Line 171: The abbreviation of the number of cases sometimes uses N, sometimes uses n (see table 1), please keep the same.
Reponses: we will do so.
- Lines 176, 186 and 187: Change the “observed” to the “actual” to keep the same in whole text.
Reponses: we will do so.
- Line 187: Change the “1959-2014” to the “1959-2015”.
Reponses: we will do so.
- Table 3: Give the UNIT of minimum temperature, and add the notes for “Tmin”.
Reponses: we will do so.
- In discussion section 4.2, the authors pointed out that their new temperature reconstruction has a higher quality than regional records. It is better to compare the correlations between the existing (TS, CB and LB in figure 6) reconstructed temperature records and other temperature records (XXA and NH). A new table should be added.
Reponses: we will add the table.
- In figure 6: What is the R and Rlow. The same in figure 8.
Reponses: we will add the notes.
- Line 268: Change the “The Atlantic Multidecadal Oscillation (AMO)” to the “The AMO”.
Reponses: we will do so.
- Line 285: “……the actual April-September AMO index……”.
Reponses: we will revise it.
- Line 291: “……and the reconstructed AMO index (annual Sea Surface Temperature anomalies for the North Atlantic) from Gray (2001)……”.
Reponses: we will do so.
- Line 294: “……the AMO index used in this study (thin line)……”.
Reponses: we will do so.
- Figure 9: The arrows in Figure 9a are not clear, please replot it.
Reponses: we will do so.
- Please double check and update the references. For example, Zhu et al., 2020a (CATENA) should be update to 2021.
Reponses: we will double check the references.
Citation: https://doi.org/10.5194/cp-2021-2-AC2
-
RC2: 'Comment on cp-2021-2', Anonymous Referee #2, 28 Feb 2021
The manuscript ‘Overcoming model instability in tree-ring-based temperature reconstructions using a multi-species method: A case study from the Changbai Mountains, northeastern China’ is a test of a multi-species approach might improve tree-ring based reconstructions of past climate. The work indicates that, indeed, multiple species can improve the potential for stability issues in reconstructing past climate from tree rings. The grammar is generally good; only a few places where it can be improved – please double check. There are a couple more steps in the analysis that would further strengthen the work.
The primary concern is that this approach is not as novel as stated in the manuscript. First, (García-Suárez et al. 2009) wrote about the value of a multi-species approach. Specifically, they wrote, “Apparent instability in the relations to climate of species like beech. However, we have found that combinations of species are better able to reconstruct climate than single species. Particularly robust are the reconstructions that use ash and beech simultaneously as predictors.” Soon after (Pederson et al. 2013) tested this with a greater number of species and developed a table showing the potential for using multiple species in a reconstruction and discussed findings that indicate species replication might stronger than chronology replication for the improvement and stability of reconstruction statistics. (Alexander et al. 2019) followed up that work by explicitly testing the hypothesis that increased species replication would improve temperature reconstructions. (Maxwell et al. 2011) has demonstrated this, too. Though not discussed in a similar way, the many drought atlases are mutli-species reconstructions and (Frank and Esper 2005) discuss the climate response of a multi-species high elevation network. There is a good amount of groundwork available in the literature for the manuscript developed here.
I am glad to see this kind of testing being replicated in other regions. What is new in this manuscript is that this is being tested in a different region and, especially, in a monsoonal climate. I am just surprised that one or a few of these earlier works are not acknowledged in this manuscript. It has been in the literature for at least 12 years.
To be clear, the author’s work here is of value to science because of the region and species. The should, however, acknowledge previous work on a multi-species approach.
Minor Comments
Line 19 – replace ‘greatly’ with ‘can’. It is not evident that all reconstructions are greatly reduced due to instability. End the sentence after the phrase paleoclimate reconstructions and start a new sentence with this remaining text, “This is most notable in areas…”
Line 20 – replace “an original” with “a test” as noted above, this is not exactly original.
Line 32 – “climates are” should be “climate is”
Line 37 – instead of “Tree rings, especially ring widths,”, one could write “Information derived from tree-ring records…”
Line 39 – instead of “, which” can replace with “and”
Line 41 – another citation to go along with Fritts is (Cook and Kairiukstis 1990).
Line 58 – “hypothesize” here should be “hypothesis”
Line 152 – an R of 0.824 is incredible! I appreciate the authors put in the figure with first differences. They should state the R of that relationship (if they have not. If I have missed it, I apologize).
Figure 3 – it looks like PK does a lot of the work here. It would be interesting to make a PCA of all chronologies to see which ones and which species contribute the most for each period tested.
Figure 4b – thank you for showing the first differences. Please note in the manuscript that the trend is doing most of the work with the R2 results.
Figure 4c - adding in uncertainty to 4c is recommended. One approach would be to calculate the root mean squared error of the model and show two levels of uncertainty. One or two times the root mean squared error, for example.
Line 195 – after and 1956-1982) it would be helpful to the readers if you add something like “according to how we defined warm/cold.
Figure 5 – suggestion: what are the spectral properties prior to 1980 vs the entire record? It would be interesting to see how much warming has or has not changed the spectral frequency.
Figure 6 - synchronization of temperatures and trends is a characteristic of global warming. Could highlight that fact in the text.
Line 272 – consider trying cross spectral analyses to see at what frequencies your recon and the AMO are in sync.
Line 315 – Given the general concern I have at the start of the review, rewrite the first sentence as, “We have tested how mixing tree species might improve the accuracy….”
References
Alexander, M. R., J. K. Pearl, D. A. Bishop, E. R. Cook, K. J. Anchukaitis, and N. Pederson. 2019. The potential to strengthen temperature reconstructions in ecoregions with limited tree line using a multispecies approach. Quaternary Research:1-15.
Cook, E. R., and L. A. Kairiukstis. 1990. Methods of Dendrochronology, Applications in the Environmental Sciences. Kluwer.
Frank, D., and J. Esper. 2005. Characterization and climate response patterns of a high-elevation, multi-species tree-ring network in the European Alps. Dendrochronologia 22:107-121.
García-Suárez, a. M., C. J. Butler, and M. G. L. Baillie. 2009. Climate signal in tree-ring chronologies in a temperate climate: A multi-species approach. Dendrochronologia 27:183-198.
Maxwell, R. S., A. E. Hessl, E. R. Cook, and N. Pederson. 2011. A multispecies tree ring reconstruction of Potomac River streamflow (950–2001). Water Resources Research 47:1-12.
Pederson, N., A. R. Bell, E. R. Cook, U. Lall, N. Devineni, R. Seager, K. Eggleston, and K. P. Vranes. 2013. Is an Epic Pluvial Masking the Water Insecurity of the Greater New York City Region? Journal of Climate 26:1339-1354.
Citation: https://doi.org/10.5194/cp-2021-2-RC2 -
AC3: 'Reply on RC2', Liangjun Zhu, 17 Apr 2021
Dear editor, dear referees,
We are so sorry for this late reply. Thank you for your constructive comments on our manuscript. The comments were valuable for helping us to revise and improve our paper, as well as the important guiding significance to our future research. We will do our best to improve the MS according your suggestions.
On behalf of the authors
Yours sincerely
Liangjun
Reply to Anonymous Referee #2:
The manuscript ‘Overcoming model instability in tree-ring-based temperature reconstructions using a multi-species method: A case study from the Changbai Mountains, northeastern China’ is a test of a multi-species approach might improve tree-ring based reconstructions of past climate. The work indicates that, indeed, multiple species can improve the potential for stability issues in reconstructing past climate from tree rings. The grammar is generally good; only a few places where it can be improved – please double check. There are a couple more steps in the analysis that would further strengthen the work.
The primary concern is that this approach is not as novel as stated in the manuscript. First, (García-Suárez et al. 2009) wrote about the value of a multi-species approach. Specifically, they wrote, “Apparent instability in the relations to climate of species like beech. However, we have found that combinations of species are better able to reconstruct climate than single species. Particularly robust are the reconstructions that use ash and beech simultaneously as predictors.” Soon after (Pederson et al. 2013) tested this with a greater number of species and developed a table showing the potential for using multiple species in a reconstruction and discussed findings that indicate species replication might stronger than chronology replication for the improvement and stability of reconstruction statistics. (Alexander et al. 2019) followed up that work by explicitly testing the hypothesis that increased species replication would improve temperature reconstructions. (Maxwell et al. 2011) has demonstrated this, too. Though not discussed in a similar way, the many drought atlases are mutli-species reconstructions and (Frank and Esper 2005) discuss the climate response of a multi-species high elevation network. There is a good amount of groundwork available in the literature for the manuscript developed here.
I am glad to see this kind of testing being replicated in other regions. What is new in this manuscript is that this is being tested in a different region and, especially, in a monsoonal climate. I am just surprised that one or a few of these earlier works are not acknowledged in this manuscript. It has been in the literature for at least 12 years.
To be clear, the author’s work here is of value to science because of the region and species. They should, however, acknowledge previous work on a multi-species approach.
Reponses: Thank you for your constructive comments on our manuscript. These comments are valuable for helping us to improve our paper. We will add those literature in to the background of our work and do our best to revise the MS according your valuable suggestions.
Minor Comments
Line 19 – replace ‘greatly’ with ‘can’. It is not evident that all reconstructions are greatly reduced due to instability. End the sentence after the phrase paleoclimate reconstructions and start a new sentence with this remaining text, “This is most notable in areas…”
Reponses: we will do so.
Line 20 – replace “an original” with “a test” as noted above, this is not exactly original.
Reponses: we will do so.
Line 32 – “climates are” should be “climate is”
Reponses: we will do so.
Line 37 – instead of “Tree rings, especially ring widths,”, one could write “Information derived from tree-ring records…”
Reponses: we will do so.
Line 39 – instead of “, which” can replace with “and”
Reponses: we will do so.
Line 41 – another citation to go along with Fritts is (Cook and Kairiukstis 1990).
Reponses: we will do so.
Line 58 – “hypothesize” here should be “hypothesis”
Reponses: we will do so.
Line 152 – an R of 0.824 is incredible! I appreciate the authors put in the figure with first differences. They should state the R of that relationship (if they have not. If I have missed it, I apologize).
Figure 3 – it looks like PK does a lot of the work here. It would be interesting to make a PCA of all chronologies to see which ones and which species contribute the most for each period tested.
Reponses: we will do so.
Figure 4b – thank you for showing the first differences. Please note in the manuscript that the trend is doing most of the work with the R2 results.
Reponses: we will do so.
Figure 4c - adding in uncertainty to 4c is recommended. One approach would be to calculate the root mean squared error of the model and show two levels of uncertainty. One or two times the root mean squared error, for example.
Reponses: we will do so.
Line 195 – after and 1956-1982) it would be helpful to the readers if you add something like “according to how we defined warm/cold.
Reponses: we will do so.
Figure 5 – suggestion: what are the spectral properties prior to 1980 vs the entire record? It would be interesting to see how much warming has or has not changed the spectral frequency.
Reponses: we will do so.
Figure 6 - synchronization of temperatures and trends is a characteristic of global warming. Could highlight that fact in the text.
Reponses: we will do so.
Line 272 – consider trying cross spectral analyses to see at what frequencies your recon and the AMO are in sync.
Reponses: we will do so.
Line 315 – Given the general concern I have at the start of the review, rewrite the first sentence as, “We have tested how mixing tree species might improve the accuracy….”
Reponses: we will do so.
Citation: https://doi.org/10.5194/cp-2021-2-AC3
-
AC3: 'Reply on RC2', Liangjun Zhu, 17 Apr 2021
Status: closed
-
RC1: 'Comment on cp-2021-2', Anonymous Referee #1, 02 Feb 2021
The authors presented a 259-year long (from AD 1757 to 2015) April-September minimum temperature record reconstructed from tree-ring widths from six sites in Changbai Mountains of northeastern China. Different from traditional method that mainly use a single tree species for dendroclimatological reconstructions, the authors reconstructed the target temperature using tree rings combined from two species. They demonstrated the appropriateness of the method by comparing the tree growth-climate relationships in two single species and their combinations, and showing that the association of tree rings with regional April-September minimum temperature was stronger using data from combination of the two species than from each of the single species. From the temperature reconstruction, they identified the warm and cold periods in the past and discussed the linkage of the regional temperature variability with Atlantic Multidecadal Oscillation (AMO). I am not expertised in AMO. The following comments are mainly about tree rings.
My major concern is on the biological basis of combining tree-ring data from the two species.
- Line 144: “Significant correlations between P. koraiensis and F. mandshurica chronologies, however, only exist for some sites”. What’s the reason for these un-correlations? And what is the biological rational for combining the tree-ring data from such un-correlated site chronologies?
- Line 153: “However, there is a differential temporal instability in the growth-climate relationship between P. koraiensis and F. mandshurica” and Fig. 3. It seems to me that the differential temporal instability in the growth-climate relationships between the two species was not removed in the combined regional chronology, rather, it was simply mixed. It is still unclear how the correlation coefficients between tree rings and temperature increased from 0.672 and 0.762 for the two single species to 0.824 for mixture of the two species. If this question is not answered, it is hard to be applied for other studies. What I guess is that the combined chronology was smoothed by the mixing which favored correlation coefficients for time series with obvious trends.
- Curiosity: As shown in Fig. 3, the lower correlations in growth-climate relationship appeared for high values of the species PK and for low values of the species FM. I suggest the authors to have a try for removal of these high values of PK and low values of FM (the level of high or low depends on authors’ definition) in development of the combined chronology. The biology for this practice is that, when PK grows wide (and FM grows narrow), factors other than April-September minimum temperature join together to limit tree growth, thus the wide rings in PK (and narrow rings in FM) no longer act as reliable signal for the target temperature. I am not sure about the correctness of this suggestion but the authors could have a try. This means you are taking actions (rather than simple mixing) to deal with the temporal instability.
Minor points:
- Title: please consider if the “overcoming model instability” and “multi-species” are appropriate.
- Line 40: “however”, is there a turning between the two sentences?
- Line 42: “reduction in tree-ring indices and temperature sensitivity”, reduction in tree-ring indices or reduction in sensitivity of ...?
- Line 53: “Pinus” or “P.”.
- Line 58: “frequent”?
- 1, the map at right: the site name of the top line: a typo?
- 2: right bottom: how about draw the trend line from the year 1986?
- Line 125: “greater than (less than) high than”: a typo?
- 3: The level of significance for correlation coefficients should be tested differently for time series with a trend and without a trend.
- Line 222: “consistent” with
- Line 231:”possibility” or reliability?
- In the early period of the reconstruction (close to AD 1757), how many sites are from PK and how many from FM?
Citation: https://doi.org/10.5194/cp-2021-2-RC1 -
AC1: 'Reply on RC1', Liangjun Zhu, 17 Apr 2021
Dear editor, dear referees,
We are so sorry for this late reply. Thank you for your constructive comments on our manuscript. The comments were valuable for helping us to revise and improve our paper, as well as the important guiding significance to our future research. We will do our best to improve the MS according your suggestions.On behalf of the authors
Yours sincerely
LiangjunReply to Anonymous Referee #1:
The authors presented a 259-year long (from AD 1757 to 2015) April-September minimum temperature record reconstructed from tree-ring widths from six sites in Changbai Mountains of northeastern China. Different from traditional method that mainly use a single tree species for dendroclimatological reconstructions, the authors reconstructed the target temperature using tree rings combined from two species. They demonstrated the appropriateness of the method by comparing the tree growth-climate relationships in two single species and their combinations, and showing that the association of tree rings with regional April-September minimum temperature was stronger using data from combination of the two species than from each of the single species. From the temperature reconstruction, they identified the warm and cold periods in the past and discussed the linkage of the regional temperature variability with Atlantic Multidecadal Oscillation (AMO). I am not expertised in AMO. The following comments are mainly about tree rings.
My major concern is on the biological basis of combining tree-ring data from the two species.
1. Line 144: “Significant correlations between P. koraiensis and F. mandshurica chronologies, however, only exist for some sites”. What’s the reason for these un-correlations? And what is the biological rational for combining the tree-ring data from such un-correlated site chronologies?
Reponses: P. koraiensis and F. mandshurica are different species and they have different response to main climate factors. The results of tree growth-climate relationships show a contrasting sensitive pattern of tree growth to minimum temperature (Tmin). We found that the FM is more limited by Tmin in the high Tmin period, while the PK more like limited by Tmin when low Tmin period. The biology for this practice is that, when the Tmin is high (warming period), factors other than April-September Tmin join together to limit tree growth of PK, thus the wide rings in PK no longer act as reliable signal for the target temperature. While the Tmin is low (cold period), other factors may join together with Tmin to limit tree growth of FM, thus the narrow rings in FM no longer act as reliable signal for Tmin. This is the biological rational for we combining the tree-ring data of two species even there do not correlated in some sites.
2. Line 153: “However, there is a differential temporal instability in the growth-climate relationship between P. koraiensis and F. mandshurica” and Fig. 3. It seems to me that the differential temporal instability in the growth-climate relationships between the two species was not removed in the combined regional chronology, rather, it was simply mixed. It is still unclear how the correlation coefficients between tree rings and temperature increased from 0.672 and 0.762 for the two single species to 0.824 for mixture of the two species. If this question is not answered, it is hard to be applied for other studies. What I guess is that the combined chronology was smoothed by the mixing which favored correlation coefficients for time series with obvious trends. Curiosity: As shown in Fig. 3, the lower correlations in growth-climate relationship appeared for high values of the species PK and for low values of the species FM. I suggest the authors to have a try for removal of these high values of PK and low values of FM (the level of high or low depends on authors’ definition) in development of the combined chronology. The biology for this practice is that, when PK grows wide (and FM grows narrow), factors other than April-September minimum temperature join together to limit tree growth, thus the wide rings in PK (and narrow rings in FM) no longer act as reliable signal for the target temperature. I am not sure about the correctness of this suggestion but the authors could have a try. This means you are taking actions (rather than simple mixing) to deal with the temporal instability.
Reponses: we will try your suggestion to test if it is work.
Minor points:
4. Title: please consider if the “overcoming model instability” and “multi-species” are appropriate.
Reponses: we will revise the title.
5. Line 40: “however”, is there a turning between the two sentences?
Reponses: we will do so.
6. Line 42: “reduction in tree-ring indices and temperature sensitivity”, reduction in tree-ring indices or reduction in sensitivity of ...?
Reponses: we will do so.
7. Line 53: “Pinus” or “P.”.
Reponses: we will do so.
8. Line 58: “frequent”?
Reponses: we have deleted it.
9. 1, the map at right: the site name of the top line: a typo?
Reponses: we will replot the figure.
10. 2: right bottom: how about draw the trend line from the year 1986?
Reponses: we will plot new figure.
11. Line 125: “greater than (less than) high than”: a typo?
Reponses: we will do so.
12. 3: The level of significance for correlation coefficients should be tested differently for time series with a trend and without a trend.
Reponses: we will do so.
13. Line 222: “consistent” with
Reponses: we will do so.
14. Line 231: ”possibility” or reliability?
Reponses: we will do so.
15. In the early period of the reconstruction (close to AD 1757), how many sites are from PK and how many from FM?
Reponses: 3 sites for FM and 2 sites for PK can reach the AD 1757.Citation: https://doi.org/10.5194/cp-2021-2-AC1
-
CC1: 'Comment on cp-2021-2', Feng Chen, 05 Feb 2021
General remarks:
The manuscript presented a multiple species method that significantly improves the accuracy of tree-ring-based climate reconstructions in areas with unstable growth-climate relationships. And also the linkage of the regional temperature of Changbai Mts in northeast China to aspects of the AMO index is one of the highlights of the MS. Overall, this paper is well-organized, and the results and conclusions are well supported by data. It provided a new idea for the development of tree-ring-based climate reconstruction. I suggest this MS being accepted by CP if the coauthors can address all reviews’ concerns and taking into account of comments in CPD (if they are reasonable).
Below I present some suggestions that the authors could find useful when revising the manuscript.
Specific comments:
- Figure 2: In the captions, the authors use the abbreviation “Tmin” and “Tmax” to represent the minimum and maximum temperature. However, subscripts are not used in the Figure 2a and Table 3.
- Line 134: “Atlantic Multi-decadal Oscillation (AMO)” should be “Atlantic Multidecadal Oscillation (AMO)”.
- Line 171: The abbreviation of the number of cases sometimes uses N, sometimes uses n (see table 1), please keep the same.
- Lines 176, 186 and 187: Change the “observed” to the “actual” to keep the same in whole text.
- Line 187: Change the “1959-2014” to the “1959-2015”.
- Table 3: Give the UNIT of minimum temperature, and add the notes for “Tmin”.
- In discussion section 4.2, the authors pointed out that their new temperature reconstruction has a higher quality than regional records. It is better to compare the correlations between the existing (TS, CB and LB in figure 6) reconstructed temperature records and other temperature records (XXA and NH). A new table should be added.
- In figure 6: What is the R and Rlow. The same in figure 8.
- Line 268: Change the “The Atlantic Multidecadal Oscillation (AMO)” to the “The AMO”.
- Line 285: “……the actual April-September AMO index……”.
- Line 291: “……and the reconstructed AMO index (annual Sea Surface Temperature anomalies for the North Atlantic) from Gray (2001)……”.
- Line 294: “……the AMO index used in this study (thin line)……”.
- Figure 9: The arrows in Figure 9a are not clear, please replot it.
- Please double check and update the references. For example, Zhu et al., 2020a (CATENA) should be update to 2021.
Citation: https://doi.org/10.5194/cp-2021-2-CC1 -
AC2: 'Reply on CC1', Liangjun Zhu, 17 Apr 2021
Dear editor, dear referees,
We are so sorry for this late reply. Thank you for your constructive comments on our manuscript. The comments were valuable for helping us to revise and improve our paper, as well as the important guiding significance to our future research. We will do our best to improve the MS according your suggestions.
On behalf of the authors
Yours sincerely
Liangjun
Reply to CC:
General remarks:
The manuscript presented a multiple species method that significantly improves the accuracy of tree-ring-based climate reconstructions in areas with unstable growth-climate relationships. And also the linkage of the regional temperature of Changbai Mts in northeast China to aspects of the AMO index is one of the highlights of the MS. Overall, this paper is well-organized, and the results and conclusions are well supported by data. It provided a new idea for the development of tree-ring-based climate reconstruction. I suggest this MS being accepted by CP if the coauthors can address all reviews’ concerns and taking into account of comments in CPD (if they are reasonable).
Below I present some suggestions that the authors could find useful when revising the manuscript.
Specific comments:
- Figure 2: In the captions, the authors use the abbreviation “Tmin” and “Tmax” to represent the minimum and maximum temperature. However, subscripts are not used in the Figure 2a and Table 3.
Reponses: we will keep the same.
- Line 134: “Atlantic Multi-decadal Oscillation (AMO)” should be “Atlantic Multidecadal Oscillation (AMO)”.
Reponses: we will do so.
- Line 171: The abbreviation of the number of cases sometimes uses N, sometimes uses n (see table 1), please keep the same.
Reponses: we will do so.
- Lines 176, 186 and 187: Change the “observed” to the “actual” to keep the same in whole text.
Reponses: we will do so.
- Line 187: Change the “1959-2014” to the “1959-2015”.
Reponses: we will do so.
- Table 3: Give the UNIT of minimum temperature, and add the notes for “Tmin”.
Reponses: we will do so.
- In discussion section 4.2, the authors pointed out that their new temperature reconstruction has a higher quality than regional records. It is better to compare the correlations between the existing (TS, CB and LB in figure 6) reconstructed temperature records and other temperature records (XXA and NH). A new table should be added.
Reponses: we will add the table.
- In figure 6: What is the R and Rlow. The same in figure 8.
Reponses: we will add the notes.
- Line 268: Change the “The Atlantic Multidecadal Oscillation (AMO)” to the “The AMO”.
Reponses: we will do so.
- Line 285: “……the actual April-September AMO index……”.
Reponses: we will revise it.
- Line 291: “……and the reconstructed AMO index (annual Sea Surface Temperature anomalies for the North Atlantic) from Gray (2001)……”.
Reponses: we will do so.
- Line 294: “……the AMO index used in this study (thin line)……”.
Reponses: we will do so.
- Figure 9: The arrows in Figure 9a are not clear, please replot it.
Reponses: we will do so.
- Please double check and update the references. For example, Zhu et al., 2020a (CATENA) should be update to 2021.
Reponses: we will double check the references.
Citation: https://doi.org/10.5194/cp-2021-2-AC2
-
RC2: 'Comment on cp-2021-2', Anonymous Referee #2, 28 Feb 2021
The manuscript ‘Overcoming model instability in tree-ring-based temperature reconstructions using a multi-species method: A case study from the Changbai Mountains, northeastern China’ is a test of a multi-species approach might improve tree-ring based reconstructions of past climate. The work indicates that, indeed, multiple species can improve the potential for stability issues in reconstructing past climate from tree rings. The grammar is generally good; only a few places where it can be improved – please double check. There are a couple more steps in the analysis that would further strengthen the work.
The primary concern is that this approach is not as novel as stated in the manuscript. First, (García-Suárez et al. 2009) wrote about the value of a multi-species approach. Specifically, they wrote, “Apparent instability in the relations to climate of species like beech. However, we have found that combinations of species are better able to reconstruct climate than single species. Particularly robust are the reconstructions that use ash and beech simultaneously as predictors.” Soon after (Pederson et al. 2013) tested this with a greater number of species and developed a table showing the potential for using multiple species in a reconstruction and discussed findings that indicate species replication might stronger than chronology replication for the improvement and stability of reconstruction statistics. (Alexander et al. 2019) followed up that work by explicitly testing the hypothesis that increased species replication would improve temperature reconstructions. (Maxwell et al. 2011) has demonstrated this, too. Though not discussed in a similar way, the many drought atlases are mutli-species reconstructions and (Frank and Esper 2005) discuss the climate response of a multi-species high elevation network. There is a good amount of groundwork available in the literature for the manuscript developed here.
I am glad to see this kind of testing being replicated in other regions. What is new in this manuscript is that this is being tested in a different region and, especially, in a monsoonal climate. I am just surprised that one or a few of these earlier works are not acknowledged in this manuscript. It has been in the literature for at least 12 years.
To be clear, the author’s work here is of value to science because of the region and species. The should, however, acknowledge previous work on a multi-species approach.
Minor Comments
Line 19 – replace ‘greatly’ with ‘can’. It is not evident that all reconstructions are greatly reduced due to instability. End the sentence after the phrase paleoclimate reconstructions and start a new sentence with this remaining text, “This is most notable in areas…”
Line 20 – replace “an original” with “a test” as noted above, this is not exactly original.
Line 32 – “climates are” should be “climate is”
Line 37 – instead of “Tree rings, especially ring widths,”, one could write “Information derived from tree-ring records…”
Line 39 – instead of “, which” can replace with “and”
Line 41 – another citation to go along with Fritts is (Cook and Kairiukstis 1990).
Line 58 – “hypothesize” here should be “hypothesis”
Line 152 – an R of 0.824 is incredible! I appreciate the authors put in the figure with first differences. They should state the R of that relationship (if they have not. If I have missed it, I apologize).
Figure 3 – it looks like PK does a lot of the work here. It would be interesting to make a PCA of all chronologies to see which ones and which species contribute the most for each period tested.
Figure 4b – thank you for showing the first differences. Please note in the manuscript that the trend is doing most of the work with the R2 results.
Figure 4c - adding in uncertainty to 4c is recommended. One approach would be to calculate the root mean squared error of the model and show two levels of uncertainty. One or two times the root mean squared error, for example.
Line 195 – after and 1956-1982) it would be helpful to the readers if you add something like “according to how we defined warm/cold.
Figure 5 – suggestion: what are the spectral properties prior to 1980 vs the entire record? It would be interesting to see how much warming has or has not changed the spectral frequency.
Figure 6 - synchronization of temperatures and trends is a characteristic of global warming. Could highlight that fact in the text.
Line 272 – consider trying cross spectral analyses to see at what frequencies your recon and the AMO are in sync.
Line 315 – Given the general concern I have at the start of the review, rewrite the first sentence as, “We have tested how mixing tree species might improve the accuracy….”
References
Alexander, M. R., J. K. Pearl, D. A. Bishop, E. R. Cook, K. J. Anchukaitis, and N. Pederson. 2019. The potential to strengthen temperature reconstructions in ecoregions with limited tree line using a multispecies approach. Quaternary Research:1-15.
Cook, E. R., and L. A. Kairiukstis. 1990. Methods of Dendrochronology, Applications in the Environmental Sciences. Kluwer.
Frank, D., and J. Esper. 2005. Characterization and climate response patterns of a high-elevation, multi-species tree-ring network in the European Alps. Dendrochronologia 22:107-121.
García-Suárez, a. M., C. J. Butler, and M. G. L. Baillie. 2009. Climate signal in tree-ring chronologies in a temperate climate: A multi-species approach. Dendrochronologia 27:183-198.
Maxwell, R. S., A. E. Hessl, E. R. Cook, and N. Pederson. 2011. A multispecies tree ring reconstruction of Potomac River streamflow (950–2001). Water Resources Research 47:1-12.
Pederson, N., A. R. Bell, E. R. Cook, U. Lall, N. Devineni, R. Seager, K. Eggleston, and K. P. Vranes. 2013. Is an Epic Pluvial Masking the Water Insecurity of the Greater New York City Region? Journal of Climate 26:1339-1354.
Citation: https://doi.org/10.5194/cp-2021-2-RC2 -
AC3: 'Reply on RC2', Liangjun Zhu, 17 Apr 2021
Dear editor, dear referees,
We are so sorry for this late reply. Thank you for your constructive comments on our manuscript. The comments were valuable for helping us to revise and improve our paper, as well as the important guiding significance to our future research. We will do our best to improve the MS according your suggestions.
On behalf of the authors
Yours sincerely
Liangjun
Reply to Anonymous Referee #2:
The manuscript ‘Overcoming model instability in tree-ring-based temperature reconstructions using a multi-species method: A case study from the Changbai Mountains, northeastern China’ is a test of a multi-species approach might improve tree-ring based reconstructions of past climate. The work indicates that, indeed, multiple species can improve the potential for stability issues in reconstructing past climate from tree rings. The grammar is generally good; only a few places where it can be improved – please double check. There are a couple more steps in the analysis that would further strengthen the work.
The primary concern is that this approach is not as novel as stated in the manuscript. First, (García-Suárez et al. 2009) wrote about the value of a multi-species approach. Specifically, they wrote, “Apparent instability in the relations to climate of species like beech. However, we have found that combinations of species are better able to reconstruct climate than single species. Particularly robust are the reconstructions that use ash and beech simultaneously as predictors.” Soon after (Pederson et al. 2013) tested this with a greater number of species and developed a table showing the potential for using multiple species in a reconstruction and discussed findings that indicate species replication might stronger than chronology replication for the improvement and stability of reconstruction statistics. (Alexander et al. 2019) followed up that work by explicitly testing the hypothesis that increased species replication would improve temperature reconstructions. (Maxwell et al. 2011) has demonstrated this, too. Though not discussed in a similar way, the many drought atlases are mutli-species reconstructions and (Frank and Esper 2005) discuss the climate response of a multi-species high elevation network. There is a good amount of groundwork available in the literature for the manuscript developed here.
I am glad to see this kind of testing being replicated in other regions. What is new in this manuscript is that this is being tested in a different region and, especially, in a monsoonal climate. I am just surprised that one or a few of these earlier works are not acknowledged in this manuscript. It has been in the literature for at least 12 years.
To be clear, the author’s work here is of value to science because of the region and species. They should, however, acknowledge previous work on a multi-species approach.
Reponses: Thank you for your constructive comments on our manuscript. These comments are valuable for helping us to improve our paper. We will add those literature in to the background of our work and do our best to revise the MS according your valuable suggestions.
Minor Comments
Line 19 – replace ‘greatly’ with ‘can’. It is not evident that all reconstructions are greatly reduced due to instability. End the sentence after the phrase paleoclimate reconstructions and start a new sentence with this remaining text, “This is most notable in areas…”
Reponses: we will do so.
Line 20 – replace “an original” with “a test” as noted above, this is not exactly original.
Reponses: we will do so.
Line 32 – “climates are” should be “climate is”
Reponses: we will do so.
Line 37 – instead of “Tree rings, especially ring widths,”, one could write “Information derived from tree-ring records…”
Reponses: we will do so.
Line 39 – instead of “, which” can replace with “and”
Reponses: we will do so.
Line 41 – another citation to go along with Fritts is (Cook and Kairiukstis 1990).
Reponses: we will do so.
Line 58 – “hypothesize” here should be “hypothesis”
Reponses: we will do so.
Line 152 – an R of 0.824 is incredible! I appreciate the authors put in the figure with first differences. They should state the R of that relationship (if they have not. If I have missed it, I apologize).
Figure 3 – it looks like PK does a lot of the work here. It would be interesting to make a PCA of all chronologies to see which ones and which species contribute the most for each period tested.
Reponses: we will do so.
Figure 4b – thank you for showing the first differences. Please note in the manuscript that the trend is doing most of the work with the R2 results.
Reponses: we will do so.
Figure 4c - adding in uncertainty to 4c is recommended. One approach would be to calculate the root mean squared error of the model and show two levels of uncertainty. One or two times the root mean squared error, for example.
Reponses: we will do so.
Line 195 – after and 1956-1982) it would be helpful to the readers if you add something like “according to how we defined warm/cold.
Reponses: we will do so.
Figure 5 – suggestion: what are the spectral properties prior to 1980 vs the entire record? It would be interesting to see how much warming has or has not changed the spectral frequency.
Reponses: we will do so.
Figure 6 - synchronization of temperatures and trends is a characteristic of global warming. Could highlight that fact in the text.
Reponses: we will do so.
Line 272 – consider trying cross spectral analyses to see at what frequencies your recon and the AMO are in sync.
Reponses: we will do so.
Line 315 – Given the general concern I have at the start of the review, rewrite the first sentence as, “We have tested how mixing tree species might improve the accuracy….”
Reponses: we will do so.
Citation: https://doi.org/10.5194/cp-2021-2-AC3
-
AC3: 'Reply on RC2', Liangjun Zhu, 17 Apr 2021
Supplement
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 1,005 | 529 | 59 | 1,593 | 133 | 50 | 41 |
- HTML: 1,005
- PDF: 529
- XML: 59
- Total: 1,593
- Supplement: 133
- BibTeX: 50
- EndNote: 41
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1