ENSO Signals Recorded by Ash Tree Rings in Iberian Riparian Forests

Round 1

Reviewer 1 Report

ENSO affects global climate at inter-annual to multi-decadal scales, but it is need to study how the ENSO effect on the regional climate. In this manuscript, the authors discuss the teleconnection between ENSO and the regional climate using tree ring data. It is important to understanding the climate and atmospheric dynamics. However, I think that it need to add the spatial correlation analysis between tree-ring width index and SST, so as to better understand the ENSO how to effects the climate in the study area. In addition, why just choose 12 years to calculate the correlation coefficient between ash growth and mean water table depth? Which 12 years?

Author Response

• We correlated ash growth with table depth for 12 years because we only had this period available for these data. In the revised ms. we have indicated this 12-year period corresponding to 2006-2017, which was already presented in the original ms.
• As requested we have calculated the spatial correlations with SSTs. However, since we are dealing with a teleconnection related to the ENSO and driven by changes in STT across the Equatorial Pacific, the correlations are difficult to interpret but they are fully congruent with the proposed teleconnection. We have added these correlations in the revised ms. as suggested by the reviewer.

Reviewer 2 Report

This study presents correlation analyses in the attempt to establish a link between ENSO in the equatorial Pacific, and river flow and riparian forest growth in the Ebro basin. The study is overall interesting and could become a nice contribution to the literature on the ENSO-Europe teleconnection. However, I have some comments that I ask the author to consider for a possible revision of the current manuscript.

My main concern is that it is known that ENSO teleconnection with Europe depends, for the warm phase of ENSO, on the “type” of El Niño. The changed teleconnection has been illustrated by Graf and Zanchettin (2012, https://doi.org/10.1029/2011JD016493). This seems to be neglected in the paper, and at least a discussion on this relevant aspect should be provided in the manuscript, especially as far as a tendency toward more “Central Pacific” (or CP) events has been observed in the recent decades, possibly linked with global warming.

Also, the ENSO teleconnection with European hydroclimate is modulated by the state of the Atlantic, as illustrated for instance in the review paper by Rodriguez-Fonseca et al. (2016, https://doi.org/10.3390/atmos7070087). This should also be discussed more in detail, as ENSO and NAO, for instance, are also interlinked. In this regard, the manuscript does not provide any discussion about how the remote forcing of ENSO could affect hydroclimate in the study region. I recommend a stronger introduction on this relevant aspect.

Then, I wonder if correlation analysis is the best approach to identify the teleconnection, as I expect it embeds a lot of noise/effect of covariates especially under neutral ENSO conditions and, as suggested by the author in the discussion section (line 234-235), correlation only captures linear relations. This may well not be the case and one should account also for the different manifestations of the warm phase of ENSO. A composite analysis could give clearer results. In fact, despite significant, a correlation of 0.4 means that only about 16% of normalized variance is shared between the two time series. Anyway, as a positive aspect of the analysis, it appears that the author accounted for serial dependence in the data. But, I wonder if the original time series were detrended before the correlation analysis, as typically accounting for autocorrelation changes the significance, not the value of the correlation coefficient. Despite the long-term trend contained in the series appears small (if any, looking at Figure 2), I would recommend removing all trends as part of data preprocessing since ENSO is typically relevant for interannual variability. In any case, a more detailed explanation of the statistical analyses would be welcomed.

Line 184-186: this seems to go against the conclusion of a significant effect of ENSO on western European climate, unless there is a threshold effect or, as I suggest above, the different types of ENSO are considered (these events are all “classical” El Niños). Again, a composite analysis differentiating between CP and classical events could contribute to get clearer results. Otherwise, I recommend some discussion on the “two types” of El Niño.

Line 223: La Niña

Line 226: in the study of Mediterannean areas

Line 227: “climate shifts” can be interpreted as sudden, permanent changes in the mean state. Is this what the author means?

Author Response

Comments and Suggestions for Authors

This study presents correlation analyses in the attempt to establish a link between ENSO in the equatorial Pacific, and river flow and riparian forest growth in the Ebro basin. The study is overall interesting and could become a nice contribution to the literature on the ENSO-Europe teleconnection. However, I have some comments that I ask the author to consider for a possible revision of the current manuscript.

• I thank you for your positive comments on the study.

My main concern is that it is known that ENSO teleconnection with Europe depends, for the warm phase of ENSO, on the “type” of El Niño. The changed teleconnection has been illustrated by Graf and Zanchettin (2012, https://doi.org/10.1029/2011JD016493). This seems to be neglected in the paper, and at least a discussion on this relevant aspect should be provided in the manuscript, especially as far as a tendency toward more “Central Pacific” (or CP) events has been observed in the recent decades, possibly linked with global warming.

Also, the ENSO teleconnection with European hydroclimate is modulated by the state of the Atlantic, as illustrated for instance in the review paper by Rodriguez-Fonseca et al. (2016, https://doi.org/10.3390/atmos7070087). This should also be discussed more in detail, as ENSO and NAO, for instance, are also interlinked. In this regard, the manuscript does not provide any discussion about how the remote forcing of ENSO could affect hydroclimate in the study region. I recommend a stronger introduction on this relevant aspect.

Then, I wonder if correlation analysis is the best approach to identify the teleconnection, as I expect it embeds a lot of noise/effect of covariates especially under neutral ENSO conditions and, as suggested by the author in the discussion section (line 234-235), correlation only captures linear relations. This may well not be the case and one should account also for the different manifestations of the warm phase of ENSO. A composite analysis could give clearer results. In fact, despite significant, a correlation of 0.4 means that only about 16% of normalized variance is shared between the two time series. Anyway, as a positive aspect of the analysis, it appears that the author accounted for serial dependence in the data. But, I wonder if the original time series were detrended before the correlation analysis, as typically accounting for autocorrelation changes the significance, not the value of the correlation coefficient. Despite the long-term trend contained in the series appears small (if any, looking at Figure 2), I would recommend removing all trends as part of data preprocessing since ENSO is typically relevant for interannual variability. In any case, a more detailed explanation of the statistical analyses would be welcomed.

• Note that tree-ring width data were standardize and detrended so they show no long term trend. Neither the SOI nor the flow series shown in Fig. 2 presented any significant trend so I think it’s sensible keeping the analyses without detrending them. Nevertheless, I have indicated their trends in the revised ms. I am fully aware that accounting for the autocorrelation changes the significance, not the value of the correlation value. Note that the analyses were focused on the year-to-year variability as you noted and indicated.
• Following a suggestion made by reviewer 1 we calculated spatial correlations between ash ring-width indices and January SSTs, and they were strong and positive thus supporting our analyses based on the ENSO. Further, they are in agreement with El Niño classic events centered in the Equatorial Pacific, towards the coasts of Southern America.
• Regarding your comments on the ENSO, I have discussed the non-stationary nature of its links with European and Mediterranean rainfall as suggested to better show hoe the ENSO could impact rainfall and river flow in the study area (I have cited the ref. by Rodriguez-Fonseca et al. 2016). I have also discussed on the role played by different El Niño (CP vs. classical) events on teleconnections over Europe. Nevertheless, note that this is not a climatic study but, instead, it is focused on climate impacts on riparian forests, a key component of river ecosystems (the ms. is submitted to the journal Water, a journal “on water science and technology, including the ecology and management of water resources”). Therefore, I prefer keeping the climate part not too long or detailed. I have also better introduced forcings involved on ENSO-hydroclimate relationships.

Line 184-186: this seems to go against the conclusion of a significant effect of ENSO on western European climate, unless there is a threshold effect or, as I suggest above, the different types of ENSO are considered (these events are all “classical” El Niños). Again, a composite analysis differentiating between CP and classical events could contribute to get clearer results. Otherwise, I recommend some discussion on the “two types” of El Niño.

• I added the suggested discussion. I also observed that the correlation between ash ring-width indices and the SOI was higher considering only EP El Niño events than considering CP events, which are supposed to influence more strongly winter conditions over Europe (at least temperature). This could be caused because during the last decades EP events are less frequent but very intense. Nevertheless, longer tree growth series would be needed to test these ideas, but, regrettably, most riparian forests do not have old trees. The presented series are too short to perform strong composite analyses.

Line 223: La Niña

Line 226: in the study of Mediterannean areas

• These two mistakes were corrected.

Line 227: “climate shifts” can be interpreted as sudden, permanent changes in the mean state. Is this what the author means?

• I mean transitory changes or shifts; this was rephrased by adding the adjective “transitory”.

Reviewer 3 Report

Generally, this manuscript is well organized with meaningful scientific implications and well written in English. The research hypothesis, the sampling design, the field observation and measurements, are all up to the criterion and standard with a sound scientific base. Thus, this manuscript is suitable for publication in the journal without much revision. 

I only have a minor concern about the method part.  In the ‘Abstract’, the author introduced a chronology of ash trees was built to quantify ‘radial growth rates’. But in the method part, the author has not well explained how to calculate radial growth rates. Only a site mean series of ring-width indices was built. In my point of view, ring-width indices and radial growth rates are two different things. The author need to give more explanations about this issue. 

Author Response

Generally, this manuscript is well organized with meaningful scientific implications and well written in English. The research hypothesis, the sampling design, the field observation and measurements, are all up to the criterion and standard with a sound scientific base. Thus, this manuscript is suitable for publication in the journal without much revision. 

• I thank you for your positive comments.

I only have a minor concern about the method part.  In the ‘Abstract’, the author introduced a chronology of ash trees was built to quantify ‘radial growth rates’. But in the method part, the author has not well explained how to calculate radial growth rates. Only a site mean series of ring-width indices was built. In my point of view, ring-width indices and radial growth rates are two different things. The author need to give more explanations about this issue. 

• I agree with your remark. I have replaced the term “growth rate” by “ring-width or growth indices” throughout the ms., excepting when dealing with raw ring-width data (in mm), which can be considered growth rates.

Round 2

Reviewer 2 Report

The author replied to all my questions and the revised version is publishable.