The authors developed a new precipitation reconstruction in western Mediterranean using tree-ring series from five sites and two pine tree species. This reconstruction extends the length of the hydroclimatic records for over 500 years, and focuses on quantitative precipitation rather than drought index. Multiple precipitation datasets and critical growth period beyond fixed monthly aggregations looks useful for identifying robust climate signals and improving the explained variance of reconstruction. This well-replicated tree-ring width based precipitation reconstruction indeed provides us a baseline for evaluating hydroclimatic extremes and related ecosystem and water resource variability. But there are still some problems should be resolved before it is considered for publication.

Main concerns:

1. From the Title and Abstract, we expected to see sufficient evidences about “recent intensification of hydroclimatic extremes”, but there is few in Results and Discussion about it. Most of the results are about the reconstruction development and comparisons with rogation-derived drought indices.
2. From Fig.7, we can see a little bit increase in frequency and intensity of wet and drought extremes after 2000 CE, but it is not proven. The variance of the reconstructed precipitation should be a good indicator to show it. Additionally, the running RABR should be shown in Fig. 6 to help evaluating the impacts of sample depth and inter-series correlations on variance of precipitation.
3. Line 340-407, there are so many results (or discussion) about the comparisons between precipitation reconstruction and rogation-derived drought event, even including some correlations (Line 361, 366, 373,…), but no one figure and table to show these results. By using the method of giving the examples (a lot of “For instance”) to show the alignment of tree-ring based precipitation reconstruction with rogation ceremony records is not sufficient to support the precipitation influence on ecosystem and society, as maybe more disagree years happened.
4. The organization of the manuscript is poor. (1) First, there are some repeat information in Discussion part. Such as, Line 416-419, “The resulting regional residual chronology demonstrates strong internal coherence (Rbar = 0.273), similar to those from other Mediterranean hydroclimatic reconstructions such as Esper et al. (2021); Klippel et al. (2018); Tejedor et al. (2016) (Rbar = 0.28; 0.31; 0.29). Our chronology retains a reliable common signal back to 1505 CE, as indicated by an SSS value consistently exceeding the 0.85 threshold (Buras, 2017; Cook and Kairiukstis, 1990; Wigley et al., 1984)”, which is repeat with Results. (2) Line 489-492, “Visual inspection of Fig. 7 highlights periods characterized by distinct wet or dry conditions, as well as shifts in variability. For instance, the reconstruction identifies notable drought periods, with years like 1526, 1527, 1879, 1931, 2005 and 2012 falling below the 5th percentile, and exceptionally wet periods, with years such as 1534, 1546, 1575, 1645, 1716, 1940, 2010 and 2013 exceeding the 95th percentile.” should be represented in Results part, as actually there is no discussion about it here. (3) The paragraph of Line 478-487 should be moved to the second paragraph from bottom as a summary to highlight the key aspects of this paper. Now, it is in the middle of discussion and disturbed the discussion about reconstructed precipitation.

Minor problems:

1. About the Title, “precipitation extremes” is more exact than “hydroclimatic extremes” to highlight the study gap.
2. Line 89, “June is the month with the highest pluviosity, followed by May” is inconsistent with Fig.1B, which shown precipitation in May is the highest, followed by April.
3. Tree-ring series from five sites and two species were used for developing one chronology. How about the correlations between sites and species, and the uniformity of five chronologies at high frequency and low frequency variability? These informations could be plotted in Supplementary materials.
4. Line 412-413, “capture the critical moisture accumulation phase influencing annual growth (late summer, autumn, winter and spring/early summer),” how to understand the autumn and winter precipitation influence on tree radial growth considering trees dormancy in winter.
5. Line 445, “comparing our Fig. 6 extremes with their findings”, should be Fig. 7, right?
6. Line 472-474, “The lack of strong correlation in earlier periods… influenced by biological memory, …” seems inconsistent with the feature of RES chronology with “pre-whitening to reduce autocorrelation”.
7. 5 is not clear and takes up too much space.

The authors developed a new precipitation reconstruction in western Mediterranean using tree-ring series from five sites and two pine tree species. This reconstruction extends the length of the hydroclimatic records for over 500 years, and focuses on quantitative precipitation rather than drought index. Multiple precipitation datasets and critical growth period beyond fixed monthly aggregations looks useful for identifying robust climate signals and improving the explained variance of reconstruction. This well-replicated tree-ring width based precipitation reconstruction indeed provides us a baseline for evaluating hydroclimatic extremes and related ecosystem and water resource variability. But there are still some problems should be resolved before it is considered for publication.

Main concerns:

1. From the Title and Abstract, we expected to see sufficient evidences about “recent intensification of hydroclimatic extremes”, but there is few in Results and Discussion about it. Most of the results are about the reconstruction development and comparisons with rogation-derived drought indices.
2. From Fig.7, we can see a little bit increase in frequency and intensity of wet and drought extremes after 2000 CE, but it is not proven. The variance of the reconstructed precipitation should be a good indicator to show it. Additionally, the running RABR should be shown in Fig. 6 to help evaluating the impacts of sample depth and inter-series correlations on variance of precipitation.
3. Line 340-407, there are so many results (or discussion) about the comparisons between precipitation reconstruction and rogation-derived drought event, even including some correlations (Line 361, 366, 373,…), but no one figure and table to show these results. By using the method of giving the examples (a lot of “For instance”) to show the alignment of tree-ring based precipitation reconstruction with rogation ceremony records is not sufficient to support the precipitation influence on ecosystem and society, as maybe more disagree years happened.
4. The organization of the manuscript is poor. (1) First, there are some repeat information in Discussion part. Such as, Line 416-419, “The resulting regional residual chronology demonstrates strong internal coherence (Rbar = 0.273), similar to those from other Mediterranean hydroclimatic reconstructions such as Esper et al. (2021); Klippel et al. (2018); Tejedor et al. (2016) (Rbar = 0.28; 0.31; 0.29). Our chronology retains a reliable common signal back to 1505 CE, as indicated by an SSS value consistently exceeding the 0.85 threshold (Buras, 2017; Cook and Kairiukstis, 1990; Wigley et al., 1984)”, which is repeat with Results. (2) Line 489-492, “Visual inspection of Fig. 7 highlights periods characterized by distinct wet or dry conditions, as well as shifts in variability. For instance, the reconstruction identifies notable drought periods, with years like 1526, 1527, 1879, 1931, 2005 and 2012 falling below the 5th percentile, and exceptionally wet periods, with years such as 1534, 1546, 1575, 1645, 1716, 1940, 2010 and 2013 exceeding the 95th percentile.” should be represented in Results part, as actually there is no discussion about it here. (3) The paragraph of Line 478-487 should be moved to the second paragraph from bottom as a summary to highlight the key aspects of this paper. Now, it is in the middle of discussion and disturbed the discussion about reconstructed precipitation.

Minor problems:

1. About the Title, “precipitation extremes” is more exact than “hydroclimatic extremes” to highlight the study gap.
2. Line 89, “June is the month with the highest pluviosity, followed by May” is inconsistent with Fig.1B, which shown precipitation in May is the highest, followed by April.
3. Tree-ring series from five sites and two species were used for developing one chronology. How about the correlations between sites and species, and the uniformity of five chronologies at high frequency and low frequency variability? These informations could be plotted in Supplementary materials.
4. Line 412-413, “capture the critical moisture accumulation phase influencing annual growth (late summer, autumn, winter and spring/early summer),” how to understand the autumn and winter precipitation influence on tree radial growth considering trees dormancy in winter.
5. Line 445, “comparing our Fig. 6 extremes with their findings”, should be Fig. 7, right?
6. Line 472-474, “The lack of strong correlation in earlier periods… influenced by biological memory, …” seems inconsistent with the feature of RES chronology with “pre-whitening to reduce autocorrelation”.
7. 5 is not clear and takes up too much space.

Overview

This paper describes a regression-based 500+ year reconstruction of seasonal-total precipitation for the Iberian Range of eastern Spain from tree-ring data of two Pinus tree species and five sites collected (recently updated) by the authors. The reconstruction stands out compared with previous reconstructions of dendroclimatic variables for its great length, strength of calibration signal, and calibration with precipitation rather a drought index. A key finding is recent intensification of hydroclimatic extremes consistent with climate change projections. The reconstruction is touted as a baseline for evaluating ecosystem resilience and water resource vulnerability. The paper is well organized, clearly written and contains excellent and appropriate graphics and tables. The paper could benefit from clarification of a few points, but is overall a strong contribution.

Major comments

1. The major conclusion, about the intensification of hydrologic extremes, is quite convincing. Fig 2 shows a large increase of sample size with time, which is typical of may tree-ring chronologies. Usually it is a good idea in these cases to implement “variance stabilization” in chronology development: adjust the time-varying chronology variance for the changing sample size (number of trees or cores). If variance stabilization was used, you should report that in the methods. If not, I suggest doing a quick check to see if it makes a difference to conclusions. Actually, it may strengthen your argument, because variance stabilization generally shrinks the chronology variance early on relative to the variance at full sample size.

2. Distinction of “hydroclimate” from “precipitation.” Water availability, which is mentioned in several places in the paper, is a function of not just precipitation but also of evapotranspiration. The authors claim that a strong point of this paper compared with previous works is that it addresses precipitation rather than PDSI or some other type of drought index. This may be a strong point in terms of synoptic climatology and delivery of precipitation to the region, but not necessarily in terms of water resources availability or ecosystem stress. It seems likely that the tree growth of these drought sensitive species would not better reflect the combined stress of low precipitation and high temperature than the stress of low precipitation alone. A natural question is whether the impressive calibration strength (R-squared) for the reconstruction model might not be even more impressive for some sort of drought index that incorporating influence of evapotranspiration. I think the paper could benefit from some comparison statistics. If checking against and index such as SPEI or PDSI, the standard rather than the residual chronology might actually be worth looking at because the strong temperature trend associated with regional warming is a low-frequency signal that could have been removed by autoregressive modeling during standardization. Such additional analysis, if done, would mainly be a discussion point, as readers may wonder why the calibration of a drought-sensitive chronology is not stronger with SPEI, say, than with precipitation.. 

3. I appreciate the rigorous daily climate analysis, which resulted in identification of an optimum window, but wonder about the sampling variability and whether you are making too much of the difference in a annual window and a 320-day window.  In the minor comments I bring this up again and suggest adding maybe a sentence about this issue in the discussion. Almost certainly there is no significant different in correlation for the selected 320-day window and the highest 365-day (annual) window. 

Minor comments

1. Title: Consider substituting “precipitation” for “hydroclimate,” in the title, because precipitation is what has been reconstructed. Precipitation is just one aspect of hydroclimate. Runoff and streamflow are the sum of net precipitation (P-ET) and are what I think of as key components of hydroclimate, though this is an arguable distinction.

2. L77. I disagree that precipitation provides a “more direct measure of past water availability than some drought index. Net precipitation (P-ET) is one possible drought index, and is actually more relevant to water availability the precipitation alone. Of course, precipitation is more directly link weather delivery systems that P-ET, which depends on vegetation and other land surface factors. 

3. L90. "pluviosity" is an overblown word when used here for “precipitation, ” which explicitly is what is shown in the climate diagram, and what is measured in a rain gauge.

4. L90. On the climate diagram I see May followed by April, not June followed by May, as the months of highest precipitation

5. L92. Looks to me like Feb is a drier month month than Aug. The statement about July and Aug being driest month applies only if just considering summer .

6. L114. “Campaigns at…”

7. L124. Standardization description needs a bit more information. Was the ratio or difference method used for converting ring widths to indices?  Was the site chronology computed as an arithmetic mean or biweight mean of core indices?  Was ariance stabilization applied to adjust variance changes in site chronology to time-varying sample size (see Major comments)? Did you compute both standard and residual versions of the chronology, and why did you select the residual version for the reconstruction

8. L 128. How many trees are represented by the 173 series? I’m assuming probably more than one core per sampled tree.

9. L 153. “ with12-” ….insert a space

10. L159. “robust” -- I assume the daily window selected is robust to selected segment of the climate-chronology overlap (e.g., approximately same day window if analysis repeated on separate halves of the record)

11. L 163. “linear transfer function model” --- the statistical reconstruction method could be described more directly as “simple linear regression of the target predictand on the site chronology.”

12. L208 “Such a low AR1 value indicates that the standardization effectively removed most of the tree-ring memory persistence inherent in tree growth, yielding a time series suitable for robust correlation analysis with external environmental variables, such as climate.” Yes, as long as the target predictand also has no autocorrelation. Is that so for precipitation in this region? Also, in regression, analysis of residuals check usually included first order autocorrelation of regression residuals (e.g., by Durbin Watson statistic). It is assumed in regression that there is no significant lag-1 autocorrelation in the regression residuals.

13. Fig 1 caption. “grid cells” – would help the interpretation to indicate the resolution of grid for the precipitation. .

14. Fig 2 caption. Specify that the “number of samples” is cores or trees.

15. Fig 3. For consistency, give the grid resolution for all of the datasets in the labels along y axis.

16. Fig 4 caption. The “red dashed line” is not the residual chronology, but the reconstruction based on it.

17. Fig 5 caption. There seems to be a wide range of day windows with high correlation, or with dark red shading. How much lower is the correlation for the "best" annual (365-day) period that the correlation for selected 320-day window (r=0.749)? Could this just be a result of sampling variability? Perhaps you can add a sentence or two on this in the discussion. In a related question I wondered whether the same 320 day window is identified if use different sub-periods (e.g., first and last halves) of the record.