the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 May 2026
Technical note: Method effects on isotope-based inference of apple tree water sources
Abstract. Stable hydrogen and oxygen isotopes are widely used to trace plant water sources, but extraction and analytical choices can bias interpretation. Using an apple orchard on the Shandong Peninsula as a field example, we compared cryogenic vacuum extraction (CVE) with centrifugation, laser spectroscopy with isotope-ratio mass spectrometry (IRMS), and tested how apparent xylem δ²H offsets affect MixSIAR source apportionment. Across plant organs, water became progressively enriched from roots and branches to leaves and fruit. In paired branch samples measured by IRMS, centrifugation yielded δ²H values about 10‰ higher than CVE, whereas δ¹⁸O differences were small (≈1–2‰). Laser spectroscopy produced systematically higher δ¹⁸O than IRMS for plant extracts (typically 1–3‰), while non-plant source waters agreed closely between instruments. A δ²H offset correction shifted inferred uptake from shallow to deeper soil water. These results show that methodological choices can alter isotope-based inference of plant water sources and should be explicitly evaluated and reported.
- Preprint
(1066 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2026-2258', Christina Hackmann, 18 Jun 2026
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-2258/egusphere-2026-2258-RC1-supplement.pdfCitation: https://doi.org/
10.5194/egusphere-2026-2258-RC1 -
AC1: 'Reply on RC1', Ying Zhao, 08 Aug 2026
Response to Referee 1 - Christina Hackmann
General assessment
The referee considers the Technical Note timely and potentially useful, particularly because it integrates extraction-method, analytical-method, and source-apportionment sensitivity analyses in one field study.
Author response: We appreciate this positive assessment. We will preserve the integrated technical-note structure while strengthening reproducibility, uncertainty reporting, and the distinction between observation and mechanism. We will consistently refer to an apparent or method-dependent isotope offset unless a causal mechanism is directly supported.
Comment 1 - LMWL definition and uncertainty
Please report the precipitation sample number and temporal coverage, the LMWL parameters, whether samples were event based or composite, whether amount weighting was used, and how LMWL uncertainty may propagate into LC-excess and the estimated δ²H deficit.
Author response: We agree. The revised Methods will report the exact number of precipitation samples or events, collection dates, sampling basis (individual event, daily, or composite), whether precipitation-amount weighting was used, the LMWL equation, R², and uncertainty in the slope and intercept. We will state explicitly that the precipitation data were used to construct the LMWL and were not entered as a MixSIAR source endmember. We will evaluate how uncertainty in the LMWL parameters propagates into LC-excess and the empirical δ²H-offset estimate. The approximately +8‰ adjustment will not be presented as a uniquely correct value; it will be evaluated as one sensitivity scenario alongside 0, +5, and +10‰.
Comment 2 - Hydrological representativeness of the five-day campaign
Please clarify whether 8-12 July 2021 was dry, wet, or typical and provide concise antecedent precipitation, irrigation, soil-moisture, and evaporative-demand context.
Author response: We agree. We will describe the 8-12 July 2021 campaign as a short, intensively sampled hydrological snapshot rather than as representative of the full growing season. Using the available field and meteorological records, we will report antecedent precipitation, irrigation history, soil-water content, and atmospheric evaporative demand where available. We will characterize the period as dry, wet, typical, water-stressed, or well-watered only if those records support such a classification; otherwise, we will state the limitation explicitly. The Discussion will explain that hydrological state may affect sap availability, the balance between mobile and tissue-associated water pools, extraction performance, and the inferred ecohydrological response.
Comment 3 - Other extraction approaches
Please acknowledge other soil- and plant-water extraction approaches, place the findings in the broader comparison literature, and avoid extrapolating the present results to all extraction systems.
Author response: We agree. We will add a concise, non-speculative paragraph placing the present CVE-centrifugation comparison in the broader extraction-method literature, including pressure-chamber extraction, direct-vapour equilibration or analysis, and other mechanical or thermal approaches. We will consider the comparative studies suggested by the referee (Orlowski et al., 2016; Millar et al., 2018; Zuecco et al., 2020; Wen et al., 2023; and Jeon et al., 2024). We will explicitly state that the direction and magnitude of the differences observed here cannot be generalized to all extraction systems, tissues, species, or water contents, although the comparative and sensitivity-analysis framework can be applied more broadly.
Comment 4 - Species limitation
Please acknowledge that the test was limited to apple trees and discuss, without speculation, whether other crops or forest species could yield similar or different results.
Author response: We agree. We will add a limitations paragraph stating that this study evaluates one woody crop species at one site and during one short hydrological window. Species differences in xylem anatomy, tissue water content, storage, sap chemistry, and hydraulic status may change both the water pools recovered and the degree of analytical interference. We will therefore present apple as a case study and call for comparable paired-method and offset-sensitivity tests across crop and forest species.
Responses to the supplementary comments attached to RC1
Supplementary comment 1 - Method overview table
Add a table showing each compartment, extraction method, analytical method, and preferably sample number.
Author response: Agreed. We will add a table cross-tabulating sample compartment, measured depth interval (0-20, 20-40, 40-60, 60-80, and 80-100 cm for soil water), extraction method, analytical platform, pairing structure, and sample number. Exact counts will be verified against the laboratory records before revision. The table will also distinguish measurements made on aliquots of the same extract from comparisons based on matched but separately collected material.
Supplementary comment 2 - Figure citations
The manuscript lacks references to the figures.
Author response: Agreed. Every figure will be cited in numerical order at the relevant point in the Results and Methods. Captions will be rewritten to stand alone and to define extraction and analytical-method abbreviations consistently.
Supplementary comment 3 - Heating method
How were 205 °C and 100 °C reached in the cryogenic extraction line?
Author response: We will specify the exact heating configuration used in the cryogenic extraction line from the laboratory records, including how the temperature-controlled silicone-oil heating unit reached and maintained 205 °C for soil and 100 °C for plant tissues, while the receiving vial was cooled with liquid nitrogen. We will also report temperature monitoring, extraction duration, vacuum threshold, and the recovery criterion in sufficient detail for reproduction.
Supplementary comment 4 - Basis for the ~10‰ and 1-3‰ differences
Were these estimates visual, averages, or statistical results?
Author response: We agree that the present wording is insufficient. The revised manuscript will report paired sample numbers, mean or median paired differences, 95% confidence intervals, and effect sizes. For matched data, individual-pair plots and Bland-Altman agreement analyses will replace or supplement regressions and pooled means. The text will identify exactly which sample types each estimate applies to.
Supplementary comment 5 - Results statement in Methods/Results
A sentence about spectral interference needs a reference and may belong in the Discussion.
Author response: Agreed. The sentence interpreting the pattern as consistent with organic spectral interference will be moved from the descriptive Results to the Discussion and supported by West et al. (2011), Penna et al. (2012), and Cui et al. (2021). Results will report only observed cross-instrument differences and quality-control outcomes.
Supplementary comment 6 - Data used for source apportionment
Which extraction and analytical data were used in MixSIAR?
Author response: We will state this explicitly. The plotted MixSIAR analysis used branch xylem water as the mixture and three source endmembers: shallow soil water (<60 cm, obtained by pooling the measured 0-20, 20-40, and 40-60 cm intervals), deeper soil water (>60 cm in this study, represented by the measured 60-80, and 80-100 cm intervals), and groundwater. Precipitation was used only to construct the LMWL and was not entered into MixSIAR. Irrigation water was also not included as a separate endmember because it reaches the trees after infiltration and mixing within the soil profile and is therefore represented within the soil-water endmembers. Accordingly, the model did not estimate a separate irrigation-water contribution. We will report the extraction and analytical platform used for every mixture and source input and provide the exact model-input table and script.
Supplementary comment 7 - “Sap-targeted methods”
Add an example such as centrifugation.
Author response: Agreed. We will revise the phrase to “sap-targeted methods, such as centrifugation or pressure-chamber extraction.”
Supplementary comment 8 - Scope of the δ²H effect
Does the δ²H bias apply generally or only to particular sample types?
Author response: We will restrict this statement to the evidence available here: matched apple branch xylem samples. We will avoid presenting the observed difference as a universal CVE bias and will instead use terms such as ‘branch-specific extraction-dependent δ²H difference’ or ‘apparent method-dependent offset’.
Supplementary comment 9 - Meaning of post-processing
What is meant by post-processing?
Author response: We will replace this general term with a precise description of the spectral-quality-control procedure: software-based inspection of absorption spectra and contamination diagnostics, the flagging threshold used, the number of samples flagged, and whether flagged samples were corrected, remeasured, or excluded. If complete diagnostics were not archived, we will state that limitation explicitly.
Supplementary comment 10 - Consequences of analytical differences
Rephrase the statement so that it recognizes the risk of flawed ecohydrological conclusions, not merely meaningful differences.
Author response: Agreed. We will state that method-dependent isotope shifts of this magnitude can change the geometry of the mixing space and may lead to incorrect conclusions about source contributions if they are ignored.
Supplementary comment 11 - Origin of the offset
Should a study-specific δ²H offset simply be accepted, or should its origin be investigated?
Author response: We agree that a study-specific offset should neither be accepted as a universal correction constant nor interpreted as evidence that one adjusted scenario is the true solution. The approximately +8‰ adjustment will be described as an empirical, study-specific sensitivity scenario; its calculation from the original dataset will be reported transparently after verification. We will compare 0, +5, +8, and +10‰ scenarios and emphasize that identifying the physical or methodological origin of the offset requires targeted controls, such as isotopically characterized reference waters, water-content manipulations, sap-targeted extraction, and species- and tissue-level comparisons. The Technical Note will therefore distinguish diagnostic sensitivity analysis from mechanistic correction.
Supplementary figure comment - Figure 1
Add panel labels, improve panel arrangement, pair corresponding methods, use a clearer design, add raw points, and check whether the rain samples are identical across panels.
Author response: Agreed. Figure 1 will be redrawn with labelled panels, aligned axes, larger plotting areas, paired ordering of method combinations, and raw observations overlaid on summary distributions. The precipitation sample set used in every panel will be reconciled; if different subsets are necessary, this will be stated explicitly in the caption.
Supplementary figure comments - Figures 2 and 3
Increase readability, use consistent design, and add raw data points.
Author response: Agreed. Symbol and label sizes will be increased, design and naming will be harmonized with Figure 1, and raw observations will be shown where sample density permits.
Supplementary figure comments - Figures 4 and 5
Compare the same samples pairwise.
Author response: Agreed. The revised figures will show the matched observations explicitly and report the corresponding differences. Where the matching structure is valid, a Bland-Altman plot will be included in the main text or Supplement to show the mean difference and limits of agreement.
Supplementary figure comment - Figure 6 and irrigation
Why is irrigation negligible although its isotope values overlap with plant water? Were different data used?
Author response: The referee correctly identifies an ambiguity created by the current presentation. The actual Figure 6 analysis was a three-endmember model comprising shallow soil water (<60 cm; pooled 0-20, 20-40, and 40-60 cm samples), deeper soil water (>60 cm; pooled 60-80, and 80-100 cm samples), and groundwater. Precipitation was used only to construct the LMWL, and irrigation water was not entered as a separate model endmember because it first infiltrates and mixes within the soil profile before plant uptake. Its influence is therefore represented within the measured soil-water endmembers. The model did not estimate a separate irrigation contribution, so any wording or labelling that suggests a negligible irrigation posterior is misleading and will be corrected in the Methods, Results, Figure 6, and its caption. We will also explain that including isotopically overlapping irrigation water and soil water as simultaneous direct endmembers would not match the uptake process represented here and could reduce source identifiability.
Supplementary formal comments
Use subscripts correctly and correct the Allen and Kirchner (2022) DOI.
Author response: Agreed. Chemical subscripts will be formatted correctly throughout. The Allen and Kirchner DOI will be corrected to https://doi.org/10.1002/hyp.14483, and all citations/DOIs will be checked.
Citation: https://doi.org/10.5194/egusphere-2026-2258-AC1
-
AC1: 'Reply on RC1', Ying Zhao, 08 Aug 2026
-
RC2: 'Comment on egusphere-2026-2258', Anonymous Referee #2, 24 Jul 2026
General comments:
This Technical Note examines a practical methodological problem in isotope-based studies of plant water sources. Using an apple orchard as a case study, the authors compare water-extraction and isotope-measurement methods and then ask whether the resulting differences affect source attribution in a mixing model. This is a suitable topic for a Technical Note in HESS. I also appreciate that the manuscript does not present either CVE or centrifugation as universally superior.
My main concern is not that the experiment needs to be repeated, but that the existing experiment and model analysis are not yet reported in enough detail. It is difficult to judge whether the method effects are consistent among paired samples, and the information provided is insufficient to reproduce Fig. 6. The sensitivity analysis suggests that a δ²H shift of roughly this magnitude can substantially change the model output, but it does not show that the adjusted scenario represents the true uptake depth of the trees. The paper could make a useful contribution if the authors provide fuller reporting of the paired comparisons, analytical quality control, and MixSIAR analysis, and narrow the conclusions accordingly. I therefore recommend major revision rather than additional field sampling.
Specific comments:
1. The approximately 10‰ difference in branch-water δ²H reported in Sect. 3.1 is one of the central empirical results, but the figures and text do not yet show how stable this difference is. Please report the actual number of paired samples and explain how material from the same branch was divided between CVE and centrifugation. It would help to show the difference for each pair and report a paired effect estimate with an uncertainty interval. This would be more informative than regression lines or pooled group means alone. If the samples span different plots, trees, or dates, please also indicate whether the method difference is broadly consistent among these groups.
2. The laser spectroscopy-IRMS comparison needs some additional methodological information. Please report how many samples of each plant material were measured with both instruments and whether the measurements used aliquots of the same extract or separately extracted samples. The Methods mention spectral screening for potential organic interference, but no screening results are presented, such as the number of flagged samples or whether any correction or exclusion was applied. The manuscript already uses the appropriately cautious phrase "consistent with sample-dependent spectral interference." Without further quality-control evidence, I would not take the interpretation beyond this wording.
3. The source classification shown in Fig. 6 is not clearly explained in the Methods. The actual soil-water sampling intervals were 0–20, 20–40, 40–60, 60–100, and 100–140 cm. Section 2.4 then defines the MixSIAR sources as 0–40, 40–80, and 80–140 cm soil water, together with precipitation, irrigation water, and groundwater, whereas Fig. 6 shows only <60 cm, >60 cm, and groundwater. Please explain the complete mapping from sampling intervals to MixSIAR sources and then to the categories displayed in Fig. 6. The 60 cm boundary needs particular attention because it is not a model-source boundary and it divides the stated 40–80 cm source. Please also explain how precipitation and irrigation water were treated and whether Fig. 6 aggregates posterior estimates or represents a separate model run with a different source structure. The supplement should provide the model inputs, sample sizes and means/variances for each source; Fig. 6 should also include credible intervals. The manuscript already commits to making the isotope data and MixSIAR input files public upon acceptance. Because these materials are needed to assess the central result, the data and analysis files or scripts should be made available during revision.
Technical correction:
Figure 1 defines LS and MS, whereas the main text generally uses IRMS, so the abbreviations are inconsistent between the text and the figure captions. Please use LS and MS/IRMS consistently in the text and figure captions and define them clearly at first use. The sample labels in Figs. 1–3 also mix plant material and extraction method, for example "Xylem," "Root," and "Branch-Centrifugal." A consistent naming scheme that identifies both plant material and extraction method would be clearer.
Citation: https://doi.org/10.5194/egusphere-2026-2258-RC2 -
AC2: 'Reply on RC2', Ying Zhao, 08 Aug 2026
General assessment
The study is suitable for an HESS Technical Note, but the experiment and model analysis require fuller reporting. The adjusted scenario should not be presented as the true uptake depth.
Author response: We agree. The revised paper will be framed consistently as a method-sensitivity study rather than as a definitive reconstruction of the trees’ true uptake depth. We will provide fuller paired-sample reporting, analytical quality-control details, the exact three-endmember model specification, scripts and inputs, posterior uncertainty, and more cautious conclusions. The existing dataset is sufficient for these reporting, reanalysis, and sensitivity tasks; no claim will be made that an adjusted scenario identifies the true uptake solution.
Comment 1 - Paired branch comparison
Report the number of paired samples, how branch material was divided, individual-pair differences, paired effect estimates and uncertainty, and consistency across plots, trees, or dates.
Author response: Agreed. We will report the exact number of matched branch replicates and clarify that the two extraction protocols were applied to separate, concurrently collected branch segments matched by tree, sampling time, and replicate because the protocols require different amounts and configurations of material; they were not repeat measurements of a single aliquot. Individual matched values will be shown, and the analysis will report the mean or median matched difference, 95% confidence interval, effect size, and consistency across plots and sampling times. A mixed-effects or stratified analysis will be used only if the matching structure and sample size support it.
Comment 2 - Laser spectroscopy vs IRMS
Report sample numbers by tissue, whether the same extract was measured on both instruments, spectral-screening outcomes, and actions taken for flagged samples.
Author response: Agreed. Table 1 will report sample numbers for each plant matrix and instrument combination. We will verify from the laboratory records whether each laser spectroscopy-IRMS comparison used aliquots of the same extract or separately extracted material and will state this explicitly. We will report the number and type of spectral-quality flags and the rules used for remeasurement, correction, or exclusion, while retaining the cautious interpretation ‘consistent with sample-dependent spectral interference’. If the archived diagnostics do not support quantitative attribution, that limitation will be stated.
Comment 3 - Source grouping and Figure 6
Explain the mapping from sampled depth intervals to model sources and Figure 6 categories; clarify precipitation/irrigation treatment; provide inputs, sample sizes, means/variances, credible intervals, data and scripts.
Author response: We agree and will correct the inconsistency. The original soil-water measurements used four intervals: 0-20, 20-40, 40-60, 60-80, and 80-100 cm. For the plotted three-endmember MixSIAR analysis, the first three intervals were pooled as shallow soil water (<60 cm), the 60-80, and 80-100 cm interval represented deeper soil water (>60 cm in this study), and groundwater formed the third endmember. The previously stated 0-40/40-80/80-140 cm grouping is inaccurate for this model and will be removed. Precipitation was used only to construct the LMWL and was not entered into MixSIAR. Irrigation water was also not a direct endmember because it reaches the trees after infiltration and mixing in soil and is therefore represented within the soil-water endmembers; the model did not estimate a separate irrigation fraction. The Supplement or repository will provide the input data, sample numbers, source means and variances, source mapping, priors, model code, chain settings, convergence diagnostics, and posterior summaries. Figure 6 will display credible intervals, and the data and scripts will be made available during revision.
Technical correction - Terminology and sample labels
Use LS and IRMS consistently and identify both material and extraction method in labels.
Author response: Agreed. ‘MS’ will be replaced by ‘IRMS’ unless a generic mass-spectrometric method is intended. Labels will follow a consistent structure, for example Branch-CVE-IRMS, Branch-centrifugation-IRMS, Root-CVE-LS, Soil-0-20-CVE-LS, and Soil-60-80-CVE-LS. All abbreviations will be defined at first use and in every relevant caption.
Citation: https://doi.org/10.5194/egusphere-2026-2258-AC2
-
AC2: 'Reply on RC2', Ying Zhao, 08 Aug 2026
-
CC1: 'Comment on egusphere-2026-2258', Pei Zhao, 29 Jul 2026
General assessment
This Technical Note addresses an important methodological issue in isotope ecohydrology: how water extraction and analytical methods influence isotope-based inference of plant water sources. The topic is timely and relevant to the HESS readership because methodological uncertainty remains one of the major challenges in interpreting plant water isotope data.
The manuscript is generally well organized and presents an interesting field dataset comparing cryogenic vacuum extraction, centrifugal extraction, laser spectroscopy, and isotope-ratio mass spectrometry. The study also explores how apparent δ²H offsets affect Bayesian source apportionment.
I believe the manuscript has the potential to become a valuable Technical Note. However, several conclusions are currently stronger than the available evidence supports, and the discussion would benefit from a clearer distinction between methodological artifacts and natural isotope heterogeneity. Most of my comments can be addressed through clarification, additional statistical analyses, and more cautious interpretation rather than new experiments.
I therefore recommend major revision.
Major comments
1. Distinguish methodological bias from extraction-dependent water pools
The manuscript frequently attributes the observed δ²H differences to a "CVE bias". However, the comparison presented here does not demonstrate that CVE is intrinsically biased.
Instead, the two extraction techniques may recover different water pools (e.g. sap water versus bulk tissue water), which could naturally differ isotopically.
I recommend using more neutral terminology throughout the manuscript, such as
extraction-dependent isotope differences
apparent extraction effects
method-dependent isotope offsets
unless direct evidence is available to demonstrate a true methodological bias.
This distinction is important because it affects how readers interpret the conclusions.
2. Better quantify the magnitude of the observed method effects
Most comparisons are currently presented graphically.
Because this manuscript is fundamentally a comparison of methods, I recommend providing quantitative statistical evidence for the observed differences.
For example,
paired comparisons,
confidence intervals,
effect sizes,
would substantially strengthen the manuscript.
If paired observations are available, a Bland–Altman analysis would provide a more informative assessment of agreement than simple regression.
3. Clarify the treatment of δ²H correction
The manuscript applies an approximately +8‰ correction before MixSIAR analysis.
The rationale for choosing this value should be explained more clearly.
Rather than arguing that one correction is "correct", I suggest demonstrating how sensitive the inferred source contributions are to the magnitude of the applied correction.
For example, comparing
no correction,
+5‰,
+8‰,
+10‰
would greatly improve the robustness of the conclusions.
4. Expand the description of the MixSIAR analysis
The current description of the Bayesian mixing model is rather brief.
Please provide sufficient methodological detail to allow readers to reproduce the analysis, including
source grouping,
priors,
model structure,
convergence diagnostics,
uncertainty estimates.
Similarly, the figures showing source contributions should include credible intervals or equivalent uncertainty measures.
5. Interpret LC-excess more cautiously
Because laser spectroscopy and IRMS show systematic differences in δ¹⁸O for plant waters, these analytical differences propagate directly into LC-excess.
Please discuss this explicitly.
It would be useful to explain whether the observed LC-excess differences primarily reflect
analytical effects,
extraction effects,
or a combination of both.
Minor comments
The abstract is clear but could emphasize more explicitly that the study evaluates methodological uncertainty rather than demonstrating a definitive biological fractionation mechanism.
Throughout the manuscript, avoid wording that implies causal mechanisms unless directly supported by the presented evidence.
For example,
"CVE introduced a δ²H bias"
could be softened to
"CVE produced systematically lower δ²H values than centrifugation."
Some discussion paragraphs are longer than necessary.
The paper would benefit from focusing more directly on
what was observed,
what can be concluded,
what remains uncertain.
Recommendation
This manuscript addresses an important methodological problem and fits well within the scope of Hydrology and Earth System Sciences Technical Notes.
The study presents useful data and practical recommendations.
After revision to better quantify uncertainty, clarify methodological interpretation, and moderate several conclusions, I believe the manuscript will make a valuable contribution to isotope ecohydrology.
Citation: https://doi.org/10.5194/egusphere-2026-2258-CC1 -
AC3: 'Reply on CC1', Ying Zhao, 08 Aug 2026
General assessment
The manuscript has potential but needs more cautious interpretation, clearer separation of artifacts and natural heterogeneity, and additional statistical reporting.
Author response: We appreciate this assessment and agree. The revised manuscript will consistently emphasize measured method-dependent differences and their propagation into inference, without assigning a unique mechanism or presenting any offset-adjusted scenario as the true uptake solution. The Discussion will be shortened and organized around what was observed, what can be defensibly inferred, and what remains uncertain.
Major comment 1 - Method bias vs water pools
Use neutral terminology because CVE and centrifugation may recover different water pools.
Author response: Agreed. We will replace broad causal statements such as ‘CVE introduced a bias’ with observations specific to the experiment, such as ‘CVE produced systematically lower δ²H values than centrifugation in matched branch samples’, ‘extraction-dependent isotope difference’, or ‘apparent method-dependent offset’. The text will distinguish analytical or procedural error from the possibility that the two techniques recover genuinely different water pools.
Major comment 2 - Quantify method effects
Add paired comparisons, confidence intervals, effect sizes, and preferably Bland-Altman analysis.
Author response: Agreed. These analyses will be added for the extraction comparison and for cross-instrument comparisons wherever the design provides valid matched observations. We will show individual matched values, report mean or median differences with 95% confidence intervals and effect sizes, and use Bland-Altman plots to evaluate agreement and limits of agreement rather than relying on correlation alone.
Major comment 3 - δ²H correction
Explain the +8‰ value and evaluate 0, +5, +8, and +10‰ scenarios.
Author response: Agreed. The approximately +8‰ value will be described as an empirical, study-specific sensitivity scenario, and its exact calculation from the original branch-water dataset and reference framework will be reported transparently after verification. We will compare model outputs under 0, +5, +8, and +10‰ scenarios. No scenario will be described as the ‘correct’ or ‘true’ uptake solution; the purpose is to quantify how plausible apparent offsets propagate into posterior source attribution.
Major comment 4 - MixSIAR reproducibility
Provide source grouping, priors, model structure, convergence diagnostics, uncertainty, and credible intervals.
Author response: Agreed. The revised Methods and Supplement will provide the complete specification of the actual three-endmember model: shallow soil water (<60 cm; pooled 0-20, 20-40, and 40-60 cm), deeper soil water (>60 cm; pooled 60-80, and 80-100 cm), and groundwater. We will report source and mixture sample numbers, means and variances, the residual and process error structure, priors, chain length, burn-in, thinning, number of chains, Gelman-Rubin and Geweke diagnostics, and posterior credible intervals. Precipitation will be identified as the dataset used to construct the LMWL, and irrigation as water represented indirectly within the soil-water endmembers rather than as a separate model source. The complete script and inputs will be deposited in a public repository during revision.
Major comment 5 - LC-excess
Discuss how cross-instrument δ¹⁸O differences propagate into LC-excess and whether LC-excess reflects analytical, extraction, or combined effects.
Author response: Agreed. LC-excess uncertainty in this study arises from both the estimated LMWL parameters and the isotope values produced by different extraction and analytical combinations. With an LMWL slope close to 8, a 1-3‰ difference in δ¹⁸O can change calculated LC-excess by approximately 8-24‰ if δ²H remains unchanged. We will quantify these propagation effects, calculate LC-excess only for methodologically comparable datasets or present method-specific values, and treat LC-excess as a diagnostic measure rather than evidence of a unique mechanism. The Discussion will state explicitly that the observed LC-excess differences may combine LMWL, analytical, extraction, internal-water-pool, and process effects.
Minor comments - Abstract, causal wording, and discussion length
Emphasize methodological uncertainty rather than definitive biological fractionation, soften causal wording, and shorten the Discussion.
Author response: Agreed. The Abstract and Conclusions will state that the study evaluates methodological uncertainty and the sensitivity of source attribution to apparent offsets; it does not demonstrate a unique biological fractionation mechanism or identify one adjusted scenario as the true uptake depth. Causal wording will be replaced by observational language, and repetitive discussion will be condensed.
Citation: https://doi.org/10.5194/egusphere-2026-2258-AC3
-
AC3: 'Reply on CC1', Ying Zhao, 08 Aug 2026
-
CC2: 'Comment on egusphere-2026-2258', Daniele Penna, 06 Aug 2026
General comment
This technical note is timely and welcome. Its main strength is the integration, within a single field study, of comparisons between cryogenic vacuum extraction (CVE) and centrifugation, between laser spectroscopy and IRMS, and of a sensitivity analysis examining how an apparent δ²H offset propagates into MixSIAR source apportionment. These methodological effects have been documented individually in the literature, but they have seldom been brought together and evaluated comparatively in this way. The results provide a clear illustration that methodological effects can be comparable to the ecohydrological signals under investigation. Hydrogen-isotope offsets have been debated for many years, and their origin and interpretation have yet to reach consensus. I therefore appreciate the authors’ appropriately cautious interpretation of the observed depletion as an apparent δ²H offset that may reflect both process-related signals and methodological artifacts, rather than attributing it exclusively to biological fractionation. Overall, I believe this Technical Note has the potential to make a useful contribution to the ecohydrological community of isotope users by demonstrating how methodological choices can affect both measured isotope compositions and the resulting ecohydrological interpretations.
Specific comments
I basically agree with most of the comments made by the previous three reviewers and will avoid repeating them. I have only a few additional comments that could further strengthen the Technical Note.
- The manuscript states that precipitation collected throughout the 2021 growing season was used to define the LMWL, but the number of precipitation samples or events is not reported. Please provide the sample number and the temporal coverage together with the LMWL parameters. It would also be useful to indicate whether individual-event or composite samples were used and whether precipitation amount weighting was considered. This information is important because the LMWL parameters enter directly into the calculation of LC-excess and into the estimation of the mean δ²H deficit used for the correction. Consequently, uncertainty in the LMWL could propagate through a substantial part of the analysis and interpretation.
- The intensive observations were conducted over five days with some sampling omitted during storm conditions. The number of samples was impressive over such a short duration but, please, clarify how representative this period was of the wider growing season. Were these days unusually dry, wet or broadly typical? Concise information on antecedent precipitation and irrigation, soil moisture and, possibly, atmospheric evaporative demand would help place the observations in their hydrological context. This could be helpful to understand whether the trees and soils were experiencing water stress or comparatively well-watered conditions. Indeed, the hydrological conditions could affect sap availability, the relative importance of mobile and tissue-associated water pools, extraction performance and, ultimately, the ecohydrological response inferred for the sampled trees.
- The Discussion Section could perhaps include a brief, non-speculative paragraph acknowledging the existence of other soil- and plant-water extraction approaches. The manuscript appropriately emphasizes that CVE and centrifugation target partly different water pools and have different practical constraints. Placing the present findings within the broader extraction-method literature would improve their general relevance, while making clear that the observed method effects cannot not automatically be extrapolated to all extraction systems. The authors could also encourage future studies to apply the same comparative and sensitivity-analysis framework to other methods. Same examples of comparative studies are Orlowski et al. (2016; doi:10.1002/hyp.10870), Millar et al. (2018; doi:10.1002/rcm.8136), Zuecco et al. (2020; doi:10.5194/hess-2020-446), Wen et al. (2023; doi:10.1016/j.jhydrol.2023.130015), and Jeon et al. (2024; doi:10.3390/f15030420).
- Finally, again refraining from speculation, some sentences or a paragraph on the possible similar/different results that could be obtaining by testing other species (another crop or forest species), and/or an acknowledgment that this test was limited to apple trees, could be useful as well.
Citation: https://doi.org/10.5194/egusphere-2026-2258-CC2 -
AC4: 'Reply on CC2', Ying Zhao, 08 Aug 2026
General assessment
The referee considers the integrated comparison timely and useful and supports the cautious interpretation of the apparent δ²H offset.
Author response: We thank the referee for this positive assessment. We will retain the integrated design and further strengthen the paper by adding reproducibility details, paired statistics, uncertainty analyses, and explicit limits on generalization.
Comment 1 - Precipitation and LMWL
Report sample number, temporal coverage, sampling type, amount weighting, LMWL parameters, and uncertainty propagation.
Author response: Agreed. We will make the same revisions described in our response to Referee 1: report the complete precipitation-sampling metadata, LMWL equation and parameter uncertainty, and evaluate propagation into LC-excess and the empirical offset. We will also state that precipitation was used to define the LMWL and was not entered as a MixSIAR source endmember, and that the approximately +8‰ case is one sensitivity scenario rather than a corrective reconstruction.
Comment 2 - Hydrological context
Clarify the representativeness and antecedent hydrological conditions of the intensive campaign.
Author response: Agreed. We will add the available antecedent precipitation, irrigation, soil-moisture, and atmospheric-demand context and describe the campaign consistently as a short intensive snapshot rather than a seasonal characterization. We will classify the sampled period as dry, wet, typical, water-stressed, or well-watered only if supported by the available records and will otherwise state the limitation explicitly.
Comment 3 - Broader extraction-method context
Discuss other approaches and avoid extrapolation to all extraction systems.
Author response: Agreed. We will add the requested non-speculative paragraph and the suggested comparative literature, explaining that the numerical differences observed between CVE and centrifugation in apple branches cannot be generalized to other extraction systems. What is transferable is the need to compare methods directly, report uncertainty, and test how plausible offsets affect source attribution.
Comment 4 - Species scope
Acknowledge that results may differ for other crops or forest species.
Author response: Agreed. We will identify the apple-tree focus as a central limitation and explain that differences in xylem anatomy, tissue water content, storage, sap chemistry, and hydraulic state may produce different outcomes in other crops and forest species. The present numerical offsets will therefore be treated as case-specific and requiring independent validation.
Citation: https://doi.org/10.5194/egusphere-2026-2258-AC4
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 189 | 65 | 20 | 274 | 13 | 14 |
- HTML: 189
- PDF: 65
- XML: 20
- Total: 274
- BibTeX: 13
- EndNote: 14
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1