the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 29 Jul 2025
Beyond two water worlds: dynamic transpiration sourcing in a mixed-species boreal forest
Abstract. The water budget of a forested catchment comprises inflows, storage pools, and outflows, each with a specific stable isotopic signature. The isotopic signature is often set by summer vs. winter precipitation events, where the subsequent use of these water sources can then be traced. Streamflow, one of the major outflows, is critically important for water supplies and flooding risk. Transpiration, a second major outflow, describes the evaporation of water vapor from within tree leaves; it reduces streamflow and is mechanistically associated with biomass production. The water sources used for streamflow and transpiration can be so isotopically distinct that they have been considered to represent two “water worlds,” with distinct controls and perhaps little mixing between them. Here we describe, on a daily time-step, the contributions of water sources used for transpiration between two species that commonly occur across the Eurasian boreal zone and have close relatives in North America. Norway spruce, which is shallow-rooted, was compared to Scots pine, which roots more deeply, and both were compared to stream water. We made these measurements in 2017, a typical summer, and compared them to 2018, a year of historic drought. Pine and spruce used distinctly different water sources. After the drought ended in 2018, the spruce switched to exclusively recent summer rainfall. Pine also switched sources, but less completely, consistent with its deeper root distributions. Streamwater was derived from residual water, with a greater representation of winter precipitation. These results support the notion that transpiration and streamwater are derived from different sources, while further dividing the transpiration between spruce and pine. They suggest modified predictions of streamflow and forest production, especially in response to extreme weather events. Models of boreal forest transpiration should be tested against these observations to determine how well they describe this water-source differentiation.
- Preprint
(602 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 14 Sep 2025)
-
RC1: 'Comment on egusphere-2025-3328', Anonymous Referee #1, 25 Aug 2025
reply
Marshall et al. present a two-year time series of stable isotopes of water in the xylem stream of two boreal forest tree species. The authors pursue three comparisons: 1) shallow vs. deep rooted species, 2) drought vs. non-drought year, and 3) water sources for streams vs. plants. It is this last point, written in terms of the “Two Water Worlds” hypothesis, that is the emphasis of the introduction and discussion.
Synoptic Comments:
This is largely a descriptive longitudinal study, with little formal statistical analysis. Not everything needs formal statistical analysis, but I believe that some opportunities to bring depth are missed. In doing so, one could provide additional insights into mechanism, which I think would be exciting. An example: It would appear that the authors have significant information on precipitation amount/stable isotopes of precipitation. How long does it take for the precipitation signal to propagate into soil, trees, or streamwater when there are distinct summer events?
There have now been many studies of “the two water worlds” hypothesis, which the authors are correct in couching as a heuristic means of proceeding. However, perhaps the turn away from this hypothesis (as noted by the authors themselves on L.56) is because it has reached its utility and there is a need for approaches that elucidate underlying mechanisms. In this sense, I was (personally) more excited by the *continuous* longitudinal nature of the study demonstrating what happens when there is drought and relief from drought. I wonder if the authors might consider giving this more weight and reframe the introduction in those terms.
Precipitation, soils, trees and streams are never all shown on the same plot; this seems like an opportunity missed, given how few studies have been able to collect all of these simultaneously!
A primary comparison of this paper is 2017 vs. 2018, which are each given their own panels. An alternative approach would be to have a single panel, with data colored by year to facilitate easier comparison. In addition (and this is would be very useful), recommend that authors shade with transparency the time period of interest (summer) to make interpretation of results easier.
Introduction
L. 30. I recommend that you modify the first sentence, which neglects to mention the rest of the planet’s fauna. Perhaps more importantly, the study is not actually about competing uses of water by vegetation and humans.
L. 43. I recommend that you consider that the citation of Brooks et al., while true, neglects the primary proposed hypothesis, which was a temporal offset in the refilling of empty soil water storage (winter) from the timing of its subsequent use.
L. 44. Similar to the comment above about Brooks et al., the description of Allen et al. is correct, but perhaps an incomplete interpretation. In fact, it would appear that soils hold different amounts of summer precipitation and in drier sites, tend to hold more winter precipitation (Allen et al. in HESS). This (and proposed climatological mechanisms) are proposed in Goldsmith et al. (2022 in GRL) and Floriancic et al. (2025 in Ecohydrology).
L. 58. I recommend that you consider whether recent studies have not discussed water sources in terms of “water worlds” because the framework may not have utility in advancing our understanding of the processes that underly the observations.
Methods
L. 80. Recommend that you provide the difference in (e.g.) summer rainfall between 2017 and 2018 as a total and as percentage of the annual. Otherwise, it is hard to contextualize the severity of the drought from Figure 1 alone, wherein there does seem to be episodic rainfall.
Figure 1. Recommend you add more dates to x-axis, particularly top panel, where it would be nice to see times more closely. Additionally, consider merging 2017 and 2018 into one panel with different colors, as that is a primary comparison.
Figure 1. Recommend reconsidering the red line. More useful than delineating a calendar year would be to delineate summer months through gray shading.
L. 84. Recommend you delineate how long sensors were in place before measurements began.
L. 105. Recommend that you specify that all isotopes are provided per mille relative to V-SMOW.
L. 109. Recommend that you offer additional details on sample handling for water samples. I assume sealed in glass/plastic vials and stored in a cool setting until analysis.
L.111. The study is actually all the stronger for at least having some measure of the soil water. It’s striking that it takes halfway through the methods to read this – it’s a much more complete picture with this in place and it should be mentioned in the introduction.
L. 113. Recommend that you compare how the depths of the soil water sampling match with what is known about the depths of soil water use by the two contrasting species.
L. 116. How does the isotopic value of water vapor in soil compare to what we would expect be available to plants. Or, in other terms, water that is more or less mobile? Recommend that the authors comment.
L. 125. Since SOI is a comparison with precipitation and xylem water reflects a potentially evaporated source water signal (soil), many studies compensate for this evaporation (e.g., see original work by Allen et al. 2019 in HESS). Recommend pursuing this approach or at least confirming that it does not change your interpretation, especially in 2018 when you would expect drought to have an impact.
L. 130. Recommend that you include details on how the precipitation isotope data were collected. What device? Are the isotope ratios amount weighted? How often were samples collected? How were they stored?
Results and Discussion
L. 136. Recommend quantifying precipitation amounts and providing comparisons.
L. 139. While this is almost certainly true, it’s so true as to be obvious. On the other hand, is physical surface evaporation a consideration in this ecosystem? Recommend revising this sentence.
L. 142. It should be relatively easy to calculate a minimum event size needed to percolate to the different sensor depths. Recommend that the authors consider as much in order to bring depth of understanding to this analysis.
L. 153-154. This number “(-10.00, SE 0.05‰)” and similar numbers are hard to interpret. Recommend specifying if it’s an annual mean and using plus/minus per typical convention, then specifying that it is standard error.
L. Figure 2. I’ve always been a little uneasy with the idea of calling it sap and in fact, this is the only place where it is referred to as such. Recommend “Xylem water” or “Xylem water vapor” for consistency with paper. Recommend indicating per mille after noting isotope ratios of particular events, for sake of clarity.
L. Figure 3. Are these continuous measurements? If not (or if average in some way), recommend adding points on the lines to clarify sampling interval.
L. 184. Why refer to it as “the xylem water used in transpiration” here and not elsewhere? This confused me; recommend you clarify if possible.
L. 188. To me, it would appear that the trees had lower SOI values at the start of the growing season in 2018 in general, as well as a change given drought. Recommend you consider commenting.
L. Figure 4. Here the seasonal origin index is >1 in late 2018, indicating that the xylem water is in excess of the summerP isotope value. This would be an argument for providing more information on the calculation of the precipitation isotope sine curve.
L. Figure 4. What is the SOI of soil water? Recommend adding this to the figure.
L. 214. More probably, an SOI near zero is an almost infinite possible mix of spring, summer, fall and winter waters.
Relevant recent literature:
Floriancic et al. (2024) Isotopic evidence for seasonal water sources in tree xylem and forest soils
Kinzinger et al. (2025) Continuous In‐Situ Water Stable Isotopes Reveal Rapid Changes in Root Water Uptake by Fagus sylvatica During Severe Drought
Brighenti et al. (2024) Snowmelt and subsurface heterogeneity control tree water sources in a subalpine forest
Sprenger et al. (2025) Opportunistic short‐term water uptake dynamics by subalpine trees observed via in situ water isotope measurements
Citation: https://doi.org/10.5194/egusphere-2025-3328-RC1 -
RC2: 'Comment on egusphere-2025-3328', Anonymous Referee #2, 05 Sep 2025
reply
This manuscript presents an excellent dataset: two years of continuous isotope measurements in precipitation, soils, trees, and stream water in a boreal mixed forest. The contrasts between spruce and pine, and between a normal (2017) and extreme drought year (2018), are highly valuable. The results go beyond the traditional “two water worlds” framing by showing that transpiration is species-specific and dynamic, shifting with drought and rewetting.
General comments
This is a strong and timely paper. The continuous isotope record across two contrasting years is a major contribution. The paper would be significantly strengthened if the authors (i) reframe the introduction toward dynamics, (ii) add quantitative analyses (event lags, rainfall thresholds, precipitation totals, unmixing), (iii) clarify methods (ideally with a schematic of the sampling design and photos of the site), and (iv) improve figure compilation.
I also wonder whether the Seasonal Origin Index (SOI) is the most informative metric. The precipitation isotopic data do not follow a clear sine-shaped seasonal pattern, which may limit interpretability. A summer–winter endmember unmixing approach applied per time step could better capture event-driven dynamics, highlight species contrasts, and allow uncertainty estimates (e.g., via Bayesian or bootstrapped mixing models). The SOI metric could still be retained for comparability.
The continuous monitoring through drought and recovery is the most exciting aspect of the study. While the “two water worlds” hypothesis provides a useful background, it may have reached its heuristic limit. The dataset is better framed in terms of temporal dynamics and species contrasts, as already suggested by Reviewer 1.
I encourage the authors to be cautious with terms such as “water pools,” which can imply complete separation of sources. Water in soils and catchments mixes continuously across depths and time. It is not black and white (not purely winter vs. summer or deep vs. shallow) but a gradient of interactions and mixtures (e.g., Dubbert et al., 2019). Acknowledging this explicitly and noting that such behavior can be simulated with mechanistic models would provide a more nuanced interpretation. For instance, Meusburger et al. (2022) showed with modeled root water uptake that both shallow- and deep-rooted species switched to deep soil water during drought but returned to recent shallow water after rewetting. An extension of this model with isotope transport (LWFBrook90.jl) can reproduce the distinct signatures between species and residual (blue) water. This suggests that physically based models can capture these dynamics without invoking a rigid “two water worlds” separation, reinforcing a process-based interpretation of water sourcing.
The manuscript is largely descriptive. While not a weakness, adding straightforward analyses would strengthen mechanistic insight. I also second the comments of Reviewer 2 regarding more quantitative evaluation. Because the soil isotope equilibration devices were only installed at 5 and 15 cm depth, please discuss how you can be confident that a meaningful isotopic gradient existed along the full profile, particularly given that spruce and pine likely access water from different depths. Providing basic information on total soil depth would also be helpful.
The observed vertical profile in soil δ¹⁸O is consistent with piston flow (line 208), but a simple quantitative check could make this conclusion more robust. For example, compare pre- and post-event soil δ¹⁸O values and volumetric water contents to estimate the fraction of soil water replaced during major rainfall events, providing quantitative support for Figure 3.
At line 215, the manuscript states that the SOI value (~0.16) in deep soil layers indicates a nearly even mixture of winter and summer precipitation and interprets this as evidence for macropore flow. This pattern may not exclusively indicate macropore flow; it could also reflect seasonal removal of isotopically heavier summer water by transpiration, leaving lighter winter-like water behind. Please discuss both processes as potential contributors.
Regarding Fig. 4, the strong distinction between SOI of streamflow and transpiration is interesting but should not be overinterpreted as evidence for fully distinct water worlds. A continuum shaped by root distribution, water availability in soil layers, and temporal dynamics of inputs and outputs is a more realistic interpretation.
Additional small comments:
- Consider combining 2017 and 2018 into single panels.
- Consider adding all compartments (precipitation, soils, trees, streams) into one composite figure (Reviewer 1 suggestion). Do the mixtures lie within the sources?
- Line 225: correct to Gessler et al., 2022.
- Clarify in Methods how precipitation isotope sampling was done (collector type, frequency, storage) to ensure reproducibility.
- Report isotope analyzer precision (e.g., ±0.1 ‰) and sample handling details (sealed vials, storage conditions) so readers can judge data quality.
- In Figure 3, if soil data are discrete, mark sampling points; this will make profiles easier to interpret.
- Add seasonal shading or vertical bars to Figure 1 for summer months to highlight the drought period visually.
- Provide average temporal resolution of stream water sampling (e.g., weekly/biweekly).
Citation: https://doi.org/10.5194/egusphere-2025-3328-RC2 -
RC3: 'Comment on egusphere-2025-3328', Anonymous Referee #3, 07 Sep 2025
reply
The manuscript "Beyond two water worlds: dynamic transpiration sourcing in a mixed-species boreal forest" presents a unique dataset of in situ measured xylem water isotopic signatures for two boreal tree species spanning two contrasting growing seasons.
The manuscript is well written, clearly structured and pleasently concise. Overall, I would recommend this paper for publication after some minor revisions.
My main point of critique for the current manuscript would be the incidental treatment of the soil water isotopic signatures. Clearly, the isotopic signatures of soil water are of paramount importance to explain the observed xylem water isotopic signatures, yet, the authors did not report any soil isotopic signatures for the first year and only measurements from 5 and 15 cm soil depth for the second growing season. It is safe to assume, that both tree species' root systems are likely to exceed these depths.
Furthermore, I am a bit sceptical about the use of BGDL 300 soil gas lances for the measurement of soil water isotopic signatures. Are there any previous publications that have used and validated these devices for the measurements of stable water isotopes? If yes, I would like to see a reference for that, if no, the manuscript should include some more details on the setup such as the material and dimensions of the BGDL 300. Did you also use heating cable for these lines? Did you do any comparisons to more estbalished ways of measuring stable water isotopes of soil water?
Looking at the results in Figs. 2 and 3, I see that after the drought ending precipitation event with -8.1‰ δ18O, which shifted the xylem isotopic signature of spruce xylem water towards values around -8‰ δ18O, somehow both of your observed soil depths did not exceed -9‰ δ18O. This would imply that spruce had to source nearly all of its water from above 5 cm of soil depths and that even 85mm of rain were not enough to flush the soil water isotopic signatures to a depth below 5 cm. Or could it be that your measurements of soil water isotopic signatures are somewhat biased?
I am fully aware that the focus of this study was on xylem water isotopic signatures and the results indicate an impressive successful long term application of the borehole technique for natural abundances of 18O. However, I think the soil water isotopic signatures within this study deserve and require a more detailed discussion. Do you trust the results? Can you recommend the use of BGDL 300 soil gas lances for soil isotopic measurements?
Apart from that, I have the following minor comments:
line 92: Please also specify the inner diameter and wall thickness of your PFA lines.
line 95: If possible, please specify the type and manufacturer of the solenoid valves.
lines 92-103: The description of the setup is a bit unclear to me: did you just draw atmospheric air from one end of the borehole through the tree and into the analyzer, or was there an additional dry air supply connected to the entry side of the borehole (as. depicted in Fig.2 of Marshall et al. (2020))?
lines 103: You say you "focus on δ18O because it becomes less biased during extractions than δ2H", but these effects are likely to be limited to cryogenic extractions, which are not part of your study. Volkmann et al. (2016) reported a missmatch between in situ δ18O meausrements with a CRDS compared to IRMS meausrements of destructive samples, while δ2H showed no such missmatch. Kinzinger et al.(2024, https://doi.org/10.1093/treephys/tpad144), using the same in situ methodology as Volkmann et al. (2016), focused their analyses to δ2H because δ18O showed a lower accuracy and higher drift. It would be interesting to know how the results for the two isotopes compare in your study. I would love to see a dual isotope plot of your in situ measurements (maybe as a supplement figure), just to get an idea on the cababilities of the borehole technique.
Fig.3: As you mentioned that there were three repetitions for the depth of 5 cm - what is the grey line showing? The mean of all three repetitions? Could you also indicate the range of the three repetitions, or just show each of the measured time series with a seperate line?
line 191: Is -0.02 actually the average streamwater SOI? In Fig. 4 it looks like the streamwater SOI is positive most of the time - or did you use a mass weighted average?
line 199: In my opinion, the dichotomy between "transpiration" and "infiltration" seems misplaced, since transpiration is also likely to be sourced from water that infiltrated into the soil. Maybe "(deep) percolation" or "ground water reacharge" would be better terms than "infiltration".
lines 245-247: Speaking of models: Could you provide (or at least archive) the data presented in Figs.1-3 as well as additional site information such as a more detailed description of the soil profile and stand properties (basal area, leaf area index, any known information on root density distributions) and (if by any chance available) sapflow or dendrometer data in such a manner, that future modellers could use them in a quantitative way? Currently, the manuscript is missing a data availability statement.
Citation: https://doi.org/10.5194/egusphere-2025-3328-RC3
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 554 | 38 | 7 | 599 | 20 | 18 |
- HTML: 554
- PDF: 38
- XML: 7
- Total: 599
- BibTeX: 20
- EndNote: 18
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1