Telling Tales of Water Journeys With Isotopic Tracers

Determining the sources of water inside plants using its isotopic composition is a long‐standing research challenge in ecohydrology. A better understanding of water sources can help improve models and ultimately contribute to more accurate forecasts of water availability, food production, carbon sequestration or ecosystem status. Over the years, several methods have been developed and applied to water source partitioning, and Gai et al. (2023, https://doi.org/10.1029/2022wr033849) provide a systematic assessment of the uncertainty of different isotopic tracers (2H, 3H, 17O, 18O) and mixing models (IsoSource, SIAR, MixSIR, MixSIAR) for an apple tree orchard on the Loess Plateau in north‐central China. For that study area, the combination of 2H and 18O with the MixSIAR mixing model is recommended. Importantly, the systematic assessment provides a framework that can be applied to select a suitable combination of tracers and mixing models for different ecosystems and climate zones. This commentary aims to provide a wider context for a selection of key results from Gai et al. (2023, https://doi.org/10.1029/2022wr033849) and highlight potential future research directions.


Introduction
Plants have a key role in the global water and carbon cycles, are the fundament of the food chain and essential for biodiversity on Earth.To live, grow and reproduce, plants require water that they extract from their environment.Vegetation also plays an important role in the retention of surface water following precipitation events and the local generation of precipitation through transpiration of water (e.g., Coenders-Gerrits et al., 2014;Miralles et al., 2020).The extraction of water from its environment requires energy and plants may adjust their water extraction strategy based on the environmental conditions (Figure 1) and competition with other plants.To improve the understanding of these complex processes and quantify the contribution of different sources of plant water, isotopic tracers of plant water are widely used (e.g., Hahm et al., 2020;Shi et al., 2023;Vilà-Guerau de Arellano et al., 2019).
The use of isotopic tracers relies on the fact that different water reservoirs have a different isotopic signature (e.g., Bowen et al., 2019;Sprenger et al., 2016).For instance, if we measure that xylem water in a plant is enriched in 2 H or 18 O, while the water from deep soil layers is not and the water close to the soil surface has a similar enrichment as the plant water, this suggests that the plant is likely extracting water from the shallow layer.Both 2 H and 18 O are widely used isotopic tracers but do not always agree, and to increase observational constraints, additional isotopic tracers can be used (e.g., 17 O or 3 H).Also, fractionation of 2 H can occur during soil water uptake or within the plant tissue (Barbeta et al., 2019) and during the cryogenic extraction (Chen et al., 2020), for which deuterium-correction methods are available.Finally, to support the water partitioning analysis, researchers often make use of isotopic mixing models (e.g., Stock et al., 2018).
In a recent study, Gai et al. (2023) performed a systematic analysis of four isotopic tracers ( 2 H, 3 H, 17 O, 18 O), two deuterium-correction methods (Barbeta et al., 2019;Chen et al., 2020) and four mixing models (IsoSource, SIAR, MixSIR, MixSIAR).The study is conducted for an apple orchard in the Loess Plateau area in China, a region that is experiencing impacts of climate change, but also from large-scale afforestation projects.These projects intend to reduce soil erosion and stimulate carbon sequestration.The region experienced an increase in grasslands, croplands and forests from 2001 to 2009, which were mostly replacing shrublands and barren lands (Fan et al., 2015).The changes in vegetation led to increased forest photosynthesis (Anema et al., 2024) and changes in the regional climate (e.g., increased local temperatures because of reduced albedo, L. Tian et al., 2022).Interestingly, the deep loess layers in the study area contain relatively high 3 H from nuclear weapon testing in the 1960s, which improves the ability of 3 H to discriminate between different soil layers.Some of the main results of the study by Gai et al. (2023) are described in the next section, and implications are discussed in a wider context in the final section of this article.et al. (2023) analyzed four isotopic tracers ( 2 H, 3 H, 17 O, 18 O), two different xylem water deuterium bias correction methods and four mixing models (IsoSource, SIAR, MixSIR, MixSIAR) for an apple orchard in the Loess Plateau.The analysis by Gai et al. (2023) builds on earlier intercomparison studies (e.g., Wang et al., 2019) and emphasis is placed on the uncertainties associated with methodological choices (e.g., including deuterium correction methods) that one could use.Uncertainty associated with the selected methods (tracers, bias correction method and mixing model) are also quantified.By providing insight into the performance of different methods, the selection of the appropriate method by new or experienced researchers is facilitated.

Gai
Within the isotopic measurement community, accurate reporting of uncertainty is common practice, whereas in some modeling studies, the analysis and reporting of uncertainties is an afterthought or underappreciated.In the systematic assessment by Gai et al. (2023), uncertainties take a central role and are systematically analyzed and reported.For instance, their Figure 9 shows the estimated partitioning from different soil depths, including uncertainty bars (although readability of those error bars in the figure is challenging due to overlap) that indicate substantial uncertainty in their results.
Besides the different isotopic tracers, deuterium correction methods and mixing models, the observational data set also covers a range of environmental conditions: data was obtained for different ages (18 and 26 years) of the apple trees and the study was performed for different seasons of the year.Also, water extracted from different soil moisture depths was considered (0-1 m; 1-3 m; 3-6 m; 6-8 m, and >8 m) in their analysis.Finally, to estimate the water isotopes in a given reservoir, multiple samples from that specific reservoir were used, providing robust estimates, in line with the earlier mentioned thorough approach.
The authors conclude that for the case studied, the water source partitioning is affected strongly by the selected isotopic tracers and mixing model, whereas the effect of the deuterium bias correction method is negligible.Overall, the combination of 2 H and 18 O with the MixSIAR model produces the best result according to the error metrics used and their expert opinion.For this combination, the study reports minor differences between the 18 and 26-year old apple trees (except for a somewhat larger contribution from shallow soil layers to the younger trees in the beginning of the rainy season).Toward the end of the rainy season the apple trees extract more water from the deeper (>3 m) soil layers.The authors stress that their recommendation for tracers and mixing model applies to the apple orchards in the Loess Plateau and that it is not certain whether this also holds for other ecosystems and climate zones.However, the systematic framework developed by Gai et al. (2023), also lends itself well for application to a wider range of climatic conditions and species.Therefore, the value of this study extends beyond the ecosystem studied here.Also, the current data set could potentially be used to further refine the recommendation of tracers and mixing models, as the observational period covers some wetter and drier months.A detailed analysis could uncover potential seasonal biases of methods and help develop more specific recommendations (e.g., an ideal tracer/model combination for dry summer months).

Outlook
This outlook section goes beyond the original scope of the paper by Gai et al. (2023) and explores connections with other water tracing methods, novel isotopic tracers, developments in isotope measurement technology and implications of changing environmental conditions.

Response to Extreme Conditions
One of the remaining questions is how the partitioning of water is changing under extreme conditions, for example, intense droughts.The topsoil layer tends to dry quickest and plants with sufficiently developed root systems are able to extract water from deeper soil layers during droughts (e.g., Buitink et al., 2020), thus shifting the sources of plant water.Also, extremely wet periods can distort the water uptake by vegetation.As these extreme conditions become more likely with a changing climate, a further exploration of these extremes would be a valuable continuation of the work by Gai et al. (2023).
The recent European droughts in the summer of 2018 (Smith et al., 2020) and 2022(van der Woude et al., 2023) have shown large impacts on vegetation, including increased fires and mortality and reduced carbon sequestration.Another recent large-scale drought occurred in the South American La Plata basin (Arias et al., 2024;Rivera et al., 2023).Impacts in this key agricultural region included food security issues and massive economic losses.With projected future climate change and more frequent extreme weather events, it is essential to develop a better understanding of water and carbon exchange at the ecosystem level.

Atmospheric Tracking of Water
Partitioning sources of water in vegetation can provide important insights in plants' behavior.Tracking water flows beyond the plants and soils can build on these insights (te Wierik et al., 2021).To track the flows of moisture in the atmosphere, isotopes can also play a role (e.g., Valdivielso et al., 2024).However, interpreting the isotopic composition of atmospheric moisture is challenging because of atmosphere mixing and fractionation during condensation and (re-) evaporation (e.g., Sun et al., 2024).Alternatively, atmospheric moisture flows can be described using moisture tracking models (e.g., van der Ent et al., 2010).Such models rely on meteorological information, including wind patterns, which is taken from reanalysis data sets or in some cases calculated by the model itself.
Moisture tracking models can provide information on vegetation dynamics through feedbacks with the atmosphere.This complements the work of Gai et al. (2023), which also considers environment-vegetation feedbacks, but focused on soil, where changes tend to be slower than in the dynamic atmosphere.For instance, Staal et al. (2023) demonstrated using such a model how reduced transpiration during drought conditions in a tropical forest resulted in lower atmospheric moisture and reductions of precipitation and photosynthesis in downstream Water Resources Research 10.1029/2024WR037033 regions.In addition, large-scale expansion of forest cover, such as is the case for the Loss Plateau region, is known to increase evaporation and moisture tracking models can indicate which fraction of this moisture is expected to precipitate back within the region itself.Hoek van Dijke et al. (2022) showed that direct evaporation increase and indirect rain fall increase due to forest cover expansion can lead to regional drying or wetting, further affecting the local vegetation.

Carbon Dioxide Exchange
The hydrological cycle is intimately linked to the carbon cycle through vegetation (Figure 1).Whereas water transpires through the stomatal pores, carbon dioxide diffuses into the leaf via these openings.Many plants can adjust their stomatal conductance to optimize trade-offs between water loss and carbon uptake.This will also affect the relative uptake rates of the rare 13 C and most abundant 12 C isotope in CO 2 : as the stomatal resistance increases, the total fractionation process associated with photosynthesis becomes less discriminatory (i.e., a less strong preference for 12 C over 13 C, e.g., Adnew et al., 2023).
Carbon dioxide also has oxygen isotopes and these oxygen isotopes can exchange with liquid water following dissolution.This exchange will occur mostly inside leaves, because of the presence of the enzyme carbonic anhydrase (Francey & Tans, 1987).Observations of the rare stable isotopes 17 O and 18 O in CO 2 thus reflect the isotopic signature of water in plants, which can greatly vary at the leaf scale.Because the oxygen isotopes in CO 2 carry information from the plant they have been proposed to study photosynthesis (Koren et al., 2019) but are also used for studying plant properties (e.g., mesophyll conductance, Adnew et al., 2023).Thus, beyond the water isotopes studied by Gai et al. (2023), isotopes in CO 2 can reveal ecohydrological processes and combining these information sources could be a next avenue of research.Gai et al. (2023) report on the preferred isotopic tracers for water source partitioning.In addition to the performance of these tracers, one could also consider other practical aspects when selecting a suitable tracer, for example, the availability of instruments and skilled personnel to perform these measurements.Practical considerations may even dictate the selection of tracers and thus overrule the recommendations by Gai et al. (2023).Further, as we have seen rapid technological development in measurement techniques over the past decades, it is possible that some tracers will become better suited for determining water sources.

Developments of Measurement Technologies
At some point, the opportunity to study "second order" isotopic tracers such as deuterium-excess, 17 O-excess and isotopic clumping becomes feasible to a larger group of researchers.Such isotopic tracers rely on small anomalies with respect to another isotopic tracer (in the case of deuterium-excess and 17 O-excess, e.g., C. Tian et al., 2018) or with respect to the statistical expectations (in the case of isotopic clumping, Eiler, 2007).Although these tracers might not be ready for extensive operational use currently, these could provide additional constraints and be considered in future research.

Future Potential of Isotopic Water Tracers
The environment is transitioning due to climate change and direct human interference in the landscape, affecting the cycling of water through soil, vegetation and atmosphere.Gai et al. (2023) demonstrated how isotopes and mixing models can be used to trace back the water sources of plant water, building on synergies between isotopic measurements and numerical models.Further development of isotope measurement technologies and modeling approaches hold strong potential for advancing our understanding of moisture flows through the soil-vegetationatmosphere continuum to improve our preparedness for weather extremes, promote terrestrial carbon sequestration and enhance food security.

Figure 1 .
Figure1.Simplified overview of water exchange between plants, soil layers and atmosphere in liquid phase (blue arrows) and changing from liquid to vapor phase (purple arrows).Also included is the intricately coupled photosynthetic carbon uptake, mediated through vegetation (red arrows).The smaller, dashed lines indicate how a plant might respond during drought conditions: a reduction of transpiration and photosynthesis following partial stomatal closure, and a shift to deeper soil layers for water extraction.The depicted processes can lead to changes in isotopic composition due to fractionation, in particular, evaporation and transpiration lead to an enrichment of the heavy 2 H and 18 O isotopes in the remaining liquid water.