Continuous in situ measurements of water stable isotopes in soils, tree trunk and root xylem: Field approval

Rationale: New methods to measure stable isotopes of soil and tree water directly in the field enable us to increase the temporal resolution of obtained data and advance our knowledge on the dynamics of soil and plant water fluxes. Only few field applications exist. However, these are needed to further improve novel methods and hence exploit their full potential. Methods: We tested the borehole equilibration method in the field and collected in situ and destructive samples of stable isotopes of soil, trunk and root xylem water over a 2.5-month experiment in a tropical dry forest under natural abundance conditions and following labelled irrigation. Water from destructive samples was extracted using cryogenic vacuum extraction. Isotope ratios were determined with IRIS instruments using cavity ring-down spectroscopy both in the field and in the laboratory. Results: In general, timelines of both methods agreed well for both soil and xylem samples. Irrigation labelled with heavy hydrogen isotopes clearly impacted the isotope composition of soil water and one of the two studied tree species. Inter-method deviations increased in consequence of labelling, which revealed their different capabilities to cover spatial and temporal heterogeneities. We applied the novel borehole equilibration method in Our experiment reinforced the potential of this in situ method for xylem isotopes in tree and roots and confirmed the of gas permeable soil probes. However, in situ xylem measurements

Rationale: New methods to measure stable isotopes of soil and tree water directly in the field enable us to increase the temporal resolution of obtained data and advance our knowledge on the dynamics of soil and plant water fluxes. Only few field applications exist. However, these are needed to further improve novel methods and hence exploit their full potential.
Methods: We tested the borehole equilibration method in the field and collected in situ and destructive samples of stable isotopes of soil, trunk and root xylem water over a 2.5-month experiment in a tropical dry forest under natural abundance conditions and following labelled irrigation. Water from destructive samples was extracted using cryogenic vacuum extraction. Isotope ratios were determined with IRIS instruments using cavity ring-down spectroscopy both in the field and in the laboratory.
Results: In general, timelines of both methods agreed well for both soil and xylem samples. Irrigation labelled with heavy hydrogen isotopes clearly impacted the isotope composition of soil water and one of the two studied tree species. Intermethod deviations increased in consequence of labelling, which revealed their different capabilities to cover spatial and temporal heterogeneities.
Conclusions: We applied the novel borehole equilibration method in a remote field location. Our experiment reinforced the potential of this in situ method for measuring xylem water isotopes in both tree trunks and roots and confirmed the reliability of gas permeable soil probes. However, in situ xylem measurements should be further developed to reduce the uncertainty within the range of natural abundance and hence enable their full potential.

| INTRODUCTION
Plant transpiration fuels the hydrological cycle by returning 35% to 90% of water from land surfaces to the atmosphere. 1,2 Therefore, transpiration greatly influences our climate and impacts water availability in consequence of land use and climate change. 3,4 However, why is the quantification of this essential water flux so uncertain? This comes down to knowledge gaps in the mechanistic functioning of root water uptake (RWU) as well as plant rooting depth that persist despite its crucial role in predicting the future of one of our most important resources. 5 A major constraint is the practical difficulty in observing below-ground processes. Moreover, the magnitude and location of RWU are the results of multiple influencing factors, such as extent of the root system and its hydraulic properties, soil water availability as well as water potential gradients, and their complex interactions. 6 In addition, RWU changes dynamically over short periods of time. 7,8 Water stable isotopes are an essential tool to shed light on hardto-observe below-ground processes and have been used for decades to trace water fluxes across the soil-plant-atmosphere continuum. 9 They help to investigate which water sources, such as water in different soil depths or groundwater, plants actively use. This has provided valuable insights into the functioning of plants and their impact on overall ecosystem water cycling (e.g., reviewed in Sprenger et al 10 and Rothfuss and Javaux 6 ). These tracers are well suited to investigate active RWU because isotope values often vary between ecosystem water pools due to physically well-understood fractionation processes during phase changes as water moves within the hydrological cycle. 11,12 Differences in isotope values, for example, between soil depths, can also be enhanced by the addition of water enriched in heavy isotopes, mostly 2 H, applied as surface irrigation 8,[13][14][15] or to specific depths in the soil profile. 16 This can be utilised to investigate the contribution of different water sources, for example, soil depths to plant water use. Furthermore, RWU and water transport in suberised plants are generally believed not to alter the isotope value for most plants 17 except for xerophytic and halophytic species. [18][19][20] Water transported in the xylem then reflects a weighted mixture of all water (with different isotope values) taken up at a specific point in time.
Lately, this long-standing principle has been questioned repeatedly, and a multitude of factors are discussed that complicate the estimation of RWU depth with water stable isotopes. These factors include, for example, fractionation during RWU, 21,22 time lag from precipitation to RWU to xylem sampling, exchange of xylem water with tree storage water and spatial and temporal heterogeneity in soils and plant xylem. 23,24 Analysis of water stable isotopes in soil and plant material has traditionally been performed using destructive sampling and subsequent extraction of water in the laboratory, with cryogenic vacuum extraction (CVE) being the most common method. 25 This captures only snapshots in time 26 and has hindered the assessment of the above-mentioned isotope effects, impairing a deeper mechanistic understanding of ecohydrological dynamics.
Following the invention of laser-based isotope spectrometry, 27 efforts were directed at developing new methods to measure water stable isotopes in precipitation, 28 soils 29-31 and plant xylem 32,33 in situ. These methodological advances hold great promise to gain a better understanding of at least some of the just-mentioned issues by enabling more frequent measurements with a lower disturbance of the system monitored. For instance, new in situ techniques have been used to show the strong short-term variations in RWU of an herbaceous species in a soil column experiment. 7 Several studies have applied in situ soil water isotope measurements in the field 8,[34][35][36] ; however, methods are still in the development phase and could benefit from further improvement.
Specifically, no uniform protocol regarding sampling setups, calibration as well as data analysis exists, 23 and it would be desirable to identify the simplest approach to get reliable data. Little progress has been made in including in situ analysis of xylem water stable isotopes in field studies, and a systematic comparison to destructive samples over an extended time period is missing. However, this is essential to assess inter-method comparability and identify strength and weaknesses associated with each methodology. The simultaneous in situ measurement of water stable isotopes in soils and plant xylem will allow us to dig deeper into the complexity of ecosystem water fluxes. 23 In this study we determined water stable isotope values in situ and via destructive sampling and subsequent CVE in both soil water and tree xylem of two tropical dry forest species. For this, we applied the borehole equilibration method 32 in a field experiment for the first time. This additionally allowed to measure the isotopic composition within root xylem repeatedly, which, to the best of the authors' knowledge, has not been performed before. Data were collected over a period of 2.5 months. During this time, we determined natural abundance δ values first and hence followed the systems reaction to labelled irrigation events. In contrast to previous studies, we measured both soil and xylem δ values in situ to obtain an unprecedented temporal resolution of obtained data and compare this with repeated destructive sampling over the entire experimental period.
The specific objectives were to:   , precipitation and destructive samples were analysed using   an autosampler and different laser spectroscopy analysers (L2130-i,   Picarro Inc., for extracted plant and soil water samples and IWA-45EP, LGR, San Jose, CA, USA for water samples).

| Manipulation of soil water isotope values by irrigating with 2 H labelled water
The experiment started in February under natural and dry conditions.
Thereafter, we manipulated the system by applying irrigation events with tap water and water artificially enriched in 2 H, that is, labelled water. Irrigation was distributed evenly using a garden hose within two plots enclosing investigated tree individuals. Their area was 211 and 139 m 2 for SM and SC, respectively. Labelling such large areas has the advantage that label is distributed equally around stems.
Due to high costs of label water and issues of practicability, often only small areas 16 or parts of the root system 33  3) were located in close proximity to each other and to soil water isotope profiles (compare Figure S1, supporting information  Figure 7). Those were sealed tightly with plasticine. All boreholes as well as isotope standards (see below) were combined in one system, which allowed for automatic switching between measurements via solenoid magnetic valves (2-Way Elec. Valve, EC-2-12, Clippard Minimatic, Cincinnati, OH, USA). Switching between measurements was controlled via a custom-made programme that was run on the CRDS analyser. To reduce total tubing length, one manifold with mounted valves was placed in each of the two plots. Both of them had a flushing port connecting the ends of the manifold mounts that enabled clearing of water vapour from the tubing path between manifolds and the CRDS analyser in between measurements and allowed us to detect and remove condensation.
To conduct a measurement of one particular borehole, one pair of Therefore, the isotopic difference between both phases can be calculated using well-established equations and depends on the temperature (T) at the location of the phase change. 30,44 To approach isotope equilibrium fractionation between liquid xylem water and sampled water vapour, air throughflow must be slow enough (depending also on trunk diameter) for sampled air to be saturated. 32 To decrease the risk of condensation, heating lines (Quintherm ILLw, 10 W/m; Quintex GmbH, Baden-Württemberg, Germany) were attached along all sample lines, from borehole outflows, over manifold mounts and hence to the CRDS analyser. With the high air temperatures at the field site, water content of sampled air was mostly too high to conduct measurements with the CRDS analyser.
Therefore, we used a second mass flow controller (Serie 358, ANALYT-MTC GmbH, Müllheim, Germany) to dilute sampled water vapour with dry air.
Type-T thermocouples (copper-constantan) were inserted via a separate-tightly sealed-small access hole to record T directly inside boreholes. Ten measurements were recorded every 15 min, for each investigated tree compartment at one individual per species.
Temperature data were stored on a data logger (CR1000X, Campbell Scientific Inc., Logan, UT, USA) and used to derive δ values of liquid water from measured vapour values.

| In situ soil water isotope probes
The stable isotope ratios of soil water were determined in situ on the basis of direct water vapour equilibration (e.g., 30,31 ). In this experiment, we used a pull-only system (e.g., 31 ) with self-made soil installed at the same depths in both profiles.

| Standardisation
In situ soil and xylem water stable isotope measurements were standardised using three water vapour standards that were integrated into the automated system. This allowed us to measure standards in the same phase and with similar preconditions as the soil and xylem δ    Over the course of the experiment, evaporation, artificial labelled irrigation and natural precipitation impacted soil water content and δ values. Figure 3 shows the timing and depths of irrigation and

| Tree xylem water δ values
The timelines of δ 2 H measured in situ in different boreholes are compared against CVE samples in

| DISCUSSION
We applied the novel borehole equilibration method for the first time in a field experiment, 32   To compare the two methods, data are summarised as boxplots next to the timeline. For this, we selected in situ measurements within a period of ±2 days around destructive sampling. The box represents the 25% and 75% quartile. Significant differences between mean values are indicated by an asterisk (Mann-Whitney U-test: P < 0.05). The timings of labelled irrigation events (orange) as well as the first precipitation event (blue) are displayed as vertical dotted lines [Color figure can be viewed at wileyonlinelibrary.com] a comparison was necessary to identify how in situ methods could be improved to reach their full potential in depicting isotope dynamics in unprecedented temporal resolution. One should keep in mind that, despite its widespread application over the past decades, recent findings have questioned the reliability of CVE. [54][55][56] Furthermore, many issues and uncertainties raised lately, for example, representation of natural heterogeneity, influence of the method used and water pools sampled, exchange of xylem water with phloem or tree storage water as well as impact of organic contamination, 23,24 apply to both in situ and destructive sampling.
Attempting to enhance novel in situ methods, this discussion focuses on issues related to their application. In addition, methods differ in the water pool they sample 57 and hence inter-method differences likely do not only arise from associated methodological difficulties.

| Soil water δ values
At natural abundance levels, in situ soil measurements overall agreed well with CVE samples (compare Figure 2). The trend to lower δ values in 5 cm soil depth measured in situ, as compared to CVE, could arise from a bigger influence of ambient air intrusion into the very dry top soil. 7,58 Noticeable differences between the two methods occurred after the application of labelled irrigation. reason but concluded that the most likely cause for the observed discrepancy was that in situ and destructive methods differ in their ability to represent spatial heterogeneity of soil water. In general, spatial isotopic heterogeneity of soil water δ values was often neglected in past studies, even though it can be substantial 59 and is likely the rule rather than the exception. 60 Quade et al 36  "good" soil probe data as well as measurements affected by condensation. In consequence, only in situ soil data with previous F I G U R E 6 Diurnal patterns for both δ 2 H (A to C) and δ 18 O (D to F) per tree compartment investigated. We used in situ data points during natural abundance measurements, that is, before the first labelled irrigation and summarised them in 3 h bins. δ values were normalised to the respective mean value of every borehole flushing was included in the final data set. This decreased temporal resolution of obtained time series by 70% and hence leaves room for method improvement. Condensation was also identified as an important source of error by Beyer et al, 23 Kübert et al, 35 Marshall et al, 32 Oerter and Bowen, 29 Quade et al 36

and Volkmann and
Weiler 31 and other coping mechanisms, for example, immediate dilution of sampled water vapour with dry air are discussed herein.
Further differences between in situ and destructive soil sampling, as well as criteria for a well-grounded choice of method, for example, cost and time expenses, technical equipment of the field site and required temporal and spatial resolution, can be found in Beyer et al, 23 Kübert et al 35  Similarly, it is expectable that trees would alter boreholes over time.
Up until now, we do not know for how long boreholes stay responsive and if, for example, measurements across a vegetation period or a whole year would be impacted by tree wound response.
This likely depends on species, ambient conditions and the set-up used and still needs to be investigated systematically. With this in mind, it is not surprising that trees complicated our measurements in unexpected ways even within our rather short experimental period.
The following summarises our experiences from the field and provides some thoughts for further developing in situ measurements of xylem δ values.
Borehole measurements were complicated by tree individuals producing liquid that drained into and blocked sample tubing. This happened even though boreholes were initially flushed with acetone in hope to kill cells that produce pitch or other substances in response to wounding, as suggested by Marshall et al. 32 We observed that SC individuals generated a brownish fluid after they ceased transpiration, that is, when SC1 exchanged its entire foliage, and produced latex with increasing stem water content at the beginning of the wet season. Based on this observation, we hypothesised that this is connected to the tension with which water is held within the tree xylem. Potentially, water is transported from cells surrounding xylem vessels to refill embolised vessels 64 at times of low transpiration and hence lowered tension pulling water upwards from the soil to the leaves. This refilling is known to even cause positive sap pressure in spring before bud break in some temperate tree species, for example, birch. 65 The issue of liquid blocking sample tubing could potentially be prevented by installing GPT into tree boreholes, similar to in situ soil probes and probes used in the method test conducted by Volkmann et al. 33 However, it remains to be tested whether produced tree sap could block membrane pores, therefore alter and delay xylem δ values and consequently introduce further uncertainties into measurements.  It also provides useful considerations for post-processing and is intended as a hands-on guide for other field scientists applying the method.
Next to the influence of condensation as described above, Organic molecules are known to affect IRIS (isotope-ratio infrared spectroscopy, including CRDS) measurements. 67 Figure 4, time series for SC2 lateral root). The higher frequency of collected data points also provided valuable insights into measurement precision and allowed for a robust estimation of associated uncertainties, unlike restricted sample amounts using traditional approaches. Measurements were also accessible in real-time directly in the field, which makes it easier to spot and solve problems. It also provides an opportunity for targeted sampling of plant physiological variables, for example, at the time of label uptake.
Maintenance of the in situ system and post-processing were time consuming. Method development should therefore aim at simplifying the approach to facilitate access to more researchers within the interdisciplinary field of ecohydrology 23 and to ideally achieve unattended, continuous measurements. At this point it cannot be definitely assessed if the latter will ever be possible or will remain an idealised conception. In addition, this dream is based on installing and maintaining rather complicated technical set-ups and deploying measurement devices directly in the field. Where this is not desired or feasible, we could also envision a semi in situ set-up where water vapour is sampled from boreholes in a comparable way as described here but then stored in inert and gastight containers, transported to the laboratory and analysed there for its isotopic composition.
Although measurements then cannot be conducted automatically, it would decrease the possibility for leaks and condensation due to shorter tubing lengths and a less complex system.

| Borehole equilibration as a novel possibility to measure δ values of root xylem water
With this data set, we showed that the borehole equilibration method allows measuring xylem δ values in roots with a minimum borehole length of 5 cm 32 and hence provides new opportunities to investigate plant water use and within plant water transport and mixing. Literature exists on measuring water fluxes in parts of the root system to disentangle hard-to-observe below-ground processes and the root system's contribution to overall tree water uptake and transpiration. 71 For example, David et al 72 monitored sap flow in the trunk as well as in superficial roots of Quercus suber over a period of 1.5 years and used the data to estimate contributions of shallow soil and groundwater to overall tree water uptake with changing seasonality.
To the best of the authors' knowledge, time series of water stable isotopes in tree roots have not previously been measured.
As expected, due to their proximity to and presumably high proportion of fine roots in the labelled top soil, δ 2 H in lateral roots of SM in general increased faster and to a larger extent than in other compartments in response to labelling events. Meanwhile, δ 2 H changed least in tap roots. An exception to this is SM2, where we could not reach the tap root located directly below the trunk and chose a different root that seemed to be extending downwards.
However, measured δ values suggested that it rather classified as a lateral root. This clearly shows that we are restricted in predicting the extent and functioning of distinct roots by assessing a small section of the entire root system and shows the possibilities that arise from determining δ values within tree roots. Like for SM, the highest δ 2 H was also found in a lateral root for SC (SC2). Surprisingly, for SC water within lateral roots and trunk xylem was not impacted markedly by labelling.
It should be noted that measurements of SM1 tap root yielded unreasonable δ values. This was especially noticeable for δ 18 O, where values were about 10‰ higher than in tap roots of other individuals (see Figure 5). Concurrently, we also observed higher δ 2 H values. The increase was, however, less remarkable due to the higher uncertainty

| CONCLUSION
We collected a unique data set in a semi-arid, tropical dry forest in the northwest of Costa Rica, illustrating both spatial and temporal heterogeneity of water stable isotopes using novel in situ methods and destructive sampling of soil and xylem water. We applied the borehole equilibration method for the first time in a field experiment and proved that in situ tree water stable isotope measurements are possible at a high temporal resolution over a period of several months. Having a consistent sampling point in the tree xylem allowed, for the first time, monitoring of water stable isotopes repeatedly in root xylem, which opens up new possibilities to investigate tree RWU.
Following multiple irrigation events with different levels of 2 Henrichment, we observed the changes in isotope values as applied water moved through the soil and the trees' roots and trunks. In our case, single irrigation events were not clearly propagated to the isotopic composition of xylem water. Therefore, the high temporal resolution recorded would not have been necessary to depict xylem isotope dynamics. However, it enabled a thorough evaluation of the methods precision when applied in the field, which is not possible to the same extent for traditional destructive sampling.
The time courses of the two methods, that is, in situ and destructive sampling, in general agreed well. For soil water, systematic inter-method differences occurred after labelling and were mainly attributed to the strong isotopic gradient as well as spatial for their many constructive comments and effort to improve the