Volume 217, Issue 1 p. 105-116
Full paper
Free Access

The effect of 18O-labelled water vapour on the oxygen isotope ratio of water and assimilates in plants at high humidity

Marco M. Lehmann

Corresponding Author

Marco M. Lehmann

Laboratory of Atmospheric Chemistry, Paul Scherrer Institute (PSI), 5232 Villigen, Switzerland

Swiss Federal Institute for Forest, Snow and Landscape Research (WSL), 8903 Birmensdorf, Switzerland

Author for correspondence:

Marco M. Lehmann

Tel: +41 44 739 2199

Email: [email protected]

Search for more papers by this author
Gregory R. Goldsmith

Gregory R. Goldsmith

Laboratory of Atmospheric Chemistry, Paul Scherrer Institute (PSI), 5232 Villigen, Switzerland

Schmid College of Science and Technology, Chapman University, Orange, CA, 92866 USA

Search for more papers by this author
Lola Schmid

Lola Schmid

Laboratory of Atmospheric Chemistry, Paul Scherrer Institute (PSI), 5232 Villigen, Switzerland

Swiss Federal Institute for Forest, Snow and Landscape Research (WSL), 8903 Birmensdorf, Switzerland

Search for more papers by this author
Arthur Gessler

Arthur Gessler

Swiss Federal Institute for Forest, Snow and Landscape Research (WSL), 8903 Birmensdorf, Switzerland

Search for more papers by this author
Matthias Saurer

Matthias Saurer

Laboratory of Atmospheric Chemistry, Paul Scherrer Institute (PSI), 5232 Villigen, Switzerland

Swiss Federal Institute for Forest, Snow and Landscape Research (WSL), 8903 Birmensdorf, Switzerland

Search for more papers by this author
Rolf T. W. Siegwolf

Rolf T. W. Siegwolf

Laboratory of Atmospheric Chemistry, Paul Scherrer Institute (PSI), 5232 Villigen, Switzerland

Swiss Federal Institute for Forest, Snow and Landscape Research (WSL), 8903 Birmensdorf, Switzerland

Search for more papers by this author
First published: 20 September 2017
Citations: 40


  • Our understanding of how temporal variations of atmospheric water vapour and its isotopic composition (δ18OV) influence water and assimilates in plants remains limited, restricting our ability to use δ18O as a tracer of ecophysiological processes.
  • We exposed oak (Quercus robur) saplings under wet and dry soil moisture conditions to 18O-depleted water vapour (c. − 200‰) at high relative humidity (c. 93%) for 5 h, simulating a fog event. We then traced the step change in δ18OV into water and assimilates (e.g. sucrose, hexoses, quercitol and starch) in the leaf lamina, main veins and twigs over 24 h.
  • The immediate δ18OV effect was highest for δ18O of leaf lamina water, but 40% lower on δ18O of main vein water. To a smaller extent, we also observed changes in δ18O of twig xylem water. Depending on the individual assimilation rate of each plant, the 18O-label was partitioned among different assimilates, with highest changes in δ18O of starch/sucrose and lowest in δ18O of quercitol. Additionally, 18O-label partitioning and allocation towards leaf starch and twig phloem sugars was influenced by the plant water status.
  • Our results have important implications for water isotope heterogeneity in plants and for our understanding of how the δ18O signal is incorporated into biomarkers.


The oxygen isotope ratio (δ18O) in plant tissues is an important tool to study plant functional responses to environmental changes, palaeoclimatic conditions and ecohydrological cycling (Barbour, 2007; Zech et al., 2014; Evaristo et al., 2015). Still, our understanding of how variations in leaf water isotopes in response to changes in water sources or environmental conditions are incorporated into organic matter is limited. Our knowledge is based mainly on leaf and tree-ring cellulose δ18O measurements (Saurer et al., 1997; Helliker & Ehleringer, 2002; Treydte et al., 2014), whereas actual δ18O data for sugars are almost absent (Schmidt et al., 2001; Werner, 2003; Lehmann et al., 2017). This leaves a gap between δ18O ratios of leaf water and associated plant biomarkers (e.g. cellulose, hemicellulose, levoglucosan).

The δ18O of organic matter is determined on the one hand by the isotopic fractionation of water by evaporation and transpiration, and on the other by biochemical isotope fractionations during and after photosynthesis. Both require careful consideration in order to reliably apply δ18O as a tool for ecophysiology (Gessler et al., 2014). The δ18O of water in plant tissues depends on the δ18O of the local source water (δ18OS), which is generally taken up by plants without isotope fractionation and transported via the transpiration stream to the leaves, where it undergoes evaporative 18O-enrichment (Farquhar et al., 2007). However, atmospheric water vapour and its isotopic composition (δ18OV) also influence leaf water isotopes (δ18OLW) and are thus an additional important consideration for isotopic variations in plant tissues. From an isotopic perspective, this should mainly be caused by a bi-directional exchange of water isotopes through open stomata (Farquhar & Cernusak, 2005; Kim & Lee, 2011; Goldsmith et al., 2017). This can be modelled according to common leaf water models (Craig & Gordon, 1965; Dongmann et al., 1974):
urn:x-wiley:0028646X:media:nph14788:nph14788-math-0001(Eqn 1)
eq and εk, equilibrium and kinetic fractionation factors; ea/ei, water vapour partial pressures outside and inside of the leaf). The influence of δ18OV on δ18OLW increases as a function of relative humidity (RH; White et al., 1994; Roden et al., 2000; Helliker, 2011; Voelker et al., 2014). Thus, atmospheric conditions with high RH such as dew, fog and rainfall should have especially strong impacts on the leaf water isotope composition (Burgess & Dawson, 2004; Helliker & Griffiths, 2007; Breshears et al., 2008; Kim & Lee, 2011), for example, wet tropical ecosystems may be expected to have high diffusion rates from atmospheric water vapour into the leaf (Lai et al., 2008; Goldsmith et al., 2013; Hu & Riveros-Iregui, 2016). As RH increases, ea/ei approaches unity and reduces the driving gradient for water loss from the leaf (i.e. transpiration). Assuming ea/ei equalling 1, Eqn 1 can be simplified to:
urn:x-wiley:0028646X:media:nph14788:nph14788-math-0002(Eqn 2)

Under such conditions, besides minor variations in εeq due to temperature changes (Horita & Wesolowski, 1994), δ18OV variations exert the sole effect on δ18OLW (Dongmann et al., 1974).

Interestingly, δ18OV is mostly assumed to be in isotopic equilibrium with δ18OS and thus is often neglected as an additional uncertainty factor in many studies (e.g. Förstel & Hützen, 1983; Keitel et al., 2006; Brandes et al., 2007; Voelker et al., 2014). However, a potential equilibrium between δ18OV and δ18OS has rarely been verified and rather a disequilibrium has been observed (Bögelein et al., 2017). Moreover, there is increasing evidence from meteorological studies that δ18OV strongly varies by up to 20‰ during a diurnal cycle and clear seasonal variations have also been observed (White & Gedzelman, 1984; Lee et al., 2006; Huang & Wen, 2014; Yu et al., 2015). Thus, δ18OV variations should be more carefully considered when interpreting δ18O of both leaf water and the resultant organic matter (Roden et al., 2000). However, only a few studies have experimentally investigated the temporal dynamics of δ18OV induced variations in δ18OLW over short timescales (e.g. hours to days; Roden & Ehleringer, 1999; Helliker & Griffiths, 2007; Kim & Lee, 2011). Those studies were focused mainly on bulk leaf water isotope ratios, whereas to our knowledge, potential varying effects of δ18OV on water and assimilates from different leaf tissues (i.e. lamina or main vein) or from twigs have thus far been neglected.

If the impact of atmospheric water vapour on leaf water isotopes is high, a large proportion of δ18OV is probably incorporated via photosynthesis into plant organic matter or individual assimilates (Roden & Ehleringer, 1999; Gessler et al., 2007). During photosynthesis, an average biosynthetic fractionation factor of c. 27‰ causes a relatively stable offset between δ18O of fresh assimilates and δ18O of the synthesis water (Sternberg & DeNiro, 1983; Roden et al., 2000). The fractionation factor is assumed to result from rapid oxygen isotope exchange between leaf water and carbonyl groups of trioses during photosynthesis in the mesophyll water of the chloroplast (Sternberg & DeNiro, 1983; Sternberg et al., 1986; Yakir & DeNiro, 1990). Moreover, the isotopic signal in assimilates derived from the mixture of xylem and atmospheric water sources in leaf water may be further altered after photosynthesis in downstream metabolic processes in different tissues, for example, via oxygen isotope exchange of carbohydrate carbonyl groups with leaf water, in hydrolysis reactions, and due to equilibrium and kinetic isotope effects in various enzymatic catalysed reactions (Schmidt et al., 2001; Lehmann et al., 2017). However, it remains poorly understood how the xylem–atmosphere mixture in leaf water is incorporated and partitioned into different assimilate pools at a compound-specific level (Lehmann et al., 2016, 2017).

Moreover, so far only a few studies have attempted to follow the oxygen isotope signal across different metabolic fractions (e.g. organic matter, water soluble compounds, bulk sugars) and plant tissues (e.g. leaves and twigs; Cernusak et al., 2003; Offermann et al., 2011; Gessler et al., 2013; Lehmann et al., 2017). Although it has been indicated that drought may influence carbon (C) storage and the oxygen isotope signal allocation into biomass (Picon et al., 1997; Pflug et al., 2015), our understanding of the rate and extent to which the δ18O signal is translocated throughout the plant under changing environmental conditions remains limited (Studer et al., 2015). This restricts the use of the δ18O of biomarkers, such as stem cellulose for climate reconstructions (Treydte et al., 2014; Pflug et al., 2015).

In order to investigate the effects of atmospheric water vapor on δ18O of plant tissue, we exposed oak (Quercus robur) saplings under wet and dry soil water regimes to 18O-depleted water vapour (c. −200‰) during a 5 h fog event at high RH (c. 93%). We then followed the 18O-label over a 24 h period into water and assimilates in leaves and twigs, using a novel approach to measure δ18O of individual carbohydrates (Lehmann et al., 2016). We hypothesized that the step change in δ18OV during the fog event is differentially affecting δ18O of water and assimilates in lamina and main vein tissues; that the 18O-label uptake from the atmosphere is differentially partitioned into individual assimilates dependent on the assimilation rate of each individual plant; and that the 18O-label partitioning into different assimilates and its allocation from the leaves towards twig phloem (i.e. sugars) depends on the plant water status.

Materials and Methods

Experimental setup

Oak (Quercus robur L.) tree seeds were collected in a forest close to Rottenschwil, Switzerland and seedlings grown in the garden at WSL Birmensdorf, Switzerland. Seedlings were then transferred to 5-l pots with standard potting soil (Ökohum, Herrenhof, Switzerland). The experiment was carried out with 2-yr-old oak saplings, with the height varying between 100 and 160 cm. Plants were located in a glasshouse with an average air temperature of 23.7 ± 3.4°C and relative humidity (RH) of 58.0 ± 8.0% (both mean ± 1 SD), and a mean maximum light intensity of c. 460 W m−2. Plants were regularly watered with tap water (δ18O = −10.8‰). After 4 wk of acclimation, half of the plants were no longer watered (dry treatment). The experiment was conducted 3 wk later, when a clear difference in predawn leaf water potential between wet and dry plants was observed. Five individuals each from the wet and dry treatments were placed in a 6 m2 A-frame tent with a 2 m height. The tent was fully enclosed with a light permeable plastic foil (Suter-Kunststoffe, Fraubrunnen, Switzerland). Fog was produced by ultrasonic mist generators (CIS Product, Courtry, France) placed in buckets of 18O-depleted water (δ18O = −367.4‰). Fans were used to facilitate distribution of humid air throughout the tent. Pots were covered by aluminum foil to prevent labelled water from entering the soil. During the 5 h fog period (09:30–14:30 h), photosynthetic photon flux density (PPFD; CR10X logger, Campbell, Leicestershire, UK with SQ110 sensors, Apogee, Logan, UT, USA), as well as RH and temperature (HP23-A loggers with HygroClip2 probes; Rotronic, Bassersdorf, Switzerland), were monitored every minute in the tent. Average PPFD was 94.2 ± 38.1 μmol m−2 s−1, whereas temperature and RH averaged 20.0 ± 0.7°C and 93.0 ± 7.9% (all mean ± SD). No apparent condensation on the leaves was observed during the fog period. After fogging, plastic foil was immediately removed from the tent to facilitate mixing with fresh air. The δ18O ratios of soil water showed no significant change after fogging.

Sampling of plant material and water vapour

In order to determine the δ18OV effects on δ18O of plant water and organic material, sampling was carried out several times during a 24 h cycle on the day of the fog event. Leaf material was harvested nine times (before fogging at 0 h, during fogging at 1, 2.5 and 4 h, and after fogging at 5, 10, 12, 21 and 24 h) in the middle of the plant crown. Leaves were separated into leaf lamina and main vein samples, then quickly transferred to sealed 12-ml exetainers (Labco, Lampeter, UK). Samples were immediately frozen in liquid N2 and stored at −20°C until further analysis.

Twig material at similar height as leaves was harvested four times (before fogging at 0 h and after fogging at 5, 10 and 24 h). Twigs were separated into xylem and bark samples. Xylem samples were transferred to exetainers and kept frozen at − 20°C until analysis. Bark samples were used for collection of phloem sugar exudates according to Gessler et al. (2013). In brief, bark (c. 9 cm length) was washed twice with deionized water to prevent contamination with water soluble substances from destroyed parenchyma cells and transferred to 15-ml centrifugation tubes filled with 12 ml deionized water. After 5 h, samples were removed and the water solutions containing the phloem sugars frozen at −20°C. Sugars were subsequently isolated by freeze-drying and the remaining pellet dissolved in deionized water.

In addition, water vapour samples were collected before and approximately once per hour from within the tent during the fog period. Air was pumped through a hose and the water vapour was trapped in a glass U-tube placed outside of the tent in an ethanol slurry that was cooled with liquid nitrogen.

Plant physiological measurements

In order to link the leaf gas-exchange of each individual plant to changes in δ18O of assimilates in response to the 18O-label uptake, we measured the net assimilation rate (An) using an infrared gas analyzer and a 6-cm2 leaf cuvette (Li-Cor 6400; Li-Cor Biosciences, Lincoln, NE, USA). Measurements were performed during (13:45 h) and after (17:15 h) the fog period. Conditions in the cuvette were maintained at a flow rate of 500 μmol s−1, 400 ppm CO2 and a light intensity of 200 μmol m−2 s−1. When stable conditions were reached, point measurements were taken and repeated five times at 10-s intervals. Among plants, leaf temperature and vapour pressure deficit (VPD) were 20.0 ± 0.1°C and 1.4 ± 0.1 kPa (all mean ± SD). Measurements were made on fully developed leaves at medium height on the plant crown. Although no condensation was observed, leaves were carefully wiped with paper tissues before measurement. To be conservative, we assume that some minor amounts of the fog water remained, potentially affecting stomatal conductance and transpiration measurements, which were therefore not used. In addition, leaf water potentials were determined on all plants before dawn (and before the fog event) at 05:00 (ψpredawn) and directly after the fog period at 14:30 (ψmidday), using a Scholander pressure chamber.

Isotopic analyses of water samples

Water from plant material was cryogenically extracted by vacuum distillation (West et al., 2006). δ18O ratios of water samples were measured by a high temperature conversion/elemental analyzer, which was coupled via a ConFlo III to a DELTAPlusXP isotope ratio mass spectrometer (TC/EA-IRMS, all supplied by Finnigan MAT, Bremen, Germany). δ18O ratios were referenced to V-SMOW. The measurement precision of the quality control standard was typically ≤ 0.3‰ (SD).

Isotopic analyses of bulk organic matter, bulk sugars and starch

All cryogenically dried leaf materials were milled to a fine powder. Bulk sugars were extracted from 60 mg of this material in 1.5 ml deionized water at 85°C for 30 min. This fraction was then further purified from ionic and phenolic substances by ion-exchange cartridges (OnGuard II A, H and P, Dionex; Thermo-Fisher Scientific, Bremen, Germany) following the protocol of Rinne et al. (2012). Starch was extracted from 100 mg leaf material in a 1.5-ml methanol/chloroform/water (MCW) solution at 70°C for 30 min, washed with MCW and water, and broken down to sugars using a standard enzymatic hydroxylation method according to Lehmann et al. (2015).

For δ18O analyses, milled bulk organic matter was directly weighed into silver capsules, whereas bulk sugar and starch aliquots were injected into silver capsules, frozen and freeze-dried. The organic material was converted to CO by pyrolysis at 1420°C (Vario pyrocube, Elementar, Hanau, Germany) and delivered via a Conflo III to a DELTAPlusXP IRMS (Finnigan). δ18O ratios were referenced to V-SMOW. The measurement precision of the quality control standard (27.6‰, cellulose) was typically ≤ 0.3‰ (SD). No considerable oxygen isotope fractionation was observed during starch sample preparation.

Compound-specific δ18O analyses of leaf sugars

Purified bulk leaf sugars were methylated according to Lehmann et al. (2016). Briefly, a c. 2-mg freeze-dried sugar pellet was mixed with 26 mg silver oxide, 400 μl acetonitrile, 40 μl methyl iodide and finally with 10 μl dimethyl sulfide to start the derivatization reaction. All samples were strongly mixed and placed in a thermo-shaker (20°C, 560 turn min−1) for 24 h in the dark. Subsequently, the samples were spiked with 20 μl water, centrifuged (10 000 g, 2 min) to remove potential water soluble residues, and the supernatants filtered by 0.45 μm syringe filters (Infochroma, Zug, Switzerland). Commercially available fructose, glucose, sucrose and proto-quercitol were used for a standard mix (≥ 97.0%; HPLC). The retention time of the quercitol peak was c. 15 s earlier than the glucose peak and thus the baseline was generally separated; all other sugar peaks were likewise separated as in previous chromatograms of methylated leaf sugars (Lehmann et al., 2016).

All compound-specific δ18O analyses were performed on a GC/Pyr-IRMS system consisting of a Trace GC Ultra gas chromatograph that was linked via a pyrolysis unit (Isolink) and a Conflo IV to a MAT 253 IRMS (all supplied by Thermo-Fisher Scientific, Bremen, Germany; for a detailed system description, see Lehmann et al., 2016). Methylated sugars were separated on a 60 m × 0.25 mm × 0.25 μm ZB-SemiVolatiles GC column (Zebron, Phenomenex, Torrance, CA, USA) and pyrolysed to CO at 1280°C in a commercial oxygen reactor (Thermo-Fisher). A liquid N2 trap was used to sequester impurities in the samples gas, improving background stability of mass 30 and thus data reproducibility. Generally, a measurement sequence consisted of five samples, which were repeated four times, and interspersed standards of different concentrations to correct for offset to TC/EA-coupling, as well as for drift and amount dependencies (Lehmann et al., 2016). The average measurement precision (SD) for individual sugars was ≤ 0.4‰ for fructose and glucose, ≤ 0.5‰ quercitol and ≤ 0.6‰ for sucrose. No significant oxygen isotope fractionation was observed for bulk and individual sugars (Lehmann et al., 2017).

Calculations and statistical analyses

All δ18O ratios of water and assimilates were related to the prefog natural abundance (∆δ18O) to emphasize the treatment effect according to the following equation:
urn:x-wiley:0028646X:media:nph14788:nph14788-math-0003(Eqn 3)
18Osample, δ18O ratio of a sample at a time point after the start of the fog event; δ18Oprefog, corresponding δ18O ratio before the start of the experiment). Linear mixed effects models with soil moisture treatment, time and their interaction as fixed effects, and individual plants as a random effect, were used to test for significant differences in ∆δ18O ratios of water and assimilates, water potentials, and in assimilation rates. Where there was no significant interaction between treatment and time, we reran the model without the interaction term. Analysis of covariance (ANCOVA) was used to identify if soil moisture treatments significantly affected the slope of the relationship between An and ∆δ18O ratios. If not mentioned otherwise, means and SE are given. All statistical tests were performed in R v.3.4.0 (R Core Team, 2017).


Before the start of the fog event, the drought treatment was successfully established as demonstrated by water potential measurements (Table 1). Water potentials were significantly lower in plants under dry compared to wet soil moisture conditions (F1,8 = 47.3, < 0.001). We also observed no clear differences between predawn and midday Ψ values under both soil moisture conditions (F1,9 = 2.2, > 0.05), indicating that transpiration was suppressed during the fogging period. The net assimilation rates during the fog period tended to be lower in wet compared to dry plants, which was reversed again after the fog period. However, neither soil moisture treatments (F1,8 = 0.1, > 0.05) nor time (F1,9 = 2.6, > 0.05) showed a significant effect on An.

Table 1. Plant physiological parameters of Quercus robur saplings in wet and dry soil moisture treatments
Treatment ѰPredawn (MPa) ѰMidday (MPa) An-fog (μmol m−2 s−1) An-after (μmol m−2 s−1)
Wet − 0.2 ± 0.1 − 0.2 ± 0.1 1.7 ± 0.9 4.0 ± 1.1
Dry − 1.1 ± 0.3 − 0.8 ± 0.3 3.1 ± 0.3 2.9 ± 0.3
  • Parameters include predawn (ѰPredawn) and midday (ѰMidday) water potentials on the day of fogging and the net photosynthetic assimilation rate during (An-fog) and after (An-after) a 5 h fog period. Mean values ± SE are given.

18O-label allocation into water and assimilates from different plant tissues

First, we followed the isotope signal of the 18O-labelled water vapour (δ18OV) into δ18O of water in different tissues over the course of 24 h (Fig. 1). ∆δ18OV ratios (i.e. δ18O of water vapour during fogging – δ18O of water vapor before fogging, Eqn 3) were on average −196 ± 22‰ during the 5 h fog period. The ∆δ18O ratios in leaf lamina water (∆δ18OLLW), main vein water (∆δ18OMW) and twig xylem water (∆δ18OXW) decreased during the course of the fog period. ∆δ18OLLW showed the strongest decrease of 160‰ as average of both soil moisture treatments after 4 h, with a change rate of c. 42‰ h−1, nearly reaching equilibrium with ∆δ18OV by the end of the fog period. ∆δ18OMW showed a pattern similar to ∆δ18OLLW; the maximum decrease, however, was c. 40% lower. Moreover, ∆δ18OXW showed a decrease of 8‰ as average of both soil moisture treatments after the fog period. After the conclusion of the experiment, the ∆δ18O of the different tissue water quickly increased again; however, ratios at 24 h were on average 13‰ (∆δ18OLLW), 6‰ (∆δ18OMW) and 3‰ (∆δ18OXW) lower compared to prefog ratios across treatments. Overall, ∆δ18O of any tissue water was not affected by soil moisture treatments (> 0.05), but was affected by time (< 0.001; Table 2).

Details are in the caption following the image
Oxygen isotope ratios (∆δ18O) of water in different Quercus robur plant tissues before, during and after exposure to a 5 h fog event with a 18O-depleted water vapour source. Mean values and ± SE are given (= 4–5 individuals per treatment).
Table 2. Statistical results for oxygen isotope ratios (∆δ18O) of different water and assimilates in Quercus robur saplings tested by linear mixed effects models
Δδ18O Treatment Time
Leaf lamina water F1,8 = 0.9 F8,72 = 91.4**
Main vein water F1,8 = 0.1 F8,72 = 76.6**
Twig xylem water F1,8 = 1.2 F3,27 = 22.4**
Leaf lamina OM F1,8 = 0.0 F7,63 = 20.5**
Leaf lamina sugar F1,8 = 0.7 F7,63 = 34.0**
Main vein OM F1,8 = 0.5 F3,27 = 31.8**
Main vein sugar F1,8 = 0.2 F3,27 = 38.1**
Phloem sugar F1,8 = 3.6* F3,27 = 23.3**
Glucose F1,8 = 0.0 F3,25 = 41.1**
Fructose F1,6 = 1.0 F3,20 = 20.0**
Sucrose F1,6 = 0.3 F3,21 = 19.7**
Quercitol F1,7 = 0.4 F3,22 = 9.8**
Starch F1,7 = 3.0* F3,23 = 15.9**
  • F-values with df are given for treatment and time. Treatment indicates soil moisture conditions (wet, dry). Values are given in bold if significant (*, P ≤ 0.1; **, P < 0.001). No significant interaction between both effects was observed for any parameter. OM, organic matter.

We then followed the 18O-label into the total leaf organic matter and bulk sugar fraction (Fig. 2). The 18O-label signal of the leaf water was clearly incorporated into photosynthetic assimilates, as demonstrated by a strong linear decrease in ∆δ18O ratios of total leaf organic matter and bulk sugars during the 5 h fog period. ∆δ18O of bulk leaf lamina sugars showed a strongest decrease of 16‰ as average of both soil water treatments after 5 h, with a change rate of c. 3.3‰ h−1, whereas ∆δ18O ratios of bulk main vein sugars showed a decrease of 11‰. In comparison to ∆δ18O of bulk sugars, the decrease in ∆δ18O of total leaf organic matter was c. 60% lower in both, the leaf lamina and the main vein. After the fog period, ∆δ18O ratios of organic matter and bulk sugars in leaf lamina and main vein stayed relatively constant over the remaining course of the experiment, with a slight tendency to increase. Still, ∆δ18O ratios of organic matter and bulk sugars clearly remained below prefog ratios at the last harvest. ∆δ18O ratios of organic matter and bulk sugars in leaf lamina and main vein were not affected by soil moisture treatments (> 0.05), but by time and, thus, by the fog event (< 0.001; Table 2). We further traced the 18O-label into ∆δ18O of twig phloem sugars. The latter was affected by the soil moisture treatment (P ≤ 0.1) and time (P < 0.001), with about a two-fold higher maximum decrease in wet (8.4‰) than in dry plants (4.5‰) 24 h after fogging start.

Details are in the caption following the image
Oxygen isotope ratios (∆δ18O) of organic matter (OM) and bulk sugars in different Quercus robur plant tissues before, during and after exposure to a 5 h fog event with a 18O-depleted water vapour source. Mean values and ± SE are given (= 5 individuals per treatment).

18O-label partitioning into different assimilates

We also analysed the 18O-label partitioning towards different individual assimilates of the leaf lamina on a compound-specific level (i.e. sucrose, hexoses, quercitol and starch; Fig. 3). Similar to the bulk sugar fraction, a clear decrease in ∆δ18O was observed for the individual sugars, with the strongest decrease in sucrose of c. 29‰ as average of both soil moisture treatments at 5 h (after the end of the fog period) and a change rate of c. 5.7‰ h−1. In comparison, ∆δ18O ratios of fructose and glucose showed a decrease of c. 20‰ as average of both soil moisture treatments, and ∆δ18O of quercitol dropped only by 4‰. ∆δ18O ratios of all individual sugars were not affected by soil moisture treatments (> 0.05), but were affected by time and, thus, by the fog event (< 0.001; Table 2). By contrast, ∆δ18O of starch was affected by the soil moisture treatment ( 0.1) and time (< 0.001), with about a two-fold higher maximum decrease (34.7‰) in dry compared to wet plants (19.8‰) 10 h after fogging start. In addition, results of one-way ANOVAs showed that the magnitude of the 18O-label partitioning was different among assimilates under wet (F4,80 = 4.7, < 0.01) and dry soil moisture conditions (F4,80 = 7.8; < 0.001) 5 h after fogging start. However, only δ18O ratios of leaf starch showed a response to the plant water status.

Details are in the caption following the image
Oxygen isotope ratios (∆δ18O) of individual sugars and starch from the leaf lamina of Quercus robur saplings before, during and after exposure to a 5 h fog event with a 18O-depleted water vapour source. Mean values and ± SE are given (= 4–5 individuals per treatment).

Relationships between An and ∆δ18O ratios of different leaf assimilates

Finally, we found that the photosynthetic assimilation rate was an important physiological parameter related to changes in the ∆δ18O of the different leaf sugar pools across both soil moisture conditions (Fig. 4; Table 3). Five hours after fogging start, we observed strong and significant relationships between An and ∆δ18O of starch, bulk sugars, and the individual sugars fructose, glucose and sucrose (r2 = 0.5–0.8,  0.01), but not for total organic matter and quercitol (> 0.05). The slopes, reflecting the 18O-label of water vapour incorporated per unit photosynthesis, were steepest for starch and sucrose, whereas the slopes for bulk sugars, fructose and glucose were less pronounced. The lowest slopes were observed for total organic matter and quercitol. However, there were also differences in slope depending on plant water status for starch, as indicated by the significant interaction between assimilation rate and soil water treatment effect (Table 3).

Details are in the caption following the image
Relationship between net photosynthetic assimilation rate (An) and oxygen isotope ratios (∆δ18O) of different assimilates across Quercus robur saplings from wet and dry soil moisture conditions (= 7–10 individuals) 5 h after fogging start. Please refer to Table 3 for statistics.
Table 3. ANCOVA results for oxygen isotope ratios (∆δ18O) of different leaf assimilates in Quercus robur saplings as a function of net photosynthesis (An) and treatment
∆δ18O r 2 Slope Intercept A n Treatment An × Treatment
Organic matter 0.3 − 1.1 − 4.7 ns ns ns
Starch 0.5 − 8.8 − 5.0 0.003 ns 0.006
Bulk sugars 0.8 − 4.8 − 4.9 0.002 ns ns
Sucrose 0.8 − 9.5 − 4.1 0.009 ns ns
Fructose 0.6 − 5.6 − 4.2 0.010 0.025 ns
Glucose 0.7 − 6.8 − 4.8 0.011 ns ns
Quercitol 0.1 − 0.3 − 3.1 ns ns ns
  • Treatment indicates soil moisture conditions (wet, dry). Regression coefficient (r2), slope and intercept are given. P-values are shown for An, treatment and their interaction. Nonsignificant results (> 0.05) are indicated by ‘ns’.


The 18O-labelled fog had a clear effect on leaf water isotopes within the first hour, whereas the strongest decrease was observed in leaf lamina water oxygen isotopes (Δδ18OLLW) after 4 h of fogging (Fig. 1). To test if the observed isotope patterns are in agreement with model predictions, we compared modelled and measured δ18O ratios. When assuming 100% relative humidity (RH) in the tent (Eqn 2), we would expect a c. 10‰ difference between δ18OLLW and the oxygen isotope ratio of water vapour (δ18OV) under steady-state conditions due to equilibrium fractionation between vapour and liquid water (εeq). The fact that the modelled δ18OLLW ratio of −208‰ was lower compared to the measured minimum δ18OLLW ratio of −153‰ indicates, however, that this model does not fully apply to our conditions and setup. Therefore, we also applied the steady-state Craig–Gordon model (Eqn 1) using the average RH of 93% during the fog event. This resulted in a modelled δ18OLLW ratio of −191‰, which was, however, still lower compared to the measured minimum δ18OLLW ratio of −153‰. In theory, a RH of 77.5% was necessary to fully predict the measured value. We need to consider that some uncertainty may be inherent within δ18OV measurements, integrating temporal variations within the fog.

Nevertheless, both model results show that the leaf water was close to, but not fully, in isotopic equilibrium with δ18OV. Given the observed temporal dynamics of ∆δ18OLLW, we estimated that steady-state conditions would have been reached after c. 5–6 h under constant fog conditions (Fig. 1), which is consistent with previous findings on leaf water steady-state in tree species (Roden & Ehleringer, 1999). In addition, we calculated a mean residence time of c. 2.5 h for leaf lamina water from the exponential change in ∆δ18OLLW from 4 to 24 h (Ruehr et al., 2009). This value lies well between values determined for other tree species ranging between c. 0.5 and 5 h (Simonin et al., 2013; Dubbert et al., 2014; Volkmann et al., 2016), indicating that the turnover of water pools in the studied oak saplings are comparable to other species.

δ18OV variations differentially influence δ18O of leaf and twig water at high RH

The temporal dynamics of ∆δ18OLLW and main vein water oxygen isotopes (∆δ18OMW) during and after the fog event were similar, but the magnitude of the decrease in ∆δ18OMW during fogging was clearly dampened by 40% (Fig. 1). This demonstrates that water samples from different leaf tissues are differentially affected by δ18OV variations. δ18OV mainly influences δ18O of the leaf lamina water. A consistent proportion of this water is continuously mixed with main vein water, causing the temporal dynamics in δ18O to be similar between both tissues (Fig. 1). The proportion is probably determined by anatomical differences among the leaf tissues, for example length and complexity of intravenous pathways within the leaf (Farquhar & Gan, 2003). Additionally, the dampening in ∆δ18OMW might be caused by dilution due to relatively higher amounts of water present in the main vein compared to the lamina or by mixing with xylem water (Cernusak et al., 2016). Normally, water moves by advective transport from the xylem to the leaf lamina, whereas the diffusive transport in the opposite direction is comparably smaller (Cuntz et al., 2007). However, the high RH and the suppressed transpiration rate in our experiment allowed the leaf lamina water isotopologues to influence those in the main vein by diffusive rather than by advective transport. We assume that an increase in the length of the fog period would have caused δ18OMW to fully equilibrate with δ18OLLW. Alternatively, in the case of remaining low amounts of transpiration and thus of a small, but constant, influx of unlabelled new xylem water, consistently higher δ18OMW ratios compared to δ18OLLW ratios would have been present.

Furthermore, because we observed no effect of fogging on δ18O of soil water, we conclude that the step change in δ18OV was also causing the decrease of 8‰ in xylem water oxygen isotopes (∆δ18OXW) (Fig. 1). This decrease indicates a back-diffusion of leaf lamina and main vein water into the twig xylem when negligible transpiration occurs. Moreover, it is known that phloem water carries part of the leaf water isotopic signal and is thus 18O-enriched compared to xylem water (Cernusak et al., 2003). Exchange of 18O-enriched phloem water with xylem water under low transpiration rates might also partly explain the decrease in ∆δ18OXW observed. Thus, under field conditions with high humidity and suppressed transpiration rates, ∆δ18OXW could still change by 0.8‰ if we assume a δ18OV variation of 20‰ over short timescales, and a comparable effect of this variation on the xylem water pool as in our study (see Fig. 5). In addition, ∆δ18O of leaf water and twig xylem water at 24 h (i.e. 19 h after the end of the fog event) were still 3–13‰ lower in our experiment compared to the prefog ratios. This suggests that information from a previous fog event remains for at least a day in the particular tissue water isotopic composition even though transpiration should have recovered after the end of the fog event.

Details are in the caption following the image
Magnitude of water vapour oxygen isotope ratio (δ18OV) influence on δ18O of water and assimilates of plants (here Quercus robur saplings) at high humidity. δ18OV is differentially affecting water from different leaf tissues, with a higher effect on δ18O of leaf lamina than on δ18O main vein water. In addition, δ18O of twig xylem water was influenced by δ18OV. The δ18OV variations in leaf water are than incorporated into assimilates. However, the extent of the 18O-label uptake is much lower and varies among the different assimilates dependent on photosynthetic assimilation rate and on the plant water status of each individual plant. δ18OV mainly affected δ18O of leaf sucrose and starch, but also δ18O of phloem sugars. Our experimental step change caused a change in δ18OV of c. − 200‰ during a 5 h fog event, whereas the natural diurnal δ18OV variations can amount to c. 20‰, thus being 10% of the variation we applied here. We assume that this 10% effect size scales linearly to water and assimilate pools in leaves and twigs (White & Gedzelman, 1984; Lee et al., 2006; Huang & Wen, 2014; Yu et al., 2015). Under this assumption, the 20‰ variation in δ18OV will still cause considerable changes of 0.3–16‰ in δ18O of different water and assimilate pools at high humidity.

Overall, our results demonstrate that δ18OV was differentially affecting δ18O of different leaf and twig water pools at high RH (Fig. 5). Our findings contribute to observed isotopic gradients and patchiness in leaf water isotopes and may also be important for studies using twig xylem water as an indicator for δ18O of source water (Gan et al., 2002; Evaristo et al., 2015; Feakins et al., 2016).

δ18OV variations are differentially partitioned into different assimilates

The change in oxygen isotope ratio of leaf water (δ18OLW) associated with the change in δ18OV was rapidly incorporated into total leaf organic matter and bulk sugars within 1 h after fogging start (Fig. 2). The lower 18O-label incorporation into total leaf organic matter compared to sugars is probably caused by various structural compounds within the bulk material that were not influenced by the short-term δ18OV variations (e.g. cellulose), dampening the 18O-label effect. Similar to the varying δ18OV effect on water samples from different tissues, we also observed a trend towards greater changes in δ18O of leaf lamina sugars compared to main vein sugars. We assume that this was driven by 18O-label differences in tissue water and by differences in net assimilation rates (An).

Interestingly, 18O-label from the fog was differentially partitioned among the individual leaf lamina assimilates (Fig. 3). Sucrose and starch pools showed the highest 18O-label incorporation in both soil water treatments. This was also true for fructose and glucose, but to a lesser extent. Our results are in agreement with other labelling studies using 13C-labelled CO2 (Streit et al., 2013; Galiano Pérez et al., 2017), indicating that carbon (C) and O allocation in plants are linked at the carbohydrate level. The observed pattern can be explained by the metabolic fluxes during photosynthetic C assimilation. Freshly assimilated C in the form of trioses-phosphates is used partly to build up the transitory starch pool in the chloroplast, whereas another part of the trioses-phosphates is exported to the cytosol, where it is mainly (but not exclusively) used for the production of the sucrose (Buchanan et al., 2015). In fact, Δδ18O ratios of sucrose, hexoses and starch showed strong relationships with An across both soil moisture treatments, with clear differences in slopes and thus 18O-label uptake per unit photosynthesis (Fig. 4; Table 3). This demonstrates that the photosynthetic assimilation rate is a major driver for the incorporation of the oxygen isotopic signal into different sugar pools. Thus, other potential pathways for 18O-label uptake (e.g. via post-photosynthetic processes such as oxygen isotope exchange via carbonyl groups) are of minor importance or their extent is also directly related to An.

The partitioning of the 18O-label depends also on the metabolic activity (i.e. turnover time) of an assimilate pool. The activity of a specific sugar pool may change with species, due to differences in anatomy and physiology of the plant. Because leaves of trees (e.g. oak leaves) need to be able to efficiently distribute recently fixed C within the plant, the mobile sucrose pool and the starch pool (i.e. C storage) are potentially more active than the hexose pool (Fig. 3). However, this might be different in other growth forms and tissues were hexoses are the primary C pool, for example to facilitate cell elongation, cell wall thickness or for osmotic regulation (Simard et al., 2013; Lehmann et al., 2015).

In contrast to the other sugars, quercitol showed less 18O-label uptake and no relationship with An. Quercitol can be found in different oak (Rodriguez-Sanchez et al., 2010) and eucalypt species (Merchant et al., 2007; Arndt et al., 2008). It is chemically similar to common alditols such as myo-inositol and pinitol in other tree species (Gessler et al., 2013; Streit et al., 2013; Rinne et al., 2015). These compounds are known to be produced in the early growing season, probably functioning as osmolytes and increasing plant resistance against cold temperatures (Arndt et al., 2008; Rinne et al., 2015). The alditol concentrations in tree species can account for > 25% of the total sugar fraction, however, these pools show very low turnover times throughout a year and thus their isotopic composition barely reflects information about environmental conditions (Streit et al., 2013; Rinne et al., 2015). We infer that the low turnover most likely explains the low 18O-incorporation in quercitol in our study (Fig. 3). In summary, our results demonstrate that a step change in δ18OV accompanied by high RH is rapidly incorporated and differentially partitioned into key assimilates. We assume that our findings are also relevant at natural δ18OV variations, with associated implications for isotopic biomarkers dependent on leaf water (Fig. 5).

Plant water status influences oxygen isotope allocation

The slower response to changes in δ18Ov of phloem sugars compared to leaf sugars (Fig. 2), which was observed independent of water status, supports previous findings suggesting a temporal delay between sugar assimilation in the leaves and sugar loading into the phloem (Gessler et al., 2013). Interestingly, whereas ∆δ18O of bulk or individual leaf sugars showed no differences between the soil moisture treatments, we observed that twig phloem sugars were more 18O-labelled under wet than under dry conditions (Fig. 2; Table 2). This suggests that sugar loading into the phloem was reduced under drought (Ruehr et al., 2009). Less leaf-to-phloem sugar export may not only lead to a decrease in biomass production (e.g. narrower tree-ring width) but also to dampened oxygen isotope allocation to the respective tissue. This may explain the often observed decoupling of δ18O ratios between leaf sugars and stem cellulose (Offermann et al., 2011; Gessler et al., 2014; Pflug et al., 2015). Additionally, a reduction in phloem loading causes the leaf sugars to be incorporated into alternative sinks such as leaf starch. This hypothesis is supported by our results as we observed more 18O-label partitioning into starch under dry than under wet conditions (Fig. 3; Table 2). However, significantly lower starch concentrations in oak plants under drought might interfere this 18O-label effect (Picon et al., 1997). Thus, the plant water status varies the 18O-label partitioning into different assimilates and influences the allocation of the oxygen isotopic signal from the leaves towards potential sinks within the plant.

Implications and conclusions

Here we demonstrate that a step change in δ18OV by c. −200‰ during a 5 h fog event clearly affected δ18O of water and assimilates from leaves and twig tissues. In comparison, natural δ18OV variations amount to c. 20‰ over the short-term, thus being only c. 10% of the variation we applied here (Lee et al., 2006; Huang & Wen, 2014; Yu et al., 2015). Still, if we assume that this 10% effect size scales linearly to δ18O of water and assimilate, this translates into a change of 16‰ in leaf lamina water, 3‰ in sucrose or starch and 0.8‰ in twig xylem at high humidity (Fig. 5).

We show that partitioning and allocation of δ18OV variations into assimilates are dependent on the photosynthetic assimilation rate and on the plant water status of each individual plant. These traits are known to strongly differ among species from different functional groups and it must be expected that the degree to which δ18OV affects the isotopic composition of assimilates may substantially vary among growth forms (Lai et al., 2008). Because our results are limited to oak, a species survey might help to differentiate plant species according to their susceptibility for δ18OV variations to be imprinted in organic matter.

The immediate effect (< 1 h) of the 18O-labelled water vapour and temporal highly variable δ18OV variations in nature will cause constantly nonsteady-state conditions in δ18O ratios of leaf water and assimilates, which impedes modelling of those ratios. Accordingly, field studies applying nonsteady-state models often make better predictions of δ18OLW (Farquhar & Cernusak, 2005; Helliker & Griffiths, 2007; Lai et al., 2008; Song et al., 2015) and δ18O of transpiration (Welp et al., 2008; Lai & Ehleringer, 2011; Dubbert et al., 2014).

Information on meteorological changes (e.g. air mass movements, changes in RH/temperature and extreme weather events) will be imprinted on the isotopic composition of water and assimilates via accompanied strong diurnal or seasonal δ18OV variations (Lee et al., 2006; Huang & Wen, 2014; Yu et al., 2015). Although this information may also be imprinted by soil water isotopes, δ18OV directly influences leaf water within minutes, causing the information about climate to be imprinted almost immediately and making it more directly accessible. Evaluating the impact of water vapour vs soil water isotopes on δ18O of organic matter, particularly for tree rings, is a challenge for future research.

We introduce fogging as a useful tool to investigate oxygen isotope allocation in water and organic samples from different metabolic fractions and plant organs. It is easy to apply and could be used in combination with 13CO2 labelling to disentangle C and O metabolic pathways, sink–source relationships and compound turnover times, as well as to study hydrogen and oxygen isotope fractionation processes.

Finally, our findings have important implications for ecophysiological and hydrological sciences using oxygen (and potentially also hydrogen) isotopes of various plant biomarkers such as cellulose, hemicellulose and n-alkanes to decipher temporally and spatially integrated information about past environmental conditions and associated plant functional responses (Hepp et al., 2015; Tuthorn et al., 2015). This is especially important for organic molecules derived from plants, sediments, or soils in regions and seasons that experience high humidity conditions (e.g. wet tropical regions, cloud forests, ecosystems subjected to morning dew and storms, such as mountainous regions or in proximity to coasts). However, larger scale isotope-based applications such as dendrochronological approaches, partitioning studies and global vegetation models may also benefit from including δ18OV effects (West et al., 2008; Treydte et al., 2014; Evaristo et al., 2015; Keel et al., 2016).


We thank assistants from ETH Zurich, PSI and WSL for their experimental and technical contributions during the experiment. We acknowledge Kathrin Streit (WSL) for fruitful discussions. This study was financed by the Swiss National Science Foundation (SNF, no. 200020_150003). G.R.G. was funded by the European Community's Seventh Framework Program (FP7/2007-2013) under grant no. 290605 (COFUND: PSI-FELLOW) and by the SNF (no. 31003A_153428). R.T.W.S. acknowledges the funding for the CSIA instrumental support by the SNF (REQIP, no. 206021_128761).

    Author contributions

    M.M.L., G.R.G., A.G., M.S. and R.T.W.S. planned and designed the research; M.M.L., G.R.G., L.S., M.S. and R.T.W.S. performed the experiments; M.M.L., L.S. and M.S. analysed data; and all authors contributed to and accepted the final version of the manuscript.