Biochemical effects of salinity on oxygen isotope fractionation during cellulose synthesis



  • The current isotope tree ring model assumes that 42% of the sucrose oxygen exchanges with stem water during cellulose synthesis and that the oxygen isotope biochemical fractionation is c. 27‰. However, previous studies have indicated that this model can overestimate the cellulose oxygen isotope ratio of plants under salinity or water stress. Saline stress increases soluble carbohydrates and osmolytes, which can alter exchange and biochemical fractionation during cellulose synthesis.
  • To test the effect of salinity as well as the synthesis of osmolytes on exchange and biochemical fractionation, we grew wild-type and a transgenic mannitol synthesizer Arabidopsis thaliana hydroponically with fresh and saline water. We then measured the oxygen isotope ratios of leaf water, stem water and stem cellulose to determine the effects on exchange and biochemical fractionation.
  • Biochemical fractionation did not change, but oxygen isotope exchange was twice as high for plants grown in saline water relative to freshwater-treated plants (0.64 and 0.3, respectively). Mannitol (osmolyte) synthesis did not affect exchange or biochemical fractionation regardless of salinity.
  • Increases in salinity increased oxygen isotope exchange during cellulose synthesis, which may explain the overestimation of cellulose δ18O values under saline conditions.


Tree ring and stem cellulose oxygen isotope ratios can potentially indicate past climate change, such as shifts in relative humidity (RH) and temperature, as well as past changes in ecohydrological processes, such as shifts in the source of water used by plants. However, our understanding of how physiological and biochemical processes modulate the plant stem cellulose oxygen isotope ratio is still in its infancy. The greater our understanding of these processes, the greater will be our ability to interpret climatic and ecohydrological changes from the oxygen isotope ratio of stem cellulose. Currently, it is known that the bulk of the leaf and stem water oxygen isotopic signal is incorporated into stem cellulose by the exchange between oxygen from cellulose precursors, such as hexoses and trioses (Sternberg, 2009) and leaf and stem water. Tree ring cellulose records both the signature of the leaf water oxygen isotope ratio (δ18OLW), which is sensitive to RH among other factors, and the oxygen isotope signal of stem water (δ18OSW; (Eqn 1)), which is related to the isotopic composition of precipitation. The incorporation of the stem water oxygen isotope ratio in the stem cellulose molecule initiates as sucrose, carrying the isotopic signature of leaf water, is translocated from the leaf to the stem and cleaves into glucose and fructose. These molecules (glucose and fructose) are isomeric and interchange from one form to the other rapidly (Hill et al., 1995). Recent findings have indicated that this exchange can occur during phloem loading at the leaf and along the pathway from the leaf to the final destination, where sucrose is finally converted to cellulose (Gessler et al., 2013).

The participation of fructose in a futile cycle (Fig. 1) generates several carbonyl groups that can undergo hydration (oxygen from carbon 2, 3, 4 and 5) and exchange with cell water (Hill et al., 1995). Hydration of carbonyl groups involves the addition of oxygen from the surrounding water molecules (Fig. 2). When the carbonyl group dehydrates, the double bond from the carbonyl group is restored, but the remaining oxygen can be from either the original molecule or the water surrounding it (Sternberg & DeNiro, 1983) (Fig. 2). The carbonyl groups from trioses exchange oxygen with stem water rapidly, especially those of dihydroxyacetone which reach equilibrium in < 20 s (Model et al., 1968; Reynolds et al., 1971). When this exchange reaches equilibrium, the δ18O value of a carbonyl group is c. 27‰ higher than that of the water surrounding it (Sternberg & DeNiro, 1983; Sternberg, 1989; Yakir & DeNiro, 1990). The isotope enrichment of the carbohydrate oxygen above that of water is known as the biochemical fractionation (εbio).

Figure 1.

Scheme of the metabolic pathways that use the same pool of soluble carbohydrates as investigated in our study. Dashed circle represents the futile cycle in which oxygen from trioses exchanges with oxygen of stem water. DHAP, dihydroxyacetone phosphate; F-1.6-P, fructose-1,6-diphosphate; F-6-P, fructose-6-phosphate; G-1-P, glucose-1-phosphate; G-3-P, glyceraldehyde-3-phosphate; G-6-P, glucose-6-phosphate.

Figure 2.

Carbonyl hydration/dehydration reactions. When the double bond is restored, only one oxygen remains in the molecule and it can be the original oxygen or the oxygen from the surrounding water. When equilibrium is reached, the carbohydrate becomes 27‰ more 18O enriched than source water.

The oxygen isotope ratio of stem water, by the above exchange, is passed on to fructose and glucose, the cellulose precursors. However, not all of the fructose molecules undergo the above futile cycle and not all of the oxygen attached to the hexose carbon exchanges with water during cellulose synthesis. Consequently, the labeling of oxygen by stem water during cellulose synthesis is incomplete and corresponds to c. 42% of all the oxygen found in the cellulose molecule (Sternberg et al., 1986; Roden et al., 2000; Sternberg & Ellsworth, 2011), the rest of which holds the isotopic signature of leaf water. Roden et al. (2000) represented the proportion of stem and leaf water oxygen isotopic contribution to δ18OCELL, assuming that εbio is a constant of 27‰, with the following equation:

display math(Eqn 1)

where δ18OCELL represents the oxygen isotope ratio of stem cellulose.

However, recent observations have indicated that the proportion of oxygen exchange between stem water and carbohydrates (abbreviated as pex) and the biochemical fractionation factor (εbio) during cellulose synthesis may vary. For example, temperature can affect εbio (Sternberg & Ellsworth, 2011). Waterhouse et al. (2002) demonstrated that the best fit between δ18O values of precipitation reconstructed from tree ring cellulose oxygen isotope ratios and observed values occurred only when pex and εbio were assumed to be 46% and 30‰, respectively, values which are different from the average observed values of 42% and 27‰ in (Eqn 1) (Cernusak et al., 2005). In addition, when a plant is under water stress, either from a lack of water or salinity, the oxygen isotope ratio of its biomass, including cellulose, is often lower than that expected according to the above model (Verheyden et al., 2004; Roden et al., 2005; Zhou, 2005). Any reconstruction of paleoclimate, when there is a possibility of salinity or drought stress, will have to take these effects into account.

In this study, we have attempted to gain a better understanding of salinity effects in the recording of oxygen isotope ratios of stem and leaf water in δ18OCELL. First, we hypothesize that salt stress will change the amount of oxygen exchange between carbohydrates and source water (pex) and/or change εbio during cellulose synthesis. Second, we hypothesize that the synthesis of mannitol, a common response by several plant species to salinity, could alter pex and/or εbio, independent of salinity. To test our first hypothesis, we compared Arabidopsis thaliana cultivated hydroponically under salt water and freshwater. To test our second hypothesis, we used different M6PR transgenic lines of A. thaliana that code for mannitol synthesis (Zhifang & Loescher, 2003), and compared the oxygen isotope ratios of stem cellulose from the M6PR transgenic lines with those of wild-type (WT) A. thaliana grown hydroponically under both saline and freshwater.

Materials and Methods

Experimental design

In the following experiments, WT A. thaliana (L.) Heynh (Columbia, Lehle Seeds cat# WT-02-38-02, seed lot# 206-440) represented the control plants. The hydroponic culture tested the effects of salinity and mannitol synthesis in biochemical isotopic fractionation of WT and seven lines of transgenic M6PR (M1, M2, M3, M4, M5, M6, M7) obtained from Dr Wayne Loescher's laboratory (Department of Horticulture, Michigan State University, East Lansing, MI, USA). These transgenic lines have the celery M6PR gene, which codes for mannose-6-phosphate reductase, transferred into their DNA and transcribed continually. They produce mannitol to the approximate concentration of 2.5 μmol g−1 FW. An assay of one of these lines (M2) also showed the presence of mannitol in the stem, whereas no mannitol was detected in the tissue of the WT (Zhifang & Loescher, 2003). In celery, the mannose-6-phosphate reductase enzyme participates in the reduction of mannose-6-phosphate to mannitol-1-phosphate, which is the last precursor in the mannitol synthesis pathway (Fig. 1).

The hydroponic culture was located in the laboratory, at room temperature (c. 21°C), with WT and M6PR transgenic lines randomly organized within each of the fresh and salt (50 mM, c. 3 ppt) water treatment. Seeds were placed in rockwool (RockWool, Leeds, AL, USA) and watered from the top until seed germination. To prevent algal growth, the light and nutrient solutions were introduced after seed germination. The nutrient solution was prepared according to Huttner & Bar-Zvi (2003) and the pH was adjusted to 6.5. Distilled water was added to the nutrient solution as needed to maintain its original volume and keep contact with the rockwool. The solution was aerated using an aquarium pump and collected weekly to measure the oxygen isotope ratios of water under each treatment. The light irradiance was maintained at 120 μmol s−1 m−2 using four 20-W bulbs at a distance of 8 cm from the plants. All plants were harvested on the same day, after the development of a flower stalk, c. 51 d after seed germination.

Leaf and stem water and stem cellulose extraction

We define stems as the flower stalk, after the leaves, flowers and seedpods have been removed. In both experiments, we had five replicates for each A. thaliana line with each replicate consisting of a pool of five plant samples. This procedure was necessary to reach the minimum amount of water and cellulose required for stable isotope analysis. For each replicate, leaves were separated from the stem of five plants and placed separately into a glass tube, which was sealed and refrigerated. Water was extracted from the leaves and stems and analyzed isotopically by the method of Vendramini & Sternberg (2007). After water extraction, pooled stem samples from each water extraction were ground and homogenized for cellulose extraction. The stem samples were processed to holocellulose according to the method of Leavitt & Danzer (1993) and treated with 17% NaOH to remove the hemicellulose.

Oxygen isotope analysis of leaf and stem water and stem cellulose

All isotope ratios are expressed here in terms of per mil (‰)

display math(Eqn 2)

where Rsample and Rstandard are the 18O/16O ratios of the sample and the Vienna Standard Mean Ocean Water standard (V-SMOW), respectively. Leaf and stem water oxygen isotope ratios were analyzed in a Multiflow system connected to an Isoprime mass spectrometer (Elementar, Hanau, Germany). A 5% CO2/helium gas mixture was flushed through the vials and equilibrated with water for a period of 48 h. The equilibrated CO2 gas was analyzed to derive the oxygen isotope ratios of the water as in Vendramini & Sternberg (2007). The precision of the water oxygen isotope analysis is ± 0.1‰.

The δ18OCELL values were determined by a modification of Saurer et al. (1998). Half a milligram of cellulose from each sample was pyrolyzed at 1080°C in a quartz column filled to approximately half its height with glassy carbon and topped off by a layer of (0.5 g) of nickelized carbon and nickel wool (Elementar America, Mt Laurel, NJ, USA) in a Eurovector Elemental Analyzer (Milan, Italy). Gases from the pyrolysis (mostly small amounts of hydrogen and nitrogen and larger amounts of carbon monoxide) were carried by helium, passed through magnesium perchlorate and Ascarite (Thomas Scientific, Swedesboro, NJ, USA) to absorb H2O and CO2, respectively, and separated in a 3-m, 5-Å molecular sieve column (Eurovector) held at 70°C. The oxygen isotope ratios were determined on the carbon monoxide and compared with a standard gas calibrated to two standards: Sigma cellulose, having an oxygen isotope ratio of 29.3‰ (Sauer & Sternberg, 1994), and International Atomic Energy Agency cellulose filter paper, with an isotope ratio of 32.4‰. The precision of the cellulose analysis is ± 0.3‰.

Solving for pex and εbio simultaneously

(Eqn 3) from Barbour et al. (2002) was arranged to a linear form, so that the slope and intercept from the equation represent pex and εbio (Eqn 4), respectively. Because we had the actual values of δ18OSW, we eliminated the term denoting the proportion of leaf water in stem water (px). Experimental values of δ18OCELL – δ18OLW were plotted against δ18OSW – δ18OLW, and the slope and intercept of the linear regression gave us the estimates of pex and εbio, respectively.

display math(Eqn 3)
display math(Eqn 4)

Statistical analysis

The PASW SPSS Statistics (version 18; IBM, Armonk, NY, USA) program was used to perform the analysis. Data were tested for normality (Kolmogorov–Smirnov test and Shapiro–Wilk test) and homogeneity of variance (Levene's statistic test). If normality was not reached after transformation, the data were ranked previously to the nested ANOVA, because there is no non-parametric test equivalent to nested ANOVA. A nested ANOVA was used to analyze the hydroponic experiment by first comparing the treatment effect (salt water vs freshwater) in the leaf and stem water, as well as in the stem cellulose, oxygen isotope ratios, keeping all the A. thaliana lines (WT and M6PR transgenic) nested (Table 1). Second, we used the nested information between the WT and transgenic (M6PR) lines from Table 1 to interpret the effect of mannitol synthesis genes in the leaf and stem water, as well as in the stem cellulose, isotope ratios. The regression equation for the variables in (Eqn 4) was tested for normality and homoscedasticity. From salt water and freshwater linear regression, we tested: (1) the significance between correlation coefficients (r) by transforming r to Z values and comparing them using chi-squared (Zar, 1999); (2) the difference between slopes (pex) using analysis of covariance (Zar, 1999). Because the slopes were significantly different, we used the intercept confidence interval from both equations to determine differences in εbio between salt water and freshwater treatments.

Table 1. Nested ANOVAs for ranked and non-ranked variables showing how the oxygen isotope ratios of leaf water (δ18OLW), stem water (δ18OSW) and stem cellulose (δ18OCELL) differed between salt water (S) vs freshwater (F) availability, as well as between Arabidopsis thaliana lines (WT and M6PR lines) growing under each condition
ExperimentParametersNested ANOVAsdfms F P
  1. Significant effects are shown in bold.

  2. Transgenic (M6PR) and wild-type (WT) lines of A. thaliana were both hydroponically cultivated under salt water (c. 3 ppt) and freshwater. No significant differences were observed between the δ18O of source water in the salt water and freshwater containers.

Hydroponic settingδ18OLWS × F1262275 < 0.001
Arab. lines141.00.41.0
δ18OSWS × F13.75.2 0.04
Arab. lines140.72.7 0.004
δ18OCELLS × F1668017.8 0.001
Arab. lines143750.80.7


There were no significant differences between the δ18O values of source water for salt water and freshwater hydroponic solution. Within salt water or freshwater, δ18OLW and δ18OCELL did not differ between WT and M6PR transgenic A. thaliana (mannitol synthesizers) (Table 1; Fig. 3a,c); however, δ18OSW differed between WT and one M6PR transgenic line (M5) (Tukey's post-hoc test, < 0.05). The M5 line also differed from another two M6PR lines (M2 and M6; Tukey's post-hoc test, < 0.05). Arabidopsis thaliana plants (WT + M6PR) growing under salt treatment had significantly higher δ18OLW, δ18OSW and δ18OCELL values compared with those growing in freshwater (WT + M6PR) (Table 1; Fig. 3a–c).

Figure 3.

Leaf water, stem water and stem cellulose oxygen isotope ratios for wild-type (black bar) and mannitol synthesizers (M6PR; gray bars) Arabidopsis thaliana (a–c) under different salinities. Error bars represent ± 1SE of the mean.

The correlation coefficients between δ18OCELL – δ18OLW and δ18OSW – δ18OLW within each treatment (salt water and freshwater) were significant (= 0.85, < 0.01 and = 0.59, < 0.01, respectively). Correlation coefficients also differed between salt water and freshwater treatment (χ2(1,N = 74) = 6.2, < 0.05). The slopes from the regression equation (Fig. 4), representing pex, were significantly different (t(76) = 3.68 < 0.001), with the slope twofold greater for salt water than for freshwater treatment (0.64 vs 0.3, respectively). The biochemical fractionation factor is represented by the intercept of the regression equation. Under salt water, εbio was 1‰ higher than under freshwater treatment (intercept = 28.8 and 27.9, respectively; Fig. 4); however, their confidence intervals overlapped.

Figure 4.

A plot of δ18OCELL – δ18OLW (y axis) vs δ18OSW – δ18OLW (x axis), with the slopes representing the proportion of oxygen exchange between soluble carbohydrate and water (pex) and the intercepts representing the biochemical fractionation factor (εbio) during cellulose synthesis. The relationship between these parameters is shown for Arabidopsis thaliana under freshwater (open circles and open squares) and salt water (closed circles and closed squares) treatment, merging wild-type (circles) and M6PR transgenics (squares) within each treatment.


Salinity effects on δ18OCELL

According to our present understanding of stem water oxygen isotope labeling in δ18OCELL (Roden et al., 2000), it is expected that plants using source water with similar oxygen isotopic ratios, under the same atmospheric conditions, will have similar δ18OCELL. This expectation holds if plant physiological and biochemical responses to abiotic factors do not vary and follow the mechanistic model (Roden et al., 2000) (Eqn 1). In the saline and freshwater hydroponic cultures used here, δ18O values of source water and RH did not differ; yet, δ18OCELL values were significantly different between the two treatments.

Leaves from salt-treated plants had higher δ18OLW values relative to those of freshwater plants, which contributed to increases in δ18OCELL. If we solve for δ18OCELL using (Eqn 1), where pex = 0.42 and εbio = 27, we expect that the δ18OCELL value of salt-treated plants will be 2.3‰ more 18O enriched than that of freshwater-treated plants. However, we only observed an average of 0.8‰ enrichment in δ18OCELL under salt treatment relative to the average of freshwater-treated plants. Salt treatment of both WT and mannitol-synthesizing A. thaliana caused depletion in δ18OCELL relative to the expectation of the current model (Eqn 1). This observation is consistent with several other studies showing that water stress, either by salinity or lack of water, causes a lower than expected oxygen isotope ratio of plant biomass and/or cellulose. Zhou (2005) observed that δ18OCELL values from plants under water stress were c. 7‰ less 18O enriched than expected. Ellsworth et al. (2013) observed that, under similar RH, δ18OCELL did not reflect the magnitude of differences in δ18OSW from mangrove vs freshwater plants. Although δ18OCELL values from mangrove plants were higher than those of freshwater plants, this enrichment was c. 2.7 times lower than expected if leaf properties between the plants were similar. Another study with mangroves showed that, during the dry season, when RH is low and salinity is high, δ18O of tree ring wood of Rhizophora mucronata was lower than expected (Verheyden et al., 2004). Roden et al. (2005) found that reconstructed δ18OCELL of juniper trees (Juniperus occidentales) using the mechanistic model equation ((Eqn 1); Roden et al., 2000) was overestimated under low water availability and low RH.

It is important to note that, under salt (Fig. 3a) as well as water stress (Zhou, 2005), the increase in δ18OLW is associated with leaf physiological responses to these conditions. This observation alone is relevant in paleoclimate studies as δ18OLW is used as a proxy for RH. As mentioned previously, the tree ring model (Roden et al., 2000; (Eqn 1)) assumption is that δ18OLW varies only as a function of changes in RH, which was not the case in our experiment. An understanding of the basic model explaining leaf water isotopic enrichment explains the observed increase in δ18OLW with salinity. Leaf water isotopic enrichment is, in part, dependent on the ratio of the advective flux of isotopically unenriched liquid water from the xylem towards the stomatal cavity and the diffusion of isotopically enriched water from the stomatal cavity towards the mesophyll (Barbour & Farquhar, 2004; Farquhar & Cernusak, 2005). This ratio of advective and diffusive fluxes is known as the Peclet effect (℘):

display math(Eqn 5)

where E is the transpiration rate, L is the effective path length of water moving from the xylem to the stomatal cavity, D is the diffusivity of water, which is a function of temperature, and C is the molar density of water (Barbour & Farquhar, 2004). The isotopic enrichment of leaf water at steady state is given by the following equation:

display math(Eqn 6)

where ΔLW is the oxygen isotope enrichment of leaf water and Δes is the steady-state isotopic enrichment of the stomatal cavity water as a result of evaporation (Barbour & Farquhar, 2004). Both of these are expressed as isotopic enrichment relative to the isotope ratio of the source water. This equation predicts that the lower the Peclet ratio, the higher the isotopic enrichment of leaf water. Thus, according to (Eqn 5), a decrease in transpiration or a shortening of the path length should decrease ℘ and drive up the leaf water isotopic enrichment. We propose here that salinity decreases stomatal conductance in A. thaliana, as it has been observed previously in both WT and the mannitol transgenics (Sickler et al., 2007). This will decrease transpiration and subsequently ℘, driving up leaf water isotopic enrichment.

The second important observation is that the increase in δ18OLW was not recorded in δ18OCELL as expected using the standard model equation (Eqn 1). Our regression analysis (Fig. 4) for pooled transgenic and WT lines, where the slope of the regression line is the exchange rate and the intercept is the biochemical fractionation, indicated that there were no significant changes in εbio between freshwater and salt water treatment, with salt water-treated plants having a slightly higher εbio. Further, if these differences in εbio were significant, it would make δ18OCELL of salt-treated plants even more 18O enriched. However, our regression analysis showed large changes in pex, which doubled (0.64) under salt water relative to freshwater treatment (0.3; Fig. 4).

Barbour & Farquhar (2000) proposed that, under mild water stress, pex may increase. They reasoned that the sucrose and soluble sugar synthesis would remain unchanged despite decreases in cell division and expansion. This would allow more time for the soluble sugars to undergo the exchange reactions discussed previously. Considering the successful growth of WT (salt-intolerant plant) and M6PR transgenic lines under salt treatment (c. 3 ppt), we assume the salinity treatment imposed here to be mild. We also have evidence that A. thaliana increases its soluble carbohydrate with salinity. Zhifang & Loescher (2003) tested the effect of salt on the amount of soluble carbohydrates and osmolytes in one line of transgenic A. thaliana (M2) compared with the WT. They found that, at 50 mM (c. 3 ppt), both WT and M2 transgenic A. thaliana plants increased the amount of sucrose (from 0.54 to 1.9 μmol g−1 FW and 0.7 to 2.4 μmol g−1 FW, respectively) and fructose (from 1.35 to 2.8 μmol g−1 FW and 1.5 to 2.2 μmol g−1 FW, respectively) compared with those grown in freshwater. Therefore, our observation that, under mild salt treatment, pex increases, probably because of an excess of soluble carbohydrates, is in line with the proposal of Barbour & Farquhar (2000).

Effect of osmolyte synthesis on δ18OCELL

The syntheses of mannitol and cellulose compete for the same precursor (Fig. 1), so that mannitol synthesis could potentially decrease the remaining pool of soluble carbohydrate directed for cellulose synthesis, as well as change its isotopic ratio (Fig. 1). If this were the case, we would expect differences in δ18OCELL between mannitol-synthesizing and WT A. thaliana, as a consequence of pex and/or εbio variations. However, there were no differences in δ18OCELL (Table 1, Fig. 3c), or in δ18OLW (Table 1, Fig. 3a) and δ18OSW values, between lines within each treatment. Therefore, we conclude that the amount of mannitol synthesized was not sufficient to significantly decrease the pool of carbohydrates and change its oxygen isotopic ratio. The oxygen exchange between sucrose and stem water, as well as εbio, remained similar for WT and M6PR lines within each treatment.

This study suggests that salinity, even at low concentrations (c. 3 ppt), can lead to changes in plant metabolic pathways. Some of these changes increase the oxygen exchange between carbohydrates and stem water during cellulose synthesis. The proportion of δ18OSW contribution to δ18OCELL increased c. twofold with salinity, making δ18OCELL under salt treatment 1.5‰ lower than expected from the standard tree ring model (Eqn 1). Although εbio differed between salt water and freshwater treatments, we cannot draw any clear conclusions, because the regression confidence interval overlapped in the intercept area. This study highlights the importance of fully understanding aspects of carbohydrate biochemistry, such as soluble carbohydrate pool size and turnover rate, as well as aspects of leaf water processing, before using δ18OCELL of tree rings as a true proxy of climate, sea level rise and other abiotic environmental factors.


We thank Dr Wayne H. Loescher (Department of Horticulture, Michigan State University, East Lansing, MI, USA) for providing us with the transgenic Arabidopsis thaliana seeds.