18O enrichment in phosphorus pools extracted from soybean leaves


Author for correspondence:

Verena Pfahler

Tel: +41 52 354 91 61

Email: verena.pfahler@usys.ethz.ch


  • The objective of this study was to investigate the isotopic composition of oxygen bound to phosphate (δ18O-PO4) in different phosphorus (P) pools in plant leaves. As a model plant we used soybean (Glycine max cv Toliman) grown in the presence of ample P in hydroponic cultures.
  • The leaf blades were extracted with 0.3 M trichloroacetic acid (TCA) and with 10 M nitric acid. These extractions allowed measurement of the TCA-soluble reactive P (TCA P) that is rapidly cycled within the cell and the total leaf P. The difference between total leaf P and TCA P yielded the structural P which includes organic P compounds not extractable by TCA.
  • P uptake and its translocation and transformation within the soybean plants lead to an 18O enrichment of TCA P (δ18O-PO4 between 16.9 and 27.5‰) and structural P (δ18O-PO4 between 42.6 and 68.0 ‰) compared with 12.4‰ in the phosphate in the nutrient solution.
  • δ18O values of phosphate extracted from soybean leaves grown under optimal conditions are greater than the δ18O-PO4 values of the provided P source. Furthermore, the δ18O-PO4 of TCA P seems to be controlled by the δ18O of leaf water and the activity of inorganic pyrophosphatase or other pyrophosphatases.


Phosphorus (P) is an essential macronutrient for plants. Plants take up P from the soil solution as inorganic phosphate (mainly inline image and inline image). After uptake, P is transported mainly as inorganic phosphate within the plant where it can be included in compounds (e.g. phospholipids, ATP, DNA, and RNA) involved in different metabolic processes (Bieleski, 1973). Several methods are used for studying these processes. P uptake and translocation, compartmentalization of P and turnover rates of P compounds are studied using radioisotopes of P (32P and 33P) and molecular methods (Bieleski, 1973). 31P-NMR and the isolation of plant organelles provide information about compartmentalization and speciation of P in plants (Bieleski, 1973; Schachtman et al., 1998).

An additional method that could be used to trace the biological transformations of P in plants is the analysis of the oxygen stable isotope composition of phosphate (δ18O-PO4) (Tamburini et al., 2012a). This method is based on the fact that in the environment P is mostly associated with oxygen. At earth surface temperatures and without biological activity the P-O bond is very stable and oxygen isotope exchange between phosphate and water is slow and negligible (O'Neil et al., 2003). However, biological activity can lead to cleavage of the P-O bond (Blake et al., 2005), leading to isotope exchange. Inorganic pyrophosphatase (PPase) catalyses the hydrolysis of pyrophosphate to phosphate and cleaves the P-O bond (Cohn, 1958; Blake et al., 2005). As a result of this reversible reaction, all four oxygen atoms associated with P in phosphate are exchanged with water. It was shown by Cohn (1958) that the oxygen exchange between phosphate and water is fast compared with the overall rate of the enzymatic reaction. This reaction leads to temperature-dependent oxygen isotope equilibrium between phosphate and water (Blake et al., 2005). The equilibrium can be calculated using the equation of Longinelli & Nuti (1973): [correction added after online publication 6 November 2012: minus symbols which were previously published within Eqn 1 have now been replaced by hyphens to present the equation in its correct form.]

display math(Eqn 1)

Cleavage of one or two P-O bonds occurs during the hydrolysis of phosphomonoesters and phosphodiesters by phosphomonoesterase and phosphodiesterase, respectively, and O from ambient water is incorporated into the released phosphate. To date, only the oxygen isotope fractionation related to alkaline phosphatase (APase), 5′-nucleotidase, RNase and DNase has been studied (Blake et al., 2005; Liang & Blake, 2006b). The observed fractionations are enzyme- and substrate-specific, but generally, the hydrolysis of the studied phosphomonoesters and phosphodiesters by phosphoenzymes leads to the release of lighter phosphate. Uptake by microorganisms can also lead to oxygen isotope fractionation. This has been shown for Escherichia coli, which preferentially takes up the lighter phosphate isotopologue, leaving the phosphate remaining in the solution enriched in 18O (Blake et al., 2005).

A recent study by Tamburini et al. (2012b) showed that TCA-soluble reactive P (TCA P) from plant leaves sampled along a 150-yr soil chronosequence was c. 10‰ more enriched than plant-available P extracted with anion exchange resins from the soils, except at the youngest and at the reference site of the chronosequence. They attributed this enrichment to equilibration of leaf TCA P with leaf water enriched in 18O (Dongmann et al., 1974) and postulated that this equilibrium is mediated by PPases. PPases are in fact found in different cell compartments in plants, for example, in the cytosol, vacuole and chloroplasts (Hirata et al., 2000; George et al., 2010), and also work as proton pumps (Masayoshi, 2000). In addition to inorganic PPase, other PPases such as the H+-ATPase can exchange oxygen atoms between phosphate and water (Hirata et al., 2000). It has also been shown that the hydrolysis of ATP catalysed by myosin subfragment 1 leads to a complete exchange of all oxygen atoms of the terminal phosphate group in ATP (Webb et al., 1978; Webb & Trentham, 1981). We hypothesize that the δ18O-PO4 of phosphate in leaves (i.e. extractable with TCA) is close to equilibrium with the leaf water. We further hypothesize that organic P compounds not extractable with TCA are not in oxygen isotope equilibrium with leaf water, because they are less cycled within the plant.

To test our hypotheses, soybean (Glycine max, cv Toliman) was used as a model plant. Soybean was grown in hydroponic cultures for c. 8 wk in the presence of ample P in the nutrient solution. Leaves were sampled at three different development stages: V2 (vegetative stage), R1 (beginning of flowering) and R6 (pod filling). TCA-soluble reactive P, total (HNO3 extractable) and structural P, and water were extracted from leaf blades and analysed for their δ18O. The activity of acid phosphatase was also determined in the leaf blades, as it is known to increase during plant development (Duff et al., 1994). The activity of inorganic PPase was not determined, because the exchange of oxygen between water and phosphate is faster than the enzymatic reaction catalysed by the inorganic PPase (Cohn, 1958). Therefore, we did not expect a correlation between inorganic PPase activity and the establishment of equilibrium.

Materials and Methods

Experimental set-up

Soybean (Glycine max (L.) Merr. cv Toliman) was grown in a glasshouse using hydroponic cultures. The seeds were surface-sterilized following the procedure described by Kremer et al. (2005) and were germinated in sterilized sand (size 0.7–1.2 mm) saturated with ultrapure water (ddH2O). Eighty plants were transferred to 10 nontransparent plastic pots (size 300 mm × 200 mm × 220 mm) containing 8 l of nutrient solution, with eight plants per pot. At stage R3 the 8 l pots were exchanged with boxes containing 28 l of nutrient solution (size 400 mm × 300 mm × 325 mm) because of the increased water demand of the plants. The plants were supplied with a modified Hoagland nutrient solution containing 0.5 mM KH2PO4, 5 mM KNO3, 5 mM Ca(NO3)2, 2 mM MgSO4, 0.1 mM Fe chelate, 0.05 mM KCl, 0.025 mM H3BO3, 0.002 mM MnSO4, 0.002 mM ZnSO4, 0.0005 mM CuSO4, and 0.0005 mM Na2MoO4. The KH2PO4 used for the preparation of the nutrient solution had a δ18O-PO4 of 12.4‰. The δ18O of the water provided to the plants was –10.2 ± 0.2‰ throughout the experiment. The nutrient solution was changed every 4–8 d depending on the water uptake by the plants, to maintain optimal plant growth and to minimize bacterial growth in the nutrient solution. Relative air humidity in the glasshouse ranged between 50 and 80%, with a mean of 60%. The air temperature ranged between 20 and 29°C, with a mean of 23°C. The plants were illuminated with 100 klux for 16 h d−1.

The experiment had three treatments (three sampling stages) and three replicates per treatment (three pots with eight plants per pot). Except for seed, total and structural P, all analyses were carried out in triplicate. The treatments were not randomized within the glasshouse, but arranged in blocks (see their distribution within the glasshouse in Fig. 1). The tenth pot was not used for the extraction of P from leaves, but for producing mature soybean seeds.

Figure 1.

Experimental design with illustration of the development stages of soybean. Horizontally the three sampling dates are represented by Roman numerals, while vertically the leaf position is represented by Arabic numerals. Each rectangle represents one pot with eight soybean plants. For each sampling date there were three replicates.


As the δ18O of leaf water shows a diurnal cycle (Cernusak et al., 2002), plants were always harvested at 14:00 h. Only mature trifoliate leaves were harvested at vegetative stage V2 (sampling I), at the beginning of flowering (R1; sampling II), and during pod filling (R6; sampling III) (Fehr et al., 1971). At each sampling, leaves were additionally separated depending on their position (Fig. 1). Directly after the harvest, the leaves were washed with ddH2O, dried with paper towels and stored at −80°C until further processing. Pods from the plants in the pot used for seed production were harvested at stage R8, when the pods reached maturity.

Sample preparation

Before analysis, the middle vein of each leaf was removed, because it still contains water that has the δ18O signature of the source water (Barbour et al., 2005) and possibly also of the phosphate source. The remaining frozen leaf material was crushed and divided into subsamples for leaf water extraction, determination of acid phosphatase activity and extraction of P with 0.3 M trichloroacetic acid (TCA) and 10 M HNO3. The δ18O-PO4 of the total P of soybean seeds used for this experiment and those produced during the experiment was also determined. A subsample (c. 20 g) of the seeds was ground using a porcelain mortar and pestle and P was extracted with 10 M HNO3.

Leaf water extraction

Leaf water was extracted using a simplified version of the method described by Peters & Yakir (2008). Leaves were transferred into 2-ml centrifuge tubes and pressed with a pestle. The tubes were centrifuged at 30000 g for 10 min at 4°C; the supernatant was then transferred to glass vials and 0.5–2 mg of HgCl2 was added in a solid form to the extracted leaf water to stop biological activity. We tested the method against the cryogenic vacuum distillation method described by Barnard et al. (2007) and observed no significant difference (data not shown).

Determination of acid phosphatase activity

Acid phosphatase was extracted from leaf material after Besford (1979). One gram of fresh leaf material was ground for c. 4 min at 4°C in a mortar with 20 ml of 90 mM citrate buffer at pH 4.8. The suspension was centrifuged at 26 500 g for 10 min at 2°C and the supernatant collected for the enzymatic assay. Acid phosphatase activity was determined following the protocol by Sigma-Aldrich using p-nitrophenyl phosphate as substrate. To stop the reaction, 4 ml of 0.1 M NaOH was added. The amount of hydrolysed p-nitrophenyl phosphate was determined at 410 nm using a spectrophotometer.

Extraction by 0.3 M TCA

With the extraction by 0.3 M TCA, we targeted the phosphate in leaves (Hawkins & Polglase, 2000). This extract could also contain organic P compounds that can be hydrolysed during the colorimetric essay used to measure the phosphate concentration (Broberg & Pettersson, 1988). Therefore, P extracted by 0.3 M TCA is not referred to as only phosphate, but as TCA-soluble reactive P (TCA P). One gram of fresh leaf material was weighed into a 60-ml plastic bottle and 20 ml of 0.3 M TCA was added. Each sample was homogenized subsequently for c. 45 s with a Polytron® (Kinematica AG, Luzern, Switzerland). Samples were then placed on a horizontal shaker at 4°C for 1 h. They were filtered with GF/F filters (0.7 μm nominal pore size; Whatman International Ltd.) and the residue discarded. A small subsample of the extract was taken to determine the total P content of the TCA extract. The total P content of the TCA extracts (total TCA P) was determined by wet digestion in the presence of H2SO4 following Ebina et al. (1983). The phosphate concentration of the undigested and digested samples was determined colorimetrically with the malachite green method (Van Veldhoven & Mannaerts, 1987).

Extraction by 10 M HNO3

This method was used to assess the total and structural P content in leaves. The three replicates of each sampling point were combined in order to have a sufficient amount of leaves for the extraction. The extraction of total P followed the protocol described by Tudge (1960). One gram of fresh leaf material was weighed in 50-ml centrifuge tubes and 20 ml of 10 M HNO3 was added. The samples were placed in a water bath at 50°C for 16 h. After the samples had cooled down, the solutions were transferred into 100-ml Erlenmeyer flasks and 0.3 M KMnO4 was added drop-wise until the solution turned brown. After 16 h, 0.1 M NaNO2 was added to the solution to reduce the remaining KMnO4 (Liang & Blake, 2006a). The solutions were filtered using GF/F filters and the P concentration was determined as in the TCA extracts. The P extraction of the seeds followed the same protocol. One gram of the ground seeds was weighed in a 50-ml centrifuge tube. P extracted by HNO3 from seeds is referred to as ‘seed P’.

During the extraction by 10 M HNO3, organic P compounds are hydrolysed. As strong acids might break P-O bonds in organic P compounds and not the C-O bond, unspiked and 18O-spiked HNO3 solutions were used (McLaughlin et al., 2006; Tamburini et al., 2010). Because the δ18O-H2O of the 10 M HNO3 could not be directly measured, different dilutions of the concentrated HNO3 were prepared with ddH2O and the δ18O-H2O of 10 M HNO3 was then calculated by mass balance.

Structural P

Structural leaf P is operationally defined as the difference between total leaf P and TCA P. The δ18O-PO4 of structural P was determined with two approaches. First we calculated the δ18O-PO4 values of structural P by mass balance from the concentrations and δ18O-PO4 values of TCA P and total P (Eqn 2).

display math(Eqn 2)

(A, B, C and D, the δ18O-PO4 values of total P, TCA P, total TCA P and structural P, respectively; a, b, c and d, the corresponding concentrations.) As total P was extracted from combined leaf samples (see section ‘Extraction by 10 M HNO3′), the average values for the concentration of TCA P and for the δ18O-PO4 value of TCA P were used.

This approach could only be applied for leaf samples where the TCA P concentration was not different from the total TCA P (i.e. in all leaves except those from stage V2) and the purification process for precipitating silver phosphate and the colorimetric method for determining the concentration of TCA P were selecting the same P compounds. The results for the δ18O-PO4 of structural P were furthermore checked by conducting a sequential extraction during which selected leaves (II.1–2, II.5–6, III.3–4 and III.9–11) were extracted with TCA and then the residue was extracted with 10 M HNO3. The δ18O-PO4 of the residue was measured and considered to be a direct measure of the signature of the structural P.

Purification of extracts and precipitation of silver phosphate

Purification of the 0.3 M TCA and 10 M HNO3 extracts and precipitation of silver phosphate (Ag3PO4) followed the protocol described by Tamburini et al. (2010). Phosphate in the extract was purified by precipitating and dissolving first ammonium phospho-molybdate (APM) and then magnesium ammonium phosphate, and using a cation exchange resin. Finally, Ag3PO4 was precipitated by adding a silver-ammine solution. Additionally, 1 ml of concentrated H2SO4 was added during the APM step in order to facilitate the precipitation of APM in the case of the 0.3 M TCA extracts.

Determination of δ18O of water and phosphate

The δ18O of leaf water and of the HNO3 dilutions was determined with the CO2 equilibration method (Epstein & Mayeda, 1953). From each sample, 0.2 ml was pipetted into a vacutainer, closed tightly and flushed with a gas mixture of 0.3% CO2 in helium (He). After an equilibration time of 18 h at room temperature, the samples were measured with a gas bench device (Gas Bench II; Thermo Fisher Scientific Inc., Waltham, MA, USA) coupled to an isotope ratio mass spectrometer (Delta V Plus; Thermo Fisher Scientific Inc.). The system was calibrated with the international standards Standard Mean Ocean Water (SMOW), Standard Light Antarctic Precipitation (SLAP) and Greenland Ice Sheet Precipitation (GISP).

The Ag3PO4 samples were measured using a thermal conversion elemental analyzer coupled to an isotope ratio mass spectrometer (TC/EA-IRMS; Delta V Plus Thermo Scientific Inc.). Results were calibrated against an internal Ag3PO4 standard (Acros Organics, Geel, Belgium; δ18O = 14.2‰ Vienna Standard Mean Ocean Water (VSMOW)) and two benzoic acid standards distributed by the International Atomic Energy Agency (IAEA) in Vienna (IAEA 601: δ18O = 23.1 ‰ and IAEA 602: δ18O = 71.3 ‰ VSMOW). For each run of a total of 49 samples, seven samples of the Ag3PO4 standard, and three of each benzoic acid standard were analyzed. Analytical error calculated on replicate analyses of standards was better than ± 0.4 ‰.

Oxygen isotope compositions are reported in the conventional delta notation:

display math(Eqn 3)

where R = 18O/16O and Rstandard is the VSMOW.

Calculation of equilibrium between the oxygen in phosphate and in water

The theoretical equilibrium between the oxygen in phosphate and in water was calculated with the equation of Longinelli & Nuti (1973) (Eqn 1). We used the δ18O of the leaf water and the minimum and maximum values of the air temperature of the glasshouse for calculating the δ18O-PO4 values at equilibrium. The equilibrium values obtained were then compared with the measured δ18O-PO4 values of TCA P.

Statistical analysis

Statistical analysis was performed with systat 13. A one-way ANOVA was conducted for samplings II and III. Within samplings II and III, the δ18O-PO4 values of the different leaf positions were compared with Tukey's HSD test after rejecting the null hypothesis of the ANOVA. Because of the position of the pots in the glasshouse, a site effect cannot be excluded, hindering the statistical comparison of sampling stages with each other. δ18O-PO4 values of TCA P were compared with the calculated equilibrium values using a two-sample t-test. The mean values of the δ18O-PO4 values of structural P were compared with the mean δ18O-PO4 of TCA P of the corresponding leaves using a one-way ANOVA.


Water and P uptake of the plants

Water uptake increased from 0.2 ± 0.0 l d−1 per pot during the vegetative phase of plant development to 4.6 ± 0.4 l d−1 per pot at R6. The P uptake ranged between 8.7 mg P d−1 per pot during the vegetative stage and 137.0 mg P d−1 per pot during stages R5 and R6.

P concentration in leaves

Concentrations of TCA P in the leaf blades ranged between 9.0 ± 0.3 (II.1–2) and 1.8 ± 0.1 g P kg−1 dry weight (DW) (III.9–11). P concentrations in the undigested and digested TCA extracts were similar in all samples except for I.1–2 (Fig. 2). The concentration of TCA P was highest in the oldest leaves (I.1–2) and lowest in the youngest mature leaves (III.9–11). Concentrations of total P in the leaves varied between 16.3 ± 2.2 g P kg−1 DW (I.1–2) and 3.9 ± 0.9 g P kg−1 DW (III.9–11). These results show that the plants were not suffering from P deficiency (Bergmann, 1993).

Figure 2.

Phosphorus (P) concentration of the P extracts from the soybean leaves. From left to right: trichloroacetic acid-soluble reactive P (TCA P) TCA concentration in soybean leaf blades (mean ± SD; n = 3), total P concentration in TCA extracts from soybean leaf blades (mean ± SD; n = 3), and HNO3 extract (n = 1). Roman numerals stand for the development stages, whereas Arabic numerals stand for the leaf position. Error bars represent the standard deviation. Black bars, TCA extract undigested; light grey bars, TCA extract digested; dark grey bars, HNO3 extract.

Acid phosphatase activity

Acid phosphatase activity changed between the different leaf positions within samplings II and III. The lowest activity was observed in the oldest leaves at stages R1 and R6 (II.1–2 and III.3–4; Fig. 3).

Figure 3.

Acid phosphatase activity of soybean leaves in μmol p-nitrophenylphosphate (PNPP) hydrolyzed per hour and per gram dry weight (mean ± SD; n = 3). Roman numerals stand for the development stages, whereas Arabic numerals stand for the leaf position. Error bars represent the standard deviation.

δ18O of leaf water

As expected, the leaf water was enriched in 18O by 10–16‰ compared with the water in the nutrient solution (δ18O = −10.2 ± 0.2‰) (Fig. 4). The drop in the δ18O of leaf water from sampling II to sampling III coincided with an increase in the daily water uptake of the soybean plants, recorded after having transferred the plants to bigger boxes.

Figure 4.

δ18O-H2O from the soybean leaf water (mean ± SD; n = 3). Error bars represent the standard deviation.

δ18O-PO4 of TCA P and calculated equilibrium values of δ18O-PO4

TCA P was enriched in 18O compared with the phosphate added in the solution. There was no significant difference between the δ18O-PO4 of TCA P at different leaf positions at stage R1. A significant difference was observed between III.3–4 and III.9–11 at stage R6 (Fig. 5). Except at I.1–2 and III.9–11, the δ18O-PO4 values of TCA P were within the equilibrium range with leaf water calculated using the minimum (20°C) and maximum (29°C) glasshouse air temperature (two-sample t-test, P-value < 0.05; Fig. 5).

Figure 5.

δ18O-PO4 of trichloroacetic acid-soluble reactive phosphorus (TCA P) from the soybean leaves (mean ± SD; n = 3). The range of expected δ18O-PO4 values of the calculated equilibrium with the formula from Longinelli & Nuti (1973) is given by the shaded boxes. Roman numerals stand for the development stages, whereas Arabic numerals stand for the leaf position. Asterisks indicate a significant difference between measured and calculated δ18O-PO4 values (two-sample t-test; P-value < 0.05). Error bars represent the standard deviation. Closed circles, TCA P; grey boxes indicate the equilibrium range.

δ18O-PO4 of total and structural P in leaves and of total seed P

The δ18O-PO4 of total P was almost constant except in leaves II.5–6 and III.9–11, which were up to 7‰ more positive than the other leaves (Table 1). δ18O-PO4 values of structural P were significantly different from those of TCA P (ANOVA; P-value < 0.001). The δ18O-PO4 of the total seed P was 38.2‰ for the seeds used for germination and 36.7 ‰ for the seeds produced in the glasshouse during the experiment.

Table 1. δ18O-PO4 values of trichloroacetic acid-soluble reactive phosphorus (TCA P), total P and measured and calculated structural P from soybean leaves
Development stage and leaf positionδ18O-PO4 (‰)
TCA PTotal PStructural P (calculated)aStructural P (measured)b
  1. a

    The calculated δ18O-PO4 values of structural P were determined by mass balance using the δ18O-PO4 values of TCA P and total P and the corresponding P concentrations. Differences between the measured and calculated δ18O-PO4 values of structural P could be attributable to the constraints of the calculation or because our assumptions were not correct (see the 'Materials and Methods' section).

  2. b

    The measured δ18O-PO4 values of structural P were determined by a sequential extraction of the leaves. Leaves were first extracted with 0.3 M TCA and then the residues were extracted again with 10 M HNO3. This was done only on selected leaves in order to verify the calculated values of the structural P.



The δ18O of the TCA P pool

Significant differences between the δ18O-PO4 of the TCA P and the calculated equilibrium values were only observed for two of eight sampling points (see Fig. 5). This suggests that oxygen isotope exchange between water and P was the most important process controlling the δ18O-PO4 of the TCA P. As PPases are ubiquitous in plants (Masayoshi, 2000), we propose that this exchange is probably mediated by PPases, as shown for bacteria by Blake et al. (2005). Although we could not statistically compare the different development stages with each other, we observed a declining trend in the δ18O-PO4 of TCA P from leaves I.1–2 to III.9–11, which is mainly attributable to the lower δ18O of the leaf water at later stages.

Because the fractionation between water and phosphate is temperature dependent and decreases by 0.2 ‰ per 1°C temperature increase (Longinelli & Nuti (1973)), offsets from the calculated equilibrium could be explained by differences between the recorded glasshouse air temperature and the actual leaf temperature. Transpiration tends to cool the leaf surface and a difference in temperature of up to 10°C has been reported between the surface of a leaf and the surrounding air (Lange, 1959). Sunshine, in contrast, leads to warming of the leaves (Ansari & Loomis, 1959). In our experiment the glasshouse air temperature ranged between 20 and 29°C, resulting in a 2.1‰ range in a possible equilibrium value (Fig. 5). The actual leaf temperature could be outside this temperature range. The offset observed for leaf III.3–4 could be explained by a difference between the air and the leaf temperature. However, differences in temperature alone cannot explain the offset observed in leaves I.1–2 and III.9–11, because leaf temperature would have to be c. 10 and 40°C for leaves I.1–2 and III.9–11, respectively.

Diurnal variations of the δ18O of leaf water, which have been observed to reach up to 23‰ in other plants (Cernusak et al., 2002), could also explain small differences between the δ18O values of TCA P and the equilibrium values, which were calculated using the δ18O of leaf water at the time of sampling. To minimize the impact of diurnal variations in leaf water δ18O, we always sampled at the same hour of the day. However, because it can take up to 50 h for inorganic PPase to fully equilibrate phosphate with water (Blake et al., 2005), the δ18O-PO4 of TCA P is probably carrying an integrated signal of the δ18O of leaf water over time; that is, it does not only reflect the δ18O of the leaf water at the sampling time. Values lower than equilibrium could also be related to the presence of some TCA P, which has not exchanged yet with the leaf water, that is, still having the δ18O-PO4 value of the phosphate source in the nutrient solution. An additional effect could be related to the hydrolysis of organic P compounds during the extraction with 0.3 M TCA and the purification of the extracts. Organic P compounds have a different δ18O from the phosphate (see the discussion on structural P below) causing an offset of the measured δ18O-PO4 of TCA P from the calculated equilibrium. This could explain the positive offset from the equilibrium at leaves I.1–2, because the TCA extracts from these leaves had a high content of organic P (Fig. 2).

Leaves III.9–11 showed the largest negative offset from the calculated equilibrium value. This offset was too large to be caused only by the difference between the glasshouse air temperature and the actual leaf temperature. Compared with leaves III.3–4 and III.5–6, leaves III.9–11 had a relatively high acid phosphatase activity (Fig. 3). Acid phosphatase is an important enzyme for the hydrolysis of organic P compounds, for example during the reproductive stages of plant development (Duff et al., 1994). It is not currently known how the acid phosphatase affects the δ18O-PO4, but if acid phosphatase has a similar isotope fractionation factor to alkaline phosphatase, it is possible that it causes the negative offset from equilibrium as observed in leaves III.9–11. The acid phosphatase activity in leaves III.7–8 was also high, but we did not observe a significant offset from equilibrium. We conclude that the δ18O-PO4 of TCA P in leaves III.9–11 resulted from a combination of high acid phosphatase activity, the discrepancy between air and leaf temperatures, the activity of other phosphoenzymes, and the presence of phosphate, which still has the δ18O-PO4 of the phosphate source.

All these factors have to be taken into account when investigating the processes controlling the δ18O-PO4 of TCA P. Furthermore, the study of Tamburini et al. (2012b) showed that the δ18O-PO4 of TCA P varies among plant species (from 4.5‰ for Crisium sp. to 31.4‰ for Rumex sp.) and our study shows that the δ18O-PO4 of TCA P varies through time and within the same plant species. Differences among plant species might be explained by differences in the 18O enrichment of the leaf water or a different use of P. In addition, environmental factors such as water stress could affect the δ18O-PO4 of TCA P. However, the δ18O-PO4 values of TCA P from Agrostis sp. published by Tamburini et al. (2012b) were in equilibrium with the leaf water along the chronosequence at different time-points (F. Tamburini, unpublished data). This indicates that also under different climatic conditions and under different levels of P supply, TCA P is close to equilibrium with the leaf water.

Based on these results, a quantitative estimate of the contribution of TCA P to the soil phosphate pool using oxygen isotopes will be difficult. This is a consequence of the above-mentioned variability in the plants, but also microbial activity in the soils, which can rapidly erase the signal of an easily available P source such as TCA P (Tamburini et al., 2012b). Larsen et al. (1989) also showed that the signature of 18O-labeled KH2PO4 is rapidly lost as a result of biological activity.

The structural P pool

As hypothesized, structural P is not in oxygen isotopic equilibrium with leaf water, because it appears to be heavily enriched in 18O compared with the leaf water, the phosphate source added into the nutrient solution and the TCA P. Also, seed P, which is mainly composed of phytate (Raboy et al., 1984), has an elevated δ18O-PO4. This is not surprising, as other organic compounds such as cellulose are known to be enriched in 18O by c. 27‰ compared with leaf water (Schmidt et al., 2001). During synthesis of organic P compounds, oxygen exchange between phosphate and other oxygen-containing species such as water or organic compounds could occur. Photophosphorylation of ADP, for example, leads to incorporation of one oxygen atom from water into the newly formed ATP (Avron et al., 1965) Because, to the best of our knowledge, fractionation effects of enzymes catalyzing the synthesis of organic P compounds have not yet been studied, it is not yet possible to estimate the effects of such reactions.

In order to estimate the δ18O of phosphate released by the hydrolysis of structural P, the δ18O of the ambient water and the fractionation factors of the enzymes involved have to be known. Considering that the fractionation factors for hydrolyzing phosphoenzymes like alkaline phosphatase and 5′-nucleotidase are −30 and −10‰, respectively, the hydrolysis of structural P of the soybean leaves would lead to the release of phosphate with δ18O values between 25.8 and 49.9‰.

Because the δ18O of structural phosphate is significantly higher than that of other phosphate sources, it may be possible to trace its contribution to soil P. However, more studies are necessary to characterize the possible range in composition of structural P in relation to plant-specific or environmental factors such as P or water stress. Variations in the P supply, for example, lead to a change in the relative proportions of organic P compounds such as nucleic acids or phospholipids in plants (Veneklaas 2012), which could result in different δ18O-PO4 values of structural P. Also, different plant species may have different δ18O-PO4 values of structural P independently of the environmental conditions.


Our study showed that TCA P and structural P extracted from soybean leaves are enriched in 18O compared with the phosphate source in the nutrient solution. The fact that TCA P is close to isotopic equilibrium with leaf water indicates that the δ18O of leaf water and the activity of inorganic PPase or other PPases are the most important controls on the δ18O of TCA P. The processes controlling the oxygen isotope composition of structural P are still poorly understood. Further studies targeting specific P compounds such as RNA or DNA and the determination of fractionations associated with synthesis processes are necessary to develop mechanistic models of fractionation. In addition, most of the fractionation factors associated with phosphoenzymes in plants and soils, the uptake of P by plants and transporters in plants are not yet known. With a better understanding of these fractionation mechanisms, analysis of δ18O-PO4 could help to elucidate P use by plants and the fate of P derived from plant material in the soil–plant system. Further studies with different plant species and different growing conditions are necessary to test whether the conclusions drawn from this experiment can be generalized for other plants.


The authors would like to thank ETH Zurich for funding this project (grant number: ETH-02_10-2) and Delley Seeds and Plants Ltd (Delley, Switzerland) for providing the soybean seeds. We thank S. Bishop, M. Coray Strasser, S. Ragot, and C. von Sperber for their help in the laboratory and three anonymous referees for their helpful comments on our manuscript.