Rice (Oryza sativa L. cv. Oochikara) was grown in pots in a greenhouse at Rothamsted Research under anaerobic (flooded) and aerobic conditions as described elsewhere (Xu et al., 2008). The soil was taken from the plough layer (0–20 cm depth) of an arable field at Rothamsted. It is a moderately well-drained Aquic Paleudalf (USDA classification) or Luvisol (FAO classification) with silty clay loam texture (26% clay, 53% silt, 21% sand), 2% organic C, 0.2% total N, 5.93 g kg⁻¹ amorphous Fe oxides (by ammonium oxalate extraction) and pH 6.4. The soil was air-dried, sieved to < 8 mm and homogenized. Two watering treatments (aerobic and flooded) with four pots per treatment were randomly arranged on a bench inside a greenhouse (day/night temperatures 28/25 °C, light period 16 h per day with natural sunlight supplemented with sodium vapour lamps to maintain light intensity of > 350 μmol m⁻²s⁻¹). Five pre-germinated rice seeds were planted in each pot. To grow rice under aerobic conditions, plastic pots with holes at the bottom were placed on saucers and filled with 1 kg of soil, while for anaerobic conditions pots without holes were used. Fertilisers (120 mg N·kg⁻¹ soil as NH₄NO₃/(NH₄)₂SO₄, 30 mg P·kg⁻¹ and 75.5 mg K·kg⁻¹ soil as K₂HPO₄) were added to the soil and mixed thoroughly. Deionised water was added to full saturation with 2 cm of standing water for anaerobic rice, and to 70% of the soil's water holding capacity for aerobic rice. The two watering regimes were maintained throughout the experiment by daily additions of deionised water. Two further doses of 50 mg of N (NH₄NO₃) were added to each pot at 33 and 70 days after planting. The plants were harvested at grain maturity (117 days after planting). Stems were cut at 2 cm above soil surface, rinsed with deionised water, and dried at 60 °C for 48 h. The samples were separated into straw and grain (including husk), ground, pooled in one sample per pot and then aliquots were processed as described below. Root material was not collected for analysis because a large fraction of the roots had died and degenerated when the plants reached maturity. Only the above ground shoot materials were used.

Sample preparation was carried out in Class-10 laminar flow benches within class 1000 clean room facilities. Mineral acids except for HF were purified from Analar grade by sub boiling distillation in quartz stills. Distilled HF (∼28 M Ultrapure grade) and H₂O₂ (∼30% v/v Aristar grade) were purchased from VWR. All acid dilutions used > 18.2 MΩ cm⁻¹ grade H₂O from a Millipore purification system.

Oven-dried plant samples were ground using a porcelain pestle and mortar with liquid nitrogen and passed through a 0.5 mm² sieve. Plant samples were digested using a CEM Microwave Accelerated Reaction System (MARS) with 100-mL XP-1500 Teflon vessels. 5 mL of 15 M HNO₃, 3 mL 30% H₂O₂, and 0.6 mL 28 M HF were added to approximately 0.35 g of material. The acid-solid mixture was ramped to 210 °C and 250 psi and held for 90 min. Soil samples (50 mg) were digested using the same protocol. Soil leaches were used to estimate the composition of the Zn pool available to the plants. 30 mL 0.1 M HCl was added to 0.3 g of air-dried soil, and the resulting suspension was stirred continuously for 48 h at 25 °C. The suspension was centrifuged, and the supernatant was collected.

After digestion, samples were evaporated to dryness and the residue was dissolved in ∼1 mL 7 M HCl with 0.05 mL 30% H₂O₂ added to prevent reduction of ferric Fe. The solutions were dried again and re-dissolved in 2.5 mL 7 M HCl containing 0.001% H₂O₂ by volume. These solutions were split into three aliquots for Zn and Fe concentration measurements, for isotopic analysis, and for archive. The 0.5-mL aliquots for concentration measurement were diluted to 3.5 mL and made up to 1 M HCl and the measured using a Varian quadrupole ICPMS.

The separation of Zn and Fe from the matrix for subsequent isotope ratio measurements utilized Bio-Rad Poly-Prep columns filled with 2 mL of AG MP-1 100–200 mesh anion exchange resin. The resin was conditioned with 7 M HCl (10 mL) and successively washed with 0.1 M HNO₃ (10 mL), H₂O (10 mL), 0.1 M HNO₃ (10 mL) and H₂O (10 mL). The resin was re-conditioned with 7 M HCl-0.001% H₂O₂ (7 mL) and the samples loaded (1 mL solution in 7 M HCl-0.001% H₂O₂). The majority of matrix elements were eluted using 30 mL 7 M HCl-0.001% H₂O₂. This was followed by elution of Fe using 17 mL 2 M HCl and Zn using 12 mL 0.1 M HCl. The resin was then washed with 10 mL 0.5 M HNO₃. The Zn and Fe fractions were dried down, dissolved in 0.2 mL 15 M HNO₃, and dried to drive off residual HCl. Finally, 1 mL 0.1 M HNO₃ was added and the samples refluxed for > 2 hours to ensure complete dissolution. The solutions were then ready to be analysed with the Nu Plasma MC-ICPMS.

The total procedural blank contributions, accounting for the dissolution and ion exchange procedures, were below 1% of the total Zn and Fe for the samples with the lowest concentrations. The effect of these blank contributions on the final isotopic results was therefore negligible compared to the measurement uncertainty and no blank correction was applied.

Zinc isotopes were analysed using the double spike method (Arnold et al., 2010b). An isotope spike enriched with ⁶⁴Zn and ⁶⁷Zn was added to the sample aliquot to achieve a total of 1000 ng of Zn and a spike: sample mass ratio of 1. Values are reported as δ⁶⁶Zn relative to the widely used standard JMC−Lyon, i.e. δ⁶⁶Zn_JMC–Lyon = [(⁶⁶Zn/⁶⁴Zn)_sample/(⁶⁶Zn/⁶⁴Zn)_JMC–Lyon−1] × 10³. The Fe isotope analyses were performed by standard-sample bracketing using the reference material IRMM-014 as the standard, as detailed elsewhere (Chapman et al., 2009). Values are reported as δ⁵⁶Fe relative to the IRMM-14 standard, δ⁵⁶Fe_IRMM-014 = [(⁵⁶Fe/⁵⁴Fe)_sample/(⁵⁶Fe/⁵⁴Fe)_IRMM-014−1] × 10³.

Certified reference materials (rye grass BCR-281 and blend ore BCR-027) and single element industrial solutions (Johnson Matthey and Puratronic Alpha Aesar) were analysed to determine accuracy and precision of the isotope measurements (see Table 1 for details).

The average Zn concentration of the shoot tissue grown in aerobic soil is 75.5 ± 12.3 μg·g⁻¹ (mean ± 1SD, n = 4) and approx. 2.3 higher than in rice grown in anaerobic soil (32.6 ± 3.5 μg·g⁻¹, ± 1 SD, n = 4, Table 1). All measured concentrations are well above the threshold for deficiency, which is below 10 μg·g⁻¹ (Dobermann and Fairhurst, 2000; Smolders et al., 2013), suggesting the plants were Zn sufficient in both soils. This conclusion is supported by the shoot and grain biomasses, which were not significantly different for the two water regimes, despite the difference in Zn concentration of the shoot (Table 1). The lower Zn uptake under anaerobic conditions is expected because of the reduced Zn solubility in submerged soils (Kirk, 2004). Under aerobic soil conditions, the shoots, i.e. stems and leaves, contain a greater proportion of the total plant Zn (75% versus 63% under anaerobic conditions). The total Zn concentration of the soil is 105 μg·g⁻¹ whilst the 0.1 M HCl soil leach extracted 26 μg·g⁻¹.

The Zn isotope compositions of the rice shoots, grains and the total above ground plant material are reported in Table 1. The total above ground plant value was calculated from the mass of the shoot and grain material and their corresponding Zn contents and isotope compositions. We find that the shoots of the rice plants grown under both soil regimes are isotopically significantly heavier than the bulk soil or leach (Fig. 1). The grains, in contrast, are isotopically light. The calculated Zn isotope composition of the total above ground plant material is closer to that of the shoot and soil leach and 0.19‰ heavier in the rice grown under aerobic conditions (Table 1). This difference is small but significant within the 95% confidence interval.

We note a small difference in the isotope ratios in rice grown in aerobic or anaerobic soils. The shoots of rice grown in the aerobic soil yield a heavier isotope composition than the Zn in the labile soil pool. This is balanced by a bias towards light isotopes of Zn between shoot and grain, resulting in a small net negative bias between Zn in the total plant and the Zn in the soil leach. Under anaerobic soil conditions, the enrichment with heavy isotopes between soil leach and shoot transfer is smaller, and the net light bias between total above ground plant and soil solution is larger (Fig. 1).

Bulk and soil leach fractions show isotopic compositions (δ⁶⁶Zn_JMC−Lyon) of 0.36 ± 0.04 and 0.59 ± 0.02‰ (mean ± 1 SD, n = 2), respectively (Table 1). By mass balance, the Zn fraction remaining in the solid phase has a δ⁶⁶Zn_JMC−Lyon of ca. 0.28‰. This lies well within the range of igneous rocks and/or the inferred range of geological materials (Cloquet et al., 2008). It is possible that the 0.1 M HCl leachable fraction represents Zn from polluted atmospheric deposition, which can account for the heavier isotopic signature compared to the bulk soil (Weiss et al., 2014; Gioia et al., 2008).

The isotopic fractionation between the different reservoirs is assessed employing the Δ⁶⁶Zn notation calculated using the following equation (Weiss et al., 2013):

where Δ⁶⁶Zn_A−B is the isotopic fractionation between reservoirs A and B (i.e. soil leach, bulk soil, grain, shoot or above ground plant) and δ⁶⁶Zn_A,B are the isotopic signatures of the reservoir A and B. The results are given in Table 2.

The observed fractionations between the total above ground plant and the soil leach (−0.27‰ in anoxic soils and −0.08‰ in oxic soils) are well in line with that found in rice grown in hydroponic solution (Smolders et al., 2013; Weiss et al., 2005) and in submerged soils (Arnold et al., 2010a). The signatures in these previous studies were explained by the kinetically controlled uptake of free Zn²⁺ in the hydroponic studies and by the uptake of PS–Zn complexes in the submerged soils. Complexation to a phytosiderophore ligand in the soil solution prefers isotopically heavy Zn as shown by experimental and computational work (Fujii and Albarède, 2012; Fujii et al., 2014; Jouvin et al., 2009). Zinc has low solubility in aerobic and anaerobic soils and graminaceous plant roots, including rice, therefore secrete phytosiderophores to absorb the resulting Zn-phytosiderophore complexes (Reid et al., 1996; von Wiren et al., 1996; Widodo et al., 2010). We expect that phytosiderophore secretion is more pronounced in aerobic soil to solubilise Fe (III) and this indeed accounts for the higher Zn capture and the incorporation of heavier isotopes under aerobic soil conditions observed during this study (Fig. 1). We note that under Zn depletion in the soil solution, the isotopic composition of the total plant above the surface would trend towards that of the initial soil reservoir because the amount taken up approaches the total amount available in solution. The increased uptake of Zn observed under aerobic conditions and the isotopic composition of the total plant above the surface being closer to that of the soil leachate would be in line with this interpretation. However, mass balance calculations suggest that the rice extracted only a small fraction of the Zn that was available for extraction from the soil. The soil per pot contained approximately 26 mg Zn compared to 1.26 mg and 0.62 mg that were present in the plants under aerobic and anaerobic soil conditions, respectively (Table 1).

The grains show a significant enrichment with light Zn isotopes relative to the shoot material in rice grown under aerobic and anaerobic conditions alike (Table 2). The Zn budget of the shoot is accumulated over most of the growth season, whereas the grains largely accumulate their Zn during the grain-filling period of approximately 30 days. Previous work suggests that the rapid accumulation of Zn during the relatively short grain-filling period is controlled by the remobilisation of Zn from various parts of the plant and not by increased uptake from the soil medium (Nishiyama et al., 2012; Yoneyama et al., 2010). Our data indeed agrees with a stable isotope fractionation model for a unidirectional process such as translocation from shoot to grain (Fig. 2), the Rayleigh model is valid assuming that (i) the Zn is remobilized from the shoot during·grain-loading and the soil as additional source is excluded, (ii) the Zn pool of the shoot is not significantly replenished from the soil during this time period, and (iii) the Zn transport is not quantitative. The unidirectional model equation is given by:

Plant Fe concentrations and the total amount of Fe do not differ significantly between the rice plants grown in the two contrasting soil environments (Table 1). In both treatments, the shoot Fe concentrations are well above the threshold range for deficiency, i.e., 50 to 100 μg·g⁻¹ (Dobermann and Fairhurst, 2000). The Fe isotopic composition of the 0.1 M HCl leachate is about 0.3‰ lighter in δ⁵⁶Fe_IRMM-014 than the bulk soil (Table 1). This is likely due to partial leach of selected mineral phases with an isotopically light signature and/or a kinetic mechanism whereby the lighter isotopes are readily removed (Chapman et al., 2009; Kiczka et al., 2010a; Wiederhold et al., 2006). The δ⁵⁶Fe_IRMM-014 value of the soil leach is −0.27‰ (Table 1).

The fractionations between plant tissues and the source are calculated for the Fe relative to the bulk soil as this enables us to compare our results with previous work (Guelke and von Blankenburg, 2007; Guelke-Stelling and von Blanckenburg, 2012). We find an enrichment with isotopically light Fe between bulk soil and shoot of up to 0.5‰ in aerobic and anaerobic soils alike (Table 2). The largest difference in Δ⁵⁶Fe value is between shoot and bulk soil of plants grown on aerobic soils (0.45‰). The Fe taken up into the shoots has a significant lighter isotopic composition compared to the bulk soil but only insignificantly lighter relative to the soil leach (Table 1 and Fig. 1).

A notable finding is the similarity in the Fe isotopic composition between rice grown in aerobic and anaerobic soils, and the small fractionation between reservoirs. This suggests that the different physical-chemical conditions in the soils, i.e. redox potential, oxygen fugacity, solubility, do not affect the isotope signature in the plant, despite of the different Fe redox states and the variable amounts of Fe(III)–PS complexes likely present. The extent of enrichment with light isotopes in the shoot material observed is significantly smaller than that found in previous work (Guelke and von Blankenburg, 2007; Kiczka et al., 2010b) and this observation suggests significant uptake of Fe in its reduced form. Ferrous Fe is taken up without the need for preceding phytosiderophore-facilitated solubilisation in the rice rhizosphere (Ishimaru et al., 2006).

The rice does not show significant enrichment with light isotopes in the grain in disagreement with previous work on oat (Guelke and von Blankenburg, 2010; Guelke-Stelling and von Blanckenburg, 2012). At present, this difference is not easily explainable and further work is needed. Iron, similar to Zn, is transported into the symplasm via the apoplasm before grain-loading. The lack of isotopic fractionation could suggest a quantitative transport of complexed Fe (III) from the soil solution to the grain.