Cadmium accumulation is enhanced by ammonium compared to nitrate in two hyperaccumulators, without affecting speciation

Highlight Supply of NH4 + compared with NO3 − increased plant Cd uptake, translocation and accumulation in Carpobrotus rossii and Solanum nigrum. Cd-S was the dominant Cd species, with Cd being transported as free ions in xylem.


Introduction
Although widespread in the environment, cadmium (Cd) is a non-essential element with no known physiological functions. It enters soil from a variety of sources including application of metal-containing sewage sludge, phosphate fertilizers, waste from incinerators, and other industrial wastes (Nicholson et al., 1994). Cadmium presents a risk because it accumulates readily in plants to levels that are harmful in animal and human diets. It has been recognized that due to the efficient soil-to-plant transfer of Cd, dietary intake of food from crops grown on Cd-contaminated soils is a major route of Cd exposure to human health (WHO, 2010). Therefore, effective ways to remediate Cd-contaminated soils and mitigate Cd accumulation in crops are needed.
Among the various approaches for remediation, phytoextraction has attracted substantial attention because it is costefficient and environmentally friendly (Mahar et al., 2016). Phytoextraction reduces metal concentrations in contaminated soils through the accumulation of these metals in the above-ground biomass of plants. For this purpose, hyperaccumulators are plants that can accumulate metals to levels at least 100 times that of most plant species (Baker and Brooks, 1989). Although many studies have emphasized the use of plants for remediation of Cd-contaminated soils, the main limitation of the phytoextraction approach is the long remediation time. To overcome this limitation, various strategies have been used to increase shoot biomass or increase the metal concentration in the harvestable portions of the plants.
Nitrogen (N) fertilization is an agronomic strategy to improve phytoextraction. As an essential nutrient, N fertilization increases shoot biomass of N-deficient plants, with this also potentially increasing Cd accumulation. However, the effect of N on Cd accumulation varies with the N source and the application rate (Mitchell et al., 2000;Maier et al., 2002;Wångstrand et al., 2007). Nitrate (NO 3 − ) and ammonium (NH 4 + ) are the main sources of inorganic N taken up by plants, with N uptake comprising up to 80% of the total ion uptake. Therefore, the form of N supply plays an important role in the cation-anion balance, cellular pH, and rhizospheric pH (Marschner, 2011). For NH 4 + , uptake results in an excess of cations over anions, with net extrusion of protons and acidification of the rhizosphere -potentially increasing the availability of toxic metals such as Cd (Wu et al., 1989). The opposite is true for NO 3 − , with uptake resulting in an excess of anions over cations. However, some studies have actually shown that supply of NO 3 − (rather than NH 4 + ) increases uptake of Cd and other metals. For example, supply of NO 3 − enhanced Cd and Zn accumulation in Noccaea caerulescens (formerly Thlaspi caerulescens) compared to NH 4 + , even though NH 4 + lowered rhizoshphere pH (Monsant et al., 2008;Xie et al., 2009). Similarly, using nutrient solutions, supply of NO 3 − increased Cd uptake in tomato (Solanum lycopersicum), potato (S. tuberosum), and in the hyperaccumulators N. caerulescens and Sedum plumbizincicola (Xie et al., 2009;Luo et al., 2012;Hu et al., 2013;Jönsson and Asp, 2013). These findings indicate that, other than just influencing rhizosphere pH, N form also influences Cd uptake through additional mechanisms.
It is possible that the N form also influences Cd uptake and accumulation through changes in root morphology and plant growth (Rosen et al., 1990;Bloom et al., 2002) or through changes in the membrane potential of root cells (the driving force for cation uptake). For example, it has been reported that the uptake of NH 4 + causes depolarization of cell membranes and thus reduces Cd uptake, whilst NO 3 − favors Cd uptake by hyperpolarizing membrane potential (McClure et al., 1990;Miller et al., 2001;Zaccheo et al., 2006). Furthermore, the N form may influence the expression of cation transporters, which may in turn alter Cd uptake and translocation given that Cd uptake is probably via the transport systems of other cations (Clemens et al., 2002;Luo et al., 2012).
The form of N supply may also alter the chemical speciation of Cd in plants, thereby influencing Cd translocation and accumulation. It is known that most transition metal ions in plants, including Cd, are bound by various ligands rather than being present as hydrated ions, and the chelation of Cd in plants determines its sequestration and mobility (Clemens, 2006). Moreover, uptake and assimilation of different N forms influence the production of amino acids and organic acids in plants, which in turn impacts the ligands for Cd complexation. Compared to NH 4 + , NO 3 − increases the concentrations of organic acids but decreases amino acids (Roosta and Schjoerring, 2007;Marschner, 2011;White-Monsant and Tang, 2013). Furthermore, the form of N supplied influences the distribution of organic compounds in plant tissues, with NH 4 + mainly being assimilated in roots whilst NO 3 − assimilates in both roots and shoots (Marschner, 2011). It has been reported that amino acids, peptides, proteins, and organic acids are main ligands for Cd complexation, and that Cd speciation in plants varies with species, tissues, and environmental conditions (Salt et al., 1995(Salt et al., , 1997Vogel-Mikuš et al., 2010;Tian et al., 2011). However, it remains unclear how the N form influences the speciation of Cd within plant tissueswhich in turn will influence Cd uptake, translocation and hence accumulation.
The present study aimed to examine the effect of N form on Cd uptake and accumulation in two Australian native plant species, Carpobrotus rossii and Solanum nigrum, with both species having shown potential for Cd phytoextraction (Wei et al., 2013;Zhang et al., 2014). Specifically, this study aimed to determine whether the form of N supply alters Cd accumulation and speciation in root and shoot tissues. It was hypothesized that the supply of nitrate, relative to ammonium, would increase Cd uptake and accumulation, and that the increased Cd accumulation would be associated with changes in Cd speciation within differing tissues. Synchrotron-based X-ray absorption spectroscopy was used for in situ analyses of Cd speciation in the tissues.

Plant growth
Carpobrotus rossii (Haw.) Schwantes (Aizoaceae) was collected from a rural landfill site (37°36′S, 143°35′E, Snake Valley, Shire of Pyrenees) in Victoria, Australia, while seeds of Solanum nigrum L. were collected from plants grown on the La Trobe University farm (37°72′S, 145°4′E). Seeds of S. nigrum were germinated in a solution containing 600 µM CaCl 2 and 5 µM H 3 BO 3 in a dark controlled-environment room for 5 d. For C. rossii, uniform cuttings (two nodes each cutting) were washed with tap water. The plant materials were transplanted to 5-l polyethylene pots filled with a basal nutrient solution aerated continuously. The basal nutrient solution had the following composition (µM): 500 NH 4 NO 3 ; 200 MgSO 4 ; 10 KH 2 PO 4 ; 600 K 2 SO 4 ; 600 CaCl 2 ; 20 FeNaEDTA; 5 H 3 BO 3 ; 1 MnSO 4 ; 0.2 CuSO 4 ; 0.03 Na 2 MoO 4 ; 1 ZnSO 4 . The root systems were well-developed after grown in these solutions for 15 d (C. rossii) and 30 d (S. nigrum). After this growth in basal solutions, three seedlings of C. rossii and five of S. nigrum were transferred to new pots containing the treatment solutions (below).
The experiment consisted of eight treatments and was replicated three times. The eight treatments were two plant species (C. rossii and S. nigrum) × two forms of N (500 µM (NH 4 ) 2 SO 4 and 1000 µM KNO 3 ) × two Cd concentrations (5 and 15 µM Cd). Solution pH was buffered with 2 mM MES [2-(N-morpholino)ethane-sulphonic acid] and 1.2 ml of 1 M KOH was used to adjust pH to ca. 6.0, which was maintained daily using 1 M KOH. Since sulfur (S) plays an important role in Cd binding, 500 µM K 2 SO 4 was added to the KNO 3 treatments in order to ensure constant S concentrations in all treatments [hence, the K concentration was 1210 µM in the (NH 4 ) 2 SO 4 treatments and 3210 µM in the KNO 3 treatments]; the composition of other nutrients was the same as that of the basal nutrient solution. Solutions were renewed every 3 d. Plants were grown in a controlled-environment growth room with 14-h photoperiod, a light intensity of 400 μmol m −2 s −1 , 20 °C during the day and 18 °C during the night, and 50% relative humidity during the day, and 60% relative humidity during the night.

Xylem sap collection
After growth in the treatment solutions for 14 d, xylem sap was collected according to the method of Monsant et al. (2011). Briefly, the stem was cut with a fresh razor blade, with the root stump immediately washed with deionized water and dried with paper tissue. Xylem sap was collected for 1 h using a micropipette and transferred into 1.5-ml Eppendorf tubes. The xylem sap samples from the various plants in each pot were pooled, with each treatment having three replicates. For S. nigrum, however, the volume of xylem sap collected was insufficient to permit analysis, and hence only samples from C. rossii seedlings are presented here.

Plant analysis
After collecting the xylem sap, the plants were harvested and fresh weights recorded. Roots were divided into two parts, with the first subsample immersed in ice-cold 20 mM Ca(NO 3 ) 2 for 15 min, washed with deionized water, frozen in liquid nitrogen, and stored at −80 °C. The second root subsample was immersed in ice-cold 20 mM Na 2 -EDTA for 15 min to remove Cd adhering to the root surface and then washed with deionized water and weighed. The roots were examined using a root scanner at 600 dpi (Epson Perfection 4990 Scanner, model J131B, Epson Inc.) to determine morphological parameters (length and surface area) before being oven-dried (at 80 °C in paper bags) for analysis. After washing with deionized water, shoots were blot-dried and separated into stems, old leaves and young leaves. Like the roots, the shoots were subdivided into two subsamples. The first subsample was frozen in liquid nitrogen, and stored at −80 °C, with the second subsample oven-dried. For the plants grown in solutions containing 5 μM Cd, a portion of each tissue was also freeze-dried for X-ray absorption spectroscopy analysis. The oven-dried samples were ground and digested using HNO 3 in a microwave digester (Multiwave 3000, Anton Paar). The concentrations of elements in the digests and xylem sap (0.5 ml mixed with 2.5 ml of 5% HNO 3 ) were analyzed using an inductively coupled plasma optical emission spectrometry (ICP-OES) (Perkin Elmer Optima 8000, MA, USA).

Cadmium speciation by X-ray absorption spectroscopy (XAS)
The speciation of Cd in different tissues was examined at the XAS beamline of the Australian Synchrotron (Victoria, Australia). The energy of each spectrum was calibrated by simultaneous measurement, in transmission, of a metallic Cd foil reference (K-edge at 27 711 eV). The spectra were collected in the fluorescence mode with a 100-element solid-state Ge detector. To minimize beam-induced artifacts and thermal disorder, samples were placed in a cryostat sample holder (maintained at ca. 12 K, liquid helium). The beam size was adjusted to ca. 2 × 0.5 mm.
To prepare the tissues for analysis, approximately 1-2 g frozen or freeze-dried samples were ground under liquid nitrogen using an agate mortar and pestle. The homogenized tissues were placed into a sample holder with Kapton tape windows cooled with liquid nitrogen before being transferred directly to the cryostat for analysis. For the xylem sap, treated plants were transported to the Australian Synchrotron (ca. 45 min) before growth for an additional 24 h. The sap was thus collected immediately prior to analysis. The xylem sap was mixed in 30% glycerol (see below) and loaded into a sample holder sealed with Kapton tape.
A total of 14 Cd-containing standard compounds (eleven aqueous and three solids) were analyzed using Cd K-edge XANES (X-ray absorption near edge structure) spectroscopy to allow interpretation of the XANES spectra from the experimental tissue samples. The eleven aqueous standards consisted of (i) 1 mM Cd(NO 3 ) 2 , (ii-x) 1 mM Cd(NO 3 ) 2 mixed with various ligands at a final concentration of 5 mM [phytate, histidine, citrate, malate, succinate, cysteine, glutathione (GSH), methionine (MT), and phytochelatin 2 (PC2)], and (xi) 1 mM Cd(NO 3 ) 2 mixed with 0.1% polygalacturonic acid -this being the major charged component of the cell wall. Standards were prepared using a 10 mM Cd(NO 3 ) 2 stock solution together with 50 mM solutions of the various ligands, or 1% polygalacturonic acid stock solution. All aqueous standards (other than polygalacturonic acid) were mixed in 30% glycerol to limit ice-crystal formation during cooling. The 1% of polygalacturonic acid stock solution, with 30.4 μmol ml −1 free carboxyl groups determined by titration with 0.02 M NaOH to neutrality, was prepared approximately 24 h before use and kept at 4 °C Taylor and Walter, 2012). The citrate, GSH, malate, phytate, succinate, histidine, and MT standards were adjusted to ca. pH 6, cysteine to pH 7, and PC2 to pH 10 using 0.1 M NaOH, while pH was not adjusted for the polygalacturonic acid or the 1 mM aqueous Cd(NO 3 ) 2 . Where constants were available, GEOCHEM-EZ was used to model the standard solutions, with the results indicating that >97% of Cd was complexed with citric acid, 85% with succinic acid, malic acid, and histidine. The three solid standards, CdS, CdO, and CdCO 3 , were diluted to 100 mg Cd kg −1 using cellulose.
Multiple XANES scans were performed for each sample, with two scans per standard, and either two or three scans for each plant tissue sample. Beam-induced damage was assessed using a tissue sample, with two XANES scans conducted at the same sample locationthe two scans were then compared to see if they differed. For other samples, positions were changed after each scan to reduce the risk of beam damage and to obtain representative spectra.
The XANES spectrum for each sample was energy-normalized using the reference energy of the Cd foil, with replicate spectra for each sample merged using Athena (version 0.9.22) (Ravel and Newville, 2005). Principal component analysis (PCA) of the normalized sample spectra was employed to assess how many independent components were contained in the samples, and target transformation (TT) was used to identify relevant standards for linear combination fitting (LCF) of the sample spectra. PCA and TT were undertaken using SixPack (Webb, 2005). Using PCA, the XANES spectra of the plant tissues were compared with those of the standards to evaluate the number of relevant components indicated by the indicator function (IND) reaching a minimum. The results of PCA indicated that the first four components accounted for 98.6% of the total variance of the XANES spectra, so only four components were needed to fit the data. Target transformation was then used to identify the standard spectra in the LCF analyses by selecting standards with a SPOIL value <3. Nine of the 14 reference spectra met this criterion, these being Cd-citrate, Cd-malate, Cd-succinate, Cd-polygalaturonate, Cd-GSH, Cd-cysteine, Cd-PC2, Cd(NO 3 ) 2 , and CdS (Supplementary  Table S1 at JXB online). The LCF was used to identify the relative proportions of standard spectra within the sample spectra as the XANES spectrum represents a combination of all Cd species present in the portion of sample transected by the beam, and the fitting energy range was -30 to +100 eV relative to the Cd K-edge.

Statistical analysis
The effect of N form and Cd concentration in solution on plant biomass, tissue Cd concentrations, shoot Cd content, and Cd translocation was examined using a two-way analysis of variance for each species. Significant (P≤0.05) differences between means were identified by Tukey's HSD test using GenStat v. 11 (VSN international).

Plant growth
All plants appeared healthy during the experimental period, with little visual difference between treatments. Overall, C. rossii had shoot biomass 5-fold higher than S. nigrum (Fig. 1).
Despite this overall difference in biomass production between the two species, the form of N and the Cd concentration had no significant impact upon shoot biomass for either species. The N form and Cd level in solutions only influenced the root biomass of S. nigrum, with biomass at 15 µM Cd 18% lower when supplied with NO 3 − than when supplied with NH 4 + . Moreover, increasing the Cd concentration in the solution from 5 to 15 µM decreased root biomass by 12% when supplied with NH 4 + and by 29% when applied with NO 3 − (P<0.05). Furthermore, the root length and root surface area per plant were 16-40% greater when supplied with NH 4 + than with NO 3 − , irrespective of the Cd concentration in the nutrient solution (except for the root surface area of C. rossii at 15 μM Cd) (Supplementary Table  S2).

Plant uptake and tissue Cd concentrations
Plant tissues differed significantly in Cd concentration (Fig. 2). Roots had the highest Cd concentrations in both species. In the shoots, C. rossii had highest Cd concentrations in young leaves, followed by stems and old leaves, resulting in a two-fold difference in Cd concentration between young and old leaves. In comparison, S. nigrum had similar Cd concentrations in the various above-ground tissues.
For both species, shoot Cd concentrations, especially leaf Cd concentrations were significantly higher when plants were supplied with NH 4 + than when supplied with NO 3 − (Fig. 2). For example, Cd concentrations in leaves of C. rossii were 50-60% higher when supplied with NH 4 + than with NO 3 − , irrespective of the Cd concentration in the nutrient solution (except for the old leaves at 5 μM Cd). Similarly, leaf Cd concentrations of S. nigrum were 50-70% (5 μM Cd) and 30% (15 μM Cd) higher in the plants supplied with NH 4 + than with NO 3 − . In addition, stem Cd concentrations of S. nigrum were 23% higher in NH 4 + -fed than NO 3 − -fed plants at 5 μM Cd, but did not differ between the N forms at 15 μM Cd. In addition, NH 4 + treatment tended to have higher Cd concentrations in roots of both species, although the N effect was not statistically significant for either of them.
Accordingly, shoot Cd amounts were higher in plants supplied with NH 4 + than with NO 3 − , although the effect of N form was greater for S. nigrum than C. rossii (Fig. 3a). Indeed, shoot Cd amounts of C. rossii were 20% higher in plants supplied with NH 4 + than with NO 3 − at 15 μM Cd (P<0.05), while those of S. nigrum were 60% (5 µM Cd) and 23% higher (15 µM Cd) in plants supplied with NH 4 + than with NO 3 − . Overall, for a given treatment, C. rossii  accumulated three times more Cd in shoots than S. nigrum. Similar trends were observed for total amounts of Cd taken up by plants and Cd uptake per unit root mass of both plant species (Supplementary Fig. S1).
For C. rossii, the translocation factor (i.e. the shoot-to-root concentration ratio) was not altered by the N form at 5 µM Cd, but at 15 µM Cd it was 29% higher when plants were supplied with NH 4 + (Fig. 3b). In comparison, the translocation factor of S. nigrum was 47% higher when supplied with NH 4 + than with NO 3 − only at 5 µM Cd, but not at 15 µM Cd. Overall, the translocation factors were approximately twice as high for S. nigrum than for C. rossii, and were 80% higher at 15 µM Cd than at 5 µM Cd.

Reference compounds for XAS
The Cd K-edge XANES spectra of various Cd standard compounds were first examined visually to identify the most likely forms of Cd within the plant tissues. Given that all standards examined were prepared using Cd 2+ , the Cd K-edge XANES spectra from all standards exhibited a similar absorption edge, with white-line peaks at ca. 26 719 to 26 722 eV (Fig. 4). However, the ligand to which the Cd was bound resulted in slight shifts in these white-line peaks, with Cd bound to S-containing ligands generally having white-line peaks at ca. 26 719 eV, whilst the white-line peaks for Cd bound to O-or N-containing ligands were at ca. 26 720 eV. Not only were slight shifts in the white-line peak observed, but differences in the shape of the spectra were also evident. For S-containing ligands, the spectra were comparatively flat and with distinctive features (Fukuda et al., 2008). Furthermore, the spectra of CdO and CdCO 3 exhibit characteristic spectral features, especially at ca. 26 729 and 26 736 eV.
Although the various types of ligands influenced the spectra, those obtained for Cd(NO 3 ) 2 and for Cd complexed by various carboxyl groups (e.g. Cd-citrate, Cd-malate, Cd-succinate, Cd-polygalaturonate) were all very similar in their appearance. Likewise, the spectra for many of the S-containing ligands (such as CdS, Cd-cysteine, Cd-GSH, and Cd-PC2) were similar. Given that some spectra were largely indistinguishable from each other, hereafter, the Cd  present as either free Cd [i.e. the Cd(NO 3 ) 2 standard] or as Cd complexed by carboxyl-containing compounds (i.e. citrate, malate, pectin) were grouped together and referred to as 'Cd-OH'. Similarly, Cd-cysteine, Cd-PC2, Cd-GSH, and CdS were grouped together and referred to as 'Cd-S'.

Cd speciation in plant samples using XAS
In all hydrated samples of plants grown at 15 μM Cd, the tissue Cd concentrations were sufficiently high to allow collection of XANES data with a good signal-to-noise ratio (Fig. 5). The spectra of plant samples from the 5 μM Cd treatments were collected using the freeze-dried samples due to the low Cd concentrations present. Thus the spectra of freeze-dried and frozen hydrated roots of C. rossii grown in 5 μM Cd were compared and were found to be similar, indicating that Cd speciation was not altered by the freeze-drying treatment ( Supplementary Fig. S2). In addition, the spectra for comparable tissues of plants exposed to different Cd levels showed strong similarities, and the LCF revealed that the Cd levels in the solution did not alter Cd speciation in plants.
With regards to C. rossii, the spectra for both root and shoot tissues were visually similar to that of Cd-S, whilst the spectra of xylem sap clearly differed (Fig. 5). Indeed, LCF predicted that the majority (60-91%) of Cd in all tissues (other than the xylem sap) was associated with S-containing ligands, while Cd-OH was the dominant form (87-95%) in xylem sap (with the remaining 5-13% of Cd in the xylem sap associated with S) (Table 1). Although Cd-S dominated in all tissues of C. rossii other than the xylem sap, there was generally a higher proportion (ca. 90%) of S-bound Cd in young leaves than in old leaves and stems (ca. 60-80%). Furthermore, the spectra were visually similar for comparable tissues irrespective of the N form, indicating that the N-form did not alter Cd speciation, although LCF subsequently predicted that the proportion of Cd-OH in root tissues, xylem sap, and stems of NO 3 − -fed plants was ca. 10% higher than that in NH 4 + -fed plants (Table 1).
Overall, the proportion of Cd-OH was higher in the shoots of S. nigrum (31-56%) than that observed for the shoots of C. rossii (9-40%). As observed for C. rossii, the proportion of Cd associated with S-containing ligands in S. nigrum differed between tissues, with higher levels in young leaves (ca. 70%) than in old leaves and stems (44-55%). Finally, it was again observed for S. nigrum that the form of N supplied did not influence Cd speciation in any tissues.

Ammonium enhances Cd uptake and translocation
The present study did not support our hypothesis that the supply of NO 3 − , compared to NH 4 + , would increase Cd uptake in these two plant species. Instead, NH 4 + increased Cd uptake and translocation to the shoots. It is evident that supply with NH 4 + increased both the shoot Cd concentration, especially that in the leaves, and Cd accumulation in shoots, relative to that observed when supplied with NO 3 − . This increased Cd concentration when supplied with NH 4 + was not due to a 'dilution' effect, because plant growth and biomass production were similar for the two N forms. Furthermore, the difference in shoot Cd uptake between the two N forms did not result from changes in solution pH or solution Cd speciation, because buffered solution pH was constant and Cd speciation was similar for both the NH 4 + and NO 3 − treatments (Supplementary Table S3). With regards to the different compositions of the two N-treated solutions, the potential (Ψ 0 o ) and the Cd 2+ activity ({Cd 2+ } 0 o ) at the cell membrane surface were calculated according to Wang et al. (2011), and only differed slightly between the NH 4 + and NO 3 − treatments (Supplementary Table S4). Thus, it appears that the increased uptake and accumulation of Cd in solutions containing NH 4 + were not the result of altered bioavailability of Cd or plant growth, but instead due to other processes that influenced Cd accumulation, including Cd uptake, partitioning within the root, and efficiency of translocation from root to shoot (Clemens et al., 2002;Clemens, 2006).
The higher Cd accumulation in the shoots of NH 4 + -fed plants could be partially attributed to the higher Cd uptake by the roots. It was evident that supply with NH 4 + increased not only the total Cd uptake but also the Cd uptake per unit of root biomass in both plant species ( Supplementary  Fig. S1). There are a number of possible reasons for this. Firstly, it is known that N form can alter root morphology. In the present study, although the N form did not affect root biomass of C. rossii, root length and surface area were greater when supplied with NH 4 + than when supplied with NO 3 − (Supplementary Table S2), which is consistent with the findings of Liu et al. (2015). It has been reported previously that root development is altered depending upon the N form supplied, with NH 4 + accelerating cell division and subsequent root branching (Bloom et al., 2002). Thus, the NH 4 + -fed plants in the present study might have had a higher capacity to acquire Cd. Secondly, the regulation of Zn transporters involved in Cd uptake might potentially be enhanced in roots of NH 4 + -fed plants. The uptake of Cd is likely to occur through ZIP transporters (Zn/Fe-regulated transporter-like proteins) and it has been shown that increasing concentrations of tissue N can enhance the abundance of Zn-uptake transporters in the root (Erenoglu et al., 2011). Considering the higher N concentration in the roots of NH 4 + -fed plants (Rosen et al., 1990;Hassan et al., 2005), the higher Cd and Zn uptake might be due to increased Zn transporters.
Other than this higher uptake by the roots, the higher Cd accumulation in the shoots of NH 4 + -fed plants could be partially attributed to an increased efficiency of Cd translocation to the shoots. This was evidenced by examination of translocation factors, with these being higher in NH 4 + -than NO 3 −fed plants. This is also consistent with the observation that the Cd concentration of the xylem sap in NH 4 + -fed plants was 20% higher than in NO 3 − -fed plants (11.51 and 9.03 mg L −1 at 15 μM Cd, respectively). Xylem loading of Cd has been shown to be mediated by the P-type ATPase transporter AtHMA4 in different plant species such as Arabidopsis thaliana, A. halleri, N. caerulescens and S. nigrum, and overexpression of AtHMA4 enhances Zn and Cd levels in leaves (Verret et al., 2004;Verbruggen et al., 2009;Xu et al., 2012). The addition of NH 4 + , compared to NO 3 − , might also elevate the expression of the Cd transporter AtHMA4 and then increase Cd translocation, but this still needs further study.
Our finding that NH 4 + increased Cd uptake and accumulation in both C. rossii and S. nigrum is consistent with that observed in lettuce (Lactuca sativa) (Florijn et al., 1992), but differs from that reported for tomato (Solanum tuberosum) and the hyperaccumulators S. plumbizincicola and N. caerulescens, where NO 3 − increased Cd accumulation in shoots (Xie et al., 2009;Luo et al., 2012;Hu et al., 2013). There are several possible ways to explain the apparent discrepancies between these studies. First, plant species differ in their growth response to N form. In the present study, plant biomass was not affected by N form, consistent with the result of Liu et al. (2015) when C. rossii was grown in an acid soil. In other studies, however, NH 4 + significantly decreased the growth of Solanum lycopersicum, S. plumbizincicola and N. caerulescens (Xie et al., 2009;Luo et al., 2012;Hu et al., 2013), with this decreased growth potentially altering Cd uptake. Second, plant species differ in their mechanisms of Cd detoxification. For example, in S. alfredii (which belongs to same genus as S. plumbizincicola) and N. caerulescens Cd is mainly co-ordinated with O-ligands (Küpper et al., 2004;Tian et al., 2011), whilst in C. rossii and S. nigrum Cd was mainly associated with S-ligands. Thus, it may not be unexpected that the effects of N form differ between these species, given that NO 3 − nutrition enhances the concentration of organic acids (O-ligands) (which are important for Cd uptake and tolerance in S. plumbizincicola and N. caerulescens) whilst NH 4 + nutrition increases amino acids and proteins (with Cd-S compounds important in C. rossii and S. nigrum). Third, plant species may differ in the mechanisms used to balance ion uptake when different N forms are applied. In previous studies, NO 3 − -fed plants accumulated more cations (including Cd) because of the antagonistic effects between NH 4 + and Cd 2+ (Xie et al., 2009;Hu et al., 2013), or the higher expression of IRT1 in NO 3 − -fed plants than in NH 4 +fed plants (Luo et al., 2012). However, in the present study, significantly higher concentrations of Zn and Cd were found in NH 4 + -fed plants, with concentrations of other cations, The values in brackets show the percentage variation in the calculated values. The goodness of fit is indicated by the R-factor: R-factor = Σ i (experiment -fit) 2 ⁄ Σ i (experimental) 2 , where the sums are over the data points in the fitting region.
including Fe, not affected by N form (Supplementary Table  S5).

Nitrogen form does not affect Cd speciation
Nitrogen form did not greatly affect Cd speciation in the plant tissues irrespective of treatment, except that the proportions of Cd-S in some tissues of C. rossii were slightly lower in NO 3 − -fed plants (Table 1). These results indicate that the increased accumulation of Cd when supplied with NH 4 + could not be ascribed to alterations in Cd speciation. This observation differs from what has previously been proposed, i.e. that differences in Cd concentrations and translocation in plants supplied with different N forms might result from differences in metabolites, such as organic acids and amino acids. For example, it is known that NO 3 − nutrition increases the concentration of organic acids in plants in order to maintain the intercellular pH (Marschner, 2011). Our observation that the N form did not alter speciation is similar to previous observations for the Zn-hyperaccumulator N. caerulescens that N form did not alter Zn speciation in different tissues, despite NO 3 − enhancing the production of organic acids (Monsant et al., 2011;White-Monsant and Tang, 2013). Therefore, the formation of Cd-organic acid complexes in plants may not depend on the concentrations of organic acids, but rather on efficient tonoplast transporters to facilitate Cd transport from the cytoplasm into vacuoles (Ueno et al., 2005).

Plant species and tissues differ in their Cd speciation
Although Cd-S was the dominant Cd species in the shoots of both species, the proportion of Cd-S in all the above-ground tissues was generally higher for C. rossii (60-90%) than for S. nigrum (45-70%) (Table 1). Thus, O-containing ligands tended to be more important in S. nigrum than in C. rossiithis observation for S. nigrum being similar to that reported previously (Sun et al., 2006). These differences are not unexpected, with Cd speciation reported to differ widely between species (Castillo-Michel et al., 2009;Tian et al., 2011;Huguet et al., 2012;Lefèvre et al., 2014) and even between ecotypes of the same species (Ebbs et al., 2009).
Not only did Cd speciation differ between species, but it also differed between tissues. Firstly, Cd was clearly bound to S-ligands in the roots of both species, with similar observation reported previously for other species, including corn (Zea mays), Indian mustard (Brassica juncea), and A. thaliana (Salt et al., 1995;Isaure et al., 2006;Castillo-Michel et al., 2009). The high proportion of Cd-S in roots might explain the greater Cd concentration in roots than in shoots in both species, as Cd translocation was restricted by its complexation with strong ligands and its sequestration in the vacuole of root cells. Secondly, the low proportion of Cd-S in the stems of both species is inconsistent with the finding of Küpper et al. (2004) who found a high concentration of Cd-S in stems of N. caerulescens and proposed that Cd was transported as PC or MT complexes. However, in the present study, analyses of the xylem sap of C. rossii showed that the free Cd 2+ ion dominated in the xylem sap, although concentrations in other aerial tissues were low. This indicates that Cd was transported as either the free ion or complexed with carboxyl groups before being chelated by S-containing compounds, which may explain the low Cd-S proportion in stems. Moreover, the importance of free hydrated Cd 2+ ions and/or Cd-O/N complexes for Cd translocation through xylem sap has also been found previously in other Cd-hyperaccumulating and non-hyperaccumulating species (Salt et al., 1995;Ueno et al., 2008;Hazama et al., 2015). Thirdly, young leaves tended to have a higher proportion of Cd bound to S-containing ligands than did old leaves. This decrease in Cd-S with increasing leaf age has also been reported in N. caerulescens (Küpper et al., 2004). The higher proportion of Cd-S in young than old leaves coincided with the accumulation of more Cd in the epidermal cells of young leaves but in the mesophyll cells of old leaves, both of which would protect the highly sensitive metabolic processes in the young leaves from Cd 2+ damage (Küpper et al., 1999;Koren et al., 2013).

Conclusions
This study used synchrotron-based XAS to investigate how the N form in the nutrient supply influences Cd speciation and subsequent Cd accumulation in two newly defined Cd hyperaccumulators, which had not been examined previously. It has been demonstrated that the supply of NH 4 + , relative to NO 3 − , increased the shoot Cd accumulation by 30% in two native Australian plant species, C. rossii and S. nigrum, when grown in nutrient solution. Interestingly, this increase in Cd accumulation did not result from changes in Cd speciation within the tissues. Regardless of this, Cd speciation differed between the species and between tissues, with a higher proportion of Cd-S in the shoots of C. rossii than for S. nigrum, and more Cd being stored as Cd-S in roots and shoots but translocating as Cd-OH in the xylem sap. In addition, C. rossii accumulated three times more Cd in shoots than S. nigrum. Thus, the application of NH 4 + -based fertilizers would potentially favor phytoextraction of Cd by C. rossii and S. nigrum, and C. rossii is a better candidate for phytoextraction in Cd-contaminated soils. However, further work is required to confirm these findings, including experiments in soil-and field-based conditions.

Supplementary data
Supplementary data are available at JXB online. Figure S1. Effects of N form on the total Cd accumulation and the Cd uptake per root dry weight of plants. Figure S2. Normalized Cd K-edge XANES spectra for freeze-dried roots and frozen hydrated roots of NH 4 + -fed Carpobrotus rossii. Table S1. Target transformation SPOIL values of selected reference spectra obtained by principle component analysis. Table S2. The root length and surface area per plant of the two species after 14 d of treatments. Table S3. Speciation of Cd in the nutrient solution of different treatments. Table S4. The calculated Cd 2+ activities in the bulk-phase medium, the cell membrane surface potentials, and Cd 2+ activities at the cell membrane surface in the different treatments. Table S5. Concentrations of calcium, magnesium, zinc, and iron in shoots of plants grown for 14 d in solutions containing Cd at a concentration of either 5 or 15 µM.