Introduction

Chronic Cd exposure has been related to adverse health effects, such as renal tubular dysfunction and osteomalacia (FAO/WHO 2010). For the non-smoking population, dietary intake is the main route of Cd exposure. After staple foods like cereals and potato tubers, it has been estimated that chocolate and chocolate products contribute significantly to the total dietary Cd exposure in the European population (EFSA 2012). To protect consumers, regulators worldwide have established limits on Cd concentrations in chocolate and cacao powders between 0.10 and 0.80 mg Cd kg−1 (Mercosur 2011; Ministry of Health of the Russian Federation 2011; European Commission 2014; Food Standards Australia New Zealand 2017; Codex Alimentarius Commission 2018). These regulations prompted research to reduce Cd concentrations in cacao beans, which are the major ingredient for chocolate and cacao powders (Meter et al. 2019; Vanderschueren et al. 2021). A major bottleneck that hampers the development of these mitigation strategies, is that the knowledge about the circulation of Cd in cacao is small compared to that in other plant species, notably regarding the transport of Cd towards the beans.

The circulation of Cd in mature trees has been studied but remains elusive. Based on existing knowledge from other plants than cacao plants, we know that Cd could be loaded into the beans (seeds) by potentially two pathways: a direct translocation or a reallocation from deposits in vegetative tissues (Maillard et al. 2015). In cacao, the first pathway may be a direct translocation from roots and branches via the transpiration stream (xylem) to the beans, with a possible xylem-to-phloem transfer in the rays of the branches or in the transpiring tissues surrounding the bean (pod husk and testa). The second possible pathway is that Cd is reallocated from vegetative stores, like the leaves. Reallocation from leaves to beans requires that Cd is mobile in the phloem (White and Brown 2010). In cacao, the first study on Cd pathways suggested that a substantial amount of Cd could be directly loaded from roots and branches into beans with a xylem-to-phloem transfer in the woody parts, without passage through the leaves (Engbersen et al. 2019). This conclusion was based on differences in Cd distribution among 11 cultivars. Their suggestion implied that reallocation from leaves through the phloem was rather limited for the loading of Cd in the beans. A later study with isotope data at natural abundance corroborated that Cd in the beans could be loaded from the phloem tissues of the branches (Blommaert et al. 2022). They observed no isotope fractionation between the branches and beans, whilst the isotopic signature of Cd in the leaves was heavier than in the branches and beans. The absence of isotopic fractionation between the branches and beans can indicate a direct translocation from the branch phloem to the beans, while Cd gets more and more depleted in light isotopes during its translocation via the xylem to the leaves. Yet, a follow-up study that used pulse-labelling of intact trees in the field, suggested that Cd in the beans is rather reallocated from a vegetative store of Cd in the tree than by direct transport from the roots (Vanderschueren et al. 2023). This is because the 108Cd isotope tracer that was applied to the soil arrived significantly later in the beans than in the leaves, which are fed directly by the transpiration stream. However, it remained unclear if the source of Cd was a reallocation from the leaves or if bean Cd was rather loaded from a stock in the branches. Despite the abovementioned studies, it remains thus unclear how exactly Cd circulates in cacao, and more particularly if Cd can be reallocated from vegetative stores via the phloem to developing organs like the beans or if it is rather retained in the leaves.

Isotope tracing using stable isotopes or radioisotopes is an interesting tool to study the short-term circulation of elements in plants. In the case of Cd, isotope tracing has been used to infer Cd pathways in agricultural crops (Page and Feller 2005; Yan et al. 2019, 2023) and hyperaccumulators (Wei et al. 2014). For example, in wheat, the use of the radioisotope 109Cd as a tracer showed a strong initial retention of Cd in the roots and corresponding slow release of 109Cd into the shoots (Page and Feller 2005). On the contrary, in the hyperaccumulator S. nigrum, 109Cd was rapidly transported to the shoots due to an efficient loading into the xylem in the roots (Wei et al. 2014). Once, in the leaves it was not or slowly redistributed via the phloem within the shoots. Thus, the comparison of these studies illustrates that the mobility of Cd in the phloem affects the efficiency of reallocation and varies largely among plant families (Page and Feller 2015; Sterckeman and Thomine 2020). Today, there are no studies that document the short-term circulation and phloem mobility of Cd in cacao, which could infer our understanding of how Cd is taken up and distributed, and if it can be reallocated from the leaves to developing organs.

Split-root experiments have been used in the past to reveal the circulation of trace elements in plants (Mclaughlin et al. 2000; Haslett et al. 2001; Qin et al. 2013; Page and Feller 2015; Valentinuzzi et al. 2020; Wei et al. 2022). This experimental design involves dividing the root system into two equivalent segments, each growing in distinct compartments. By introducing an isotope tracer to one compartment, the subsequent circulation of this tracer throughout the plant can be precisely tracked. Tracer movement from the labelled to the non-labelled root compartment can only occur via the phloem, either by transport through the phloem between root segments or by a downward flux of Cd from the shoot (Welch et al. 1999). Hence, such design can advance greatly our understanding of how Cd circulates in cacao.

This study was set up to identify the circulation of Cd in cacao, and if there is Cd reallocation via the phloem. To this end, we used a split-root approach with cacao seedlings. In this setup, the nutrient solution of one root compartment was labelled with 108Cd stable isotope to trace the fate of 108Cd in the plant.

Materials and methods

Experimental design

A split-root hydroponic system was set up with five-week-old Theobroma cacao L. seedlings. The roots were divided into two separate containers (Figure S1). The containers had an equal volume (330 mL) of nutrient solution either enriched with the stable isotope 108Cd (+) or not (-). There were three different treatments; (-/-), (+/-), and (+/+). In treatment (+/-), the nutrient solution in the root + compartment was labelled with 108Cd as a tracer. The other compartment (root -) had an identical solution, but no 108Cd tracer. Due to the labelling, the Cd concentrations increased in the (+) treatments by 8%, compared to those in the non-labelled (-) nutrient solution (as detailed in Sect. “Cadmium stable isotope labelling”). In treatment (-/-), both root compartments contained the nutrient solution without a 108Cd tracer. In (+/+) treatment, both compartments were labelled with 108Cd tracer. The non-labelled and fully labelled solutions served as reference treatments. The experiment took fourteen days, in this period, neither the nutrient nor the 108Cd labeling solutions were replaced. At days 0, 1, 7, 11, and 14, four plants per treatment were harvested according to a sampling strategy shown in Table 1. In total, the experiment consisted of n = 40 split-root seedlings.

Table 1 Experimental design and sampling timeline after transfer of seedlings to the experimental solutions (day 0)
Full size table

Plant growth

Three-week-old seedlings (from the genetic group Trinitario) were purchased from a nursery (Cacaobomen.nl, Harmelen, The Netherlands), where they germinated from seed. Upon arrival in the laboratory facilities in Leuven (Belgium), the seedlings were removed from their nursery substrate. Their roots were washed by immersion in deionized water to remove any adhering substrate particles. The seedlings were transplanted into 9-L vessels with continuously aerated nutrient solution for a two-week preculture in hydroponics before the split-root experiment started. The nutrient solution contained: Ca(NO3)2, 2.05 mM; MgSO4, 0.50 mM; KNO3, 1.24 mM; KH2PO4, 90 μM; H3BO3, 15 μM; (NH4)6Mo7O24.4H2O, 0.10 μM; MnSO4.H2O, 4.81 μM; CuSO4.5H2O, 1.42 μM; ZnSO4.7H2O, 5.80 μM; NaCl, 25 μM; MES (2-(N-morpholino)ethanesulfonic acid) 2 mM; NaOH, 1 mM; FeNaEDTA, 10 µM. Cadmium was added as Cd(NO3)2.4H2O together with nitrilotriacetic acid (NTA (50 μM)) (total Cd 19.6 nM). The pH of the nutrient solution was measured at the beginning of the preculture and at the beginning of the experiment. The pH was on average 6.1, and was maintained within 0.1 pH units with the MES pH buffer. Speciation calculations predicted that most Cd was complexed with EDTA (37%) and NTA (61%) ligands (Visual Minteq version 3.1). This complexation ensured that free Cd2+ was buffered at low activity (0.38 nM) and environmentally relevant concentrations. Yet, it should be noted that during the experiment, the pH was not monitored and Cd speciation could have potentially changed.

The culture vessels with the seedlings were installed in a growth chamber with a constant air humidity of 70% and a 12 h/12 h (25 °C/20 °C) day/night cycle with 200 µmol m−2 s−1 light intensity. The vessels were covered with aluminium foil to exclude light to prevent algal growth. After two weeks of preculture, individual plants were transferred from the 9 L containers to the split root device (Figure S1) of twice 330 mL of continuously aerated solution. The lateral roots were gently split and each part was transferred into separate compartments (as described in Saiz-Fernández et al. 2021). During the course of the experiment, the nutrient solution (with the 108Cd tracer) was not replaced and consequently, the volume of the nutrient solution was reduced because of evapotranspiration. Therefore, on day 7, deionized water was added to each compartment to restore the initial volume of 330 mL. The Cd concentration was not monitored during the experiment but the mass balance calculations based on the net Cd uptake in the plant suggest that almost all Cd was taken up during the experiment (details not shown).

Cadmium stable isotope labelling

A stock solution (6.3 mg Cd L−1) with an isotopic abundance of 108Cd of 71.1% was prepared from an enriched 108Cd stable isotope (Trace Sciences International, Delaware, USA) (for details on the preparation, see Argüello et al. (2020); Vanderschueren et al. (2023)). In Table S1, the isotopic abundance of natural Cd and the enriched Cd sources is depicted (Berglund and Wieser 2011). For the preparation of the labelled (enriched in 108Cd) nutrient solution (+), 260 μL of the enriched stock solution was added to 9 L of nutrient solution (19.6 nM Cd) and well mixed. The final (measured) 108Cd isotopic abundance in the (+) nutrient solution was 6.51% (Table S1). In the labelled nutrient solutions, the Cd concentrations increased (21.3 nM) compared to those in the non-labelled (-) nutrient solution (8% difference). Depending on the treatment, the compartments were filled with either 108Cd enriched (+) or natural Cd (-) nutrient solution. The isotope labelling experiment started at day 0 when the roots were immersed in the nutrient solution.

Sample collection

After each harvest (days 0, 1, 7, 11, and 14 (Table 1)), the plant organs were gently separated into the two root parts, crown (which is the interface between the split roots and the stem), stem, leaves, and newly developing leaves if available (Figure S2). Special care was taken to ensure that there was no cross-contamination from one root compartment to the other. The roots were immersed in a CaCl2 solution (2 mM) for 20 min to remove apoplastically bound Cd. All parts of the plant were rinsed in baths with deionized water and dried with paper towels to remove excess water. All plant tissues were oven-dried overnight (60 ◦C) and weighed for mass balance calculations. A subsample of each sample was ground into a powder with a mill, or with a mortar and pestle when the dry mass was not sufficient for the mill. Crowns were not included for Cd and 108Cd measurements.

Chemical analyses

Leaves and stems (400 ± 20 mg) were acid digested with 8 mL HNO3 (NORMATOM® 67–69% w/w, VWR International) in a microwave digestion system (MARS 6, CEM, Matthews, NC, USA). The dry mass of the roots and new leaves was not sufficient for microwave digestion that requires > 200 mg DW. Therefore, roots and new leaves (50 mg) were digested with 2 mL HNO3 in an open block digestion with 10 mL glass tubes.

Cadmium isotopes (108Cd, 111Cd) were measured by Inductively Coupled Plasma Mass Spectrometry (ICP-MS) (Agilent 7700x, Agilent Technologies, collision cell ORS3), with He as the collision gas. 111Cd is commonly the preferred isotope for Cd detection with ICP-MS as it has almost no isobaric interferences (Rehkamper et al. 2012).

The mass bias of the isotopic ratio (IR) 108Cd/111Cd was determined during two different ICP-MS measurement sessions. A Cd single element standard (10 μg Cd L−1, Certipur Merck 119,777) was measured and the measured IR (IRmeasured) was compared with the theoretical IR (IRtheory) to obtain the mass bias correction factor fMB (Nolan et al. 2004).

$${f}_{MB}=\frac{{IR}_{theory}}{{IR}_{measured}}$$

The mass bias factor was 1.11 in the first measurement session and 1.18 in the second session. For each session, the corresponding value was selected as the linear mass bias correction factor. The limits of quantification (LOQs) were determined as ten times the standard deviation of the measured signal in all procedural blanks (n = 22) and were 15 counts per second (cps) for 108Cd and 42 cps for 111Cd (expressed as signal). The signal in the measured plant samples in our study was on average 868 cps for 108Cd and 4019 cps for 111Cd. The IA (isotopic abundance) of 108Cd was determined on the mass-bias corrected isotopic ratio of 108Cd/111Cd and the isotopic ratios of all other Cd isotopes relative to 111Cd;

$${}^{108}Cd\, IA= \frac{\left(\frac{{}^{108}Cd}{{}^{111}Cd}\right)measured}{{\frac{{}^{106}Cd}{{}^{111}Cd}}+ \left(\frac{{}^{108}Cd}{{}^{111}Cd}\right)measured+ \frac{{}^{110}Cd}{{}^{111}Cd} + 1 + \frac{{}^{112}Cd}{{}^{111}Cd} + \frac{{}^{113}Cd}{{}^{111}Cd} + \frac{{}^{114}Cd}{{}^{111}Cd} + \frac{{}^{116}Cd}{{}^{111}Cd}}$$

Only the 108Cd and 111Cd isotopes were measured, thus we used the isotope ratios of 106Cd, 110Cd, 112Cd, 113Cd, 114Cd and 116Cd relative to 111Cd at natural abundance (Berglund and Wieser 2011). This decision was justified because the change in the isotopic ratios of these isotopes with respect to 111Cd was negligible at the maximum IA of 108Cd (6.51%) in this study (Sect. 1.1 in SI and Table S1). The analytical quality of the ICP-MS 108Cd measurements was assessed by measuring a series of 1 μg Cd L−1 solutions with increasing 108Cd IA. These solutions were prepared by mixing a certified reference Cd solution with the 108Cd stock solution used for labelling. The measured IA was compared with the theoretical IA and the data closely followed the 1:1 line, which indicated accurate IA measurements (see SI (Vanderschueren et al. 2023)). The average ± SD of the measured 108Cd IA in the plant material grown without 108Cd isotope in nutrient solution (n = 28) was 0.93 ± 0.19% (theoretical natural abundance 108Cd: 0.89%). The precision (SD) and accuracy (mean) of the natural abundance depended on the sample matrix (roots: 1.00 ± 0.20% (n = 12), leaves: 0.91 ± 0.17 (n = 8), stems: 0.84 ± 0.05% (n = 8)).

Cadmium concentration measurements with ICP-MS typically use one isotope only (here: 111Cd) and are calibrated with calibration standards at natural abundance. The 111Cd was at non-natural abundance in the labelled (+) treatments. The Cd concentrations were not corrected for the change in isotopic abundance of 111Cd in the samples at non-natural abundance. This decision was justified because the difference in isotope abundance of 111Cd between natural abundance samples (111Cd IA = 12.80%) and samples that were maximally enriched with 108Cd (111Cd IA = 12.15%) was negligible for our experiment (Table S1). Plant-certified reference materials were included in each ICP-MS measurement session. The % total Cd recoveries ranged from 86–95% (min–max) with on average 90% for NIST-1573a Tomato Leaves (certified concentration: 1.52 ± 0.07 mg Cd kg−1), n = 8 and from 91–100% with on average 95% for BCR-679 White Cabbage (certified concentration: 1.66 ± 0.07 mg Cd kg−1), n = 2.

Calculations and data analysis

Mass balances combined with isotope abundances allowed us to calculate the fraction of the 108Cd tracer in each plant organ at the different harvest times. The total Cd in each organ i (roots, crown, stem, leaf, new leaf) was calculated by multiplying the Cd concentration measured with the biomass of that organ. The sum of total Cd in each organ i represents the total Cd taken up by the plant. Crown Cd concentrations (not measured) were assumed to be similar to the stem Cd concentrations.

$$Total\;Cd_i\;\left(ng\right)=Cd_i\;\left(mg\;kg^{-1}\;DW\right)\ast biomass_i\;\left(mg\;DW\right)$$

The total Cd in each organ i was multiplied with the 108Cd IA corrected for the natural 108Cd in that organ to obtain the mass of tracer derived Cd in that tissue.

$${}^{108}Cd,{\;tracer}_i\;\left(ng\right)={total\;Cd}_i\;\left(ng\right){\lbrack{}^{108}{Cd}_i\;IA-{}^{108}Cd}_{i,nat}\;IA\rbrack$$

With 108Cdi,nat IA the average natural abundance measured in organ i in the (-/-) treatment. The mass of the tracer 108Cd in each organ i was summed to determine the total 108Cd in the different plants (108Cd, tracerplant (ng)). Subsequently, the 108Cd in each organ i was divided by the total tracer 108Cd in the plant to obtain the fraction of tracer in each plant organ.

$$f_{tracer}={}^{108}Cd,{\;tracer}_i\;\left(ng\right)/{}^{108}Cd,{\;tracer}_{plant}\;\left(ng\right)$$

All statistical analyses were performed with the software JMP-Pro version 17.0. The mean and standard deviation (SD) reported throughout the study were calculated with biological replicates (n = 4), unless otherwise specified. One-way analysis of variance (ANOVA) was conducted in the (+/-) and (+/+) treatments, to determine if there were differences in 108Cd isotopic abundances among organs on the same day or differences. Differences were tested by Student’s t-test or Tukey’s HSD post-hoc test at p < 0.05 significance level. For the root of the non-labelled compartment, leaf, and stem, differences in 108Cd isotopic abundances for the same organ compared to the first day (control) were tested with Dunnett’s post-hoc test. All data were checked to comply with ANOVA assumptions (Table S2). The normal distribution of residuals was tested with the Shapiro–Wilk test and the equality of variances with the O’Brien test. If the equality of variances was not met, the Welch test was used instead of ANOVA.

Results

Plant growth

A one-way ANOVA revealed that total plant biomass was similar among the different treatments (-/-), (+/-), (+/+) at day fourteen (p = 0.07). Thus, labelling of the nutrient solution with 108Cd did not affect total plant biomass (DW) (Figure S3). Hence, at each harvest day, we grouped the biomasses of all treatments (-/-), (-/+), and (+/+) to infer plant growth. The total plant biomass approximately doubled between day 0 (850 ± 160 mg (mean ± SD, n = 4) and day 14 (1630 ± 260 mg, n = 12). At day 0, the largest biomass fraction was in the leaves (50 ± 5% of DW, mean ± SD of all treatments (n = 4)), followed by the stem (37 ± 4%), crown (10 ± 2%), and roots (root mass fraction in one compartment was on average 1.5 ± 1%). At the end of the experiment, the stem was the largest organ (38 ± 1% (n = 12)), followed by the leaves (33 ± 2%), crown (16 ± 1%), roots (13 ± 1%), and newly developing leaves (2 ± 1%). These new leaves started to develop on day 7 of the experiment.

Cadmium concentrations in plant organs

Cadmium concentrations were measured separately in the root tissues from each compartment, leaves, stem and, if available, newly developing leaves (Table 2). Regardless of the treatment, there were no differences in Cd concentrations between the roots of the two compartments, except for day 7 in treatment (+/+), where the root Cd concentration was significantly higher in one root compartment (2.87 ± 1.25 mg kg−1) than in the other root compartment (1.26 ± 0.11 mg kg−1) (p = 0.04) (data not shown). Therefore, the Cd concentration measurements in these two root compartments were grouped for simplicity (Table 2).

Table 2 Cd concentrations (mg kg−1 DW−1) in different organs at each sampling day (t)
Full size table

Before the start of the experiment, the seedlings had already grown for two weeks in a preculture nutrient solution that contained 19.6 nM Cd. Hence, at day 0, there was already an accumulation of Cd in the roots (1.52 ± 0.52 mg kg−1), leaves (0.55 ± 0.10 mg kg−1) and stem (1.07 ± 0.54 mg kg−1). At day 14, the Cd concentrations were similar in all organs (Table 2). At the end of the experiment, the total plant Cd uptake did not differ significantly among the three treatments (1.15 ± 0.54 μg (-/-), 1.59 ± 0.43 μg (+/-), and 1.90 ± 0.07 μg Cd (+/+)) (F(2, 9) = 3.51, p = 0.08). We can thus conclude that the isotope enrichment in the (+) nutrient solution did not affect the uptake of Cd. This result is logical given the fact that the isotope labelling enriched total Cd from 19.6 to 21.3 nM, i.e. an enrichment in total Cd of only 8%. The fraction of total Cd in each organ remained similar over the course of the experiment (Fig. 1) and ranged from 5 to 11% in the roots, from 12 to 20% in the crown (under the assumption that the crown contains the same Cd concentration as the stem), from 26 to 35% in the leaves and from 39 to 46% in the stem. The newly developing leaves contained about 4% of all plant Cd.

Fig. 1
figure 1

Mass fraction of total Cd in different organs (Cdorgan (ng)/Cdplant (ng)*100%) as a function of time (day). Crown Cd concentrations were not measured but were treated as equal to the stem Cd concentrations. The partitioning of Cd in the organs remained similar over the course of the experiment. The error bars represent the standard error of the mean (n = 4 for crown, leaves, and stem and n = 8 for roots)

Full size image

Circulation of 108Cd tracer in the split-root cacao plants

The (+/-) treatment revealed a clear increase in 108Cd IA in the labelled root compartment (+) (4.70 ± 0.93%) at day 1 compared to the natural abundance (0.89%) (Fig. 2, Table S3). In addition, the 108Cd IA did not change in all other organs. Thus, the uptake of the 108Cd tracer was already observed within 24 h and accumulated in this period primarily in the root of the labelled compartment (+) (108Cd fraction, labelled roots, day 1 = 66%) (Fig. 3). At day 7, the 108Cd IA increased in the stem (2.04 ± 0.76%) compared to day 1 (Table S3). Most of the 108Cd tracer was translocated to the shoot organs and accumulated in the stem (43% in (+/-)) (Fig. 3). In contrast, the 108Cd IA in the roots of the non-labelled compartment was near natural abundance which showed that the tracer was not yet transported to the other root compartment. At day 11, the 108Cd IA in the roots of the non-labelled compartment (1.73 ± 0.23%) increased above the natural abundance at day 1 (0.98 ± 0.19%). The newly developing leaves (2.50 ± 0.41%) had an 108Cd IA similar to that of the stem (2.61 ± 0.24%), and leaves (2.22 ± 0.25%). In addition, the 108Cd IA in the labelled root decreased after day 7, likely because of reallocation of natural Cd from e.g., the non-labelled root compartment and the shoot. Finally, on day 14, there were no significant differences in 108Cd IA among stem, leaves, and new leaves as well as in the labelled and non-labelled roots (Table S3). The fraction of 108Cd tracer indicated that most of the 108Cd tracer was accumulating in the stem (around 46%), and the leaves (21%).

Fig. 2
figure 2

The isotopic abundance (108Cd IA) (%) for the (+/-) and (+/+) treatment as a function of time after labelling one root compartment with 108Cd. Error bars represent the standard error of the mean (n = 4 or n = 8 in case of the labelled roots). The results of statistical tests are shown on day 11 (indicated by the blue boxes) (Table S3). Letters a and b denote different statistical groups. In the (+/-) treatment, at day 11, the 108Cd IA in the non-labelled roots (-) differs significantly from 108Cd IA in the non-labelled roots (-) at day 1. This indicates that there is a transport of Cd via the phloem to the non-labelled root part. In the (+/+) treatment, the newly developing leaves have the same 108Cd IA as the leaves and stems and differ from the roots, which indicates that the developing leaves are rather fed from the leaves/stems via reallocation than via direct root uptake

Full size image
Fig. 3
figure 3

The fractions of the 108Cd tracer (%) in each organ relative to the total tracer quantity of 108Cd in the plant. The error bars represent the standard error of the mean (n = 4 or n = 8 in the case of the labelled roots (+) in the (+/+) treatment). The calculation of the fraction 108Cd tracer is explained in the Materials and methods section

Full size image

In the (+/+) treatment, the general trends in the circulation of the 108Cd tracer were similar to those observed in the (+/-) treatment (Fig. 2, Table S3). At day 1, the roots had already absorbed 108Cd (108Cd IA in roots: 3.83 ± 0.83%) and accumulated most of the tracer 108Cd in the plant (Fig. 3). The other organs had a 108Cd IA around natural abundance at day 1. At day 7, the 108Cd IA in the roots (6.06 ± 1.30%) was similar to the 108Cd IA of the nutrient solution (6.51%). The increase of 108Cd IA in the stem (3.73 ± 0.34%) was faster than that in the leaves (1.93 ± 0.56%). Thus, most of the translocated 108Cd tracer was accumulating in the stem (54%, Fig. 3). The new leaves sampled on day 11 had an isotopic abundance intermediate between the stem and leaves (4.08 ± 0.86%) and differed significantly from the 108Cd IA of the roots (p = 0.05). At day 11, the isotopic abundances of the leaves and stem in the (+/+) treatments doubled compared to those in the (+/-) treatments. At day 14, there were no significant differences in 108Cd IA among the different organs. At the end of the experiment, the distribution of the 108Cd tracer was very similar to the distribution of the total Cd (Fig. 1). At day 14, the new leaves contained on average 2.4% of the total 108Cd in the plant, the roots 10%, the crown 20%, the leaves 24%, and the stem 43%.

Discussion

Efficient root-to-shoot translocation and accumulation in the stem

The leaf Cd concentrations in the plant organs were in a similar range as the average leaf Cd concentration observed in a meta-analysis in cacao plantations in Latin America (2.6 mg kg−1 dry weight, n = 762) (Vanderschueren et al. 2021). The concentrations in this study were also below the critical Cd toxicity levels that were proposed for various crops in mature leaves (6–19 mg Cd kg−1 (Krämer 2010) and 5–30 mg Cd kg−1 (Kabata-Pendias 2010). The Cd dose of ca. 20 nM Cd in hydroponic solution can thus be considered as environmentally relevant.

During the experiment, the Cd mass fraction distribution (%) remained constant. Most Cd accumulated in the shoots and only a minor fraction remained in the roots (Fig. 1). The fraction of the freshly added 108Cd tracer was logically larger in the labelled roots at the beginning of the experiment than the fraction in the shoot organs (Fig. 3). At the end of the experiment, most 108Cd tracer had been translocated to the shoots. Cadmium was thus only slightly retained in the roots. An experiment in a similar time frame with wheat and white lupin showed that Cd was only slightly translocated from the roots to the shoots, and primarily retained in the roots (Page and Feller 2005; Riesen and Feller 2005). That translocation behavior denotes an excluder strategy (Sterckeman and Thomine 2020), while we observed in this study that cacao follows more of a hyperaccumulator strategy in terms of the high root-to-shoot translocation (van der Ent et al. 2013).

Cadmium is known to use nutrient pathways. Once Cd is transferred in the xylem sap of root plants, it migrates towards the shoots thanks to the transpiration flow, and it can be deposited along this pathway in xylem cell wall and parenchyma cells (Sterckeman and Thomine 2020). This deposition is more pronounced in cereals like rice, where Cd has been shown to accumulate at the shoot base and in nodes (Fujimaki et al. 2010), than in Cd hyperaccumulators, where Cd is efficiently transferred to shoots via the xylem (Ueno et al. 2008; Hu et al. 2013). In rice, it has been shown that Cd reaches the sink organs after xylem-to-phloem transfer in the nodes (Fujimaki et al. 2010). In the present study, most of the 108Cd tracer accumulated in the stem and crown (Fig. 3). This result suggests that Cd could be sequestered in the stem during its translocation to the transpiring leaves. Alternatively, this result might suggest that part of Cd was reallocated from the leaves to the stem. The sequestration of Cd in the stem on the one hand, could be driven by interactions of Cd with cell walls in the xylem and/or selective removal from the xylem to the phloem by an exchange in the rays/nodes (Fujimaki et al. 2010; Page and Feller 2015). The reallocation of Cd from leaves to stem on the other hand, would require transport by the phloem. The efficient root-to-shoot translocation, and possible sequestration in stems observed in this study correspond well with observations in mature cacao trees in the field where branches were a main organ for Cd accumulation (Engbersen et al. 2019; Blommaert et al. 2022).

Cadmium transport via the phloem to roots and newly developing leaves

The isotopic labelling split-root setup was used to precisely identify short-term Cd circulation fluxes in the (+/-) treatment. In the non-labelled roots, the 108Cd IA increased after day 11 and the 108Cd tracer fractions in the non-labelled roots increased to 6% at the end of the experiment (Fig. 2). Along the same lines, the 108Cd IA in the labelled roots decreased between day 7- 14 after an initial rise at day 1, suggesting transport of natural Cd in the opposite direction. This transport can only occur via the phloem, either by transport through the phloem between root segments or by a downward flux of Cd from the shoot to the roots (Welch et al. 1999). Consequently, this result evidenced the mobility of Cd in the phloem of cacao seedlings. Moreover, the earlier rise in 108Cd IA in the shoot compared to that in the non-labelled roots suggests that Cd was first transported to the shoot and then reallocated to the roots (Figs. 2 and 4). The reallocation of Cd from the shoots to the roots was small (6%) compared to the root-to-shoot translocation (> 75%) and was of a similar extent as observed in wheat and white lupin (Page and Feller 2005; Riesen and Feller 2005; Page et al. 2006).

Fig. 4
figure 4

A model split-root hydroponic system with cacao seedlings. In treatment (+/-), the nutrient solution in the + compartment was labelled with stable isotope 108Cd as a tracer. The non-labelled compartment (-) had an identical solution, with a similar Cd concentration but no 108Cd tracer. The results of our study indicated that the 108Cd absorbed in the roots is readily transported via xylem (blue) to the aerial organs, where it accumulates primarily in the stems. A minor contribution of 108Cd was reallocated to the roots via the phloem (red). Cadmium in the new developing leaves originated from reallocation via the phloem from the stems, crowns and/or leaves rather than from direct root uptake via the xylem

Full size image

Because the mature leaves did not gain much biomass over the course of the experiment (Figure S3), we consider them to be a net phloem source. This means that they are fed mainly by the xylem sap. The fact that the 108Cd tracer was not detected in the mature leaves after one day of treatment, suggests that Cd translocation is slower than the transpiration flow in the xylem. This phenomenon has previously been observed in rice (Fujimaki et al. 2010), and is in line with the pronounced Cd accumulation in the stem as discussed above. On the contrary, in a study of the hyperaccumulator S. nigrum, the root-to-shoot translocation was rapid and large fractions of Cd accumulated in leaves and stems (Wei et al. 2014). Additionally, there was only a poor reallocation from leaves and stems to other organs via the phloem. Thus, in the present study, Cd shows a slower xylem mobility in cacao than the Cd hyperaccumulator S. nigrum.

The newly developing leaves had a 108Cd IA that was similar compared to the stem and leaves, and significantly different from the roots (Fig. 2, Table S3 (+/+) treatment, day 11). These results indicate that the majority of Cd in the new leaves was reallocated from leaves/stems and/or crowns, as opposed to directly transported from roots. This observation is consistent with the classical representation of newly developing leaves as a phloem sink due to their need for nutrients for growth (Page and Feller 2015).

In conclusion, our experiment demonstrated an efficient transfer of Cd from the root to the stem, probably through the xylem, and a reduced xylem mobility in the shoots, probably related to a progressive deposition of Cd in the stem. We moreover evidenced the phloem mobility of Cd towards the roots and young leaves. Thus, in terms of Cd accumulation strategy, we suggest that the circulation of Cd in cacao lies between that of the hyperaccumulator S. Nigrum and agricultural crops such as wheat and white lupin. This is in line with the interpretations from a meta-study with mature cacao trees, that compared the uptake potential and partitioning of cacao with other plants, and that suggested to term cacao a moderate Cd accumulator (Vanderschueren et al. 2021).

Conclusions and perspectives

In conclusion, this study evidenced that Cd was mobile in the phloem of cacao seedlings that were grown in environmentally relevant Cd conditions. Through the phloem, Cd was translocated from shoot tissues to the roots and newly developing leaves. The dominant transport pathway of Cd was nonetheless the transpiration pathway (xylem) from roots to leaves, with a large Cd accumulation in the stems (Fig. 4). Limitations of this study are that short-term labelling results strongly depend on the developmental stage of the plant at the time of labelling, and that the seedlings do not reach a reproductive stage for the development of beans. Yet, this simplified system helped to corroborate the Cd uptake, distribution and redistribution patterns that have previously been observed in mature trees. We confirmed that there is a high Cd translocation to the above-ground tissues and that the stems (or branches in mature trees) are an important organ for the accumulation of Cd. In addition to the more direct transport pathway of Cd from roots and stems to the beans (a sink organ) as proposed by previous studies (Engbersen et al. 2019; Blommaert et al. 2022), this study shows that there is also Cd reallocation via the phloem from the stems, crowns and/or mature leaves to sink organs such as new leaves and roots.