JEQ Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 27 October 2006
Published in J Environ Qual 35:2055-2065 (2006)
DOI: 10.2134/jeq2005.0461
© 2006 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Ingwersen, J.
Right arrow Articles by Streck, T.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Ingwersen, J.
Right arrow Articles by Streck, T.
Agricola
Right arrow Articles by Ingwersen, J.
Right arrow Articles by Streck, T.
Related Collections
Right arrow Plant and Soil Interactions
Right arrow Remediation
Right arrow Sorption/Exchange
Right arrow Heavy Metals
Right arrow Soil Pollution

TECHNICAL REPORTS

Heavy Metals in the Environment

Cadmium Leaching from Micro-Lysimeters Planted with the Hyperaccumulator Thlaspi caerulescens

Experimental Findings and Modeling

Joachim Ingwersena,*, Barbara Bücherla, Günter Neumannb and Thilo Strecka

a University of Hohenheim (310), Institute of Soil Science and Land Evaluation, Biogeophysics Section, 70593 Stuttgart, Germany
b University of Hohenheim (330), Institute of Plant Nutrition, 70593 Stuttgart, Germany

* Corresponding author (jingwer{at}uni-hohenheim.de)

Received for publication December 16, 2005.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
The use of heavy metal hyperaccumulating plants has the potential to become a promising new technique to remediate contaminated sites. We investigated the role of metal mobilization in the Cd hyperaccumulation of Thlaspi caerulescens (J. & C. Presl, ‘Ganges’). In a micro-lysimeter experiment we investigated the dynamics of Cd concentration of leachate as well as Cd removal by plant uptake in four treatments: (i) Control (bare soil), (ii) T. caerulescens, (iii) nonhyperaccumulator Brassica juncea (L.) Czern. (‘PI 426308’), and (iv) co-cropping of the hyperaccumulator and nonhyperaccumulator. The experimental findings were analyzed using one- and two-site rate-limited desorption models. Co-cropping of T. caerulescens and B. juncea did not enhance metal uptake by B. juncea. Although Cd uptake of T. caerulescens was 10 times higher than that of B. juncea, the Cd concentration of leachate of the T. caerulescens treatment did not decrease below that of the B. juncea treatment. The Cd depletion in leachate was well reproduced by the two-site rate-limited desorption model. The optimized desorption coefficient was three orders of magnitude higher in the rhizosphere than in the bulk soil. Our results indicate that T. caerulescens accelerates the resupply of Cd from soil pointing to an important role of kinetic desorption in the hyperaccumulation by T. caerulescens.

Abbreviations: CAL, calcium–acetate–lactate • EDTA, ethylenediaminetetraacetic acid • EGTA, ethylene-glycol-bis(2-aminoethyl ether)–N,N,N'N'–tetraacetic acid • GUI, graphical user interface


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
THE DEMAND for remediation techniques to clean-up heavy metal contaminated soils is increasing continuously. In Europe, the area of moderately contaminated soils, that is, soils in which the content of at least one heavy metal exceeds the regulatory limit for arable land, is in the range of millions of hectares (Felix, 1997). No low-cost technique is currently available to clean up moderately contaminated soils and ensure that soils retain their fertility after metal removal. Conventional remediation techniques, for example excavation or washing with chelators, are very costly and adversely affect the gas and water regime as well as the biological activity of treated soils.

In recent years, several alternative techniques have been proposed to clean up such sites. A widely tested approach is natural phytoextraction, which uses hyperaccumulating plants to remove metals from soil. The hyperaccumulator Thlaspi caerulescens is capable of taking up high amounts of the heavy metals Cd and Zn. Despite extraordinarily high metal contents in shoots, the total removal rate remains generally low because hyperaccumulators produce low biomass. Consequently, time necessary to remediate a contaminated site is in the range of decades to centuries. Brown et al. (1994), for example, estimated that at least 28 yr of T. caerulescens cultivation would be necessary to remove all the Zn from a soil containing 2100 mg kg–1. Felix (1997) estimated a much longer period of time to remediate a site contaminated with Cd (6.6 mg kg–1). Based on results of field trials the author estimated that it would take between 89 and 186 yr to decrease the soil Cd content to the Swiss limit value for agricultural land (0.8 mg kg–1).

The mechanisms by which hyperaccumulators mobilize and take up heavy metals are still not completely understood. It is well known that plants can modify the rhizosphere to enhance the acquisition of nutrients, for example, by release of acids. This is particularly important for ions that are transported to the root surface mainly by diffusion (e.g., Fe, Zn, and P) and under nutrient-limiting conditions (Marschner, 1995). Several studies, however, have not shown evidence that natural hyperaccumulators use rhizosphere acidification to enhance metal uptake (Bernal et al., 1994; Knight et al., 1997; McGrath et al., 1997). The role of root exudates in metal hyperaccumulation by T. caerulescens is little studied and the results so far appear to be contradictory. Recently, several authors reported little or no evidence for a strong metal mobilizing ability of T. caerulescens roots (Whiting et al., 2001; Zhao et al., 2001). In contrast, Whiting et al. (1997) found, in a situation with a low soil Zn content (55 mg kg–1), that T. caerulescens increased Zn uptake of the nonhyperaccumulator T. arvensis when the plants were co-cropped. At present, the two mechanisms that are thought to be most likely involved in the hyperaccumulation by T. caerulescens are a higher Zn and Cd transporter density in root cells (Lombi et al., 2001; Pence et al., 2000) and an active proliferation of root branches in Zn/Cd-rich patches (Haines, 2002; Schwartz et al., 1999; Whiting et al., 2000).

Several models have been developed during the last decades to simulate the plant uptake of metals. The model most widely used is the Barber-Cushman model (Barber and Cushman, 1981). This model simulates the radial flow of metals from soil to plant roots taking into account diffusive and convective flow. In combination with active solute uptake described by Michaelis-Menten kinetics nutrient-depleted zones can be modeled around the root. The capacity of a soil to compensate plant removal of metals from the liquid phase is characterized by the soil buffer power; this is defined as the first derivative of the total concentration with respect to the liquid phase concentration. Typically, the soil buffer power is considered constant during a simulation. This assumption is equivalent to a linear equilibrium sorption.

Although such models are powerful tools to study and to interpret lab and field experiments, they have not yet been widely used in the research on hyperaccumulation. Sterckeman et al. (2004) have applied the Barber–Cushman model to simulate Cd uptake by T. caerulescens. Their results show that the soil buffer power and hence the sorption strength of soil is a very sensitive parameter in modeling the Cd uptake by T. caerulescens. Moreover, their data suggest that the soil buffer power depends on the extent to which the plant has depleted the soil. Seuntjens et al. (2004) presented a root-zone model that includes the effect of organic ligands on metal uptake and leaching. The model was used to simulate the effect of oxalate exudation by roots on Cu transport and bioavailability. Recently, Puschenreiter et al. (2005) published a study on root-induced changes of Ni biogeochemistry in the rhizosphere of T. goesingense. Their study is a good example for how mathematical models and experiments can be combined to get a better insight into the complex processes involved in hyperaccumulation. In an "experiment-model" loop they showed that T. goesingense takes up free Ni2+ but excludes metal-organic complexes.

Studies by McGrath et al. (1997) and Whiting et al. (2001) revealed that between 67 and 90% of the total mass of Zn accumulated by T. caerulescens were mobilized from "less soluble fractions in the soils." Desorption, that means the transfer of metals from the sorbed phase into the solution phase, might play an important role in this context. It is well known that Cd sorption–desorption might be kinetically limited (Selim et al., 1992; Hinz and Selim, 1994; Filius et al., 1998). In such a case the kinetics of metal sorption–desorption is fundamental in assessing the mobility and bioavailability of metals in soils (Sparks, 2000). The role of kinetic desorption in hyperaccumulation by T. caerulescens, however, remains unstudied. The present study helps close this gap. We conducted a micro-lysimeter experiment to monitor the depletion of the soil solution in the presence of hyperaccumlator and/or nonhyperaccumulator plants. We investigated the dynamics of Cd concentration of leachate as well as Cd removal by plant uptake in four treatments: (i) Control (bare soil), (ii) T. caerulescens, (iii) nonhyperaccumulator Brassica juncea, and (iv) co-cropping of the hyperaccumulator and nonhyperaccumulator. The latter treatment addressed the question whether rhizosphere effects of T. caerulescens can enhance Cd bioavailability in general. The experimental findings were evaluated using one- and two-site rate-limited desorption models. In parallel, in a rhizobox culture system root morphological characteristics of T. caerulescens were nondestructively evaluated and root exudates were collected and analyzed for the concentration of carboxylates.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Soil Data and Measurements
Soil was collected from the plowing horizon of a Dystric Cambisol (Food and Agriculture Organization, 1990) or Typic Haplumbrept (Soil Survey Staff, 1994). The sampling site was located within a storage basin (District 1) of the waste water irrigation area of Braunschweig (Germany). A detailed description of the study area is given in Ingwersen (2001). The soil has received considerable loads of heavy metals by irrigation and disposal of municipal waste water for up to 40 yr. The total Cd content as determined by aqua regia extraction is 4.7 mg kg–1. This value exceeds the limit value of the German sewage sludge regulation, which is 1 mg kg–1 for sandy soils. The sampled soil material was air-dried and passed through a 2-mm sieve. Subsequently, 40 kg of the air-dried soil was filled in a cement mixer and 4 g NH4NO3, 4 g MgSO4, and 4 g agricultural lime (calcium carbonate) dissolved in 3 L deionized water were added. Afterward, the soil was carefully homogenized. For equilibration the soil was stored for 2 d in covered plastic buckets. After the pre-treatment soil pH (CaCl2) was 5.3. Further important soil characteristics are: Organic C content 1.5% by mass, calcium–acetate–lactate (CAL)-extractable P (as P2O5) 48 mg kg–1, CAL-extractable K (as K2O) 18 mg kg–1. The soil texture was silty sand.

We assume that extraction of soil with a strong complexing agent such as EGTA represents the amount of Cd that participates in sorption–desorption reactions (Filius et al., 1991; Gäbler et al., 1999). The EGTA-extractable metal contents of soil were determined in duplicates as follows: 2 g soil were put into acid-rinsed 50-mL polypropylene centrifuge tubes and received 40 mL 0.025 M Na2–EGTA. The samples were horizontally shaken for 90 min at 140 rpm. Subsequently, the samples were centrifuged for 5 min at 1900 x g. The supernatant solution was transferred through a filter paper into acid-rinsed 15-mL polystyrene tubes for Cd analysis.

Moreover, we will assume that the Cd concentration of the leachate is a good approximation of the concentration of soil solution. This assumption does not hold under every experimental condition. The concentration of leachate is a flux-averaged concentration and the soil solution concentration is a resident concentration. Whether both concentrations are similar depends on the apparent dispersivity. In a theoretical study Parker and van Genuchten (1984) showed that the flux-averaged and the resident concentrations are very similar if the apparent dispersivity is smaller than 0.1 m (Parker and van Genuchten, 1984, Fig. 2). For the present experiment we estimated an apparent dispersivity ranging from 0.012 to 0.042 m. Hence, under the conditions of our experiment the concentration of leachate is a good approximation of the concentration of soil solution. In this calculation we used dispersion lengths (1–3 cm) that are in the upper range of observed values for columns of sandy soil (Boesten, 1986). The soil liquid diffusion coefficient of Cd was estimated to be 0.048 cm2 d–1 (Olsen and Kemper, 1968). The pore water velocity ranged between 10.1 mm d–1 (Control) and 7.8 mm d–1 (B. juncea monoculture) (Table 3).


Figure 2
View larger version (13K):
[in this window]
[in a new window]
  		Fig. 2. Schematic diagram of the two-site rate-limited desorption model.

 

View this table:
[in this window]
[in a new window]
  		Table 3. Transpiration, leaching, and dry matter yield in the four treatments. Plants were grown in micro-lysimeters filled with a Cd-contaminated sandy soil. The duration of the experiment was 91 d. When the plants were harvested, the average volumetric soil water content was 23%. Values are means ± SD.

 
Plant Culture and Sampling
Thlaspi caerulescens from southern France (Ganges) and B. juncea (PI 426308) were cultivated in micro-lysimeters (diam. = 12 cm, height = 12 cm), which enabled us to sample both plant material and leachate. We studied four different treatments: (i) Control (bulk soil), (ii) T. caerulescens monoculture, (iii) B. juncea monoculture, and (iv) co-cropping of T. caerulescens and B. juncea. Each treatment was performed in three replicates.

We filled the lysimeters with the prepared soil material (1.5 kg dry weight) and adjusted the bulk density to 1.3 kg L–1. Within the experimental period (13 wk), each lysimeter was irrigated in total with 312 mm of electrolyte solutions. Each day micro-lysimeters were manually irrigated with water volumes ranging from 20 to 100 mL. The irrigation schedule is shown in Fig. 1. From the first to the 35th day, we used a 0.0025 M CaCl2 solution (pH 6.5) for irrigation. Because of a low biomass production, the composition of the irrigation water was changed after the 35th day to a 0.0025 M nutrient solution (pH 7.0) containing KNO3, MgSO4, and Ca(HPO4) in a ratio of 2:2:1. At the bottom of each lysimeter, we installed a porous glass frit (max. pore diam. = 40 µm) to which we applied a constant suction of 30 hPa using a compressor combined with a pressure regulator. The leachate of each lysimeter was sampled in a 500 mL glass flask. To avoid metal sorption to the glass walls, we added 5 mL of HNO3 (13% by volume) to each flask at the beginning of each sampling step. The leachate was collected every 2 to 7 d. Based on the water balance, we calculated the daily transpiration for each lysimeter. Evaporation was estimated by the water loss in the control treatments.


Figure 1
View larger version (13K):
[in this window]
[in a new window]
  		Fig. 1. Irrigation scheme of the micro-lysimeter experiment. The diameter of micro-lysimeters was 12 cm.

 
The preparatory culture of T. caerulescens was grown for 4 wk in a climate chamber under controlled light and temperature conditions. After this time span, six plants were transferred to each lysimeter of the T. caerulescens monoculture and co-cropping treatments and were grown in a greenhouse using a 10-h photoperiod at 21°C. The short photoperiod of 10 h was chosen to avoid flowering of B. juncea. After four further weeks, three B. juncea plants, which had been grown 1 wk before in a climate chamber, were transferred to the lysimeters. After 10 d, the plants were thinned to two plants per lysimeter. After 13 wk, we harvested shoots and roots of all plants and all treatments. Roots were carefully separated from the soil and were intensively washed with distilled water. In the co-cropping treatment, roots of B. juncea and T. caerulescens could not be sampled separately. Samples of ground, dried plant material (0.5 g for shoots and 0.125 g for roots) were microwave-digested with 5 mL nitric acid and 4 mL hydrogen peroxide. Metal concentrations of plants, soil extracts, and leachates were analyzed using a graphite furnace AAS with Zeeman background compensation (Varian SpectrAA 800).

Root Observations and Collection of Root Exudates
For nondestructive root observations, minirhizotron "rhizoboxes" (25 by 10 by 1 cm) with transparent root observation windows were employed (Dinkelaker and Marschner, 1992). After an initial fertilization with 110 mg N kg–1 soil as Ca(NO3)2, 80 mg K kg–1 soil as KCl, 50 mg P kg–1 soil as Ca(H2PO4)2, and 50 mg Mg kg–1 soil as MgSO4, the rhizoboxes were filled each with 300 g soil. Germination and pre-cultivation of the plants were performed for 6 wk in a mixture of peat culture substrate (Euflor GmbH, München, Germany) and sand 70%/30% (v/v) adjusted to a moisture content of 50% (w/w). After full development of the primary leaves, the culture substrate was carefully washed from the root systems and three bundles of four to six seedlings were transplanted into each rhizobox in five replicates. Soil moisture was adjusted to 15% (w/w) and controlled gravimetrically in intervals of 2 to 3 d. The rhizoboxes were adjusted in an angle of 45° to direct root development along the root observation window. Plant cultivation was conducted under controlled conditions in a growth chamber with a 16/8 h, 23/20°C day/night cycle at a light intensity of 150 µmol m–2 s–1 and a relative humidity of 60%.

Root observations and measurements were performed with a "Zeiss Axiovision" video- microscope system (Zeiss, Oberkochen, Germany). Antimony-micro-electrodes were employed for the measurement of rhizosphere pH as described by Häussling et al. (1985). After a culture period of 4 mo the root system was well developed and pH was measured in four to seven replicates in each rhizobox. At the same point in time we collected rhizosphere soil solution using filter paper discs (5 mm diam., Schleicher & Schuell 2992, Dassel, Germany). The paper discs were put onto the root surface as desribed by Wasaki et al. (2005). The filters were re-extracted with distilled water and analyzed for LMW-organic acids by reversed-phase HPLC in the ion-supression mode on a 250 by 4 mm C18 Grom-Sil 120 ODS-5 ST column (Grom, Rottenburg, Germany) by isocratic elution with 18 mM KH2PO4 pH 2.3 at 40°C with a flow rate of 0.9 mL min–1 and direct UV detection at 215 nm (Neumann et al., 1999).

Statistics
The statistical analysis of the data was performed with SPSS Version 12.0.1 (SPSS, 2003). Treatment effects were tested by a one-way analysis of variance (ANOVA) followed by Fisher's LSD post hoc test for significant differences between treatment means. Lysimeter averages were used as replicate measurements. The LSD test was performed as a two-tailed test using a 5% significance level.

Model Description
Two different modeling approaches (Table 1 and Table 2) were used to simulate Cd depletion and replenishment in soil solution, which were monitored during the experiment by measuring Cd concentrations of leachates. Both models assume that in soil metals are located either in the soil solution or are sorbed to solids. At equilibrium, partitioning between both phases is controlled by a sorption isotherm. Metal removal by plant uptake and leaching disturbs the equilibrium and induces a transfer of metals from the sorbed phase to the solution phase (desorption). Because desorption is kinetically limited the new equilibrium state between both phases is not reached instantaneously and the solution phase concentration drops below the equilibrium concentration.


View this table:
[in this window]
[in a new window]
  		Table 1. Definition of parameters used in models.

 

View this table:
[in this window]
[in a new window]
  		Table 2. Differential equation systems of the one-site and two-site rate-limited desorption model. Parameters are given in Table 1.

 
In both models, the sorption equilibrium between the solution and sorbed phase is described by an extended Freundlich equation (e.g., Elzinga et al., 1999)

Formula 1[1]
All parameters and functions are presented in Table 1. The use of an extended Freundlich isotherm was necessary because soil pH changed during the experiment. Soil pH was linearly interpolated between pH values measured at the start and the end of the experiment.

Kinetic desorption was modeled using a compartment approach. Compartment models have been successfully used in a large number of case studies on kinetic sorption–desorption of heavy metals (e.g., Amacher et al., 1988; Selim et al., 1992; Buchter et al., 1996). The velocity of desorption was assumed to be proportional to the concentration difference between the surface of the sorbent and the solution phase. The desorption coefficient {alpha} (d–1) controls the overall velocity of the reaction (Table 2). Metal uptake by plants was reconstructed from measured data assuming that the uptake is related to transpiration. This assumption is based on the idea that under our experimental conditions (constant temperature, air humidity etc.) transpiration should be a good relative measure of the root length density. We approximated metal uptake as follows

Formula 2[2]
where {phi}total is the total plant removal (kg) of Cd. The symbol Ttotal (m3) denotes the total amount of transpiration and T(t) (m3 s–1) stands for the transpiration rate at time t (s).

Models were formulated as sets of ordinary differential equation (ODE) systems which were solved with the ODE solver Berkeley Madonna (BM) Version 8.0.1 (Macey et al., 2000).

One-Site Rate-Limited Desorption
The one-site rate-limited desorption model (Table 2, Eq. [1]–[2]) considers two stocks of metals: dissolved and sorbed metal. The model does not distinguish between rhizosphere and nonrhizosphere soil. Heavy metals are removed from the solution phase by plant uptake and leaching. The stock is replenished by rate-limited desorption of metals from the sorbed phase.

Two-Site Rate-Limited Desorption
In the two-site rate-limited desorption model (Table 2, Eq. [3]GoGo–[6]), the sorbed phase of soil is divided in a rhizosphere and a nonrhizosphere domain. The central assumption is that the desorption coefficient {alpha} of the reaction sites is different in both domains. Concentrations in the rhizosphere and nonrhizosphere domain are labeled with subscript r and subscript n, respectively. Figure 2 shows a schematic diagram of the model. Cadmium is removed from the solution phase by plant uptake and leaching and is replenished by rate-limited desorption of metals from the sorbed phase of both domains. We further assume that during the experiments the rhizosphere expands in a logistic manner (Table 2, Eq. [6]). Rhizosphere expansion implies the conversion of nonrhizosphere sites to rhizosphere sites.

Parameter Optimization
Because the automatic parameter-fitting routine implemented in BM Version 8.0.1 does not provide standard errors or correlation of parameters we used the program UCODE Version 3.061 (Poeter and Hill, 1999) for parameter optimization. UCODE performs inverse simulations by computing parameter values that minimize a weighted least-square objective function using the Levenberg–Marquardt algorithm. For optimization, data were weighted with the reciprocal value of the measurement variance. Sensitivities were calculated using forward differencing and convergence was reached if parameter values between iterations changed <1%.

UCODE requires that the applied model code (here BM) uses text (ASCII) input files, produces text output files, and can be executed in batch mode. Berkeley Madonna fulfills the second and the third requirement but not the first one. Model input parameters can be changed by the user only via the Windows graphical user interface (GUI). The automatic manipulation of model input parameters via the GUI was performed using the freeware scripting language AutoIt Version 3.0.102 (Bennett, 2004).


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Experimental Findings
The initial EGTA-extractable Cd of the investigated soil was 3.33 mg kg–1. During the experiment, soil pH decreased from 5.3 to 4.9 on average. When the plants were harvested, the average volumetric soil water content was 23%. Table 3 gives the transpiration, evaporation, leaching, and dry matter yields of the treatments. The high dry matter yield in the co-cropping treatment is due to the higher plant density there. During the period of high irrigation (the first 3 wk, see Fig. 5b) the evaporation was on average 25.2 ± 2.3 mL d–1. Afterward evaporation dropped down and was on average 15.6 ± 2.1 mL d–1.


Figure 5
View larger version (20K):
[in this window]
[in a new window]
  		Fig. 5. (a) Simulated and measured Cd concentration of the leachate of the control treatment (bare soil). Simulation was performed with a one-site rate-limited desorption model. (b) Time course of the volume flux of leachate and irrigation. The error bars indicate the standard deviations.

 
Co-cropping of B. juncea and T. caerulescens did not significantly increase Cd shoot contents of B. juncea. Cadmium contents of B. juncea shoots were 16.9 ± 1.4 and 15.4 ± 1.6 mg kg–1 for the monoculture and co-cropping treatment, respectively. For the T. caerulescens monoculture and co-cropping treatments, Cd removal ranged from 345.7 to 370.9 µg per lysimeter (Table 4). This value was about 10 times higher than that of the B. juncea monoculture. About 12% of the total Cd found in the plant material of T. caerulescens was recovered in the root biomass.


View this table:
[in this window]
[in a new window]
  		Table 4. Metal pools in plant organs, Cd output via leaching, and total Cd output for the four treatments after a growth period of 13 wk. Plants were grown in micro-lysimeters filled with a heavy-metal contaminated sandy soil. Numbers in brackets give the percentage of initial EGTA-extractable Cd (3.33 mg kg–1). The total amount of Cd in the soil of a micro-lysimeter was 4995 µg. Numbers with the same index are not significantly different ({alpha} = 0.05).

 
Cadmium output via leaching ranged from 1.7 to 2.3% of the initial EGTA-extractable metal content (Table 4). In the micro-lysimeters with plants, leaching of Cd was markedly reduced compared to the control treatment. In the B. juncea monoculture treatment, for example, Cd leaching decreased by 24% compared to the control.

Before transpiration started, the Cd concentrations of the leachates were very similar in all treatments (Fig. 3). This situation changed after transpiration started. With incipient transpiration we observed that the Cd concentrations of the leachates from the treatments with plants distinctly decreased below the concentrations of the control treatment (Fig. 3). Although Cd uptake of T. caerulescens was 10 times higher than that by B. juncea, the Cd concentration of the leachate of the T. caerulescens monoculture treatment did not drop below concentrations of the nonhyperaccumulator treatment. A similar result we observed for the co-cropping treatment. Despite the 10 times higher Cd removal, Cd concentration of leachates were at a similar level as that of the B. juncea monoculture treatment.


Figure 3
View larger version (16K):
[in this window]
[in a new window]
  		Fig. 3. (a) Cadmium concentrations of leachates of the control, the T. caerulesens and B. juncea monocultures, and the co-cropping treatment. (b) The bottom graph shows the transpiration of treatments.

 
In rhizoboxes three types of roots could be differentiated at the surface of the root observation windows after a culture period of 4 mo. Light-brown colored first-order lateral roots (diam.: 0.30 ± 0.08 mm) and white second-order laterals (diam.: 0.15 ± 0.02 mm) were spreading over the entire depth of the rhizoboxes (Fig. 4A). Shorter fine roots (diam.: 0.10 ± 0.02 mm) emerged in dense groups mainly from second-order laterals (Fig. 4B), and formed large numbers of fine, long root hairs (diam. <0.01 mm, length 1–2 mm; Fig. 4A, 4B). Fine roots and root hairs were covered with mucigel (Fig. 4C). The mucigel was obviously involved in binding and enclosing soil particles and thereby enhancing the root–soil contact (Fig. 4D). The bulk soil pH was 4.8 ± 0.02. In the rhizosphere of fine roots the pH increased to 5.2 ± 0.06.


Figure 4
View larger version (155K):
[in this window]
[in a new window]
  		Fig. 4. Root-morphological characteristics of Thlaspi caerulescens grown in rhizo-box culture on a heavy-metal contaminated sandy soil from a waste water irrigation area. (A) Dense formation of long, fine root hairs (length 1–2 mm, diam. <10 µm) along fine roots; (B) Local proliferation of fine roots, densely covered with root hairs; (C) Mucigel, covering root hairs and fine roots; (D) Binding of soil particles along fine roots by root hairs and mucigel.

 
A comparative analysis of soil solutions collected from the rhizosphere of fine lateral roots covered with root hairs (Fig. 4A, 4B) and from the bulk soil did not indicate a significant exudation of the metal-chelating carboxylates citrate, isocitrate, malate, and malonate. Only the concentration of lactate, which ranged between 50 and 100 µM, was distinctly higher in the rhizosphere soil solution than in the bulk soil solution. The lactate concentration in the bulk soil solution was in the range of the detection limit (<20 µM).

Modeling
Figure 5 shows the measured and simulated Cd concentrations of the leachate of the control treatment. The changes in the observed concentrations are clearly related to the irrigation height. High leachate fluxes coincide with decreasing Cd concentrations and vice versa. This finding strongly indicates that in our experiment the desorption of Cd from the sorbed phase was kinetically limited. The one-site rate-limited desorption model was well suited to reproduce the measured concentrations in the control (Radj2 = 0.98). To obtain a satisfactory simulation result, however, we had to run the model with two different desorption coefficients- one for the time in which micro-lysimeters were irrigated with CaCl2 solution and a second for the time in which we irrigated with nutrient solution. Without this approach, the model systematically overestimated the depletion during the first phase and systematically underestimated the depletion during the second phase. The systematic deviation between measured and simulated concentration between Day 35 and Day 60 is due to the change in the irrigation solution. For the sake of simplicity, the model assumes that the CaCl2 solution is completely displaced by the nutrient solution after Day 35. In the real system complete displacement probably required several days. All parameters used in the model are given in Table 5.


View this table:
[in this window]
[in a new window]
  		Table 5. Model parameters optimized for the micro-lysimeter experiment.

 
Applying the one-site model with the optimized parameters of a nonrhizosphere soil to the T. caerulescens treatment did not yield satisfactory results (Fig. 6a). After Day 42, that is after measurable transpiration started, the simulated Cd concentration decreased to zero within several days. The replenishment of the solution Cd through desorption was not fast enough to compensate for the removal by root uptake. The two-site rate-limited desorption model reproduced the dynamics of the Cd concentration after the beginning of transpiration and Cd uptake much better (Radj2 = 0.99) (Fig. 6b). The optimized desorption rate coefficient of the rhizosphere domain is about three orders of magnitude higher than that of the nonrhizosphere domain (Table 5). Because values of the initial fraction of the rhizosphere close to zero caused numerical instability its value was set to 0.01 (Table 2). During the simulation the rhizosphere fraction increased to 0.11. The sensitivity analysis shows that the average Cd concentration in leachate is highly sensitive to the parameters controlling the dynamics of root growth; the growth rate of the rhizosphere r and the initial fraction of rhizosphere (Fig. 7). The sensitivity of the desorption coefficients is moderate.


Figure 6
View larger version (19K):
[in this window]
[in a new window]
  		Fig. 6. Simulated and measured Cd concentration of the leachate of the T. caerulescens monoculture treatment. (a) one-site rate-limited desorption model, and (b) two-site rate-limited desorption model. The error bars indicate the standard deviations.

 

Figure 7
View larger version (18K):
[in this window]
[in a new window]
  		Fig. 7. Sensitivity analysis of the average Cd concentration of the leachate. Simulations were performed for the period from Day 0 to 91 using the two-site rate-limited desorption model. Each parameter was varied between 0.5 and 2 times the optimized value while holding the remaining parameters constant (Table 5). Simulated average Cd concentration of leachate at a relative change of 1 was 30.6 µg L–1.

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
CONCLUSIONS
 REFERENCES
 
Although the Cd uptake of T. caerulescens was 10 times higher than that of B. juncea, the Cd concentrations of the leachate never dropped below the values of the B. juncea monoculture treatment. This experimental finding suggests that T. caerulescens can accelerate the replenishment of the depleted soil solution. The experimental data were well reproduced by the two-site rate-limited desorption model. The desorption coefficient for the rhizosphere domain was three orders of magnitude higher than that for the nonrhizosphere. Although the fit of the model to the data cannot be construed as proof for the correctness of the model assumptions, it is possible to propose mechanisms that are consistent with the model and experimental observations. The hypothesis that in our experiment T. caerulescens accelerated the kinetically-limited desorption is compatible with both experimental and modeling results.

In general, it is well known that the sorption–desorption of heavy metals from soil might be kinetically limited (Amacher et al., 1988; Selim et al., 1992; Hinz and Selim, 1994; Filius et al., 1998). Kinetic sorption–desorption, however, is not found in every soil. Selim et al. (1992), for example, studied the kinetic of sorption of two soils differing in organic matter and Fe2O3 content. In the soil with high organic matter and Fe2O3 contents they observed a pronounced kinetically-limited sorption. Retention of Cd by the soil was rapid during the initial stages of reaction and was then followed by slow and continued Cd retention. Even after 10 d the equilibrium between the sorbed and the solution phase was not reached. In the other soil, which was lower in organic C and Fe oxide contents, however, they observed no significant kinetic behavior of Cd sorption.

Sterckeman et al. (2004) modeled the Cd uptake of Zea mays L. and T. caerulescens using the Barber–Cushman model. To get satisfactory simulation results, the authors had to apply different soil buffer powers for the Zea mays and T. caerulescens treatments, although the same soil was used in both treatments. The buffer power of the T. caerulescens treatments used in the simulation was up to seven times higher than in the Zea mays treatments, supporting our results. The consequence of a higher buffer power is that the soil has a higher capacity to compensate for metal removal by plant uptake.

The active uptake by a hyperaccumulator can only be efficient if the replenishment of the liquid phase is fast enough to compensate for the removal by active root uptake. The co-cropping experiment, however, provided evidence that in our experiment the observed metal-mobilizing mechanism of T. caerulescens did not generally increase Cd bioavailability. In the co-cropping treatment, Cd was not accumulated more in the B. juncea shoots than in the monoculture treatment. The mechanism was only beneficial for T. caerulescens itself. Hence, the mechanism was species specific. The formation of long (1–2 mm) and extremely fine (diam. <10 µm) root hairs (Fig. 4A, 4B), comparable with the dimensions of extra-radical endomycorrhizal hyphae (Polomski and Kuhn, 2002), may enable the exploitation of soil pores by roots T. caerulescens, which are not accessible to the roots of B. juncea. A similar exclusive effect on soil exploitation by T. caerulescens may be expected from the observed formation of rhizoheaths, that is, layers of soil particles which are embedded in mucigel around the fine roots and root hairs of T. caerulescens (Fig. 4C, 4D).

An active uptake mechanism, which is well documented for T. caerulescens (Lombi et al., 2001; Pence et al., 2000), as well as an aqueous complexation of Cd by root exudates lower the free Cd2+ concentration. A lower free Cd2+ concentration accelerates desorption. In our study we did not detect elevated concentrations of metal chelating carboxylates in the rhizosphere but of course it cannot be ruled out that other compounds than the ones analyzed are involved in aqueous complexation. In the literature the findings with regard to the question whether T. caerulescens releases root exudates that can enhance metal uptake are controversial. In a soil with low soil Zn content (55 mg kg–1), Whiting et al. (1997) found that T. caerulescens increased Zn uptake of the nonhyperaccumulator T. arvensis when both species shared the rhizosphere. In contrast, Zhao et al. (2001) reported that root exudates of T. caerulescens mobilized little Cu and Zn from Cu- or Zn-loaded resins, and little Zn, Cd, Cu, or Fe from a contaminated calcareous soil. They concluded that root exudates are not involved in Zn and Cd hyperaccumulation. However, Zhao et al. (2001) investigated only the metal-mobilizing effects of water-soluble root exudates obtained by root washing from T. caerulescens grown in hydroponic culture. More hydrophobic compounds such as certain phenols and also mucilage and mucigel are not easily caught by this technique. It has been reported already by Nambiar (1976) that stimulation of mucilage secretion in response to increased soil-mechanical impedance in dry soils can contribute to Zn2+ uptake of oat (Avena sativa L.) by facilitating Zn2+ transport from embedded soil particles to the root surface. Therefore, it remains to be established, whether the intense mucigel formation, around fine roots and root hairs of T. caerulescens (Fig. 4C, 4D), exerts similar stimulating effects on uptake of heavy metals.

In the proposed macroscopic model an active proliferation of root branches in Zn/Cd-rich domains (scavenging) (Haines, 2002; Schwartz et al., 1999; Whiting et al., 2000) would also yield a higher desorption coefficient. This process, however, depends also on the kinetics of desorption. At the beginning of the T. caerulescens monoculture treatment, the Cd concentration of the soil solution was 48.5 µg L–1. With a total soil water volume of 0.31 L, the Cd pool in the solution phase was about 15.0 µg per micro-lysimeter. The total uptake by T. caerulescens averaged 370.9 µg, that is, about 96% of the Cd taken up by T. caerulescens must have been supplied from the sorbed phase. Only 4% could have been directly scavenged by active proliferating. Similar to aqueous complexation, this mechanism can only decrease the metal concentration in the soil solution. Thus, the supply would again be limited by the achievable maximum desorption rate.

It is obvious that root length density as well as root geometry affect the speed by which a depleted soil solution is replenished. More intensive root penetration includes more soil surfaces in desorption. The presented two-site rate-limited desorption model does not implicitly take into account the root characteristics. Effects of the root system would be accounted for by the (lumped) desorption coefficient. A study by Sterckeman et al. (2004) does not indicate that T. caerulescens possesses an extraordinarily intensive soil-penetrating root system. The authors studied the Cd uptake of Zea mays and T. caerulescens. To parameterize the Barber–Cushman model, they measured the root length as well as the mean root radius of both plants. After a cultivation time of 79 d the root length of T. caerulescens was comparable to the root length of Zea mays cultivated for 12 d. The root growth rate of T. caerulescens was by a factor of four smaller than that of Zea mays. The mean root radii of the two plants were similar (T. caerulescens: 0.012 cm; Zea mays: 0.015 cm). To determine the root length, however, Sterckeman et al. (2004) separated roots from soil by washing. Fine roots and root hairs as those observed in our rhizobox experiment (Fig. 4A, 4B) are easily lost by washing. Therefore, values given for T. caerulescens by Sterckeman et al. (2004) may have largely underestimated the true root length. Similar characteristics of fine roots and root hairs as observed in the present study have recently been reported also for the metal hyperaccumulator Thlaspi goesingense (Himmelbauer et al., 2005).

Similar to results of previous studies (Bernal et al., 1994; Knight et al., 1997; McGrath et al., 1997) we found no evidence in our experiment that root-induced rhizosphere acidification is involved in metal mobilization by T. caerulescens. We observed that rhizosphere pH was even higher than the pH of the bulk soil. The reason for the finding is probably that N was supplied in the form of NO3– (Neumann and Römheld, 2002). The general decline of soil pH in micro-lysimeters from 5.3 to 4.9 during the experiment was most likely due to leaching of base cations such as Ca and Mg.

It may be possible that rhizodeposition can directly modify the sorption properties of soil. However, only little evidence points in that direction. Piccolo et al. (1996) found that some organic acids—in contrast with mineral acids, phenols, alcohols, or dipolar aprotic solvents—can alter the stereochemical hydrophobic arrangement of humic material, thereby shifting the nominal molecular size from >100 kilodalton (KD) to <25 KD. The lower the molecular size the more surfaces are available for sorption–desorption reactions. This microscopic mechanism would be reflected in a macroscopic rate-limited desorption model by a higher desorption coefficient. Note, however, that the study of Piccolo et al. (1996) was restricted to pH values, which were lower than usually found in soils.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
REFERENCES
 
Our experimental data show that T. caerulescens is able to accelerate the resupply of Cd in soil. Moreover, they suggest that besides other mechanisms such as active metal uptake or scavenging, desorption kinetics plays an important role in the hyperaccumulation of T. caerulescens. Whether the observed intense mucigel formation around fine roots and root hairs of T. caerulescens is involved in this process remains to be established. The dynamics of the observed Cd concentrations of leachates can be well reproduced using a two-site rate-limited desorption model. In this model the sorbed phase of the soil is separated into a rhizosphere domain and nonrhizosphere domain. The fraction of both domain is time-dependent and each domain has different kinetic desorption properties. In the rhizosphere domain the optimized desorption coefficient was three orders of magnitude higher than in the nonrhizosphere domain. The study shows that experimental modeling is a useful tool to get a better understanding of the mechanisms involved in the hyperaccumulation of T. caerulescens.


   ACKNOWLEDGMENTS
 
We thank Professor Römheld and Mrs. Aiyen for their support and many fruitful discussions. We also thank Rita Hierl for laboratory assistance. Special thanks to Clara Hoffmann for taking care of the rhizoboxes and our tender Thlaspi plants. The authors thank two anonymous reviewers for critical comments and suggestions, which helped to improve the manuscript. This work was financially supported by the Geschwister-Stauder foundation.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 





This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Ingwersen, J.
Right arrow Articles by Streck, T.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Ingwersen, J.
Right arrow Articles by Streck, T.
Agricola
Right arrow Articles by Ingwersen, J.
Right arrow Articles by Streck, T.
Related Collections
Right arrow Plant and Soil Interactions
Right arrow Remediation
Right arrow Sorption/Exchange
Right arrow Heavy Metals
Right arrow Soil Pollution

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS