Published November, 2001

Phytoremediation of Heavy Metal–Contaminated Soils: Natural Hyperaccumulation

versus Chemically Enhanced Phytoextraction
E. Lombi, F. J. Zhao, S. J. Dunham, and S. P. McGrath*

ABSTRACT mulating capacity. These plants have several beneficial

A pot experiment was conducted to compare two strategies of characteristics such as the ability to accumulate metals
phytoremediation: natural phytoextraction using the Zn and Cd hyper- in their shoots and an exceptionally high tolerance to
accumulator Thlaspi caerulescens J. Presl & C. Presl versus chemically heavy metals (Baker et al., 2000). On the other hand,
enhanced phytoextraction using maize (Zea mays L.) treated with many hyperaccumulator plants tend to be slow-growing
ethylenediaminetetraacetic acid (EDTA). The study used an industri- and produce low biomass, with the exception of some
ally contaminated soil and an agricultural soil contaminated with Ni hyperaccumulator species. With the plant materials
metals from sewage sludge. Three crops of T. caerulescens grown
currently available, years or decades are needed to clean
over 391 d removed more than 8 mg kg⫺1 Cd and 200 mg kg⫺1 Zn
up a contaminated site. For instance, McGrath et al.
from the industrially contaminated soil, representing 43 and 7% of
the two metals in the soil. In contrast, the high concentration of Cu
(1993), using field data, calculated that nine croppings
in the agricultural soil severely reduced the growth of T. caerulescens, of T. caerulescens would be required to decrease Zn
thus limiting its phytoextraction potential. The EDTA treatment concentration in the soil from 440 to 300 mg Zn kg⫺1.
greatly increased the solubility of heavy metals in both soils, but this Similarly, Brown et al. (1994) estimated that 28 yr of
did not result in a large increase in metal concentrations in the maize T. caerulescens cultivation would be necessary to re-
shoots. Phytoextraction of Cd and Zn by maize ⫹ EDTA was much move all the Zn from a soil containing 2100 mg Zn kg⫺1.
smaller than that by T. caerulescens from the industrially contami- Another problem with the continuous phytoextraction
nated soil, and was either smaller (Cd) or similar (Zn) from the of metals from soils is related to the fact that some
agricultural soil. After EDTA treatment, soluble heavy metals in soil metals such as Pb are largely immobile in soil and their
pore water occurred mainly as metal–EDTA complexes, which were
extraction rate is limited by solubility and diffusion to
persistent for several weeks. High concentrations of heavy metals in
soil pore water after EDTA treatment could pose an environmental root surface.
risk in the form of ground water contamination. Chemically enhanced phytoextraction has been de-
veloped to overcome these problems (Huang and Cun-
ningham, 1996; Blaylock et al., 1997; Huang et al., 1997;
Blaylock, 2000). This approach makes use of high-bio-
A large number of sites worldwide are contaminated
by heavy metals as a result of human activities.
Traditional solutions such as disposal of contaminated
mass crops that are induced to take up large amounts
of metals when their mobility in soil is enhanced by
soil in landfills account for a large proportion of the re- chemical treatments. Several chelating agents, such as
mediation operations at present. However, some of the citric acid, EDTA, CDTA, DTPA, EGTA, EDDHA,
remediation techniques currently in use will probably lose and NTA, have been studied for their ability to mobi-
economic favor and public acceptance in the near future. lize metals and increase metal accumulation in different
Therefore, new technologies based on environmentally plant species (Huang et al., 1997; Cooper et al., 1999).
friendly and low-cost processes are urgently required. Different metals have been targeted, such as Pb (Blay-
Phytoremediation of heavy metal–contaminated soil lock et al., 1997; Huang et al., 1997), U (Huang et al.,
is an emerging technology that aims to extract or inacti- 1998), 137Cs (Lasat et al., 1998), and Au (Anderson et
vate metals in soils (McGrath, 1998; Salt et al., 1998). al., 1998). However, at the moment, the most promising
It has attracted attention in recent years for the low cost application of this technology is for the remediation of
of implementation and environmental benefits. More- Pb-contaminated soils using Indian mustard [Brassica
over, the technology is likely to be more acceptable to juncea (L.) Czern.] in combination with EDTA (e.g.,
the public than other traditional methods. It has been Blaylock, 2000). Despite the success of this technology,
estimated that the market for phytoextraction of metals some concerns have been expressed regarding the en-
from soils in the USA alone was approximately $1–2 mil- hanced mobility of metals in soil and their potential
lion in 1997, with a potential to increase to $15–25 mil- risk of leaching to ground water (Cooper et al., 1999).
lion by 2000 and $70–100 million by 2005 (Glass, 2000). However, no detailed studies regarding the persistence
Two approaches have been proposed for phytoextrac- of metal–EDTA complexes in contaminated soils have
tion of heavy metals, namely continuous or natural phy- been conducted.
toextraction and chemically enhanced phytoextraction Soils contaminated with multiple heavy metals can
(Salt et al., 1998). The first is based on the use of natural present a difficult challenge for both approaches to phy-
hyperaccumulator plants with exceptional metal-accu- toextraction. Although some hyperaccumulators appear
to be capable of accumulating elevated concentrations
of several heavy metals simultaneously, there is still con-
Agriculture and Environment Division, IACR-Rothamsted, Harpen- siderable specificity in metal hyperaccumulation (Baker
den, Hertfordshire AL5 2JQ, UK. Received 1 Aug. 2000. *Corre-
sponding author (steve.mcgrath@bbsrc.ac.uk).
Abbreviations: EDTA, ethylenediaminetetraacetic acid; ICP–AES,
Published in J. Environ. Qual. 30:1919–1926 (2001). inductively coupled plasma atomic emission spectroscopy.

1919
1920 J. ENVIRON. QUAL., VOL. 30, NOVEMBER–DECEMBER 2001

et al., 2000). Also, Pb and Cu hyperaccumulators have deionized water and dried at 80⬚C for 16 h, and their dry
been noted, but largely unproven (Baker et al., 2000). weight was recorded. Dried samples were ground to ⬍0.5 mm
For chemically enhanced phytoextraction, establish- before analysis.
ment of a high-biomass crop is required before chelate Nylon-coated soil moisture samplers (Rhizosphere Research
Products, Wageningen, the Netherlands) were placed inside the
application. This may be difficult to achieve if the soil soil in each pot to collect soil pore water, following the method
is heavily contaminated with metals such as Zn, Cd, and described by Knight et al. (1998). In the pots with T. caeru-
Cu, which are usually much more bioavailable, and thus lescens, soil pore water was collected at the beginning of the
more phytotoxic, than Pb. experiment and immediately before each harvest. In the pots
In the present study, we compared natural phytoex- with maize, soil pore water samples were collected at the
traction using the well-known Zn and Cd hyperaccumu- beginning of the experiment, before each EDTA application
lator T. caerulescens, and chemically enhanced phytoex- and at five different times after the application up to 4 wk.
traction using maize and EDTA treatment. Two soils
contaminated with multiple heavy metals were used. Chemical and Statistical Analyses
The primary aim was to evaluate which approach is Total concentrations of heavy metals in the soils were deter-
likely to be more successful to remediate soils contami- mined using inductively coupled plasma atomic emission spec-
nated with multiple heavy metals. Furthermore, we in- troscopy (ICP–AES; Fisons Accuris, Ecublens, Switzerland),
vestigated the potential environmental risk associated following aqua regia digestion (McGrath and Cunliffe, 1985).
with metal mobilization by EDTA. Soil pH was measured in the 1:2.5 suspension of soil and water.
Total C and N were determined using a LECO (St. Joseph,
MI) CNS-2000. The sand, silt, and clay contents, and soil cation
MATERIALS AND METHODS exchange capacity (CEC) were determined using standard
methods (Avery and Bascomb, 1982). Soils before and after
Pot Experiment
phytoextraction were determined for extractable metals using
Two contaminated soils (0–20 cm) were collected in France 1 M NH4NO3 (Deutsch Institut für Normung, 1995).
and the UK. The French soil was contaminated due to the Soil pore water samples were analyzed for total soluble
activity of a nearby Zn smelter. The UK soil was an agricultural metal concentrations using ICP–AES. In addition, we quanti-
soil contaminated by sewage sludge applications in the 1960s. fied metal–EDTA complexes by ion chromatography (Dionex
Soils were air-dried and sieved to ⬍5 mm. Because our aim [Sunnyvale, CA] DX500), using a hydrophilic anion column
was to evaluate the potential of chemically enhanced phytoex- (Dionex IonPac AS5) and a guard column (Dionex IonPac
traction and natural hyperaccumulators to clean up contami- AG5). The eluant consisted of 2.24 mM Na2CO3 and 1.76 mM
nated soils, we did not include an uncontaminated soil as a NaHCO3, which was pumped through the columns isocrati-
control. We chose T. caerulescens (the Ganges ecotype from cally at a flow rate of 1 mL min⫺1 (Sun et al., 2001).
southern France) and maize (cv. Aviso) for the natural hyper- Subsamples of plant materials (0.2 g) were digested with a
accumulation and high biomass strategies, respectively. Thlaspi mixture of HNO3 and HClO4. Metal concentrations of the
caerulescens is a well-known Zn hyperaccumulator, and in a digests were determined using ICP–AES.
previous study we have shown that the Ganges ecotype is very Statistical analysis was performed using Genstat 5 (Genstat
efficient in accumulating Cd, as well as Zn (Lombi et al., 5 Committee, 1993). Means and standard errors are presented
2000). Indian mustard, commonly used for phytoextraction of for all data. Comparisons between croppings were made using
Pb in combination with EDTA (Blaylock et al., 1997), was t tests.
not used in this experiment. This is because a preliminary test
showed that Indian mustard suffered severe phytotoxicity in
both soils. Day and night temperatures in the glasshouse were RESULTS
16 and 12⬚C, respectively, and natural sunlight was supple- Soil Physicochemical Properties
mented with 1 KW SON-T lamps (Philips Lighting UK, Croy-
don, Surrey, UK) to maintain a minimum light intensity of The two soils had different physicochemical proper-
250 ␮mol m⫺2 s⫺1. ties and patterns of pollution (Table 1). The UK soil
We grew each plant species for three consecutive growth was mainly contaminated with Zn, Cu, Cd, and Ni,
cycles over a 391-d period, allowing about 1 mo between whereas the French soil was contaminated with Zn, Cd,
croppings. Both were resown each time. One plant of maize, and Pb in comparison with the 1986 EU Directive (Com-
or four plants of T. caerulescens, grew in each pot containing mission of the European Communites, 1986). The ag-
1 kg soil (oven-dry basis). There were four replicates for each ricultural soil from the UK had higher concentrations
treatment. At the beginning of each cropping cycle, 240, 120, of organic carbon and total N than the French soil,
and 150 mg kg⫺1 of N, P, and K, respectively, were added to
reflecting the past use of sewage sludge in the former.
the pots as fertilizer. Deionized water was used throughout.
In each cycle of maize, when it reached the early flowering
The soil pH was around 6, and differed slightly between
stage, EDTA (as disodium salt) was applied to the soil at a the two soils.
rate of 1 g per kg (2.7 mmol kg⫺1 of soil). Maize was harvested
4 wk after the EDTA treatment. To assess the effect of EDTA Plant Growth and Metal Concentrations
application on the metal uptake by maize, we also included
a control of maize without EDTA, only at the same time as
Thlaspi caerulescens plants grew healthily in the French
the first cropping ⫹ EDTA. The control was replicated four soil, but its growth was impaired in the UK soil (Table
times for each soil. All pots were arranged randomly. 2). The plants in the UK soil showed clear phytotoxic
At the end of each growth cycle (approximately 4 mo), the symptoms. The shoot biomass of T. caerulescens was 2
shoots of T. caerulescens and the shoots and roots of maize to 10 times higher in the French soil than in the UK
were harvested. Plant materials were washed thoroughly with soil. Also, shoot biomass increased with successive crop-
 LOMBI ET AL.: PHYTOREMEDIATION OF HEAVY METAL–CONTAMINATED SOILS 1921

Table 1. Selected physicochemical properties of soils used.

(6 222)
(3903)
7 814 (301)
9 187 (652)
9 101 (652)

(107)
(119)
(142)

(107)
(209)
(758)
Parameter UK soil French soil

Zn

1 050
1 213
1 832
7 529

2 150
4 929
6 277
19 426
Texture clay loam loam
Clay (⬍2 mm), % 35.9 19.9
pH 6.3 5.8
Total C, g kg⫺1 59.2 34.7

(256.4)
Total N, g kg⫺1 6.5 1.4

45.9 (2.09)

259.5 (16.2)
577.9 (87.9)
4 561 (1 804)
29.7 (3.5)

(1.4)
(0.1)
45.4 (6.2)

(8.0)

14.8 (2.6)
CEC†, cmolc kg⫺1 39.1 17.6
Total Cd, mg kg⫺1 42 19

Pb
Total Cu, mg kg⫺1

11.1
3.2

468.8
34.1
1245 78
Total Ni, mg kg⫺1

mg kg⫺1
155 16
Total Pb, mg kg⫺1 230 842
Total Zn, mg kg⫺1 1756 2920

French soil

(290.3)
(80.0)

(36.9)
† Cation exchange capacity.

12.3 (0.9)
23.9 (0.8)

(0.6)

(2.1)
9.1 (0.3)

(0.6)
(0.5)

(6.2)
Cu

12.3

41.3
5.8
6.7

145.7

126.4
245.6
764.4
pings, particularly in the third one, in the French soil.
In contrast, the biomass decreased with cropping in the

98.2 (15.3)

(19.7)

(11.7)
576.0 (54.8)

(36.9)
UK soil.

(3.8)

(2.3)
(3.1)
115.6 (4.0)

(0.9)
(0.6)
The growth of maize appeared to be normal in both

Cd

13.9
40.2

13.8
60.0
72.3
3.8
4.9

121.7
soils prior to the EDTA application in the first cycle.
The shoot biomass of maize grown on the two soils was
similar when no EDTA was applied (control; Table 2).

Table 2. Biomass and metal concentrations of T. caerulescens and maize. Standard errors are reported in parentheses.

0.44 (0.21)

0.08 (0.01)
0.6 (0.17)
Biomass

6.2 (0.2)
6.7 (0.1)

5.0 (0.8)

2.4 (0.4)
2.2 (0.3)
12.7 (0.7)

11.9 (2.3)
12.5 (1.4)
g pot⫺1
Compared with the control, the application of EDTA
at the flowering stage did not significantly affect the
shoot and root biomass of maize in the first cropping
on either soil. However, the biomass of maize at the

124.2 (25.3)
185.1 (21.1)
348.3 (20.0)

502.1 (95.1)
second and third harvest was significantly lower than

736.2 (202)
331 (15.5)
1 950 (197)
1 868 (166)
2 599 (778)

1 314 (160)
that in the first on the French soil (P ⬍ 0.05 and P ⬍

1 218 (95)
0.01, respectively). For this soil, the shoot biomass at Zn
the third harvest was only 3% of that in the first. There
was no significant effect of the EDTA treatments on
the shoot biomass of maize grown on the UK soil.

(11.3)
3.5 (0.7)
4.4 (0.5)

(0.7)
(0.5)
(0.5)
(0.1)
11.4 (7.1)

(3.5)
(3.4)
(1.5)
The concentrations of Cd, Cu, Ni, Pb, and Zn in the
Pb

0.9
1.9
2.2
0.5

20.1
20.9
13.1
30.0
shoots of T. caerulescens varied considerably across the
three harvests (Table 2). On average, the concentrations
of Zn and Cd in the shoots of T. caerulescens were 2139 (15.0)
135.6 (31.9)

(17.1)
(70.0)
(39.3)
53.5 (2.2)
71.8 (5.0)

(1.5)

(5.8)
(0.5)
(0.2)

and 116 mg kg⫺1 in the UK soil, and 8700 and 263 mg

mg kg⫺1
Ni

kg⫺1 in the French soil, respectively. The concentrations

35.6
12.6

60.8
3.9
2.1

250.3
213.3
180.3
of Cd obtained in most harvests on both soils were
UK soil

greater than the value used to define Cd hyperaccumula-

tion (100 mg kg⫺1; Baker et al., 2000), whereas the con-
680.4 (197.5)
92.2 (15.7)

(49.8)

273.3 (31.5)
563.7 (32.1)
38.7 (2.0)
27.5 (1.3)

(4.2)
(1.9)
(0.1)

806 (187)

centrations of Zn obtained on the French soil were close

Cu

to the threshold value for hyperaccumulation (10 000 mg

24.0
9.1
8.8
116.3

kg⫺1 ). The concentrations of both metals did not de-

crease with cropping (Table 2). Thlaspi caerulescens
grown on the French soil had higher concentrations of
151.1 (92.99)

0.72 (0.01)
0.81 (0.05)
59.5 (11.6)
138.1 (39.9)

4.13 (0.6)

(2.2)
(3.6)
(6.7)
(7.4)
1.4 (0.2)

Pb than plants on the UK soil. In contrast, the concentra-

Cd

tions of Cu were considerably higher on the UK soil

16.5
33.7
25.9
32.4

than those on the French soil, reflecting a much higher

level of Cu contamination in the former. For Cu, Pb,
and Ni, the concentrations in the shoots of T. caerules-
0.41 (0.2)
Biomass

2.9 (1.0)
1.7 (0.1)
1.3 (0.7)

(4.5)

3.2 (0.1)
3.7 (0.4)

1.3 (0.2)
(0.7)
(1.2)

(4.6)
g pot⫺1

cens were far below the values used to define hyperaccu-

mulation (1000 mg kg⫺1 ).
8.5
13.4
15.7

14.3

In maize, the concentrations of all heavy metals deter-

mined were much higher in the roots than in the shoots
⫹EDTA 3rd cropping

⫹EDTA 3rd cropping

⫹EDTA 2nd cropping

⫹EDTA 2nd cropping

⫹EDTA 1st cropping

⫹EDTA 1st cropping

Control (no EDTA)

Control (no EDTA)

(Table 2). In the first cycle, application of EDTA to

both soils increased the concentrations of Cd, Cu, and
Zn in the roots by one- to threefold (P ⬍ 0.05). In the
T. caerulescens

2nd cropping
3rd cropping
1st cropping

Maize shoots

Maize roots

French soil, which had a high concentration of Pb (Table

1), application of EDTA increased the concentration of
Pb in the roots by 16-fold (P ⬍ 0.001). In contrast,
application of EDTA had no significant effect on the
1922 J. ENVIRON. QUAL., VOL. 30, NOVEMBER–DECEMBER 2001

Phytoextraction of Heavy Metals

The amounts of Cd and Zn removed by maize (shoots)
and T. caerulescens are shown in Fig. 1. The amounts
of other metals extracted by both plants were small,
and are not presented. In the French and UK soils,
T. caerulescens extracted, respectively, 63 and 8 times
more Cd than maize (P ⬍ 0.001). In particular, T. cae-
rulescens removed a total of more than 8 mg Cd per kg
of soil in the three croppings in the French soil. Thlaspi
caerulescens also extracted approximately 4 times more
Zn (197 mg kg⫺1 soil) than the EDTA-treated maize
(25 mg kg⫺1 soil) from the French soil. The removal of
Zn by the two plant species from the UK soil was similar.

Changes in the Solubility of Heavy Metals in Soils

The concentrations of Cd, Ni, and Zn in the pore
waters from the UK soil, and of Cd from the French soil
appeared to increase with croppings of T. caerulescens
(Table 3). These increases were relatively small, and
may be due to the release of metals from the roots of
T. caerulescens, which were not recovered from the soils.
In comparison, the treatment with maize ⫹ EDTA
had a dramatic effect on the mobility of heavy metals
(Fig. 2). In both soils, the applications of EDTA mark-
edly increased the concentrations of metals in the soil
pore water within the first 24 h. For example, the con-
centration of Zn in the soil pore water of the UK soil
increased from 2.4 to 104 mg L⫺1 after the first applica-
tion of EDTA. Similarly, the Pb concentration in the
soil solution from the French soil increased from 0.1 to
36 mg L⫺1 within a 24-h period. Over time, the concen-
trations of the different metals in the soil pore water
followed a similar pattern. EDTA mobilized metals rap-
idly, and subsequently their concentrations in solution
Fig. 1. Removal of Cd (A ) and Zn (B ) from the French and UK soils decreased slowly.
by three croppings of T. caerulescens and maize (the latter in In the maize ⫹ EDTA treatment, there was a good
combination with EDTA). Data reported are means of four rep-
licates. agreement between the concentrations of metals in soil
pore water determined by ICP–AES and the concentra-
tions of metal–EDTA complexes determined by ion
chromatography (Fig. 3). The results indicate that most
concentrations of metals in the shoots, except for Zn in of the heavy metals in the soil pore waters were com-
the UK soil (50% increase, P ⬍ 0.01) and Pb in the plexed by EDTA.
French soil (2.5-fold increase, P ⬍ 0.05). Compared with The extractable fractions of heavy metals in soils were
the first crop cycle, the metal concentrations in both assessed using 1 M NH4NO3 (Deutsch Institut fur Nor-
roots and shoots increased successively in the two fol- mung, 1995). In the UK soil, extractability of the metals
lowing croppings. In the French soil, a much-reduced at the beginning of the experiment and after growing
biomass in the third cropping may partly explain the T. caerulescens was not significantly different. But in
large increases in the metal concentrations in maize. the French soil, where the metals were much more ex-

Table 3. Initial and preharvest metal concentrations in soil pore water of the T. caerulescens treatment. Standard errors are reported
in parentheses.
UK soil French soil
Cd Cu Ni Zn Cd Pb Zn
␮g L⫺1 mg L⫺1 ␮g L⫺1 mg L⫺1
Beginning of the experiment 15.3 (0.4) 0.9 (0.14) 0.19 (0.01) 1.5 (0.3) 32 (2.1) 120 (27) 6.2 (1.4)
After 1st cropping 20.7 (5.5) 0.6 (0.07) 0.29 (0.07) 2.2 (0.2) 52 (17) 150 (21) 10.4 (3.2)
After 2nd cropping 35.7 (6.1) 1.17 (0.37) 0.60 (0.07) 5.3 (1.1) 45 (11) 82 (32) 36.3 (9.6)
After 3rd cropping 52.0 (10.2) 0.95 (0.12) 1.13 (0.11) 6.9 (1.1) 78 (14) 25 (12) 11.2 (1.3)
 LOMBI ET AL.: PHYTOREMEDIATION OF HEAVY METAL–CONTAMINATED SOILS 1923

Fig. 2. Changes over time in metal concentrations in soil pore water extracted from the maize treatment. Arrows indicate EDTA applications,
bars represent standard errors.

tractable at the beginning of the experiment, there was the extractability of Cd, Cu, and Ni increased signifi-
a small but significant decrease in the extractable Cd cantly (P ⬍ 0.05 for Cd and Ni, P ⬍ 0.01 for Cu) in
(P ⬍ 0.001) and Zn (P ⬍ 0.05) as a result of both maize the maize ⫹ EDTA treatment, whereas Cd and Zn
and T. caerulescens cultivation (Table 4). In the UK soil decreased in the French soil (Table 4).

Fig. 3. Relationship between the concentrations of total soluble heavy metals, determined by inductively coupled plasma (ICP), and metal–EDTA
complexes determined by ion chromatography (IC).
1924 J. ENVIRON. QUAL., VOL. 30, NOVEMBER–DECEMBER 2001

Table 4. Metal availability determined using 1 M NH4NO3 extraction, at the beginning and end of the experiment. Standard errors are
reported in parentheses.
UK soil French soil
Cd Cu Ni Zn Cd Pb Zn
mg kg⫺1
Beginning of the experiment 0.35 (0.003) 1.93 (0.02) 3.63 (0.02) 38.7 (3.01) 5.2 (0.03) 10.40 (0.19) 451 (13.2)
After T. caerulescens 0.45 (0.02) 2.67 (0.15) 5.32 (0.33) 40.6 (3.9) 1.90 (0.17) 9.25 (1.94) 387 (8.1)
After maize 0.65 (0.02) 67.87 (10.2) 24.0 (3.8) 39.9 (3.9) 2.35 (0.02) 13.75 (1.4) 370 (5.0)

DISCUSSION metal concentrations in this species. Similarly, Bennett

The pattern of contaminants in the two soils used et al. (1998) showed that optimizing N fertilization of
significantly influenced the biomass of the Zn and Cd T. caerulescens increased yield without decreasing the
hyperaccumulator T. caerulescens. The UK soil was Zn concentration in the plant. Assuming a constant re-
heavily contaminated with Cu, containing 1245 mg kg⫺1 moval rate, approximately six croppings of T. caerules-
total Cu. The reduced growth of T. caerulescens on the cens would be required to decrease the total Cd in the
UK soil was most likely due to the phytotoxic effect of French soil from 19 to 3 mg kg⫺1. This very encouraging
Cu. Thlaspi caerulescens has been shown to tolerate result was obtained with an ecotype of T. caerulescens
exceedingly high levels of Zn and Cd (Brown et al., that is particularly efficient in hyperaccumulating Cd
1995; Shen et al., 1997; Lombi et al., 2000). In fact, this (Lombi et al., 2000). In the case of Zn, 40 croppings
plant species thrives on soils heavily contaminated with would be required to reduce its total concentration from
Zn and Cd, such as the French soil used in the experi- 2920 to 300 mg kg⫺1. This estimate is very similar to
ment. Apart from Zn and Cd, T. caerulescens also has that of Brown et al. (1994).
an elevated tolerance to other metals such as Pb and Brassica species are often used in chemically enhanced
Ni (Baker et al., 1994). However, it is sensitive to Cu phytoextraction (Blaylock, 2000), although other plant
(McLaughlin and Henderson, 1999). The concentrations species such as maize and pea (Pisum sativum L.) have
of Cu in the shoots of T. caerulescens grown on the UK also been used (Huang and Cunningham, 1996; Huang
soil were in the range of 28 to 92 mg kg⫺1. The threshold et al., 1997). However, Indian mustard suffered from
concentration of Cu toxicity for many crop species is severe phytotoxicity when grown on both soils used in
about 30 mg kg⫺1 (Marschner, 1995). It thus appears this study, even before EDTA was applied. Elevated
that T. caerulescens is no more tolerant to Cu than tolerance to heavy metals in soils is a prerequisite for
normal crop species. Therefore, heavy contamination a successful phytoextraction. Unlike Pb, which is usually
of Cu can seriously limit the potential for phytoextrac- low in bioavailability in soil, metals like Zn, Cd, and
tion of Zn and Cd by T. caerulescens. Phytotoxicity of Cu tend to have much higher bioavailability, and thus
Cu was probably the reason why T. caerulescens grew are more phytotoxic. This means that phytoextraction
very poorly in a soil containing 11 700 mg kg⫺1 Zn and using nontolerant cultivars of Brassica species is unlikely
3420 mg kg⫺1 Cu, in a study by Ebbs et al. (1997). This to succeed in soils contaminated with large concentra-
led Ebbs et al. (1997) to conclude that the phytoextrac- tions of Zn, Cd, or Cu.
tion efficiency for Zn was lower in T. caerulescens than Monocotyledon species are usually more tolerant to
in several Brassica species. It is possible that Cu toxicity metals than dicotyledon species (Marschner, 1995). Maize
not only inhibits plant growth, but also the uptake of was able to grow on both soils, showing only limited
Zn and Cd by T. caerulescens. Such inhibition has been signs of phytotoxicity. However, the growth of maize
observed in barley (Hordeum vulgare L.) (Beckett and on the French soil was severely stunted in the second
Davis, 1978) and Brassica species (Ebbs and Kochian, and third cropping, following the EDTA treatments in
1997), and would explain why the concentrations of Zn the first and second cropping. It is likely that Zn toxicity
and Cd in T. caerulescens grown on the UK soil in this was responsible for this inhibition of growth. The con-
study and on the soil used by Ebbs et al. (1997) were centrations of soluble Zn in the soil pore waters col-
far below the hyperaccumulation potential of this plant. lected from the French soil at the beginning of the sec-
The French soil had high concentrations of Zn, Cd, ond and third cropping were ⬎100 mg L⫺1, compared
and Pb, but low Cu, and the biomass of T. caerulescens with ⬍50 mg L⫺1 before the EDTA treatment (Fig. 2).
was comparable with that of maize. Three successive In contrast, the EDTA treatments did not decrease the
croppings of T. caerulescens removed approximately 8 biomass of maize in the second and third cropping in
and 200 mg kg⫺1 of Cd and Zn, respectively, represent- the UK soil, probably because this soil had a lower
ing 43 and 6.8% of the total Cd and Zn contents in the concentration of total Zn and a lower concentration of
French soil. It is interesting to note that the phytoextrac- soluble Zn in the pore water.
tion efficiency for Zn and Cd did not decrease with Although EDTA increased the concentrations of sol-
cropping. In the third cropping, which produced the uble metals dramatically (Fig. 2), metal uptake by maize
largest biomass, the concentrations of Cd and Zn in the was minimal (Table 2). Also, the EDTA treatment in-
shoots were also the highest or among the highest. This creased the metal concentrations in the roots far more
suggests that optimization of the growing conditions than in the shoots, suggesting that EDTA was far more
may result in an increased yield without decreasing efficient in overcoming the diffusion limitation of metals
 LOMBI ET AL.: PHYTOREMEDIATION OF HEAVY METAL–CONTAMINATED SOILS 1925

to root surface than the barrier of root to shoot translo- favoring biodegradation. This may be because metal–
cation. The results are consistent with those of Ebbs EDTA complexes with high stability constants (i.e., che-
and Kochian (1998), who showed that EDTA increased lates of Cu, Fe, Pb, and Zn) are degraded more slowly
the concentration of Zn in the shoots of Indian mustard (Satroutdinov et al., 2000) than complexes with low
by less than twofold, and did not enhance Zn uptake stability constants (i.e., chelates of Ca, Mg, and Mn). In
by oat (Avena sativa L.) and barley. Similarly, Kayser the contaminated soils used in our study, virtually all
et al. (2000) showed that NTA and elemental S increased the EDTA added formed complexes with metals of high
the solubility of Cd, Cu, and Zn in the soil by a factor stability constants, and this may contribute to the slow
of 58, 9, and 21 respectively, but accumulation of these degradation and the persistent mobility of the metals.
metals in maize, Indian mustard, and other plants was
only increased by a factor of 2 to 3. The enhancing effect CONCLUSIONS
of EDTA observed here, and by Ebbs and Kochian
(1998), was much smaller than that reported for Pb by This study demonstrated the promising potential for
Blaylock et al. (1997) and Huang et al. (1997), and for Cd and Zn phytoextraction by the ecotype of T. caerules-
Zn by Blaylock et al. (1997). The latter spiked a low- cens from southern France. Three croppings of this
Zn soil with zinc carbonate, which may be more readily plant, covering a period of 391 d, removed approxi-
solubilized than the Zn in the aged contaminated soils mately 8.3 and 200 mg kg⫺1 Cd and Zn, respectively,
used in this study. from a soil contaminated mainly with Zn (2920 mg kg⫺1 )
Our results indicate that EDTA alone increases metal and Cd (19 mg kg⫺1 ). However, co-contamination of
mobility in soil and accumulation in roots, but does not Cu (1245 mg kg⫺1 ) in an agricultural soil inhibited the
substantially increase the transfer of metals to shoots. growth of T. caerulescens, and severely limited its phy-
In their patent on induced hyperaccumulation of metals toextraction potential. It appears that T. caerulescens
in plant shoots, Ensley et al. (1999) described chemically was sensitive to Cu toxicity.
induced phytoextraction as a two-step process in which In chemically enhanced phytoextraction, we used maize
plants first accumulate metals in their roots and then, instead of Indian mustard, because the latter was not
by application of an inducing agent, enhanced transfer tolerant to the heavy metals presented in the two soils,
of the metals to the shoots occurs. This transfer is due even before EDTA applications. The EDTA treatment
to disrupting the plant metabolism that regulates the greatly increased the concentrations of soluble metals,
transport of metal to the shoots. Ensley et al. (1999) but increased metal extraction by maize shoots to a
maintain that chelating agents (such as EDTA) and much smaller degree. Maize also suffered severe phyto-
acids increase both solubility of metals in the soil and toxicity in the second and third cropping in the industrial
root to shoot transfer of metals. However, our results soil. Thlaspi caerulescens was far more efficient than
indicated that transfer of metals to the shoots was not maize ⫹ EDTA treatments in phytoextracting Cd from
greatly enhanced by EDTA. It is not clear whether both soil types. Thlaspi caerulescens also extracted much
application of an herbicide, as proposed by Ensley et al. more Zn than maize ⫹ EDTA treatments from the
(1999), enhances metal accumulation in shoots through low-Cu soil, whereas both approaches extracted similar
disruption of plant metabolism. amounts of Zn from the high-Cu soil.
Compared with natural phytoextraction using T. cae- In chemically enhanced phytoextraction, the dramati-
rulescens, EDTA-enhanced phytoextraction of Cd and cally increased metal solubility and the persistence of
Zn by maize was much less efficient from the French metal–EDTA complexes in soil pore water may pose
soil. The amounts of Cd and Zn removed by maize ⫹ an environmental risk of leaching to ground water.
EDTA over the three croppings were only 1.6 and 11.5%,
respectively, of those removed by T. caerulescens. Natu- ACKNOWLEDGMENTS
ral phytoextraction with T. caerulescens performed We gratefully acknowledge financial support from DG XII
poorly because of the high Cu concentrations in the UK of the European Commission for the PHYTOREM Project.
soil. Even so, its removal of Cd was still eight times We thank A.R. Crosland for his help with ICP analysis, and
greater than that by maize ⫹ EDTA. The amounts of J.L. Morel and C. Schwartz for providing the French soil.
Zn removed by T. caerulescens and by maize ⫹ EDTA IACR-Rothamsted receives grant-aided support from the Bio-
from the UK soil were similar. technology and Biological Sciences Research Council of the
The marked increases in soluble metals in soil pore United Kingdom.
water following EDTA applications may pose a serious
concern in terms of potential leaching of heavy metals REFERENCES
to ground water. Furthermore, the concentrations of Anderson, C.W.N., R.R. Brooks, R.B. Stewart, and R. Simcock. 1998.
soluble metals remained very high several weeks after Harvesting a crop of gold in plants. Nature 395:553–554.
Avery, B.W., and C.L. Bascomb. 1982. Soil survey laboratory meth-
the application of the chelating agent. In fact, 5 mo after ods. Soil Survey Tech. Monogr. 6. Rothamsted Exp. Stn., Harpen-
EDTA application (which was the time interval between den, UK.
EDTA applications), metal–EDTA complexes were still Baker, A.J.M., S.P. McGrath, R.D. Reeves, and J.A.C. Smith. 2000.
found in the soil pore water, indicating the persistence Metal hyperaccumulator plants: A review of the ecology and physi-
ology of a biochemical resource for phytoremediation of metal-
of metal–EDTA complexes. This is in agreement with polluted soils. p. 85–107. In N. Terry and G. Banuelos (ed.) Phyto-
the work of Hong et al. (1999), who reported that EDTA remediation of contaminated soil and water. Lewis Publ., Boca
is relatively biologically stable even under conditions Raton, FL.
1926 J. ENVIRON. QUAL., VOL. 30, NOVEMBER–DECEMBER 2001

Baker, A.J.M., R.D. Reeves, and A.S.M. Hajar. 1994. Heavy metal Huang, J.W., M.J. Blaylock, Y. Kapulnik, and B.D. Ensley. 1998.
accumulation and tolerance in British populations of the metallo- Phytoremediation of uranium-contaminated soils: Role of organic
phyte Thlaspi caerulescens J. & C. Presl (Brassicaceae). New Phy- acids in triggering uranium hyperaccumulation in plants. Environ.
tol. 127:61–68. Sci. Technol. 32:2004–2008.
Beckett, P.H.T., and R.D. Davis. 1978. The additivity of the toxic Huang, J.W., J. Chen, W.B. Berti, and S.D. Cunningham. 1997. Phyto-
effects of Cu, Ni and Zn in young barley. New Phytol. 81:155–173. remediation of lead-contaminated soils: Role of synthetic chelates
Bennett, F.A., E.K. Tyler, R.R. Brooks, P.E.H. Gregg, and R.B. in lead phytoextraction. Environ. Sci. Technol. 31:800–805.
Stewart. 1998. Fertilisation of hyperaccumulator to enhance their Huang, J.W., and S.D. Cunningham. 1996. Lead phytoextraction: Spe-
potential for phytoremediation and phytomining. p. 249–260. In cies variation in lead uptake and translocation. New Phytol. 134:
R.R. Brooks (ed.) Plants that hyperaccumulate heavy metals: Their 75–84.
role in phytoremediation, microbiology, archaeology, mineral ex- Kayser, A., K. Wenger, A. Keller, W. Attinger, H.R. Felix, S.K. Gupta,
ploration and phytomining. CAB Int., Wallingford, UK. and R. Schulin. 2000. Enhancement of phytoextraction of Zn, Cd
Blaylock, M.J. 2000. Field demonstration of phytoremediation of lead and Cu from calcareous soil: The use of NTA and sulfur amend-
contaminated soils. p. 1–12. In N. Terry and G. Banuelos (ed.) ments. Environ. Sci. Technol. 34:1778–1783.
Phytoremediation of contaminated soil and water. Lewis Publ., Knight, B.P., A.M. Chaudri, S.P. McGrath, and K.E. Giller. 1998.
Boca Raton, FL. Determination of chemical availability of cadmium and zinc using
Blaylock, M.J., D.E. Salt, S. Dushenkov, O. Zakharova, C. Gussman, inert soil moisture samplers. Environ. Pollut. 99:293–298.
Y. Kapulnik, B.D. Ensley, and I. Raskin. 1997. Enhanced accumula- Lasat, M.M., M. Fuhrmann, S.D. Ebbs, J.E. Cornish, and L.V. Koch-
tion of Pb in indian mustard by soil-applied chelating agents. Envi- ian. 1998. Phytoremediation of a radiocesium-contaminated soil:
ron. Sci. Technol. 31:860–865. Evaluation of cesium-137 bioaccumulation in the shoots of three
Brown, S.L., R.L. Chaney, J.S. Angle, and A.J.M. Baker. 1994. Phyto- plant species. J. Environ. Qual. 27:165–169.
remediation potential of Thlaspi caerulescens and bladder campion Lombi, E., F.J. Zhao, S.J. Dunham, and S.P. McGrath. 2000. Cadmium
for zinc- and cadmium-contaminated soil. J. Environ. Qual. 23: accumulation in populations of Thlaspi caerulescens and Thlaspi
1151–1157. goesingense. New Phytol. 145:11–20.
Brown, S.L., R.L. Chaney, J.S. Angle, and A.J.M. Baker. 1995. Zinc Marschner, H. 1995. Mineral nutrition of higher plants. Academic
and cadmium uptake by hyperaccumulator Thlaspi caerulescens Press, London, UK.
grown in nutrient solution. Soil Sci. Soc. Am. J. 59:125–133. McGrath, S.P. 1998. Phytoextraction for soil remediation. p. 261–287.
Commission of the European Communities. 1986. Council Directive In R.R. Brooks (ed.) Plants that hyperaccumulate heavy metals.
86/278/EEC. On the protection of the environment and in particular CAB Int., Wallingford, UK.
of the soil when sewage sludge is used. Off. J. Eur. Community McGrath, S.P., and C.H. Cunliffe. 1985. A simplified method for the
L181 (Annex A):6–12. extraction of metals Fe, Zn, Cu, Ni, Cd, Pb, Cr, Co and Mn from
Cooper, E.M., J.T. Sims, S.D. Cunningham, J.W. Huang, and W.R. soils and sewage sludge. J. Sci. Food Agric. 36:794–798.
Berti. 1999. Chelate-assisted phytoextraction of lead from contami- McGrath, S.P., C.M.D. Sidoli, A.J.M. Baker, and R.D. Reeves. 1993.
nated soils. J. Environ. Qual. 28:1709–1719. The potential for the use of metal-accumulating plants for the in
Deutsch Institut für Normung. 1995. Soil quality extraction of trace situ decontamination of metal-polluted soils. p. 673–676. In H.J.P.
elements with ammonium nitrate solution. DIN 19730. Beuth Ver- Eijsackrs and T. Hamers (ed.) Integrated soil and sediment re-
lag, Berlin, Germany. search: A basis for proper protection. Kluwer Academic Publ.,
Ebbs, S.D., and L.V. Kochian. 1997. Toxicity of zinc and copper to Dordrecht, the Netherlands.
Brassica species: Implications for phytoremediation. J. Environ. McLaughlin, M.J., and R. Henderson. 1999. Effect of zinc and copper
Qual. 26:776–781. on cadmium uptake by Thlaspi caerulescens and Cardaminopsis
Ebbs, S.D., and L.V. Kochian. 1998. Phytoextraction of zinc by oat halleri. p. 886–887. In Proc. of the 5th Int. Conf. on the Bio-
(Avena sativa ), barley (Hordeum vulgare ), and Indian mustard geochemistry of Trace Elements, Vienna. 11–15 July 1999. Int. Soc.
(Brassica juncea ). Environ. Sci. Technol. 32:802–806. for Trace Element Res., Vienna.
Ebbs, S.D., M.M. Lasat, D.J. Brady, J. Cornish, R. Gordon, and Salt, D.E., R.D. Smith, and I. Raskin. 1998. Phytoremediation. Annu.
L.V. Kochian. 1997. Phytoextraction of cadmium and zinc from a Rev. Plant Physiol. Plant Mol. Biol. 49:643–668.
contaminated soil. J. Environ. Qual. 26:1424–1430. Satroutdinov, A.D., E.G. Dedyukhina, T.I. Chistyakova, M. Witschel,
Ensley, B.D., M.J. Blaylock, S. Dushenkov, N.P.B.A. Kumar, and Y. I.G. Minkevich, V.K. Eroshin, and T. Egli. 2000. Degradation of
Kapulnik. 1999. Inducing hyperaccumulation of metals in plant metal–EDTA complexes by resting cells of the bacterial strain
shoots. U.S. Patent 5 917 117. Date issued: 29 June. DSM 9103. Environ. Sci. Technol. 34:1715–1720.
Genstat 5 Committee. 1993. Genstat 5. Claredon Press, Oxford, UK. Shen, Z.G., F.J. Zhao, and S.P. McGrath. 1997. Uptake and trans-
Glass, D.J. 2000. Economic potential of phytoremediation. p. 15–31. port of zinc in the hyperaccumulator Thlaspi caerulescens and the
In I. Raskin and B.D. Ensley (ed.) Phytoremediation of toxic met- non-hyperaccumulator Thlaspi ochroleucum. Plant Cell Environ. 20:
als. John Wiley & Sons, New York. 898–906.
Hong, P.K.A., C. Li, S.K. Banerji, and T. Regmi. 1999. Extraction, Sun, B., F.J. Zhao, E. Lombi, and S.P. McGrath. 2001. Leaching
recovery and biostability of EDTA for remediation of heavy metal– of heavy metals from contaminated soils using EDTA. Environ.
contaminated soil. J. Soil Contam. 8:81–103. Pollut. 113:111–120.