Advances in agronomy vol 75

P1: FCG/FMZ P2: FUY
Advances in Agronomy PS097-FM December 12, 2001 14:19 Stylefile version:April 24, 2000
Contributors
Numbers in parentheses indicate the pages on which the authors’ contributions begin.
E. BEN-DOR (173), The Remote Sensing and GIS Laboratory, Department of
Geography and the Human Environment, Tel-Aviv University, Ramat Aviv,
Tel-Aviv 69978, Israel
E. LOMBI (1), Agriculture and Environment Division, IACR-Rothamsted,
Harpenden, Herts AL5 2JQ, United Kingdom
C. LIU (135), Institute of Geographic Sciences and Natural Resources Research, The
Chinese Academy of Sciences, Beijing 100101, China
S. P. McGRATH (1), Agriculture and Environment Division, IACR-Rothamsted,
Harpenden, Herts AL5 2JQ, United Kingdom
A. PICCOLO (57), Dipartimento di Scienze Chimico-Agrarie, Universit`a Degli
Studi Di Napoli “Federico II,” 80055 Portici, Italy Via Universita 100, Naples,
Italy
H. WANG (135), State Key Laboratory of Water Environment Simulation, Key
Laboratory for Water and Sediment Sciences, Ministry of Education, Beijing
Normal University, Beijing 100875, China
L. ZHANG (135), CSIRO Land and Water, Canberra Laboratory, P.O. Box 1666,
Canberra, ACJ 2601, Australia
F. J. ZHAO (1), Agriculture and Environment Divison, IACR-Rothamsted,
Harpenden, Herts AL5 2JQ, United Kingdom
vii
P1: FCG/FMZ P2: FUY
Advances in Agronomy PS097-FM December 12, 2001 14:19 Stylefile version:April 24, 2000
Preface
Volume 75 contains four outstanding reviews dealing with phytoremediation,
issues related to water use in China, humic substances, and remote sensing. Chap-
ter 1 is an extensive review on phytoremediation of metals, metalloids, and radionu-
clides including discussion on phytoextraction technologies, hyperaccumulator

plants, and chemically induced phytoextraction and phytovolatilization. Chapter 2
covers the conservation and use of water in Chinese agriculture including engi-
neering, economic, and agronomic aspects and considerations. Chapter 3 presents
advances in understanding the structure of humic substances, particularly the con-
cept of a supramolecular structure. Analytical and molecular scale evidence for
this latter structure are presented as well as discussions on the role of humic su-
perstructures in soils. Chapter 4 presented frontiers in quantitative remote sensing
of soil properties including principles, methods, mechanisms, and limitations.
I thank the authors for their first-rate reviews.
D
ONALD
L. SPARKS
ix
12/05/2001 09:33 AM Agronomy-V. 75 PS097A-01.tex PS097A-01.xml APserialsv2(2000/12/19) Textures 2.0
PHYTOREMEDIATION OF METALS,
M
ETALLOIDS, AND RADIONUCLIDES
S. P. McGrath, F. J. Zhao, and E. Lombi
Agriculture and the Environment Division
IACR-Rothamsted, Harpenden, Herts
AL5 2JQ, United Kingdom
I. Introduction
A. Risks of Metals and Metalloids in Soils
B. The Need for Cleanup of Contaminated Soils
C. Phytoextraction, Phytomining, and Removal Technologies
II. Phytoextraction Using Hyperaccumulator Plants
A. Metal Hyperaccumulators
B. Phytoextraction Using Hyperaccumulator Plants
C. Mechanisms of Metal Hyperaccumulation
III. Chemically Enhanced Phytoextraction

A. Potential Applications
B. Chemically Enhanced Phytoextraction of Lead
C. Chemically Enhanced Phytoextraction of Other Heavy Metals
D. Chemically Enhanced Phytoextraction of Radionuclides
E. Chemically Enhanced Phytomining
F. Chemically Enhanced Phytoextraction versus Natural Hyperaccumulation
G. Possible Concern Relating to the Use of Chelating Agents
IV. Phytovolatilization
A. Selenium
B. Mercury
V. Summary and Future Directions
References
Phytoremediation is a developing technology that can potentially address the
problems of contaminated agricultural land or more intensely polluted areas
affected by urban or industrial activities. Three main strategies currently exist to
phytoextract inorganic substances from soils using plants: (1) use of natural hy-
peraccumulators; (2) enhancement of element uptake of high biomass species by
chemical additions to soil and plants; and (3) phytovolatilization of elements, which
often involves alteration of their chemical form within the plant prior to volatiliza-
tion to the atmosphere. Concentrating on the techniques that potentially remove
inorganic pollutants such as Ni, Zn, Cd, Cu, Co, Pb, Hg, As, Se, and radionu-
clides, we review the progress in the understanding of the processes involved and
1
Advances in Agronomy, Volume 75
Copyright
C
2002 by Academic Press. All rights of reproduction in any form reserved.
0065-2113/02 $35.00
12/05/2001 09:33 AM Agronomy-V. 75 PS097A-01.tex PS097A-01.xml APserialsv2(2000/12/19) Textures 2.0
2 S. P. McGRATH ET AL.

the development of the technology. This includes the advances made in the study
of the physiology and biochemistry of metal uptake, transport and sequestration by
hyperaccumulator plants, as well as the investigation of the processes occurring in
soil and plant systems subject to the chemicalenhancement approach. Enoughwork
has been carried out inthe last decadeto allow some assessment of the situations and
elements in which phytoremediation is likely to be most successful. However, we
also identify where there is lack of knowledge. Finally, the likely future directions
for research and application are discussed.
C
2002 Academic Press.
I. INTRODUCTION
Phytoremediation can be loosely defined as the use of plants to improve the envi-
ronment. Obviously this is an enormous subject and here we will concentrate on the
phytoremediation of metals, metalloids, and radionuclides. Phytoremediation of
organic compounds in soil and water is a related and rapidly expanding area, which
is covered elsewhere (Kruger et al., 1997; Salt et al., 1998; Wenzel et al., 1999).
It is very appropriate to review this subject at this time because it was around
1990 that the first field experiments began examining phtyoremediation of metals
and Se (reported in Ba˜nuelos et al., 1993; McGrath et al., 1993); and now a
decade has passed. We will examine the different strategies that have evolved for
phytoremediation and the progress that has been made on the physiology of metal
accumulation. On a more practical level, the attempts at field application will be
evaluated, and the likely future directions of the science and technology will be
discussed.
A. RISKS OF
METALS AND METALLOIDS IN SOILS
Metals and metalloids such as As and Se can pose risks when they build up in
soils due to many forms of anthropogenic influences. Some such as Zn, Cu, Mn,
Ni, Se, Co, Cr, and Mo are essential for living organisms, and therefore deficiency
situations exist either because of very low total amounts of these metals in soil or

because of low bioavailability caused by soil chemical conditions. In these cases,
when metals are added, there may be positive biological responses in terms of
growth and health of organisms. However, these metals and those that are thought
to be nonessential such as Pb, As, Hg, and Cd tend to build up in soils; and when
their bioavailability becomes high, toxicity can result. These negative effects can
occur in soil microbes, soil fauna, higher animals, plants, and humans. A further
12/05/2001 09:33 AM Agronomy-V. 75 PS097A-01.tex PS097A-01.xml APserialsv2(2000/12/19) Textures 2.0
PHYTOREMEDIATION 3
threat is from radionuclides such as those of U,
137
Cs,
90
Sr, and
3
H in soil and water
(Negri and Hinchman, 2000).
Of course, these elements may occur at elevated concentrations quite naturally
in soils and waters. In these cases there may be “effects” on biodiversity and
on animal and human health. Examples would be metal-tolerant vegetation that
has evolved on metal-mineralized soils (Baker and Proctor, 1990), the effects on
human health due to excess Se (Yang et al., 1983), and Cd accumulation in tissues
of white-tailed ptarmigan (Lagopus leucurus) in the Colorado Rocky Mountains,
resulting in toxicity (Larison et al., 2000). In these cases, it may not be possible
or desirable to clean up the soils, but there may be a role for plants in reducing the
exposure of biota to these elements, for example, by reduced uptake and exclusion
from tissues, or removing elements like Se in geogenically laden water (Ohlendorf
et al., 1986; Wu et al., 1995). Indeed, where these natural hot spots occur, there
may be specialized fauna and flora, like metallophyte vegetation, which may be in
need of preservation (Reeves and Baker, 2000).
Metals and metalloids enter soils and waters due to many processes including

atmospheric deposition from industrial activities or power generation; disposal
of wastes such as sewage sludge, animal manures, ash, domestic and industrial
wastes or by-products; irrigation and flood or seepage waters and the utilization
of fertilizers, lime, or agrochemicals. Radionuclides may build up in some areas
due to deliberate or accidental releases related to their use for energy production
or for military purposes. Unlike nitrate or chloride, many of these elements are
relatively strongly retained in the surface of soils and do not readily leach, causing
the accumulation that may ultimately pose a threat to humans and biota. However,
under some conditions, small amounts of these elements do leach and can be an
issue in waters, particularly those used for irrigation or drinking. Key examples
here would be radionuclides, As, Se, and Cr (Chiou et al., 1995; Kimbrough et al.,
1999; Negri and Hinchman, 2000; Ohlendorf et al., 1986).
Under these conditions, phytoremediation is an important developing technol-
ogy for removal of these elements from either soil or water. It has the potential to
be low cost and to be applicable to large areas where other methods may be too
expensive and where the concentrations of contaminants are too small for other
methods to be effective or economically viable.
B. THE NEED FOR CLEANUP OF CONTAMINATED SOILS
There isa long history of contaminationaccumulating in soils due tothe practices
mentioned earlier. Public and political pressure to reverse this situation and clean
up areas only occurs when critical levels are reached. Leaving aside the methods
of deriving critical levels for microbes, animals, plants, and humans, once these
exist, they provide a direct stimulus for cleanup.
12/05/2001 09:33 AM Agronomy-V. 75 PS097A-01.tex PS097A-01.xml APserialsv2(2000/12/19) Textures 2.0
4 S. P. McGRATH ET AL.
Using various ways of defining “contaminated” land, it has been estimated that
in the European Union alone, there are potentially 1,400,000 contaminated sites
(ETCS, 1998). Not all of these will be contaminated with metals or metalloids,
but this gives an indication of the scale of the problem as it may exist worldwide.
For example, trace elements are present at high concentrations at 65% of the con-

taminated Superfund sites for which the US EPA has signed Records of Decision
(US EPA, 1997). Indeed, some areas are not included in these assessments, such
as those with low-level contamination due to atmospheric deposition or to the use
of chemicals in agriculture. For example, the use of phosphate sources that are
contaminated with Cd for agricultural fertilizers may result in crops that contain
more than the allowed concentrations of Cd in foodstuffs (Commission of the
European Communities, 2001). It is unlikely that these areas are included in the
previously described estimates, as they focus more on urban and industrial land.
However, for the sustained practice of agriculture with inputs of fertilizers, sewage
sludge, and animal manures, there may be a role for plants in removing the small
excess amounts of metals such as Cd, Zn, and Cu from soils, perhaps on a long
rotational basis. Use of low-Cd phosphate is already taking place, while removal of
Cd from phosphate rock is still not considered economically feasible (Oosterhuis
et al., 2000). The average concentration in phosphate fertilizers in Europe is still
138 mg Cd kg
−1
P (ERM, 1997). In comparison, the background level of cadmium
is 0.3 mg Cd kg
−1
soil or less in most agricultural soils in Europe. Concentrations
in soils are increasing because the inputs are not balanced by the output in terms
of removal by crops and leaching out of the ploughing layer (Eriksson et al., 1996;
Kofoed and Klausen, 1983). Thus it is likely that phytoremediation will be needed
for continued agriculture in the future.
C. P
HYTOEXTRACTION
,PHYTOMINING, AND
REMOVAL TECHNOLOGIES
Our focus in this review is on the methods that remove metals and metalloids
from soil. This can be achieved by phytoextraction or phytovolatilization, depend-

ing on the element considered. A variant of phytoextraction, which applies when
the extracted elements are of high value, is phytomining. In the latter case, the
aims are to derive a “bio-ore” from the burning of the plant material and to profit
from the energy released by combustion of the biomass and the value of the ore
itself. The recycling of elements that are bioconcentrated during phytoextraction
will not be discussed, and the disposal options for plant biomass will depend on
the market for the elements concerned.
Related technologies exist or are under development, such as phytostabiliza-
tion. This is when the plants are used essentially to stabilize contaminated land
or the pollutants present in soil and in so doing prevent or reduce erosion, water
12/05/2001 09:33 AM Agronomy-V. 75 PS097A-01.tex PS097A-01.xml APserialsv2(2000/12/19) Textures 2.0
PHYTOREMEDIATION 5
flow, and flow of pollutants. In this case, metal-tolerant species that do not take
up large quantities of metals are often used. In this review, for reasons of space,
we chose to focus on remediation of contaminated soils, while preventing con-
tamination of groundwater, and not remediation of contaminated water itself. That
subject is covered elsewhere (Dushenkov and Kapulnik, 2000; Terry and Ba˜nuelos,
2000).
The efficiency of phytoextraction is ultimately the product of a simple equation:
biomass × element concentration in biomass. Both factors are important, but it
is easy to show that high concentrations in the above-ground material are very
important. Harvesting roots or other below-ground organs is difficult and prevents
regrowth if the “crop” is a perennial one. The increasing yield from 2 to 20 t ha
−1
,
which is probably a biological maximum for an annual plant or harvestable from
a perennial one, has little influence on the removal rate below about 1000 mg kg
−1
of an element in the plant dry matter (Fig. 1). Therefore, maximizing concentra-
tions in the plant seems to be the obvious strategy for increasing efficiency, while

optimizing yields by agronomic means. However, it must be kept in mind that this
thinking relates to very pollutedsoils thatrequire hundreds of kilograms per hectare
to be removed. For elements like Cd where relatively small removals (<1kgha
−1
)
Figure 1 Modeled removal of an element from soil by crops, showing the dependence on the
concentration in the biomass and the effect of yield. All amounts relate to above-ground material.
12/05/2001 09:33 AM Agronomy-V. 75 PS097A-01.tex PS097A-01.xml APserialsv2(2000/12/19) Textures 2.0
6 S. P. McGRATH ET AL.
are important, or radionuclides where small quantities are of concern, this may not
be so true. But the principle still applies to Cd, i.e., the scope for maximizing Cd
concentration is in the magnitude of hundreds (hyperaccumulators, see following),
whereas the scope for maximizing biomass is <10).
Two strategies exist for obtaining plant biomass with high metal concentrations:
(1) use of natural hyperaccumulators and (2) the enhancement of uptake of metals
by normally non-accumulating species by applying various chemicals that increase
uptake. These are discussed in the following sections.
II. PHYTOEXTRACTION USING
HYPERACCUMULATOR PLANTS
A. M
ETAL HYPERACCUMULATORS
1. Definition
Based on the relationship between metal concentrations in shoot and in soil,
Baker (1981) proposed that plants growing on metalliferous soils can be grouped
into three types: (1) excluders, where metal concentrations in the shoot are main-
tained at a low level across a wide range of soil concentration, up to a critical
soil value above which the mechanism breaks down and relatively unrestricted
root-to-shoot transport results; (2) accumulators, where metals are concentrated in
above-ground plant parts from low to high soil concentrations; and (3) indicators,
where uptake and transport of metals to the shoot are regulated so that internal

concentration reflects external levels, at least until toxicity occurs.
Exclusion of metals from the shoots is by far the most common strategy em-
ployed by many metal-tolerant species. On the other hand, metal accumulation
can occur in some plant species that grow mainly on metalliferous soils. Reeves
and Baker (2000) traced the earliest qualitative observation, Zn accumulation in
Viola calaminaria in the Zn-rich soils in the Aachen area between Germany and
Belgium, to A. Braun in 1855. Some 30 years later, Baumann (1885) showed that
both Viola calaminaria and Thlaspi calaminare (later called Thlaspi caerulescens)
growing over the calamine deposits contained over 1% Zn (10,000 µgg
−1
)inthe
shoot dry matter. Exceedingly high accumulations of Se in Astragalus plants and of
Ni in Alyssum bertolonii were discovered in the 1930s and 1940s, respectively (see
Brooks, 1998). Brooks et al. (1977) first introduced the term “hyperaccumulators”
to describe plants capable of accumulating more than 1000 µgNig
−1
on a dry
leaf basis in their natural habitats. The criterion for defining Co, Cu, Pb, and Se
hyperaccumulation is also 1000 µgg
−1
in shoot dry matter, whereas for Zn and
12/05/2001 09:33 AM Agronomy-V. 75 PS097A-01.tex PS097A-01.xml APserialsv2(2000/12/19) Textures 2.0
PHYTOREMEDIATION 7
Mn the threshold is 10,000 µgg
−1
, and for Cd 100 µgg
−1
(Baker et al., 2000;
Brooks, 1998). Although these criteria are quite arbitrary, in general the concentra-
tions of metals in hyperaccumulator plants are about 100- to 1000-fold higher than

those in normal plants growing on soils with background metal concentrations, and
about 10- to 100-fold higherthan most other plants growing on metal-contaminated
soils.
In addition to the exceedingly high accumulation of metals in the shoots, hyper-
accumulator plants are also characterized by a shoot-to-root metal concentration
ratio of >1, whereas non-hyperaccumulator plants generally have higher metal
concentrations in roots than in shoots (Baker, 1981; Baker et al., 1994a,b; Brown
et al., 1995a; Gabbrielli et al., 1990; Homer et al., 1991a; Kr¨amer et al., 1996; Shen
et al., 1997; Zhao et al., 2000). A highly efficient transport of metals from roots
to shoots is one of the key features associated with all hyperaccumulator plants.
Metal hyperaccumulation is a rare phenomenon in terrestrial higher plants. To
date, about 400 plant species have been identified as metal hyperaccumulators,
representing <0.2% of all angiosperms (Baker et al., 2000; Brooks, 1998). It is
foreseeable that the number of hyperaccumulator plants will increase as more
geobotanical surveys are carried out worldwide. On the other hand, some of the
hyperaccumulator species reported earlier may not be confirmed as true hyperac-
cumulators, but may be incorrectly identified due to contamination or analytical
errors (see following). Details of different metal hyperaccumulator species and
their geographical distributions have been documented elsewhere (see Baker and
Brooks, 1989; Baker et al., 2000; Brooks, 1998; Reeves and Baker, 2000). The
following sections give only brief descriptions of the key features of different metal
hyperaccumulators.
2. Nickel Hyperaccumulators
Ni hyperaccumulatorsare the most numerous among hyperaccumulating species
of plants, with a current total number of 318 taxa distributed mainly in the
tropical to warm temperature regions of the world (Baker et al., 2000; Reeves
and Baker, 2000). The richness of Ni hyperaccumulator species is probably
due to the widespread occurrence of Ni-rich ultramafic (serpentine) soils and
the long history of geobotanical studies of ultramafic floras. Some of the well-
known Ni hyperaccumulators are in the genus Alyssum L. (Brassicaceae), al-

though the most remarkable example is perhaps Sebertia acuminata (Sapotaceae),
a New Caledonian tree that can grow to a height of about 10 m. Jaffr´e et al.
(1976) showed that this plant produces a blue-green latex containing 11.2%
Ni on a fresh weight basis (25.7% on a dry weight basis). A mature tree of
Sebertia acuminata was estimated to contain 37 kg Ni (Sagner et al., 1998). The
other species that has recently attracted attention is Berkheya coddii, which can
12/05/2001 09:33 AM Agronomy-V. 75 PS097A-01.tex PS097A-01.xml APserialsv2(2000/12/19) Textures 2.0
8 S. P. McGRATH ET AL.
accumulate Ni to more than 1% and is tall, fast-growing, and productive (Morrey
et al., 1989). These are the attributes that are ideal for phytoremediation or phyto-
mining (see following).
3. Zinc and Cadmium Hyperaccumulators
In comparison to Ni hyperaccumulators, far fewer plant species have been
reported that are able to hyperaccumulate Zn and Cd. Baker et al. (2000) listed
11 taxa of Zn hyperaccumulator plants, whereas Reeves and Baker (2000) also con-
sidered two otherspecies (Thlaspi ochroleucum and Polycarpaea synandra), which
did not reach the criterion of 10,000 µgZng
−1
to be hyperaccumulators. The best
known examples of the Zn hyperaccumulators are Thlaspi caerulescens (formerly
called T. alpestre) and Arabidopsis halleri (formerly named Cardaminopsis
halleri), both belonging to the Brassicaceae family. In the case of Thlaspi
ochroleucum, Shen et al. (1997) showed that its Zn accumulation and toler-
ance are considerably lower than those in T. caerulescens in hydroponic cultures.
T. ochroleucum also accumulates more Zn in roots than in shoots, thus behaving in
that sense rather like a non-hyperaccumulator. For Cd, T. caerulescens is the
only known hyperaccumulator (Ernst, 1974; Reeves and Bakers, 2000), although
recent hydroponic experiments showed that A. halleri is capable of accumulat-
ing >1000 µgCdg
−1

in the shoots without suffering from phytotoxicity (K¨upper
et al., 2000), and for this reason may be classified as a Cd hyperaccumulator.
Whether A. halleri accumulates over 100 µgCdg
−1
in any of its natural habitats
(the criterion for defining Cd hyperaccumulation) is probably determined by the
Cd concentration in the soil.
4. Copper and Cobalt Hyperaccumulators
Twenty-eight and 37 taxa of Co and Cu hyperaccumulator plants, respectively,
have been reported (see Brooks, 1998; Reeves and Baker, 2000). These plants are
mainly distributed in the Shaban Copper Arc of the Democratic Republic of Congo
(formerly Za¨ıre). Some of these plants can hyperaccumulate both metals. However,
there have been few experimental studies on the ability of these plants to accumu-
late metals, and therefore whether or not they can truly hyperaccumulate Cu and
Co remains to be confirmed. A recent study using hydroponic cultures showed
that Haumaniastrum katangense and Aeollanthus biformifolius (Lamiaceae), both
of which have been described as Cu and Co hyperaccumulators, did not hyperac-
cumulate Cu and Co in the shoots, but rather behaved as typical metal excluders
(K¨ohl et al., 1997). Contamination of plant samples with dust is a possibility when
sampling and analyzing wild plants from their natural habitats. This can cause
large errors if the dust happens to be rich in the metals to be analyzed. Reeves
and Baker (2000) gave an example in which 0.2 mg of malachite (a secondary
12/05/2001 09:33 AM Agronomy-V. 75 PS097A-01.tex PS097A-01.xml APserialsv2(2000/12/19) Textures 2.0
PHYTOREMEDIATION 9
Cu mineral) included as a dust with 100 mg of plant tissue genuinely containing
10 µgCug
−1
is enough to raise the apparent Cu concentration to >1150 µgg
−1
.

5. Lead
Pb hyperaccumulation is rare, primarily because Pb is very insoluble in soil.
Non-accumulating plants such as Brassica juncea and Zea mays have been shown
to hyperaccumulate Pb in the shoots once Pb solubility in the soil was greatly
enhanced with synthetic chelates such as EDTA (see following). This chemically
induced hyperaccumulation should not be confused with the natural hyperaccu-
mulation discussed in this section. Fourteen taxa have been reported to be Pb
hyperaccumulators with Pb concentration in the shoots varying from 1000 to
20,000 µgg
−1
(Reeves and Baker, 2000). Similar to the situation with Cu/Co
hyperaccumulators, Pb uptake and translocation in these reported hyperaccumu-
lators have not been researched often under controlled conditions. Results from
a field survey showed that Thlaspi rotundifolium spp. cepaeifolium from a Pb/Zn
mining area in northern Italy contained up to 8200 µgPbg
−1
in the shoots (Reeves
and Brooks, 1983). However, our results using hydroponic and soil experiments
(unpublished) and those of Huang and Cunningham (1996) indicate that this plant
does not hyperaccumulate Pb in the shoots. In the roots, a large accumulation of
Pb occurs in the apoplast, principally as lead phosphate deposits. This type of Pb
accumulation does not represent a true uptake by roots.
6. Selenium and Arsenic
The best known examples of Se hyperaccumulators are probably in the genus
Astragalus (Leguminosae). In the 1930s, O. B. Beath and his colleagues found at
least 13 taxa of Astragalus containing more than 1000 µgSeg
−1
in the shoot dry
matter in the Colorado Plateau in the United States (see Brooks, 1998; Reeves and
Baker, 2000). The high Se concentrations in these plants caused serious disease

in the grazing cattle and sheep. A number of other plants have been identified
as Se hyperaccumulators in other regions of the world, including a Venezuelan
tree, Lecythis ollaria (Lecythidaceae), which has a Se concentration in nuts of
up to 18,200 µgg
−1
and is therefore toxic to humans and animals (Aronow and
Kerdel-Vegas, 1965).
Arsenic accumulation by terrestrial plants is a very rare phenomenon, and no As
hyperaccumulators have been reported until recently. Ma et al. (2001) discovered
that Brake fern (Pteris vittata) growing on an As contaminated soil contained
3280–4980 µgAsg
−1
in the shoot (fronds) dry matter, compared to 2–23 µg
As g
−1
in 13 other species growing on the same soil. In a greenhouse experiment,
Brake fern growing on a soil amended with 1500 µgAsg
−1
accumulated up to
22,630 µgAsg
−1
in the fronds in 6 weeks. It is also highly efficient in transporting
12/05/2001 09:33 AM Agronomy-V. 75 PS097A-01.tex PS097A-01.xml APserialsv2(2000/12/19) Textures 2.0
10 S. P. McGRATH ET AL.
As from roots to shoots. This plant is hardy and fast growing, and thus could be
potentially used in the phytoremediation of As.
7. Other Metals
Reeves and Baker (2000) listed nine species from New Caledonia that had at
least one specimen containing above 10,000 µgMng
−1

in the shoots. These may be
considered as Mn hyperaccumulators. In general, however, Mn hyperaccumulation
has not been researched very much.
Thallium (Tl) is extremely toxic to animals and humans, although soil conta-
mination with Tl is rare. Unusual hyperaccumulation of Tl (>500 µgg
−1
shoot dry
weight) has been reported in two species from southern France, Iberis intermdia
and Biscutella laevigata, both belonging to the Brassicaceae family (Anderson
et al., 1999; Leblanc et al., 1999). Growing on soils with a total of Tl up to
40 µgg
−1
, Iberis intermdia and Biscutella laevigata contained up to 4000 and
14,000 µgTlg
−1
in the shoot dry weight. The plant-to-soil concentration quotient
was mostly greater than 10 (Anderson et al., 1999), suggesting true hyperaccumu-
lation rather than the possibility of soil contamination on plant specimens.
B. PHYTOEXTRACTION USING HYPERACCUMULATOR PLANTS
1. Zinc and Cadmium
The concept of using hyperaccumulator plants to take up and remove heavy
metals from contaminated soils was first discussed by Chaney (1983). However, it
was until the early 1990s that field experiments were carried out to test the potential
of phytoextraction of metals with hyperaccumulator plants. A field-based exper-
iment was conducted in 1991–1993 in sewage sludge-treated plots at Woburn,
England, with the total Zn in soil varying from 124 to 444 mg kg
−1
and the total
Cd varying from 2.8 to 13.6 mg kg
−1

(Baker et al., 1994a; McGrath et al., 1993,
2000). This experiment compared metal extraction efficiency of different hyperac-
cumulator plant species, including several populations of the Zn hyperaccumulator
T. caerulescens. In 1991, two populations of T. caerulescens (Prayon from Bel-
gium and Whitesike from the UK) produced shoot biomass yields of 3.6–4.5 t ha
−1
(dry weight) and accumulated 2000–4300 µgZng
−1
dry weight in the shoots
(Fig. 2).Biomass increased to 7.5–7.8 tha
−1
in 1992,but shootZn concentration de-
creased to 500–2200 µgg
−1
. The concentrations of Zn in the shoots were consider-
ably lower than the 10,000 µgg
−1
value used to define hyperaccumulation, because
the soil was only slightly or moderately contaminated with Zn. Nevertheless, these
concentrations werestill 10- to20-fold higher thanthose ina number ofnormal crop
species growing on the same plots (McGrath et al., 2000). Both the concentration
12/05/2001 09:33 AM Agronomy-V. 75 PS097A-01.tex PS097A-01.xml APserialsv2(2000/12/19) Textures 2.0
PHYTOREMEDIATION 11
Figure 2 Concentrations (a) and total uptake (b) of Zn in the shoots of T. caerulescens (the Prayon
population) grown on different plots of a long-term sewage sludge experiment at Woburn, England, in
1991 and 1992.
of Zn and the total Zn removal in the shoots of T. caerulescens increased with
increasing soil Zn (Fig. 2). In the plots with total soil Zn >300 mg kg
−1
, total Zn

removals by a single crop of T. caerulescens were between 10 and 25 kg ha
−1
.
Being a wild plant, T. caerulescens is not easy to grow under field conditions, and
a substantial variation in individual yields exists. It is possible to improve biomass
production of hyperaccumulator plants through optimization of agronomic in-
puts (Bennett et al., 1998). To estimate the maximum potential removal of Zn by
an optimized crop of T. caerulescens, McGrath et al. (1993, 2000) made model
calculations based on the uptake by the largest two rows of the plants observed
in the field experiment. The calculated maximum potential removals of Zn were
25–50 kg ha
−1
, two to three times higher than the average values shown in Fig. 2.
With the optimized removal rate, it would take 7–14 crops of T. caerulescens to
reduce total soil Zn from 440 to 300 µgg
−1
. This compares favorably with over
800 croppings with Brassica napus (oilseed rape) and more than 2000 croppings
with Raphanus sativus (raddish) (McGrath et al., 1993).
In the field experiment at Woburn described previously, the Cd concentration
in the shoots of two populations of T. caerulescens (Prayon and Whitesike) was
<40 µgg
−1
, and the Cd removal varied from a few to about 300 g ha
−1
(McGrath
et al., 2000). The Whitesike population was more efficient than Prayon in Cd
uptake. Although the rate of Cd removal by the Whitesike population is small
compared to the inputs of Cd from sludge on this experimental site, it is far greater
than inputs of Cd from the use of phosphate fertilizers and from the atmospheric

deposition. Later, samples of a population of T. caerulescens from southern France
(Ganges) was also grown in the Woburn experiment. This population was found
12/05/2001 09:33 AM Agronomy-V. 75 PS097A-01.tex PS097A-01.xml APserialsv2(2000/12/19) Textures 2.0
12 S. P. McGRATH ET AL.
Figure 3 Cadmium concentrations in three different ecotypes of T. caerulescens grown on
different plots of a long-term sewage sludge experiment at Woburn, England.
to be far superior to the other two in Cd uptake (Fig. 3), with Cd concentration
in the shoots reaching 500 µgg
−1
in the highest Cd plot (13.6 µgCdg
−1
in the
soil; Lombi et al., 2000a). This work shows that there is a substantial potential to
screen for more efficient ecotypes for metal uptake.
Also in 1991–1992, Brown et al. (1995b) compared Zn and Cd uptake by
T. caerulescens (Prayon) and two other non-hyperaccumulators on long-term
sewage sludge plots in Beltsville, Maryland, which have total soil Zn and Cd
up to 181 and 5.5 µgg
−1
, respectively. They found the concentrations of Zn in
T. caerulescens varying from about 1000 up to 4000 µgg
−1
, about 10-fold greater
than those in Silene vulgaris and lettuce. Similar to the results obtained from the
Woburn experiment, the Prayon population of T. caerulescens did not take up sig-
nificantly more Cd than the other two plants. Brown et al. (1995b) did not present
biomass yields on a unit area basis; therefore it was not possible to calculate metal
extraction rates.
Robinson et al. (1998) grew T. caerulescens (the Ganges population from south-
ern France) both in pots and in mine waste in fields. They found that a single

fertilized crop could remove 60 kg Zn ha
−1
and 8.4 kg Cd ha
−1
. Because bioac-
cumulation coefficients (plant/soil metal concentration quotients) were in general
higher for Cd than for Zn, phytoremediation using T. caerulescens would be fea-
sible for low levels of soil Cd, but not feasible to remediate the extremely high Zn
concentrations (40,000 µgg
−1
) found in the mine wastes.
The main constraints for using T. caerulescens to phytoextract Zn and Cd
from contaminated soils are the slow growth rate, its rosette growth habit, and
its generally small biomass production. Several approaches may be taken to tackle
these constraints. Significant differences in zinc concentration and plant size were
found among sib families within the same populations of T. caerulescens, but the
12/05/2001 09:33 AM Agronomy-V. 75 PS097A-01.tex PS097A-01.xml APserialsv2(2000/12/19) Textures 2.0
PHYTOREMEDIATION 13
characters of the zinc concentration in shoots and plant size were not correlated
(Pollard and Baker, 1996). In one population, the variations in these characters
appeared to be heritable. The results indicate that selecting large-sized plants of
T. caerulescens does not lead to reduced Zn concentration in the shoots as a result
of a growth dilution, and also that good attributes for phytoextraction found in
individual plants may be heritable. The other approach is to transfer the metal
hyperaccumulation traits from hyperaccumulators to other high biomass plants,
through either hybridization or genetic engineering. Recently, Brewer et al. (1999)
produced several somatic hybrids between T. caerulescens and the high biomass
crop oilseed rape (Brassica napus). These hybrids produced a larger biomass
than T. caerulescens and had an erect growth habit that is suitable for mechan-
ical harvesting. The hybrids were able to accumulate and tolerate Zn and Cd at

levels that are toxic to Brassica napus, although their ability to accumulate met-
als appeared to be lower than T. caerulescens. In the future, it may be possible
to introduce genes responsible for metal hyperaccumulation and internal toler-
ance to high biomass, and preferably non-food, crops. Genetic engineering has
been successfully applied to transform poplar trees for the volatilization of Hg
(see following).
2. Nickel
Compared to Thlaspi caerulescens, many Ni hyperaccumulators yield a much
higher biomass. Nicks and Chambers (1995) found that a natural crop of the
Californian Ni hyperaccumulator Streptanthus polygaloides was capable of pro-
ducing up to 100 kg Ni ha
−1
(worth $550 ha
−1
at the time). Incineration of the crop
could produce combustion energy for electricity generation, yielding additional re-
turn. They concluded that the return to a farmer growing a “crop of Ni” would be
roughly comparable, or superior to that obtained for a crop of wheat. Such an oper-
ation of growing hyperaccumulators over a low-grade ore body or mineralized soil
for the prodcution of bio-ore has been termed phytomining (Brooks et al., 1998).
Robinson et al. (1997a,b) investigated the Ni extraction potential of two other
Ni hyperaccumulators. Dry biomass yields of Alyssum bertolonii and Berkheya
coddii reached 9 and 22 t ha
−1
, respectively. The Ni yields were in the range
of 70–100 kg ha
−1
. They concluded that the phytomining of Ni could be eco-
nomically feasible at many sites worldwide. Their studies also showed that both
hyperaccumulators responded markedly to fertilizer additions without a reduction

in the Ni concentration in shoot tissues. The concentration of Ni in the shoots of
B. coddii was found to correlate with extractable Ni in the soil. Unless the ex-
tractable concentration is perfectly buffered by soil solid phases, this means that
the rate of Ni extraction will decrease as the number of cropping increases. For
phytoremediation of Ni-contaminated soils, Robinson et al. (1997b) estimated that
four croppings with B. coddii would be required to bring the total soil Ni from
250 µgg
−1
to below the EU guideline value of 75 µgg
−1
.
12/05/2001 09:33 AM Agronomy-V. 75 PS097A-01.tex PS097A-01.xml APserialsv2(2000/12/19) Textures 2.0
14 S. P. McGRATH ET AL.
3. Other Metals
Phytomining for Tl using Tl hyperaccumulators, such as Iberis intermdia and
Biscutella laevigata, has also been suggested (Anderson et al., 1999).
C. MECHANISMS OF
METAL HYPERACCUMULATION
Hyperaccumulation of heavy metals is likely to involve several steps, including
metal transport across root cell plasma membranes, xylem loading and transloca-
tion, and sequestration of metals in specialized leaf cells or vacuoles. Sequestration
may involve both physical compartmentation and complexation with ligands. Be-
cause of the multiple processes involved, metal hyperaccumulation is likely to
be controlled by multiple genes. Indeed, a recent study on crosses between the
Zn hyperaccumulator Arabidopsis halleri and the non-accumulating, non-tolerant
species Arabidopsis petraea showed that metal accumulation and tolerance in hy-
peraccumulators are likely to be genetically independent traits (Macnair et al.,
1999). A true metal hyperaccumulator must posses both genetic traits for its hy-
peraccumulation potential to be realized. Another interesting observation is that
metal hyperaccumulation appears to be a constitutive property; i.e., populations

of the same hyperaccumulator species collected from both metalliferous and non-
metalliferous sites are able to hyperaccumulate the metal similarly (Bert et al.,
2000; Boyd and Martens, 1998a; Reeves and Baker, 1984). This raises an inter-
esting question regarding the evolution of metal hyperaccumulation, which still
awaits convincing answers. A number of hypotheses have been advanced to explain
the ecological functions of metal hyperaccumulation. Among them the hypothesis
that metal hyperaccumulation plays a role in defence against pathogens and her-
bivores has gained support from recent studies (see reviews by Boyd (1998) and
Pollard et al. (2000)).
There has been rapid progress in our understanding of the processes involved
in metal hyperaccumulation in recent years. However, the whole picture is still far
from complete.
1. Metal Uptake
a. Zinc and Cadmium
A number of studies using hydroponic culture have demonstrated the extraor-
dinary ability of T. caerulescens (Baker et al., 1994b; Brown et al., 1995a; Shen
et al., 1997) and Arabidopsis halleri (Zhao et al., 2000) to take up Zn. Even with a
low concentration of Zn in the nutrient solution (e.g., 1 µM), T. caerulescens was
able to accumulate much higher concentrations of Zn in the shoots than in other
12/05/2001 09:33 AM Agronomy-V. 75 PS097A-01.tex PS097A-01.xml APserialsv2(2000/12/19) Textures 2.0
PHYTOREMEDIATION 15
non-hyperaccumulating plants (Shen et al., 1997). Zn uptake by T. caerulescens
from non-contaminated or slightly contaminated soils was always greater than
that by non-hyperaccumulator plants such as Brassica napus or Raphanus
sativus (Knight et al., 1997; McGrath et al., 1993). These results suggest that
T. caerulescens possesses a highly efficient system in Zn uptake and/or translo-
cation. Lasat et al. (1996) studied the kinetics of
65
Zn influx in the roots of
T. caerulescens and a non-accumulator Thlaspi arvense. They found that the

maximum influx velocity, V
max
, was 4.5-fold higher in T. caerulescens than in
T. arvense. Recently, Pence et al. (2000) cloned a gene in T. caerulescens, ZNT1,
which encodes a high-affinity Zn transporter that is a member of the ZIP (ZRT,
IRT-like Protein) family of metal transporters. They found that ZNT1 was highly
expressed in the roots of T. caerulescens even when the plants had a high Zn status,
whereas in T. arvense ZNT1 was expressed at far lower levels, and the expression
was stimulated by Zn deficiency. Both physiological and molecular studies show
that the Zn hyperaccumulator T. caerulescens has a constitutively high density
of Zn transporter(s) on root plasma membranes. Thus, at least for Zn and Cd
(see following), an uptake rate higher than that in non-accumulating plants is one
of the reasons for metal hyperaccumulation.
In the case of Cd, it is often assumed that its uptake is unspecific and inadvertent
via transporters for other essential nutrients. Members of the ZIP gene family are
capable of transporting metals including Fe, Zn, Mn, and Cd (Guerinot, 2000). The
Fe transporters such as IRT1 (a member of ZIP) and Nramp have been shown to
be capable of transporting several metals including Cd in Arabidopsis thaliana
(Eide et al., 1996; Korshunova et al., 1999; Thomine et al., 2000). Using point
mutation to substitute a single amino acid residue of IRT1, Rogers et al. (2000)
demonstrated altered selectivity of the transporter for different metals. In addition,
there is some evidence that Ca channels may also mediate transport of Cd across the
plasma membrane in wheat, albeit at a low level of affinity (Clemens et al., 1998).
In the Zn hyperaccumulator T. caerulescens (Prayon), Pence et al. (2000) found
that the Zn transporter ZNT1 can also mediate a low-affinity uptake of Cd. The high
expression of ZNT1 may explain the accumulation of Cd in the Prayon ecotype of
T. caerulescens. However, the level of Cd accumulation in the Prayon population
has been shown to be much lower than that in the Ganges population (Lombi
et al., 2000). This difference cannot be explained by the difference in Zn uptake,
because the two populations show similar characteristics of Zn uptake both in soil

experiments and in short-term hydroponic studies of
65
Zn influx (Lombi et al.,
2000, 2001a). We found that the V
max
for
109
Cd and the Cd concentration in xylem
sap were approximately fivefold higher in Ganges than in Prayon (Fig. 4). In
addition, at an equimolar concentration, Zn inhibited Cd uptake by Prayon but
not by Ganges (Lombi et al., 2001a). Furthermore, Ca and its channel blocker
La appeared to inhibit Cd uptake in the Prayon ecotype but not in the Ganges
ecotype (Zhao et al., unpublished results). These results provide physiological
12/05/2001 09:33 AM Agronomy-V. 75 PS097A-01.tex PS097A-01.xml APserialsv2(2000/12/19) Textures 2.0
16 S. P. McGRATH ET AL.
Figure 4 Concentration-dependent
109
Cd uptake kinetics in two contrasting populations of
T. caerulescens.
evidence that a high-affinity Cdtransporting systemexists in the Ganges population
of T. caerulescens. Iron deficiency further enhanced Cd uptake by the Ganges
population, but not by the Prayon population, and had little effect on Zn uptake in
both populations (Lombi et al., unpublished). However, whether or not the high-
affinity Cd transportingsystem observed inthe Gangespopulation belongs to one of
the metal transporters in the ZIP family (Guerinot, 2000) remains to be elucidated.
b. Nickel
Although Ni hyperaccumulators are the most numerous among plant species
known to show hyperaccumulation, the mechanisms of Ni uptake by these plants
are still poorly understood. The molecular mechanisms of Ni transport in Ni hy-
peraccumulators are largely unknown. By comparing the Ni hyperaccumulator

Thlaspi goesingense with the non-accumulator Thlaspi arvense, Kr¨amer et al.
(1997a) found that the rates of Ni uptake and root-to-shoot translocation were
similar between the two species, as long as the concentration of Ni in the uptake
solution was below the toxicity threshold. However, because T. goesingense was
much more tolerant to Ni than T. arvense, the former species could continue to take
up Ni when the concentration of Ni in the solution was high and above the toxicity
threshold for T. arvense. These authors suggested that Ni tolerance alone was suffi-
cient to explain the Ni hyperaccumulator phenotype observed in hydroponically
grown T. goesingense when compared with the Ni-sensitive non-hyperaccumulator
T. arvense. This conclusion is clearly very different from that for the Zn and
Cd hyperaccumulation in T. caerulescens. It also conflicts with the observation
that hyperaccumulation and tolerance are separate genetic traits (Macnair et al.,
1999). It is not known whether the conclusion drawn by Kr¨amer et al. (1997a) is
12/05/2001 09:33 AM Agronomy-V. 75 PS097A-01.tex PS097A-01.xml APserialsv2(2000/12/19) Textures 2.0
PHYTOREMEDIATION 17
also applicable to other Ni hyperaccumulators, particularly those in the Alyssum
genus. Although T. goesingense is a known Ni hyperaccumulator, the level of
Ni accumulation observed in this plant is lower than that observed in Alyssum.
For example, both Alyssum bertolonii and Alyssum lesbiacum readily accumulate
over 10,000 µgNig
−1
in the shoots in hydroponic experiments (Gabbrielli et al.,
1990; Kr¨amer et al., 1996), whereas T. goesingense tend to accumulate <10,000 µg
Ni g
−1
under similar conditions (Kr¨amer et al., 1997a). In a pot experiment, we also
found thatT. goesingense was less tolerant to Ni thanA. bertolonii and A. lesbiacum
(K¨upper et al., 2001).
2. Metal Translocation from Root to Shoot
One of the key features distinguishing metal hyperaccumulators from non-

hyperaccumulators is the extremely efficient translocation of metals from roots
to shoots. For example, translocation of
65
Zn from roots to shoots was approx-
imately 10-fold greater in T. caerulescens compared with the non-acumulator
T. arvense, over a 96-h uptake period (Lasat et al., 1996; Fig. 5). In contrast, the
roots of T. arvense retained more
65
Zn than the roots of T. caerulescens. This may
be partly explained by a smaller sequestration of metals in the root vacuoles of
hyperaccumulators than in non-hyperaccumulators (Lasat et al., 1998a). It is also
possible that hyperaccumulators have a more efficient xylem loading.
Translocation of Ni from roots to shoots may involve specific ligands in some
hyperaccumulator species. Kr¨amer et al. (1996) found that exposing several Ni
hyperaccumulator species of Alyssum to Ni elicited a large and proportional in-
crease in the levels of histidine in the xylem sap. Histidine in the xylem sap
was shown to be coordinated with Ni, although the concentration of histidine
was enough to complex only about 25% of the Ni in the xylem sap. A similar
response was observed in two Ni hyperaccumulator species outside the genus
Alyssum, Streptanthus polygaloides and Berkheya coddii (Smith et al., 1999). In
contrast, exposing the Zn hyperaccumulator Arabidopsis halleri to Zn, or the Mn
hyperaccumulator Grevillea exul var. exul to Mn, did not result in increased his-
tidine in the xylem saps. The histidine response may not be universal in all Ni
hyperaccumulator species. Persans et al. (1999) did not observe any Ni-inducible
responses in terms of histidine concentrations in the roots, shoots, and xylem sap of
T. goesingense, nor did they find any regulation by Ni of three cDNAs encoding
the enzymes involved in the histidine biosynthetic pathway.
3. Mechanisms of Metal Tolerance in Hyperaccumulator Plants
Thlaspi caerulescens and Arabidopsis halleri have been shown to accumulate
25,000–30,000 µgZng

−1
in the shoot dry matter without growth reduction or
showing phytotoxicity (Brown et al., 1995a; Shen et al., 1997; Zhao et al., 2000).
12/05/2001 09:33 AM Agronomy-V. 75 PS097A-01.tex PS097A-01.xml APserialsv2(2000/12/19) Textures 2.0
18 S. P. McGRATH ET AL.
Figure 5 Time course of
65
Zn accumulation in roots (a) and shoots (b) of T. caerulescens and
T. arvense. (From Lasat et al., 1996, with permission from the American Society of Plant Physiologists).
Similarly, Alyssum lesbiacum accumulated up to 20,000 µgNig
−1
in the shoots
without reduction in plant biomass (Kr¨amer et al., 1996). These concentrations
contrast dramaticallywith toxicitythresholds ofabout 500µgZng
−1
and 10–50 µg
Ni g
−1
in many crop species (Marschner, 1995). Clearly, metal hyperaccumulators
must also be hypertolerant to the metals they accumulate. Furthermore, whereas
metal excluders may be able to tolerate metals present in the substrate, metal hy-
peraccumulators must be tolerant to metals present both externally and internally.
Tolerance mechanisms employed by metal hyperaccumulators most likely involve
internal detoxification, which may be achieved through (1) cellular and subcellular
compartmentation and (2) complexation with ligands.
a. Cellular and Subcellular Compartmentation
Compartmentation of heavy metals may be achieved at both cellular and sub-
cellular levels. Studies using energy dispersive X-ray microanalysis (EDXMA)
12/05/2001 09:33 AM Agronomy-V. 75 PS097A-01.tex PS097A-01.xml APserialsv2(2000/12/19) Textures 2.0
PHYTOREMEDIATION 19

Figure 6 EDXMA line-scan of a cross section of a T. caerulescens leaf showing preferential
accumulation of Zn in the epidermal cells (top panel) and concentration of Zn in the single-cell saps
extracted from epidermal and mesophyll cells of leaves of T. caerulescens (bottom panel). (Adapted
from K¨upper et al., 1999, with permission from the American Society of Plant Physiologists).
clearly show that Zn is preferentially accumulated in the leaf epidermal cells in
Thlaspi caerulescens (K¨upper et al., 1999; V´azquez et al., 1992, 1994) (Fig. 6),
except forthe guard and subsidiary cells in the stomatal complex (Frey et al., 2000).
In contrast, P is preferentially distributed in the leaf mesophyll cells (K¨upper et al.,
1999), thus avoiding precipitation of zinc phosphate in leaves which would cause
12/05/2001 09:33 AM Agronomy-V. 75 PS097A-01.tex PS097A-01.xml APserialsv2(2000/12/19) Textures 2.0
20 S. P. McGRATH ET AL.
Pdeficiency (Zhao et al., 1998). Further studies using single-cell sap sampling
(K¨upper et al., 1999; Fig. 6) and ultrathin cryosectioning (Frey et al., 2000) in-
dicate that Zn is sequestered in the vacuoles in the epidermal cells. K¨upper et al.
(1999) also observed a strong correlation between the Zn concentration and the
size of leaf epidermal cells. Because larger cells tend to be more vacuolated, this
suggests that the size of vacuoles may be an important factor determining Zn
distribution in leaf tissues.
Preferential accumulation of Ni in the leaf epidermal cells has also been
observed in several Ni hyperaccumulators including Senecio coronatus (Mesjasz-
Przybylowicz et al., 1994), Alyssum lesbiacum (Kr¨amer et al., 1997b), four
other species of Alyssum, and three other Ni hyperaccumulators from Greece
(Psaras et al., 2000), Alyssum bertolonii, Alyssum lesbiacum, and Thlaspi
goesingense (K¨upper et al., 2001). All Alyssum species examined have dense non-
glandular stellate trichomes on the leaf surfaces. Kr¨amer et al. (1997b) found
high concentrations of Ni in the trichomes of A. lesbiacum. However, this was
not confirmed in a recent study, which showed that Ni is excluded from the
trichomes in all Alyssum species examined (Psaras et al., 2000). These differ-
ences may be caused by differences in methodology, particularly sample prepa-
ration. Methods used to dehydrate fresh plant tissues (e.g., freeze-substitution

or freeze-drying) prior to X-ray microanalysis (V´azquez et al., 1992, 1994)
or proton microproble analysis (Kr¨amer et al., 1997b; Mesjasz-Przybylowicz
et al., 1994) may cause redistribution or losses of elements, and thus artifacts.
In contrast, X-ray microanalysis conducted on frozen hydrated tissues (Frey
et al., 2000; K¨upper et al., 1999) is more likely to provide a true picture of
the patterns of elemental distribution in plant cells. However, one drawback
of the latter method is a lack of resolution for observation at the subcellular
level.
Preferential distribution of Zn and Ni in leaf epidermal cells and reduced ac-
cumulation of metals in the mesophyll cells and those in the stomatal complexes
may be useful mechanisms to protect photosynthesis, which takes place in the
mesophyll cells, and the function of stomata. However, this pattern of metal distri-
bution is not observed in the Zn hyperaccumulator Arabidopsis halleri. A. halleri
has trichomes on the leaf surface, and the basal compartments of the trichomes
are highly enriched with Zn (and Cd, if present) (K¨upper et al., 2000; Zhao et al.,
2000). But the epidermal cells other than trichomes in A. halleri leaves are very
small and do not accumulate Zn or Cd. Instead, with increasing Zn/Cd accumula-
tion, mesophyll cells (most likely the vacuoles) become an increasingly important
sink of the metals (K¨upper et al., 2000).
A common feature of the hyperaccumulators studied so far is the sequestration of
metals in the vacuoles (Frey et al., 2000; K¨upper et al., 1999, 2000; V´azquez et al.,
1992, 1994). The Ni hyperaccumulator T. goesingense was found to sequester more
Ni in the vacuoles in leaf cells than the non-accumulator T. arvense (Kr¨amer et al.,
12/05/2001 09:33 AM Agronomy-V. 75 PS097A-01.tex PS097A-01.xml APserialsv2(2000/12/19) Textures 2.0
PHYTOREMEDIATION 21
2000). The difference in vacuolar sequestration has been proposed as an important
reason for differences in tolerance to Ni between T. goesingense and T. arvense
(Kr¨amer et al., 2000). In the non-hyperaccumulator Silene vulgaris, Chardonnens
et al. (1999) also found that the naturally selected Zn-tolerant ecotype had an
enhanced Zn uptake across tonoplasts compared to the Zn-sensitive ecotype. This

suggests that metal transport across tonoplasts, and hence vacuolar sequestration,
may be a key component of metal tolerance mechanisms.
b. Complexation with Ligands
Complexation of metals with ligands results in decreased free ion activity and
thus decreased toxicity. There is evidence for the role of ligands in detoxifying
metals in hyperaccumulators. For example, histidine has been shown to be involved
in the tolerance of Ni by several Alyssum species, possibly through the formation
of the Ni–histidine complex in root cells and in xylem saps (Kr¨amer et al., 1996).
In many Ni hyperaccumulators, there is an apparent association of Ni and organic
acids such as citrate, malate, and malonate (Brooks, 1998). For example, Lee et al.
(1977, 1978) showed a strong correlation between Ni and citrate in the leaves of 17
plant species from New Caledonian which varied in their ability to accumulate Ni,
and that the purified aqueous extracts of these plants contained Ni as a Ni–citrate
complex. In the latex collected from Sebertia acuminata, citrate was found to com-
plex about 40% of the Ni present (Sagner et al., 1998). Ni–citrate complexes also
existed in the aqueous extracts of the leaves of the Philippine hyperaccumulators
Dichapetalum gelonioides, Phyllanthus palawanensis, and Walsura monophylla
(Homer et al., 1991b). In several Alyssum species from Europe, however, the con-
centrations of citrate were low, and malate and malonate appeared to be the main
counter ions for Ni in the aqueous extracts (Brooks, 1998; Brooks et al., 1981;
Lee et al., 1978). As pointed out by Brooks (1998), an apparent association of
Ni and organic acids in the aqueous extracts of plant leaves does not necessarily
mean that they were originally complexed together in the plant materials. It is also
important to remember that malate and malonate have relatively low affinity for
Ni and other heavy metals, and therefore the complexes of Ni and these organic
anions are not very stable. In comparison, citrate has a higher affinity for Ni and
other heavy metals, and the metal citrate complexes are expected to be stable in
the vacuoles which have a pH of about 5.5.
In the Zn hyperaccumulator Thlaspi caerulescens, Mathys (1977) proposed
that malate plays an important role in shuttling Zn from the cytoplasm to vac-

uoles. This hypothesis is mostly likely incorrect, because malate has a rather
low affinity for Zn and the Zn–malate complex would be unstable in cy-
toplasm. Analysis of leaves of Thlaspi caerulescens and Arabidopsis halleri
showed that malate and citrate concentrations did not vary greatly (Shen et al.,
1997; Zhao et al., 2000). High concentrations of malate in Thlaspi caerulescens
leaves appear to be a constitutive property, and may only explain the charge
12/05/2001 09:33 AM Agronomy-V. 75 PS097A-01.tex PS097A-01.xml APserialsv2(2000/12/19) Textures 2.0
22 S. P. McGRATH ET AL.
balance between cations and anions rather than a mechanism for metal complexa-
tion (Tolr`a et al., 1996). In the roots of Arabidopsis halleri, the concentration of
citrate was found to increase with increasing Zn (Zhao et al., 2000). Recently, Salt
et al. (1999) used X-ray absorption spectroscopy to investigate Zn binding ligands
in T. caerulescens. The X-ray absorption spectra of liquid N
2
frozen tissues were
compared to the spectra of a number of model compounds. They found that 38, 9,
16, and 12% of the total Zn (5050 µgg
−1
dry weight) in the shoots was complexed
with citrate, oxalate, histidine, and cell wall, respectively, with the remaining 26%
as aqueous free cation. The citrate–Zn complex also accounted for 21% of the to-
tal Zn in the xylem sap, with the remaining amount apparently present as the free
cation. In the roots of T. caerulescens that contained 1320 µgZng
−1
dry weight,
histidine complexed 70% of the total Zn. Malate was found not to be involved
in Zn complexation in the roots, shoots, or xylem sap. These results suggest that
histidine may be important in detoxifying Zn in the root cells; however, organic
acids, particularly citrate, may be involved in the root-to-shoot transport and stor-
age of Zn in the leaf vacuoles. It is important to point out that although organic

acids such as citrate may play a role in metal chelation and storage in vacuoles,
they account for neither the metal specificity nor the species specificity of metal
hyperaccumulation (Kr¨amer et al., 1996).
Little is known about the ligands responsible for chelating Cd, Cu, and Co
in metal hyperaccumulators. In the non-accumulator Arabidopsis thaliana, phy-
tochelatins play an important role in the binding of Cd and the tolerance to Cd
(Cobbett, 2000). It is not known whether phytochelatins are also involved in the
detoxification of Cd in Cd hyperaccumulators. There is no evidence for a role of
phytochelatins in Zn or Ni tolerance and hyperaccumulation (Sagner et al., 1998;
Salt and Kr¨amer, 2000). Kr¨amer et al. (1996) also observed an increase in the
concentration of histidine in the xylem sap of Alyssum lesbiacum on exposure to
Co, similar to the response to Ni.
4. Rhizosphere Aspects of Metal Acquisition
by Hyperaccumulator Plants
Rhizosphere processes are important aspects of plant nutrient acquisition, for
ions that primarily reach the root surface by diffusion (e.g., Fe, Zn, and P) and par-
ticularly under nutrient-limiting conditions (Hinsinger, 1998; Marschner, 1995).
In some cases, differences between plant species or between cultivars within the
same species in nutrient acquisition from soils have been attributed to their differ-
ent abilities to modify the rhizosphere microenvironment. Furthermore, the rooting
pattern and associations between roots and microorganisms may also be important
factors in determining nutrient acquisition by plants. Do heavy metal hyperac-
cumulators employ rhizosphere-related mechanisms to achieve or enhance metal
accumulation?
12/05/2001 09:33 AM Agronomy-V. 75 PS097A-01.tex PS097A-01.xml APserialsv2(2000/12/19) Textures 2.0
PHYTOREMEDIATION 23
Figure 7 Root proliferation of T. caerulescens grown in rhizoboxes of previously uncontaminated
soil with (+) or without (−) ZnO added at 1000 mg Zn kg
−1
soil. (Photos courtesy of S. N. Whiting,

The University of Melbourne Australia).
a. Rooting Pattern
Two recent studies (Schwartz et al., 1999; Whiting et al., 2000) provided an
important insight into the rooting pattern of the Zn/Cd hyperaccumulator Thlaspi
caerulescens. These studies showed that T. caerulescens responded positively to
Zn in soil. The roots of this plant predominantly colonized a zone of Zn-polluted
soil that was embedded in an uncontaminated agricultural soil (Schwartz et al.,
1999). The addition of sparingly soluble ZnO to one half of a rhizobox contain-
ing an uncontaminated soil stimulated proliferation of lateral roots in the metal-
amended half, whereas root biomass in the unamended half was reduced (Fig. 7)
(Whiting et al., 2000). The plants consistently allocated about 70% of their to-
tal root biomass and root length, and about 70% of the current assimilate (
14
C)
into the metal-enriched soil. When both halves of soil were either unamended
or amended with ZnO, even distribution of root biomass occurred. In contrast,
the non-hyperaccumulator T. arvense tended to restrict root growth in the metal-
enriched soil and proliferate in the unamended soil. Addition of CdS appeared to
have a similar stimulating effect on the root growth of a Cd hyperaccumulating
population of T. caerulescens (Whiting et al., 2000), whereas Pb had no clear
effect (Schwartz et al., 1999). These results suggest that roots of T. caerulescens
are able to sense and actively forage in the Zn- or Cd-rich patches in soil. Because