In-situ phytoextraction of Ni by a native population of Alyssum murale on an ultramafic site (Albania)

Abstract

Ultramafic outcrops are widespread in Albania and host several Ni hyperaccumulators (e.g., Alyssum murale Waldst. & Kit.). A field experiment was conducted in Pojske (Eastern Albania), a large ultramafic area in which native A. murale was cultivated. The experiment consisted in testing the phytoextraction potential of already installed natural vegetation
(including A. murale) on crop fields with or without suitable fertilisation. The area was divided into six 36-m2 plots, three of which were fertilised in April 2005 with (NPK+S). The soil (Magnesic Hypereutric Vertisol) was fully described
as well as the mineralogy of horizons and the localisation of Ni bearing phases (TEM-EDX and XRD). Ni availability was also
characterised by Isotopic Exchange Kinetics (IEK). The flora was fully described on both fertilised and unfertilised plots
and the plant composition (major and trace elements) and biomass (shoots) harvested individually were recorded.

The soil had mainly two Ni-bearing phases: high-Mg smectite (1.3% Ni) and serpentine (0.7% Ni), the first one being the source
of available Ni. Ni availability was extremely high according to IEK and confirmed by Ni contents in Trifolium nigriscens Viv. reaching 1,442mgkg−1 (A new hyperaccumulator?). Total biomass yields were 6.3tha−1 in fertilised plots and 3.2tha−1 in unfertilised plots with a highly significant effect: fertilisation increased dramatically the proportion of A. murale in the plots (2.6tha−1 vs. 0.2tha−1). Ni content in the shoots of A. murale reached 9,129mgkg−1 but metal concentration was not significantly affected by fertilisation. Phytoextracted Ni in total harvest reached 25kg
Niha−1 on the fertilised plots. It was significantly lower in unfertilised plots (3kg Niha−1). Extensive phytomining on such sites could be promising in the Albanian context by domesticating already installed natural
populations with fertilisation.

Full text

Plant Soil (2007) 293:79–89
DOI 10.1007/s11104-007-9245-1
123
REGULAR ARTICLE
In-situ phytoextraction of Ni by a native population
of Alyssum murale on an ultramaWc site (Albania)
Aida Bani · Guillaume Echevarria ·
Sulejman Sulçe · Jean Louis Morel · Alfred Mullai
Received: 17 October 2006 / Accepted: 12 March 2007 / Published online: 11 April 2007
© Springer Science+Business Media B.V. 2007
Abstract UltramaWc outcrops are widespread in
Albania and host several Ni hyperaccumulators (e.g.,
Alyssum murale Waldst. & Kit.). A Weld experiment
was conducted in Pojske (Eastern Albania), a large
ultramaWc area in which native A. murale was culti-
vated. The experiment consisted in testing the phy-
toextraction potential of already installed natural
vegetation (including A. murale) on crop Welds with
or without suitable fertilisation. The area was divided
into six 36-m2 plots, three of which were fertilised in
April 2005 with (NPK + S). The soil (Magnesic
Hypereutric Vertisol) was fully described as well as
the mineralogy of horizons and the localisation of Ni
bearing phases (TEM-EDX and XRD). Ni availability
was also characterised by Isotopic Exchange Kinetics
(IEK). The Xora was fully described on both fertilised
and unfertilised plots and the plant composition
(major and trace elements) and biomass (shoots) har-
vested individually were recorded.
The soil had mainly two Ni-bearing phases: high-
Mg smectite (1.3% Ni) and serpentine (0.7% Ni), the
Wrst one being the source of available Ni. Ni availabil-
ity was extremely high according to IEK and con-
Wrmed by Ni contents in Trifolium nigriscens Viv.
reaching 1,442 mg kg¡1 (A new hyperaccumulator?).
Total biomass yields were 6.3 t ha¡1 in fertilised plots
and 3.2 t ha¡1 in unfertilised plots with a highly sig-
niWcant eVect: fertilisation increased dramatically the
proportion of A. murale in the plots (2.6 t ha¡1 vs.
0.2 t ha¡1). Ni content in the shoots of A. murale
reached 9,129 mg kg¡1 but metal concentration was
not signiWcantly aVected by fertilisation. Phytoex-
tracted Ni in total harvest reached 25 kg Ni ha¡1 on
the fertilised plots. It was signiWcantly lower in unf-
ertilised plots (3 kg Ni ha¡1). Extensive phytomining
on such sites could be promising in the Albanian con-
text by domesticating already installed natural popu-
lations with fertilisation.
Keywords In-situ phytoremediation ·
Hyperaccumulator plant · Nickel · Phytomining ·
Bioavailability · Serpentine
Introduction
Serpentine substrate covers large areas in Balkans,
more than in any other part of Europe which could
justify the development of phytomining activities as an
alternative to local agriculture on such unproductive
A. Bani · S. Sulçe
Agro-Environmental Department,
Agricultural University of Tirana, Kamez, Albania
A. Bani · G. Echevarria (&) · J. L. Morel
Laboratoire Sols et Environnment,
INPL-ENSAIA/INRA UMR 1120, Nancy-Université,
2 Avenue de la Forêt de Haye, BP 172,
54505 Vandoeuvre-les-Nancy, France
e-mail: Guillaume.Echevarria@ensaia.inpl-nancy.fr
A. Mullai
Institute of Biological Research, Tirana, Albania

80 Plant Soil (2007) 293:79–89
123
lands. Biodiversity in the area is high, with a great
number of interesting local and regional endemics.
More than 300 endemic taxa occur on serpentine in
Balkans. The high number of endemics indicates the
importance of serpentine habitats as centers for Xoris-
tic diVerentiation and speciation. Serpentine areas in
Balkans exist in large blocks or as small outcrops.
UltramaWc substrates in Albania cover 30% of the
areas and they extend towards NC & SE. UltramaWc
outcrops host a rich vegetation (Shallari et al. 1998)
including a certain number of endemics and suben-
demics (Stevanovic et al. 2003) such as trans-regional
endemics (e.g., Polygonum albanicum, Bornmuellera
baldacii, Alyssum markgraWi or Viola dukadjinica) or
regional endemics (e.g., Alyssum smolikanum, Viola
albanica, Festucopsis serpentine).
Hyperaccumulation is a mechanism that is
believed to allow plants to survive on serpentine soils
(Brady et al. 2005). Hyperaccumulators are deWned as
plants that contain in their tissue more than
1,000 mg kg¡1 dry weight of Ni, Co, Cu, Cr, Pb, or
more than 10,000 mg kg¡1 dry weight of Zn, or Mn
(Baker and Brooks 1989). Aside from metal toler-
ance, hyperaccumulation is thought to beneWt the
plant by means of allelopathy, defense against herbi-
vores, or general pathogen resistance (Boyd and
JaVré, 2001; Boyd and Martens 1998; Davis et al.
2001). There are at least 400 known Ni hyperaccumu-
lators. The Ni hyperaccumulator Alyssum murale
Waldst. & Kit., is widely present on ultramaWc
regions and industrial areas of Albania. The need to
manage the Ni polluted soils necessitates the study of
the behaviour of this hyperaccumulator plant in con-
ditions in situ. Previous studies clearly evidenced a
great variability in phytoextraction potential by diVer-
ent Albanian populations of A. murale depending on
the site of collection (Chardot et al. 2005; Massoura
et al. 2004; Shallari et al. 1998). This variability prob-
ably results from the interrelationships between the
ecophysiology of speciWc populations and the edaphic
characteristics of their native site.
Phytoextraction employs metal hyperaccumulator
plant species to transport high quantities of metals
from soils into the harvestable parts of roots and
aboveground shoots (Kumar et al. 1995, Chaney et al.
1997). EVective phytoextraxtion requires both plant
genetic ability and the development of optimal agro-
nomic management practices (Li et al. 2000). It has
been well-documented that modifying soil fertility
may aVect the eYciency of phytoextraction of heavy
metals such as Ni, Zn, Co, Cd with a single crop
(Bennett et al. 1999; Li et al. 2003a; Kukier et al.
2004). But the eVects of fertilisation are less predict-
able when dealing with a native vegetation cover. In
the case of phytomining, the use of native Xora
(including local populations of hyperaccumulators)
with limited agronomic practices (extensive phytoex-
traction) could be an alternative to intensively man-
aged crops (Li et al. 2000) provided that Ni
bioavailability in soils is high and not limiting over
time and that hyperaccumulator cover is reasonably
eYcient. However, there is an evident need of fertili-
sation and phytoextraction yield improvement (Rob-
inson et al. 1997) to achieve suYcient Ni extraction.
Such extensive practices of phytoextraction could be
more easily implemented in a country such as Albania
(small surfaces, limited investment capacity of
farmers).
This paper reports the Wrst Wndings of an in situ
experimental work aiming at studying the response of
a native vegetation cover, including A. murale, to
agronomic practices in the view of optimising the
phytoextraction yield of Ni. The objectives were (i) to
optimise extensive phytoextraction methods that
would be adapted in the Albanian context and (ii) to
understand the relationships between the ecophysiol-
ogy and the mineral nutrition of stands of native A.
murale and other species, and the speciWc properties
of a soil which displays high Ni availability. The soil
of the site was therefore carefully studied to identify
the factors inXuencing plant growth and Ni uptake.
Then, 36-m2 experimental plots were designed to
study the eVect of mineral fertilisation on species dis-
tribution, speciWc biomass yield, mineral nutrition
and Ni phytoextraction.
Materials and methods
Site and soil characterisation and experimental
plot design
An in situ Weld experiment was conducted in Pojske
(500 m) (Pogradec, East of Albania), a wide ultra-
maWc area in which native A. murale populations
were grown. The experiment was carried out in spring
2005. The experimental area was a colluvial
downslope (10–15%) above the lake surrounded by

Plant Soil (2007) 293:79–89 81
123
ultramaWc hills. The parent material was colluvium of
ultramaWc and magnesite origin. Climatic data for
Pojske in 2005 are given in Table 1. The soil proWle
was described and samples were taken from the three
horizons identiWed in the Weld (0–20 cm, 20–35 cm,
>50 cm). Analyses of Ni-bearing phases with trans-
mission electron microscopy, coupled with X-ray
spectroscopy (TEM-EDX) techniques, were per-
formed to identify the minerals and their elemental
composition. X-Ray diVraction (XRD) was run on the
50 m fraction to determine mineralogy. Ni availabil-
ity in diVerent soil samples of surfaces was character-
ised by Isotopic Exchange Kinetics (IEK) (see
complete details and procedure in Echevarria et al.
1998). DTPA-extractable Ni were determined using
the method of Lindsay and Norvell (1978). Concen-
tration of Ni in soil extracts were determined by
plasma emission spectrometry (ICP).
The experimental site was already covered by
spontaneous native ultramaWc vegetation in March
2005 (an abandoned cropped Weld which had hardly
received any fertiliser in the past). The experimental
area was divided into six 36-m2 plots, three of which
were fertilised in April 2005 with the following
regime: 120 kg ha¡1 (of each of the following ele-
ments N, P, K and S) with NH4NO3, K2SO4 and CaH-
PO4. The rest of the plots did not receive any
fertiliser.
Soil surface samples (0–20 cm) were taken on
each plot before fertilisation, during vegetation period
(May) and after harvest (June), for full characterisation
(including mineral fertility and Ni availability). Soil
samples were air dried and sieved (2 mm) prior to
analyses.
Fertilisation pot trial
A greenhouse pot experiment was undertaken simul-
taneously in which A. murale was treated with diVer-
ent combinations of nutrients to determine the
optimal fertilisation regime on this soil and to orien-
tate the fertilisation on the Weld. Four addition treat-
ments of 0:0:0, 50:40:40, 100:80:80, 120:120:120 kg
of N:P:K per ha (1 kg per ha corresponds to
0.33 mg kg¡1. The experiment was conducted in a
well-ventilated unheated greenhouse. Soil from Ap
horizon in Pojske similar to that of the experimental
plot was sieved (2 mm) and airdried before experi-
ment. For each fertilisation treatment, four replicates
of 1-kg pots were made. Five plants were sown per
pot. Pots were daily watered to 100% of the water
holding capacity and cultivated for 3 months after
germination. After 3 months, plants were harvested
(cut at 1 cm above ground surface) and the plant sam-
ples were rinsed in deionised water before drying at
80°C for 24 h.
Plant identiWcation and harvest
on experimental plots
The Xora of each plot was fully described in June
2005 prior to harvest (with the help of European
Flora). About 12 plant species were sampled at site
(Table 5) and the plant composition was carefully
Table 1 Monthly climatic
parameters at Pojske
in 2005
Month Max
temperature
(°C)
Min
temperature
(°C)
Average
temperature
(°C)
Rainfall
(mm)
Potential
evapotranspiration
Penman (mm)
Jan 5.6 3.8 0.9 62.0 4.0
Feb 4.8 ¡5.3 ¡0.2 71.0 1.2
Mar 9.8 ¡1.5 4.1 77.4 16.5
Apr 14.6 2.5 8.6 33.9 34.5
May 21.4 7.4 14.4 55.1 57.2
Jun 24.0 9.5 16.8 30.2 67.2
Jul 26.8 11.9 19.4 19.5 77.2
Aug 27.0 9.8 18.4 13.5 73.2
Sep 22.3 7.5 14.9 47.3 59.2
Oct 16.4 3.2 9.8 61.4 40.0
Nov 11.3 2.3 6.8 80.4 27.0
Dec 7.7 ¡0.7 3.5 179.4 14.1

82 Plant Soil (2007) 293:79–89
123
recorded. Alyssum murale Waldst. & Kit., Chrysopo-
gon gryllus (L.) Trin., Trifolium nigriscens Viv. were
the most frequent species in site, the other species
were much rarer. The shoots of plants of the three
dominant species were individually collected and bio-
mass yields were carefully recorded for each plot. The
rest of the biomass from other species was pooled
together. Total root biomass (all species) was also
sampled on 900-cm2 areas located in the center of
each plot. For each plot and species (1-m2 surface in
the centre of the plot), plant samples were taken,
rinsed with deionised water and dried at 80°C for
24 h.
Plant sample mineralisation, elemental analyses and
statistical analysis of data
Trace metal contents in plants samples were analysed
by plasma emission (ICP) spectrometry after diges-
tion of plant samples in microwaves. A 0.25-g DM
plant aliquot was digested by adding 8 ml of 69%
HNO3, 2 ml H2O2. Solution were Wltered and adjusted
to 25 ml with 0.1 M HNO3. CertiWed reference mate-
rials (V463 maize) were analysed in order to control
the data quality. We have also double-checked the
high concentrations with dilutions of some Alyssum
samples.
Digestion of plants samples, to determine N con-
tents in harvested biomass, was made in digestion
tubes using 4 ml of 98% H2SO4, 1 g of potassium sul-
fate-catalyst mixture, 6 ml H2O2. Digestion tubes
were heated in digestion block to 350°. After diges-
tion, samples were diluted to 25 ml with deionised
water. Total N was determined using Kjeldhal
method. Analyses of variance and Newman–Keuls
Test were used to test signiWcance of the diVerences
between treatments.
Results
Edaphic properties of the site
The parent material is strongly weathered serpentine
mixed with colluvium. The pattern of pedogenesis is
strongly related to climate. The main phase in the
weathering process is recombination of Si and Mg to
smectite clays at the base of slopes, where excess
SiO2 recrystallises to form quartz and excess Mg is
precipitated. This phase occurs in dry conditions and
leads to the formation of soils that are hypermagnesic,
including vertisols dominated by smectites. Soils on
down slopes are less shallow (>50 cm) than upslope.
They are stony and have numerous fragments of
weathering rock on the surface. The clay content is
high, but there is still a substantial proportion of sand
(Table 2). The soil had a neutral pH (ranging from 6.9
to 7.3) and a high CEC, which was dominated by
Mg2+ ions: there was relatively a low concentration of
Ca2+. In the soil, total Ca concentration was 0.24–
0.32%, whilst total Mg concentration was 6.7–8.3%.
The Mg:Ca ratio was therefore extremely high: 29.
According to FAO WRBSR (1998) the soil at the
experimental site was classiWed as Magnesic (Hyper-
magnesic) Hypereutric Vertisol.
Localisation of Ni-bearing phases
and Ni availability
XRD analyses of the three horizons revealed that
chlorite, smectite and serpentine (undetermined type)
were the three predominant minerals in the soil. TEM
and EDX observations and analyses showed that the
last two were both Ni-bearing phases (Fig. 1). It was
also shown that serpentines were mainly of the antig-
orite or lizardite type with very scarce chrysotile par-
ticles. Chlorite and serpentine were the main primary
minerals occurring in soils derived from ultramaWc
rocks and smectite was the main secondary phyllosili-
cate occuring in soil. Two types of smectites could be
distinguished according to their Mg and Al content.
The average Ni concentration in Mg-rich smectites
was highest (1.3%). It was lower in serpentines
(0.7%) and in Al-rich smectites (0.5%) (Decarreau
et al. 1987) (Table 3). The highest Ni concentrations
were observed in some particles of a Mg-rich and Al-
poor smectite also rich in Mn (19.5%), thus reaching
4.9% Ni in weight (Fig. 1A).
Ni availability as assessed by IEK can be described
with the intensity factor (i.e., Ni concentration in
solution) on one hand and the labile pools in the solid
phase that are distributed according to their average
time of exchange with the soil solution (i.e., Et). Both
are important for the deWnition of the possibility of
root absorption of Ni by plants. Both intensity and
labile pools were extremely high in this soil (Table 4).
On average on the plots prior to fertilisation, CNi
values were 0.25 mg l¡1 and instantaneously

Plant Soil (2007) 293:79–89 83
123
exchangeable Ni (E0–1 min) was higher than
100mgkg
¡1. Medium term labile pools were also
high and all the parameters showed an extremely high
and well-buVered Ni availability in this soil (Echevar-
ria et al. 2006). The DTPA-extractable Ni has con-
Wrmed the results of IEK, The amount of Ni extracted
by DTPA roughly corresponds to the same amount
(Fig. 2) as the exchangeable Ni (E0–1 min). DTPA-
extractable Ni was higher after plant harvest than dur-
ing plant growth but this was only a trend with no sig-
niWcant diVerence.
EVect of fertilisation on biomass production
by A. murale and the other dominating species
A. murale (Brassicaceae), C. gryllus (Poaceae) and
T. nigriscens (Fabaceae) were the most frequent spe-
cies in this site but other species were reported on the
plots although their contribution to biomass produc-
tion was negligible (Table 5). The overall vegetation
responded dramatically to fertilisation by doubling
the biomass yield. The results showed that in ferti-
lised plots we obtained a total biomass of 6.3 t ha¡1
(dry weight) whereas in unfertilised plots it was only
of 3.2 t ha¡1 (Table 6) thus showing a highly signiW-
cant diVerence (P< 0.01) between fertilised and unf-
ertilised plots. However, the contribution of each of
the species varied according to fertilisation. In unf-
ertilised plots, C. gryllus accounted for most of the
biomass whereas, in fertilised plots, A. murale was
the main contributor. A. murale biomass abundance
was dramatically increased from 6.1% to 40.7%
whereas C. gryllus biomass abundance decreased
from 77.5% to 54.5%. T. nigriscens dramatically
decreased in abundance in fertilised plots (16.3–
4.8%). In greenhouse experiment we found that the
increase in biomass production (P<0.01) was
aVected by fertilisation and shoot biomass yield did
not aVect shoot Ni concentration, and therefore total
amount of phytoextracted Ni was increased. The sig-
niWcant diVerences were between unfertilised pots
and fertilised pots, but there were no signiWcant diVer-
ences among fertilised pots.
EVect of fertilisation on Ni concentration
in plant tissues and on phytoextraction yield
The chemical analyses of plant tissue showed the
diVerences between three dominant species of
Table 2 Physico-chemical characteristics of the three horizons of the soil at experimental site
ND: not determined
Particle size distribution (g kg¡1) pH (water) Organic C C/N CEC (Metson) Exchangeable cations Total major elements Total trace elements
Clay Silt Sand Ca2+ Mg2+ K+Na+Ca Mg Fe Co Cr Mn Ni
Horizon g kg¡1gkg
¡1cmol + kg¡1gkg
¡1Mg kg¡1
AP (0–25 cm) 447 239 314 6.9 27 10 39.2 5.39 41.3 0.51 0.07 2.9 72 98 215 1,420 2,260 3,440
BS (35–50 cm) 532 167 301 7.3 33 15 42.5 5.56 55.8 0.39 1.36 ND ND ND ND ND ND ND
BC 438 139 423 7.5 35 8.7 43.3 1.63 51.4 0.4 0.12 1.2 79 118 224 1,210 2,110 3,500

84 Plant Soil (2007) 293:79–89
123
experimental site. There were diVerences for nitrogen
concentration. The mean N concentration in A. murale
was 0.85% in unfertilised plots and 1% in fertilised
plots. C. gryllus had the lowest value 0.60%, whilst
T. nigriscens had the highest value 1.5%. Potassium
concentrations in A. murale shoots were 1.5% (previ-
ous study on Pojske plants collected close to the site).
Ca concentrations in shoots were about 0.6% for A.
murale whilst for C. gryllus the value was 0.24% and
for T. nigriscens it was 0.86% (Table 7). For Mg, the
mean concentration for A. murale was the lowest
(0.28%) in fertilised plots, for C. gryllus it reached
0.52% and the highest value was found in T. nigris-
cens (0.81%). A. murale showed a diVerent uptake
pattern in response to the high concentration of Mg
Fi
g.
1A
&
B
: TEM p
i
c-
tures (right) and EDX spec-
tra (left) of typical Mg-rich
smectite (A) and non-chrys-
otile serpentine (B) found in
the two lower soil horizons
of the Magnesic (Hypera-
magnesic) Hypereutric Ver-
tisol at experimental site in
Pojske (Albania)
Table 3 Ni concentrations in the main Ni-bearing minerals
of the soil (all three horizons) at experimental site. Results are
given as mean §standard deviation
Ni concentration (%)
Al-rich smectites (n= 10) 0.54 §0.18
Mg-rich smectites (n= 19) 1.30 §1.04
Serpentines (n= 16) 0.73 §0.34
Table 4 Ni available compartments in surface soils (averaged
sample) prior to culture as assessed by Isotopic Exchange Kinet-
ics Methods (IEK)
Total soil Ni (mg kg¡1)3,440
CNi (Water soluble Ni in mg l¡1)0.251
E
0–1 min (Ni exchangeable within
1 min in mg kg¡1)
124
E
1 min–24 h (Ni exchangeable between
1 min and 24 h in mg kg¡1)
119
E
24 h–3 mo (Ni exchangeable between
24 h and 3 mo in mg kg¡1)
119
E
3mo–1yr
(Ni exchangeable between
3 mo and 1 yr in mg kg¡1)
46

Plant Soil (2007) 293:79–89 85
123
and the low concentration of Ca in soils than the two
other species. It had a much lower Mg:Ca ratio in
shoots than C. gryllus and T. nigriscens. So A. murale
had a Mg:Ca quotient much lower than 1, the highest
being 2.2 for C. gryllus. Fertilisation tended to
decrease this ratio in all three species.
The mean Ni content in the above ground part of A.
murale was 9,129 mg Ni kg¡1 for fertilised plots and
8,483 mg Ni kg¡1 for unfertilised plots. The fertilisa-
tion increased slightly the Ni concentration in plant
tissue but with no signiWcant eVect. The Ni concentra-
tions in above ground parts of A. murale were on
average 2.9 times higher than in the soil. This ratio
(i.e., transfer factor) shows an indication of optimal
hyperaccumulation conditions due to high availability
of Ni. The higher concentration of Ni was found in
above ground parts of plants than roots, inverse was
found for Mn and Fe. It was observed that the concen-
tration of Ni in C. gryllus and T. nigriscens tissue was
extremely high (>1,000 mg kg¡1 in some plots for T.
nigriscens) for such plants that are not considered as
Ni-hyperaccumulators. The Cr contents in these two
plants was reasonably low (between 6 mg kg¡1 and
9mgkg
¡1). The hypothesis of soil particle contami-
nation of plant aerial parts was therefore discarded.
There were strong diVerences in Ni phytoextrac-
tion yield (Table 6) between fertilised plots and unf-
ertilised plots. The total phytoextraction yield (Sum
of phytoextraction by the three species) was 24.9 kg
Ni ha¡1 in the fertilised plots, whilst it was 3.1 kg
Ni ha¡1 in unfertilised plots with a highly signiWcant
diVerence (P< 0.001). The contribution of A. murale
was 22.6 kg ha¡1 (91%) in fertilised plots and that of
C. gryllus was 2.2kgha
¡1 (8.7%). In unfertilised
plots, the low total phytoextraction yield was
recorded with a much lower contribution of A. murale
(54%). The relative and net increase in biomass pro-
duction of A. murale was the main reason for increase
of phytoextraction yield since the Ni concentration in
shoots was not signiWcantly aVected by fertilisation.
The phytoextraction yield in pots in greenhouse
experiment was 9.9 mg Ni/pot in the unfertilised
treatment, 17.2 mg Ni/pot in the 50:40:40 kg ha¡1
NPK treatment, 19.9 mg Ni/pot in 100:80:80 kg
ha¡1NPK treatment, and 19.0 mg Ni/ pot in the
120:120:120 kg ha¡1 NPK treatment (Table 8).
Discussion
Soil characteristics and Ni availability to Alyssum
murale
The serpentine soil in Pojske is a clayey and Mg-satu-
rated soil that contains elevated levels of heavy met-
als, such as Mn, Ni, Cr, Co that are typical from
ultramaWc environments. The three soil horizons are
characterised by extremely high magnesium-to-cal-
cium ratios which are toxic to unadapted plant spe-
cies. In general, the Mg:Ca ratio observed in
ultramaWc soils was between 2.5 and 47 (Proctor
1971). Soil mineralogy was simple and homogenous
Fig. 2 Evolution of the DTPA-extractable Ni during culture in
surface soil samples of the fertilised and unfertilised plots.
There were no signiWcant diVerences between fertilised and unf-
ertilised soils and between May and June extractions (ANOVA,
P= 0.05)
0
20
40
60
80
100
120
140
1
Fertilised
)
l
ios gk/ gm( iN e
l
batcartxe-A
PTD
May June
Unfertilised
2
Table 5 List of plant species collected on the experimental
plots in late June 2005 prior to harvest
Species Family
Dominant species
A
lyssum murale Waldst. et Kit Poaceae
Chrysopogon gryllus (L.) Trin. Brassicaceae
Trifolium nigriscens Viv. Fabacaeae
Accompaning species
Cichorium intybus L. Compositae
L
otus corniculatus L. Fabacaeae
A
egilops geniculata Roth. Poaceae
Viola arvensis Murray Violaceae
Petrorhagia prolifera L. Caryophyllaceae
Vicia villosa Roth. Fabacaeae
D
asypyrum villosum (L.) P. Condargy Poaceae
Centaurea solstitialis L. Asteraceae
L
olium perenne L. Poaceae

86 Plant Soil (2007) 293:79–89
123
through the soil proWle. Although soil pH was quite
high (neutral to slightly alkaline), Ni availability in
this soil was very high and therefore mainly inXu-
enced by the Ni-bearing minerals. This is in agree-
ment with a previous study which showed that Ni
availability in the soil is generally much higher when
Ni is associated with poorly crystallised Fe oxides or
high charge phyllosilicates such as smectites despite
of high pH values (Massoura et al. 2006). We suggest
in the present case that the availability of Ni from pri-
mary clay minerals (e.g., serpentines) was low and
attributed to the presence of Ni in the crystal lattice.
Subsequently, the Ni in secondary clay minerals
(smectites) was probably sorbed onto the mineral sur-
faces or located on internal exchangeable sites, and
therefore, its availability was very high. This soil is
then highly suitable for proWtable phytoextraction
(e.g., phytomining) as it will supply over the long
T
a
bl
e
6
B
i
omass pro
d
uct
i
on an
d
p
h
ytoextract
i
on y
i
e
ld
(
per
ha) of the three main species grown on fertilised and unfertilised
plots. Values for each species within a plot are given as mean
values §standard deviation. For the sum of biomass and
p
h
ytoextract
i
on y
i
e
ld
(i
.e.,
‘
Tota
l’)
,
diV
erent
l
etters
i
n
di
cate sta-
tistical diVerence at the P< 0.01 level for biomass and at the
P< 0.001 for phytoextraction yield (ANOVA and Newman
–
Keuls Test)
Species Plots Biomass (t ha¡1) Ni phytoextraction yield
(kg ha¡1)
A
lyssum murale Fertilised 2.56 §0.70 22.60 §0.44
Chrysopogon gryllus 3.43 §0.35 2.17 §0.15
Trifolium nigriscens 0.30 §0.20 0.17 §0.12
Total 6.30a 24.93a
A
lyssum murale Unfertilised 0.20 §0.44 1.67 §0.12
Chrysopogon gryllus 2.53 §0.32 0.83 §0.49
Trifolium nigriscens 0.53 §0.12 0.60 §0.30
Total 3.27 b 3.10 b
Table 7 Concentration of Ni, Mn, Cr Fe, Ca and Mg in harvested parts of plants on fertilised and unfertilised plots. Results are given
as mean values §standard deviation as there were no signiWcant diVerences between fertilised and unfertilised plots (ANOVA)
Species Plots Ni (mg kg¡1)Mn (mgkg
¡1) Cr (mg kg¡1) Fe (%) Ca (%) Mg (%)
A
lyssum murale Fertilised 9,129 §1,275 10 §2 – 0.04 §0.04 0.60 §0.13 0.28 §0.07
Chrysopogon gryllus 648 §50 52 §25 8.7 §0.4 0.19 §0.09 0.24 §0.03 0.52 §0.08
Trifolium nigriscens 727 §497 22 §48.3§0.6 0.11 §0.05 0.86 §0.12 0.67 §0.18
Roots (all species) 1,132 §73 29 §1ND 0.13§0.06 0.31 §0.04 0.24 §0.06
A
lyssum murale Unfertilised 8,483 §477 5.5 §2.2 – 0.02 §0.02 0.57 §0.16 0.49 §0.14
Chrysopogon gryllus 337 §148 35.0 §17.4 7.5 §9.0 0.18 §0.08 0.17 §0.05 0.44 §0.12
Trifolium nigriscens 910 §585 24.7 §8.5 6.4 §3.1 0.11 §0.05 0.68 §0.05 0.81 §0.11
Roots (all species) 896 §199 35.0 §10.0 ND 0.21 §0.03 0.37 §0.02 0.30 §0.10
Table 8 Biomass production, Ni content in shoots of A. murale
and phytoextraction yield as aVected by fertilisation level in pot
experiments with the surface soil from Pojske. Values with the
same letter indicate no signiWcant diVerence (ANOVA, New-
man–Keuls test at the P= 0.05 level)
Fertilisation treatment (N:P:K) Biomass yield (g kg¡1) Ni concentration (mg kg¡1) Ni phytoextracted (mg kg¡1)
0:0:0 0.87 c 11,385 a 9.9 b
50:40:40 2.04 b 8,454 ab 17.2 a
100:80:80 2.95 a 6,772 b 19.9 a
120:120:120 3.04 a 6,261 b 19.0 a

Plant Soil (2007) 293:79–89 87
123
term large and accessible quantities of labile Ni with a
high buVer capacity (Echevarria et al. 2006; Li et al.
2000).
The presence of high Ca concentrations in soils
may inhibit both Ni and Co hyperaccumulation by
Alyssum. Acceptable Ca concentrations in soil were
reported to range from 0 to such a value that
exchangeable soil Ca is less than 20% of exchange-
able soil Mg. (Chaney et al. 1998). In our soil, the
Ca:Mg ratio was close to 13% in the surface horizon
and therefore matched these recommended values.
For phytomining with A. murale, the optimal pH was
deWned within the following range: 4.5–6.2; and pref-
erably between 5.2 and 6.2. In a previous study, it was
shown that pH favoured Ni phytoextraction yield by
A. murale on an industrially contaminated soil
(Kukier et al. 2004). But this eVect was not observed
during the same study on a serpentine soil in which
the dominant Ni-bearing phases seemed to be Fe
oxides. The phases that bear available Ni in our soil
(i.e., high-Mg smectites) are diVerent from that previ-
ous serpentine soil, and are not believed to strongly
modify their Ni retention capacity with pH. More-
over, the pH in the soil is neutral and the A. murale
population used here is native to this soil. We there-
fore believed that soil conditions were optimal for Ni
phytoextraction with A. murale.
EVect of fertilisation on species abundance
and biomass production
Alyssum murale, C. gryllus and T. nigriscens were the
most frequent spontaneous species in this site.
According to what was expected on such soils with
low potassium and phosphorus availability, the over-
all vegetation responded dramatically to fertilisation
by doubling the biomass yield. Unexpectedly, the
contribution of each of the species varied according to
fertilisation and A. murale was highly favoured by
fertilisation compared to the two other species. It is
obvious that N fertilisation tended to decrease the bio-
mass production of the Leguminosa in this competi-
tion context, however, the decrease of abundance of
the Poacea was a surprise. In this experimental study
of phytoextraction with natural hyperaccumulator
stands, the next step in improving hyperaccumulator
biomass would be to apply selective herbicides to
control the population of C. gryllus. This practice has
been implemented in 2006 and the plots have now all
been subdivided to include a new treatment with an
anti-monocots herbicide.
The fertilisation treatment seemed to control quite
eYciently the population of T. nigriscens and less that
of C. gryllus. In terms of ecological adaptation, A.
murale is therefore a more competitive species than
the two others. However, anti-monocots herbicide
treatments were included in the plots in 2006 to allow
for the full development of A. murale (unpublished
data). Soil management practices (fertilisation herbi-
cide treatment) have been applied from others in ser-
pentine and Ni contaminated soil to improve Ni
phytoextraction. Li et al. 2003b in developing a com-
mercial technology using hyperaccumulator plant
species (A. murale) to phytoextract Ni reached a bio-
mass of 20 t ha¡1. In our experiment (extensive tech-
nology), in which we stimulated natural vegetation
stands with fertilisation (and later with herbicide
treatment), A. murale biomass reached 3.7 t ha¡1 and
was therefore much lower than with intensive man-
agement (Li et al. 2003b) but we need more refer-
ences on several years before making reliable
comparisons with other in-situ experimental data.
Plant nutrition and Ni uptake by A. murale
Walker et al. (1954) surmised that serpentine-tolerant
species survive on soils with depleted levels of Ca
because they are still able to absorb quantities of Ca
without taking up excessive quantities of Mg. The Ca
concentration in plant tissue of A. murale was higher
than Mg concentration, whilst in the two other species
it was inverse. This conWrms the ability of A. murale
to accumulate Ca and its positive response to Ca fer-
tilisation (Kukier et al. 2004). Fertilisation seems to
improve plant uptake of Ca and to limit excess uptake
of Mg for all species. This is probably a reason for the
extreme increase in biomass yield after fertilisation. It
was observed that the concentration of Fe and Mn in
above ground parts of A. murale were lower in com-
parisons with two others species. In analysing tissues
of a number of serpentine endemics from Zimbabwe,
Brooks and Yang (1984) found the concentration of
Mg in plant tissue to be inversely proportional to the
concentration of other nutrients: Fe, Co, Mn. These
data suggest that the uptake of Mg comes at a cost to
the plant so that the uptake of other element nutrients
is forfeited. So, Brooks and Yang (1984) proposed
that the heightened level of Mg in serpentine soils and

88 Plant Soil (2007) 293:79–89
123
its antagonistic behaviour toward other elements
could be the most important factor in serpentine
syndrome.
The low Cr concentration in T. nigriscens and C.
gryllus and trace quantities of this element in A.
murale conWrmed Brooks (1987) Wndings, that ser-
pentine plants without exception contain only trace
quantities of Cr. The low Cr content in T. nigriscens
and C. gryllus and the high Ni:Cr ratio in plant tissues
(much higher than the soil Ni:Cr ratio) conWrmed the
root uptake origin of Ni in these plants. Ni uptake by
the two non-hyperaccumulator species was therefore
extremely and surprisingly high. In fact, T. nigriscens
should be added to the list of Ni hyperaccumulators
with a maximal concentration of 1,442 mg kg¡1
observed in this in-situ experiment. However, it is
clear from the bioavailability data that these two
species in other conditions could display much lower
Ni concentrations in their tissues as opposed to a truly
Ni hyperaccumulator such as A. murale. Neverthe-
less, C. gryllus contributed roughly to 10% of the Ni
phytoextraction yield in the fertilised plots. Ni
concentrations in A. murale were high, but not
excessively, with regard to the high level of Ni avail-
ability in this soil. The latter species is genetically
highly variable as it is widespread all over ultramaWcs
in the Balkans and Turkey. The Albanian populations
of A. murale collected in situ varied from 1,508 to
8,463 mg kg¡1 (Shallari et al. 1998) except in one site
in Prrenjas (Roger Reeves, personal communication).
For phytoextraction purposes, the reasons that deter-
mine the level concentration of Ni in the total biomass
harvested on the Weld should be investigated.
EVect of fertilisation on Ni phytoextracion yield
and perspectives for future research
Alyssum murale was evidenced with great phytoex-
traction potential in situations where native vegeta-
tion stand is enhanced with simple low-cost
agronomic actions (fertilisation). The fertilisation
nearly doubled the total biomass harvested but dra-
matically increased by 10- to 15-fold the phytoextrac-
tion yield. Shoot Ni in Weld-grown plants reached 1%
and was not aVected or diluted by fertilisation. Phy-
toextracted Ni in harvested biomass reached 28 kg
Ni ha¡1 in one of the three fertilised plots with an
average of 25 kg Ni ha¡1. With a price of Ni at 30
USD per kg, a reasonable income per ha would be
750 USD, provided that all the Ni contained in the
biomass is recovered. This is why we started investi-
gating the performance of low-cost phytoextraction
with limited agronomic actions adapted to the Alba-
nian context to see if extensive phytoextraction was
feasible as an alternative to intensively managed phy-
toextraction (Li et al. 2003b). The Wrst answer after
one year of experiment is positive, however, we need
to better understand the relationships between fertili-
sation and species distribution to clearly deWne pest
control and its consequences on phytoextraction yield
by stands of A. murale alone. Also, we need to
improve the level of Ni concentration in the shoots of
A. murale shoots through agronomic practices and
possibly to compare it with sown accessions of A.
murale.
Acknowledgement The Authors wish to thank the French
Embassy in Tirana (Albania) for the doctoral scholarship of
Aida Bani.
References
Baker AJM, Brooks RR (1989) Terrestrial higher plants which
hyperaccumulate metalic elements—A review of their dis-
tribution, ecology and phytochemistry. Biorecovery 1:81–
126
Brady KU, Kruckeberg AR, Bradshaw HD (2005) Evolutionary
ecology of plant adaptation to serpentine soils. 36:243–266
Bennett FA, Tyler EK, Brooks RR, Greg PEH, Stewart RB
(1999) Fertilisation of hyperaccumulators to enhance their
potential for phytoremediation and phytomining. In:
Brooks RR (ed) Plants that hyperaccumulate heavy metals.
CAB International, Wallingford, Oxon, UK, pp 249–259
Boyd RS, JaVré T (2001) Phytoenrichement of soil content by
Sebertia acuminata in New Caledonia and the concept of
elemental alelopathy. S Afri J Sci 97:535–538
Boyd RS, Martens SN (1998) The signiWcance of metal
hyperaccumulation for biotic interactions. Chemoecolo-
gy 8:1–7
Brooks RR, Yang XH (1984) Element levels and relationships
in the endemic serpentine Xora of the Great Dyke, Zimba-
bwe and their signiWcance as controlling factors for this Xo-
ra. Taxon 33:392–399
Brooks RR (1987) In: Dudley TR (ed) Serpentine and its vege-
tation. Dioscorides, Portland, OR, 454 pp
Chaney RL, Malik M, Li YM, Brown SL, Brewer EP, Angle JS,
Baker AJM (1997) Curr Opin Biotechnol 8:279–284
Chaney RL, Angle JS, Baker AJM, Li YM (1998) Method for
phytomining of nickel, cobalt and other metals from soils.
US Patent, No. 5,711,784
Chardot V, Massoura ST, Echevarria G, Reeves RD, Morel JL
(2005) Phytoextraction potential of the nickel hyperaccu-
mulators Leptoplax emarginata and Bornmuellera tymp-
haea. Int J Phytoremediation 7:323–335

Plant Soil (2007) 293:79–89 89
123
Davis MA, Boyd RS, Cane JH (2001) Host switching does not
circumvent the Ni-based defense of the Ni hyperaccumula-
tor streptanthus polygaloides (Brassicaceae). S Afr J Sci
97:554–557
Decarreau A, Colin F, Herbillon A, Manceau A, Nahon D, Pa-
quet H, Trauth-Badaud D, Trescases JJ (1987) Domain
segregation in Ni-Fe-Mg smectites. Clays Clay Miner
35:1–10
Echevarria G, Leclerc-Cessac E, Fardeau JC, Morel JL (1998)
Assessment of phytoavailability of Ni in soils. J Environ
Qual 27:1064–1070
Echevarria G, Massoura T, Sterckeman T, Morel JL (2006)
Assessment and control of the bioavalability of nickel in
soils. Environ Toxicol Chem 25:643–651
Kumar PBAN, Dushenkov V, Motto H, Raskin I (1995) Phy-
toextraction: the use of plants to remove heavy metals from
soils. Environ Sci Technol 29:1232–1238
Kukier U, Peters CA, Chaney JS, Angle JS, Roseberg RJ (2004)
The eVect of pH on metal accumulation in two Alyssum
species. J Environ Qual 32:2090–2102
Li YM, Chaney RL, Angle JS, Baker AJM (2000) Phytoremedi-
ation of heavy metal contaminated soilsK. In: Wise DL
(ed) Bioremediation of contaminated soils. Marcel Dekker,
New York, pp 837–884
Li YM, Chaney RL, Brewer E, Angle JS, Nelkin J (2003a) Phy-
toextractrion of nickel and cobalt by hyperaccumulator
Alyssum species grown on nickel-contaminated soils.
Environ Sci Technol 37:1463–1468
Li YM, Chaney RL, Brewer E, Roseberg R, Angle JS, Baker
AJM, Reeves RD, Nelkin J (2003b) Development of a
technology for commercial phytoextraction of nickel:
economic and technical considerations. Plant Soil
249:107–115
Lindsay WL, Norvell WA (1978) Development of DTPA soil
test for zinc, iron, manganese, and copper. Soil Sci Soc Am
J 42:421–428
Massoura ST, Echevarria G, Becquer T, Ghanbaja J, Leclerc-
Cessac E, Morel JL (2006) Nickel bearing phases and
availability in natural and anthropogenic soils. Geoderma
136:28–37
Massoura ST, Echevarria G, Leclerc-Cessac E, Morel JL (2004)
Response of excluder, indicator, and hyperaccumulator
plants to nickel availability in soils. Aust J Soil Res
42:933–938
Proctor J (1971) The plant ecology of serpentine. III. The inXu-
ence of a high magnesium/calcium ratio and high nickel
and chromium levels in some British and Swedish serpen-
tine soils. J Ecol 59:827–842
Robinson BH, Chiarucci A, Brooks RR (1997) The nickel hy-
peraccumulator plant Alyssum bertolonii as a potential
agent for phytoremediation and phytomining of nickel. J
Geochem Explor 59:75–86
Shallari S, Schwartz C, Hasko A, Morel JL (1998) Heavy metals
in soils and plants of serpentine and industrial sites of
Albania. Sci Total Environ 209:133–142
Stevanovic V, Tan K, Iatrou G (2003) Distribution of the en-
demic Balkan Xora on serpentine I—obligate serpentine
endemics. Plant Syst Evol 242:149–170
Walker RB (1954) The ecology of serpentine soils. II. Factors
aVecting plant growth on serpentine soils. Ecology
35:258–266