Environ. Sci. Technol.
2008, 42, 21622167
Phylogeny and Growth Strategy as
Predictors of Differences in Cobalt
Concentrations Between Plant
Species
N. J. WILLEY* AND J. WILKINS
Faculty of Applied Sciences, University of the West of
England, Coldharbour Lane, Frenchay,
Bristol BS16 1QY, United Kingdom
Received June 22, 2007. Revised manuscript received
December 3, 2007. Accepted December 17, 2007.
Analyses reported here quantify the contribution of plant
phylogeny and plant growth strategy to soil-to-plant transfer
of Co. Estimated relative mean (ERM) Co concentrations in shoots
of 241 species of flowering plant were derived using a
residual maximum likelihood (REML) analysis. There were
significant differences in, and a loge-normal frequency distribution
of, ERM Co concentrations between species. A significant
percentage of interspecies variance could be assigned
to taxonomic categories above the species, (Family and above
21.5%; Order and above 12.22%). Time-series analysis of
ERM Co concentrations ordered in the species-sequence of
the Angiosperm Phylogeny Group (APG II (2003)) revealed
significant autocorrelation with an increase from Commelinid
Monocot to Asterid Eudicot and a pronounced peak in the Core
Eudicots. ERM Co concentrations categorized by plant
growth strategy sensu Grime (2001) showed an increase
toward stress-tolerant strategies. Plant species are not, therefore,
independent units of Co concentration factors derived
from higher levels of biological organization exert significant
effects. These effects can provide the basis of new techniques
for selecting plant species for biotechnologies and for
predicting the exposure of organisms to Co. They show that
plant phylogeny and growth strategy might help refine predictions
of soil-to-plant transfer of a variety of pollutants, and suggest
research that might link molecular and higher level processes in
contaminated soil-plant systems.
Introduction
The stable isotope 59Co and the radioisotope 60Co are
pollutants of terrestrial ecosystems59Co via industrial
activities and 60Co from nuclear installations. 60Co is also an
important constituent of radioactive waste. The most com-
mon entry point of Co to terrestrial food chains is uptake
from soil through plant roots. Soil-to-plant transfer of Co is
the focus of continuing research for both nutrition (13) and
pollution/phytoremediation (4). There are refined predictions
of the availability of Co to plants from soils (57), and a
detailed understanding of the mechanisms mediating Co
uptake in roots (8), impacting on Co tolerance in plants (9)
and affecting compartmentation within shoots (2). There are
also established differences between plant species in Co
concentrations after similar root exposures (10). Such dif-
* Corresponding author e-mail:Neil.Willey@uwe.ac.uk.
2162 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 6, 2008
ferences contribute significantly to soil-to-plant transfer of
Co and, hence, both to the potential of species in phyto-
technologies and to ecotoxicological assessments of exposure
through foodchains. However, although soil availability and
root uptake of Co are being explored in detail, there is less
research focused on understanding interspecies differences
in Co concentration following similar exposure. In fact, there
are reported concentration data for only a small proportion
of the worlds domesticated food crops/cultivars and for a
very small proportion of the approximately 300 000 wild
species of flowering plants (angiosperms). There is no meth-
od to predict values for uninvestigated species or cultivars,
and models for pollution prevention and remediation are
based primarily on quantification of soil availability. This
neglect of interspecies differences arises because of the
difficulty in amassing data sets with sufficient interspecies
comparisons and the lack of identified plant factors on which
to base predictions.
There have been significant changes in the past decade
in understanding the phylogeny (evolutionary history) of
angiosperms (11, 12). It is now clear that differences between
plant species in many traits are affected by the phylogenetic
position of the species and that new phylogenies have clarified
knowledge of these effects. For example, and pertinently for
an inorganic pollutant such as Co, characteristic mineralogies
can be detected in certain clades (branches) of the an-
giosperm phylogeny (1316). Further, it has been shown that
there are phylogenetic influences on the concentration in
plants of the stable inorganic pollutants Cu, Zn, Ni, Pb, Cd,
and Cr (17) and the radioactive inorganic pollutants 134/137Cs
(18, 19), 89/90Sr (20), 109Ru (21), and 36Cl (22). These results,
based on ANOVAs coded with recent angiosperm phylog-
enies, have begun to provide an understanding of how the
effects of high-level biological organization constrain the
exposure of the biosphere to pollutants from the soil. It is an
understanding based on evolutionary influences that have
seldom been considered in environmental technologies or
assessments. An influence of phylogeny on 59Co concentra-
tions was detectable in a recent analysis of plant element
composition across four different field sites (16). Here we
focus on Co, and we include different Co isotopes, a
multispecies comparison from controlled conditions, and a
statistical analysis that provides a novel, graphical quanti-
fication of relative concentration across the flowering plant
phylogeny. Further, for Co and for almost all other elements,
there are no previous analyses comparing, as we do here, the
effect in intervarietal differences to those higher up the
phylogeny.
Grimes plant growth strategy theory describes, for plant
species in the established phase of growth, three primary
and four secondary growth strategies and makes some explicit
predictions about their influence on mineral concentrations
and ecosystem processes (23). There is now much empirical
evidence that Grimes growth strategy theory can be used to
make reasonable predictions of plant traits such as response
to nutrient supply and response to stress (24). In the aftermath
of the deposition of Chernobyl-derived radioactive fall-out
on Britain, Grime predicted the now widely acknowledged
persistence of 137Cs in British upland vegetation (25). It has
recently been shown that Grimes plant growth strategies
can be linked not just to the persistence of 137Cs but also to
the relative concentrations of 137Cs in different plant species,
and that together with the influence of phylogeny it can be
used to make general predictions of relative 137Cs concentra-
tions in angiosperms (19). 137Cs is the only pollutant for which
the influence of growth strategies on relative concentrations
10.1021/es071531r CCC: $40.75 2008 American Chemical Society
Published on Web 02/14/2008
has been investigated. In contrast to evolutionary effects
manifest through phylogeny, growth strategies reflect more
recent plant adaptations to particular types of environment.
They are, therefore, an additional aspect of high-level
biological organization superimposed on that of phylogeny.
Growth strategy theory provides a new test of the importance
of levels of organization above the species to determining
the transfer of pollutants in the soil-plant system.
To investigate the influence of phylogeny and growth
strategy on relative Co concentrations in angiosperms, a
database with concentration values for many species is
necessary. This is condition-specific if derived empirically in
one set of conditions. A statistical method has been developed
based on residual maximum likelihood (REML) analysis to
make estimates of relative concentrations in species from
different data sets. It provides databases that include
concentration data for sufficient species to enable the effects
of higher levels of biological organization to be tested (1322).
Here we report experiments that grew 32 plant species and
9 cultivars, analyzed their mean shoot Co concentrations,
and then used REML analysis to construct a database of this
experimental data plus similar literature data to give esti-
mated relative mean (ERM) Co concentrations for 241 species
of angiosperm. We use the database to test the hypotheses
that angiosperm phylogeny and plant growth strategy affect
Co uptake by plants, and we show that these factors have
sufficient influence to help refine predictions of the relative
concentrations of 59/60Co in the shoots of important categories
of flowering plants.
Experimental Methods
Experimental Data. Thirty-two species and nine cultivars
were chosen to complement those for which data existed in
the literature. The species included both domesticated and
wild species from across the angiosperm phylogeny (Table
S1) sourced from, respectively, Kings Seeds (Essex, United
Kingdom) and Chiltern Seeds (Cumbria, United Kingdom).
Five cultivars of Beta vulgaris (Italian Chard, Cheltenham
Green, Perpetual Spinach, Rhubarb Chard, Mangel Holm-Sidak posthoc tests. ERM concentrations included
Wurzel; Kings Seeds, Essex, United Kingdom) and 4 cultivars
of Cicer arietenum (G130, C214, H358, H208; ICRISAT, (27) (Table S1), which were averaged for the three primary
Patencheru, AP, India) were grown. For each of the 32 species
and 9 cultivars, five replicate 12 cm diameter pots with a
single plant were grown in Levingtons F2 compost in a
greenhouse with 22 C 16 h light/15 C 8 h night. Species/
varieties were grown in three successive batches with five
replicate pots of Aster Michaelmas Daisy in batches 1 and
2, and five replicate pots of Celosia x flamingo purple in
batches 2 and 3 providing link species. In each batch, species/
varieties were grown in a randomized block design. At 5 weeks
old, plants in each batch were radiolabeled in a randomized
block design in an arena supplied with supplementary light
at 350 E m-1 s-1, with 50 mL of 300 M CoCl2 radiolabeled
with 1125 kBq 57Co L-1 added to the substrate surface of each
pot, and harvested after 5 h. We assume that this saturated
the substrate homogenously because in almost every pot
this procedure resulted in excess solution that was caught
by saucers and often partially reabsorbed during radiola-
beling. Plant shoots were dried at 80 C, ground, and counted
for 57Co -emissions with appropriate blanks, standards, and
background correction in an LKB Wallac CompuGamma 1282
-counter (NaI(Tl) detector). Plants at radiolabeling were in
the exponential, established phase of their growth and had
not flowered.
Literature Data. We found 54 data sets from 28 publica-
tions in which concentrations (which ranged from 0.01 to
426 g/g) of a Co isotope in above-ground green shoots had
been compared in at least 2 species of angiosperm under
similar conditions (Table S1). Data sets in which foliar
contamination might have occurred were not included. The
literature provided data for 214 species. There were 5 species
in common with the experimental data, so there were data
for 241 species in all, consisting of 568 concentration values
from 3 + 54 ) 57 data sets.
Statistical Analyses. Two-way ANOVA of experimental
data using batch and species as factors was carried out on
SigmaStat 3.0. The program of previously reported analyses
for other pollutants including radionuclides (1822), heavy
metals (17), and inorganic nutrients (13, 14) was used to
relativise Co concentrations across data sets. A REML analysis
of loge-transformed concentration values relativizes data for
each species across the 54 literature data sets plus the 3
experimental batches by treating data sets as blocks (fixed
factors) and their 241 species as treatments. This was run in
Genstat for Windows 5th ed. release 4.2 (VAG International,
Oxford, United Kingdom) with units and nomenclature of
original authors and, as expected, produced relative con-
centrations for species that in some instances were positive
and in others negative (26). Blocking data sets in this way
adjusts the absolute differences in values arising from the
different edaphic conditions of each data set to provide a
database of ERM concentrations in treatments (species).
A KolmogorovSmirnov test was run on SigmaStat 3.0 for
Windows (Systat Software UK Ltd., Hounslow, United
Kingdom) to test for normality of ERM concentrations for
the 241 species. Time-series analysis in APG II (2003)
sequence of the ERM concentrations was performed in Systat
11.0 for Windows (Systat Software UK Ltd., Hounslow, United
Kingdom). ERM concentrations were coded using a recent
angiosperm phylogeny (APG II, 2003) (12) with 7 levels down
the taxonomic hierarchy (Class, Subclass, Group,
Superorder, Order, Family, Genus), and seven-way ANOVA
was run in Genstat for Windows 5th ed. release 4.2. The
taxonomic categories Class, Subclass, Group, and Superorder
were used in a nominal sense only because they derive from
a Linnaean hierarchy that does not have a clear relationship
with recent angiosperm phylogenies. A one-way ANOVA was
run on SigmaStat 3.0 for Windows on Group values, with
values for 26 exemplar species of Grimes growth strategies
strategies and four secondary strategies and were contour
plotted using SigmaPlot 9.0 for Windows (Systat Software
UK Ltd., Hounslow, United Kindom).
Results
ERM concentrations in the 241 species ranged from -6.31
(Arundinaria tecta) to 9.01 (Arenaria marcescens), with a
standard error of 0.16 (Table S1) and a frequency distribution
that was a negatively skewed bell-curve that failed the
Kolmogorv-Smirnov normality test (KS distribution ) 0.113,
P ) 0.001) (Figure 1). ERM concentrations for some species
were based on numerous absolute concentration values from
a variety of different environmental conditions whereas
others were made from a single absolute concentration value
(Table S1). However, despite the differences in the accuracy
of the estimates that this produced, the ERM concentrations
in Table S1, together with two-way ANOVA of experimental
data (F ) 4.67, P < 0.001; Figure 2), strongly indicate that
there are significant differences between species in relative
Co concentrations. Given that the absolute values used for
the estimates in Table S1 are loge-transformed prior to REML
analysis, it seems likely that, from a given Co availability in
soil, absolute interspecies differences in Co concentrations
in plants might vary by several orders of magnitude. There
was some difference between cultivars in Co concentrations,
but ANOVA indicated that these were not significant.
Time-series analysis of ERM concentrations in APG II
(2003) sequence showed significant autocorrelation (Figure
VOL. 42, NO. 6, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 2163
FIGURE 1. Frequency distribution of estimated relative mean Co FIGURE 2. Mean Co concentrations in green shoots of 32 species
concentrations in 241 species of flowering plant following loge- of flowering plants following 5 h of exposure to 50 mL of 300 M
transformation and REML analysis of 568 absolute concentration CoCl2 radiolabeled with 1125 kBq 57Co L-1 added to the substrate
values. surface of pot (n ) 5, (1 SE). Species numbers: 1, Allium sativum;
2, Allium cepa ailsa craig; 3, Cucumis sativus telegraph
improved; 4, Amaranthus paniculatus; 5, Mentha spicata; 6, Violax
TABLE 1. Results of ANOVA for Mean Relative Concentrations Swiss giants mixed; 7, Borago officinalis bianca; 8, Gallium
of Co in 241 Species of Plants Coded Using the Phylogeny of vernum; 9, Dianthus caryophyllus giant chabaud; 10, Antirrhinum
APG II (2003)a scarlet giant; 11, Aster michaelmas daises mixed; 12, Euphorbia
lathyrus; 13, Papaver commutatum; 14, Helianthemum numularium;
Cumulative 15, Linum lewisii; 16, Fraxinus excelsior; 17, Aesculus hippo-
df SS x 106 %SS MS x 106 VR castenum; 18, Veronica spicata; 19, Commelina coelestris; 20,
Rhuem tataricum; 21, Begonia cordifolia; 22, Fragaria vesca; 23,
Class 2 3.17 1.29 1.58 1.29 Phoenix dactylifera; 24, Juncus effusus; 25, Juncus ensifolius; 26,
Subclass 1 0.49 1.49 0.49 0.40 Carex nigra; 27, Lolium perenne; 28, Zingerber officinarum; 29,
Group 3 19.00 9.22 6.33 5.17 Celosia x flamingo purple; 30, Digitalis purpurea; 31, Beta
Superorder 6 4.06 10.87 0.68 0.55 vulgaris; 32, Cicer arietinum. (F ) 4.67, P < 0.001 for ANOVA of
Order 15 3.31 12.22 0.22 0.18 species).
Family 20 22.87 21.53 1.14 0.93
Genus 118 100.90 62.60 0.86 0.70
residual 75 91.94 100.02 1.23
total 240 245.70
a
df, degrees of freedom; SS, sums of squares; MS,
mean square; VR, variance ratio.

3 inset) indicating that species-values were not sequentially

independent but were correlated to nearby values. Nearby
values derive from species that are phylogenetically related.
ANOVA coded with a recent angiosperm phylogeny (APG II,
2003) for ERM concentrations in 241 species showed that
there was significant variation at the genus level and above,
that is, species are not independent samples for Co con-
centration but have differences that depend to an extent on
phylogenetic factors. About 12% of variance was at the level
of Order and above, there being especially significant effects
at the level of the Group (VR ) 5.17; Table 1). Figure 3
describes an upward trend in ERM concentrations but the
most notable feature is the high ERM concentrations in the
core eudicot Group (Figure 4).
There were 26 species in the database reported here whose
growth strategies have been defined experimentally (27). FIGURE 3. Estimated relative mean Co concentrations in 241
When plotted onto the triangular representation of growth species of plants from REML analysis of 568 concentration values
strategies used by Grime (25, 27), there is a trend toward plotted in the species order of Angiosperm Phylogeny Group II
high ERM Co concentrations in plants of the stress-tolerator (2003), smoothed with a moving mean over 10 values and dashed
lines showing (3 SE of overall mean. (Inset: Results of
strategy (Figure 4).
autocorrelation using time-series analysis on plotted values with
P ) 0.05 thresholds up to lag of 50).
Discussion
To describe interspecies differences in Co concentrations by composed of three data sets of experimental concentration
simultaneously measuring uptake in replicates of hundreds measurements derived to complement 54 data sets previously
of species is time-consuming and impractical. Further, if reported in the literature. Data has been relativized from
identical conditions can be attained for all species in such different numbers of replications across a variety of condi-
an experiment, then it would generate results that are specific tions (Table S1). Further, plant exposure time to Co differs
to that set of conditions. The database reported here is between data sets (there being a bias toward acute exposures),

2164 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 6, 2008


FIGURE 4. Average estimated relative mean Co concentrations in
plants of three primary (C ) Competitor, S ) Stress-tolerator, R )
Ruderal) and four secondary growth strategies sensu Grime (2001)
contour plotted onto Grimes triangular representation of growth
strategies. (Carex nigra (SC/S), Carex panacea (S) Juncus effusus
(C/SC), Anthoxanthum odoratum (SR/CSR), Dactylis glomerata (CSR/
C), Deschampsia caespitosa (CSR/S), Lolium perenne (CR/CSR),
Molinia caerulea (SC), Phleum pratense (S/CSR), Poa pratensis
(CSR), Poa trivialis (CSR/CR), Chenopodium album (R/CR), Trifolium
pratense (CSR), Trifolium repens (CSR/CR), Fragaria vesca (CSR),
Sanguisorba minor (S), Helianthemum nummularium (S), Calluna
vulgaris (SC), Erica cinerea (S), Vaccinium myrtillus (SC), Galium
verum (CSR/S), Fraxinus excelsior (C), Plantago lanceolata (CSR),
Digitalis purpurea (CR/CSR), Daucus carota carota (SR/CSR),
Campanula rotundifolia (S)) [Species with growth strategies in
zone A tend to accumulate nutrient resources; species in zone D
tend to convert nutrient resources into growth].
as does plant age and type (there being a bias toward young,
herbaceous nonaquatic plants in the database). These
differences in conditions, exposure, and plant status are the
weakness and the strength of the database because, respec-
tively, accuracy of the estimates of mean relative concentra-
tions varies but estimates are mean values across a range of
experimental/plant combinations and thus provide general
estimates across a variety of conditions. There is also a
limitation with the REML procedure described; it does not
take account of any interactions between interspecies
differences and environmental conditions, potentially con-
founding the estimation of relative mean concentration. If
relative concentrations change with conditions, then variance
would increase and make it less likely that we would find the
effects reported here. If there is a correlation between
phylogeny and environmental conditions (which might arise
through adaptation of taxa to particular soil types for
example), then the effects described might not be phylo-
genetic. However, there is much data from controlled
conditions in Table 1, and based on field samples, it has
been reported that such correlations are not significant in
any case (16). So, despite the above limitations, we suggest
that the effects detected are real but that they apply most
securely to young herbaceous plants acutely exposed to Co.
This is itself frequently an important pattern of exposure to
Co, but given that most plants take up the majority of nutrient
ions during the exponential growth phase (28), in which most
of the plants in the database were exposed, it is likely also
to reflect longer term exposure patterns.
Analyses of the experimental data, the largest interspecies
comparison yet reported for Co under controlled conditions,
confirm that there are significant differences between species
in Co concentration after identical exposures. Intervarietal
comparisons suggest that if such differences exist then they
are of lower magnitude than interspecies differences. This
accords with recent reports that, in contrast to other macro-
and micronutrients, there are no intervarietal differences in
Lens culinaris in Co concentrations after similar exposures
(29). So, the great range of Co concentrations reported here
suggests that, even if Co is highly available in a soil, there are
some plant species that will take it up to relatively low
concentrations, that is, there is a significant plant-derived
contribution to difference in soil-to-plant transfer. To make
accurate predictions of soil-to-plant transfer of 59/60Co for all
combinations of soil type and plant species it is, therefore,
necessary to be able to predict not just availability in a soil
but also the species contribution to differences in transfer.
Figure 1 suggests that although ERM Co concentrations fail
the test of normality they have an essentially normal but
skewed distribution. In fact, the database includes a few
species with particularly high concentrations (Supporting
Information Table S1; Figure 1), and although there are no
Co hyperaccumulating species of plants included in Table
S1, a few species are skewing the frequency distribution.
Figure 1 therefore suggests that concentrations in different
plant species in the field after similar exposure will be
approximately loge-normally distributed. This is the frequency
distribution reported from field collected samples in the only
other study that has described it (16). Knowledge of the
attributes of this frequency distribution is useful for modeling
differences, especially using probabilistic modeling, in soil-
to-plant transfer and has been noted for some other
radionuclides, although for many fewer plant species (30).
It is part of an increasing body of evidence that loge-normal
frequency distributions characterize soil-to-plant transfer
across numerous plant species for many elements (16) and
pollutants (1922).
There are at present no methods to predict relative
concentrations of Co in plant species. To do this it is useful
to isolate plant factors that explain some of the variance in
ERM Co concentrations. Figure 3 and Table 1 show that there
is a significant phylogenetic signal in ERM Co concentration
that might aid the prediction of relative Co concentrations
in plants. At the Ordinal level and above, with 12% of the
cumulative sum of squares, Co concentrations have phylo-
genetic effects greater than those on N (3.3%) and P (6.8%)
(14), similar to those on Cs (15%) (19) and Sr (15%) (20), but
less than those for Pb (20%), Cr (23%), Cu (24%) (16), Na
(23%) (13), Cd (27%) (15), Cl (35%) (20), Ru (39%) (19), Zn
(44%), Ni (46%) (15), K (49%) (14), or Ca (63%) (13). At the
family level and above, 21.5% of the variance is associated
with phylogeny (Table 1) a value very nearly the same as
that reported (20.7%) in the only other report of effects of
phylogeny on Co concentrations in plants (16). Of the
taxonomic categories above the species, there is reasonable
replication in the data reported here in each of the Group
categories, and it seems very clear (Figure 3) that, in general
terms, the Commelinid Monocots have lower than average
relative mean Co concentrations in them, that Core Eudicots
have higher than average, and Rosids have average concentra-
tions. One-way ANOVA of Group values confirmed the existence
of significant differences between Groups (F ) 14.77, P < 0.001)
with both Core Eudicots and Asterids having significantly higher
concentrations than non-Commelinid Monocots, Commelinid
Monots, and Rosids (Holm-Sidak posthoc test at P ) 0.05).
Interestingly, the phylogeny of protein transporters that mediate
the uptake of ions in plant roots is currently being described
(e.g., via PlantsT database: http://plantst.genomics.purdue.edu/
plantst/html/phylo.html). If phylogenetic effects in relative Co
concentrations and the phylogeny of transporters that mediate
Co uptake become known in sufficient detail, then they might
VOL. 42, NO. 6, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 2165
provide a useful test of the links between molecular effects and
those at higher levels of biological organization.
The three primary and four secondary growth strategies
defined by Grime (23) have mineral accumulation strategies
as an axis of specialization, primarily because nutrient
deficiency is an important stress to plants. Grimes growth
strategy theory has recently been reported to be a better
predictor of response to stress than those based on resource-
ratio hypotheses (24). Given that many inorganic pollutant
ions are chemically closely related to plant mineral nutrients,
or even are plant mineral nutrients but in above optimal
amounts, it seems likely that the growth strategies of plants
sensu Grime might influence the uptake of inorganic
pollutants by plants. The concentrations of Cs in flowering
plants have been shown to be influenced by plant growth
strategy, with stress-tolerant ruderals having the highest
concentrations (19). Figure 4 here suggests that stress tolerant
plants have the highest relative concentrations of Co. It is
notable that for neither Cs nor Co do plants of competitive
strategies have high concentrations but rather those that fall
within zone A of the growth strategy triangle where there is
a tendency to accumulate nutrient resources. Growth strategy
theory can be used to predict numerous aspects of ecosystem
processes, including those that impact on nutrient dynamics
(23, 24); linking pollutant behavior to plant growth strategy
might thus be the basis of improved predictions of pollutant
dynamics in ecosystems. It is, however, necessary to know
growth strategies of many plant species to do this in a variety
of ecosystems. The growth strategies of 286 ordinal species
has been defined using an extensive set of experiments (27),
but simpler methods have been developed that, in theory,
enable the growth strategy of many more plant species to be
predicted (31). In the 54 literature data sets used here, values
are not given for the variables necessary to predict growth
strategies. However, the link between growth strategy and
Co concentration we have established shows that predicted
growth strategies can provide a new perspective on the effects
of high levels of biological organization on pollutant move-
ment in the soil-plant system.
We conclude that, from a given set of environmental
conditions, there are significant interspecies differences in
the concentrations to which plants accumulate Co and that
a significant proportion of these differences can be predicted
from plant phylogeny. Further, we suggest that much of the
remaining variance should not just be ascribed to site
conditions but that other plant factors such as growth strategy
can be significant. We predict that, for a given soil Co
availability, plant species on the Core Eudicot clades of the
stress-tolerant growth strategy will have the highest Co
concentrations and that Commelinid Monocots of the
competitive strategy will have the lowest. We suggest that
the magnitude of variation in relative concentrations, the
frequency distribution, and the effects of phylogeny and
growth strategy reported will be useful in predicting the plant-
derived differences in soil-to-plant transfer of Co. They are,
therefore, a useful complement to models that predict the
availability of Co in soil and the compartmentation of Co
within shoots. For a number of inorganic heavy metal and
radioactive pollutants it is now established that phylogenetic
effects in plant concentrations exist and that for some
inorganic pollutants there are links to growth strategies.
Differences in plant traits impacting on organic pollutant
degradation and uptake have been investigated for genera
within a family (29), between species (30), and between
varieties (31) but not yet within recent phylogenetic frame-
works. We suggest that the influence of phylogeny and growth
strategy on plant traits impacting on many pollutants,
including organic pollutants, is worth further investigation.
As the foundation of terrestrial foodchains, the soil-plant
system plays a key role in determining the exposure of
2166 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 6, 2008
terrestrial organisms to pollutants. The ability to predict
pollutant behavior in the soil-plant system is crucial not
just for calculating exposure but also to plant-based envi-
ronmental technologies. Such technologies are rapidly
becoming part of the tool-kit of environmental engineers
(35, 36). There is more knowledge of pollutant behavior in
soils than of the influence of plant characteristics, especially
of factors related to higher levels of biological organization
in plants. This is unfortunate because plant factors have often
been shown to significantly influence soil-to-plant transfer
of pollutants. The methods we report here and the effects we
establish show that influential factors derived from high levels
of biological organization can be identified and that they
can be useful for quantifying the contribution that any plant
taxon might make to soil-to-plant transfer of a pollutant.
This can help to refine predictions of the exposure of
organisms and provides a basis for species selection for plant-
based environmental technologies. For environmental bio-
technologies, biodiversity is an exploitable resource, but most
species used are selected on the basis of their occurrence at
contaminated sites. The plant factors identified here might
help mine plant biodiversity for taxonomic units, and
ultimately for genes, useful to plant-based environmental
technologies. They reflect evolutionary processes and growth
strategies linked to vegetation processes/ecosystem proper-
ties and are thus a reminder that fundamental mechanisms
important to ecotoxicology and environmental health occur
at high levels of biological organization.
Acknowledgments
We would like to thank the UK Food Standards Agency for
funding, Judy Brown of UWE, Bristol for help with radio-
analysis, and A. Mead of Warwick HRI for developing the
REML program.
Supporting Information Available
Table S1 provides ERM Co concentrations for all 241 species
in phylogenetic layout and includes details of all data sources
and data extracted from them. This material is available free
of charge via the Internet at http://pubs.acs.org.
Literature Cited
(1) Shinonga, T.; Ambe, S. Translocation of radionuclides of Co,
Zn, Se, Rb, Y, Tc and Re into organs of tomato plants via roots.
Biol. Trace El. Res. 2005, 104, 7182.
(2) Riesen, O.; Feller, U. Redistribution of nickel, cobalt, manganese,
zinc and cadmium via the phloem in young and maturing wheat.
J. Plant Nutr. 2005, 28, 421430.
(3) Page, V.; Feller, U. Selective transport of zinc, manganese, nickel,
cobalt and cadmium in the root system and transfer to the
leaves in young wheat plants. Ann. Bot. 2005, 96, 425434.
(4) Pilon-Smits, E. Phytoremediation. Ann. Rev. Plant Biol. 2005,
56, 1539.
(5) Ohnuki, T.; Tanaka, T. Migration of radionuclides controlled by
several different migration mechanisms through a sandy soil
layer. Health Phys. 1989, 56, 4753.
(6) Schimmack, W.; Bunzl, K.; Dietl, F.; Klotz, D. Infiltration of
radionculides with low mobility (137Cs and 60Co) into a forest
soil: Effect of the irrigation intensity. J. Environ. Radioact. 1994,
24, 5363.
(7) Yasuda, H.; Uchida, S.; Muramatsu, Y.; Yoshida, S. Sorption of
manganese, cobalt, zinc, strontium and cesium onto agricultural
soils: statistical analysis on effects of soil properties. Water, Air,
Soil Pollut. 1995, 83, 8596.
(8) Williams, L.; Mills, R. F. P-1B-ATPases an ancient family of
transition metal pumps with diverse functions in plants. Trends
Plant Sci. 2005, 10, 491502.
 (9) Arrivault, S.; Senger, T.; Krmer. The Arabidopsis metal tolerance
protein AtMTP3 maintains metal homeostasis by mediating Zn
exclusion from the shoot under Fe deficiency and Zn oversupply.
Plant J. 2006, 46, 861879.
(10) Gerzabek, M. H.; Strebl, F.; Temmel, B. Plant uptake of
radionuclides in lysimeter experiments. Environ. Pollut. 1998,
99, 93103.
(11) Soltis, P. S.; Soltis, D. E.; Chase, M. W. Angiosperm phylogeny
inferred from multiple genes as a research tool for comparative
biology. Nature 1999, 402, 402404.
(12) The Angiosperm Phylogeny Group. An update of the Angiosperm
Phylogeny Group classification for the orders and families of
flowering plants: APG II. Bot. J. Linn. Soc. 2003, 141, 399436.
(13) Broadley, M. R.; Bowen, H. C.; Cotterill, H. L.; Hammond, J. P.;
Meacham, M. C.; Mead, A.; White, P. J. Variation in the shoot
calcium content of angiosperms. J. Exp. Bot. 2003, 54, 116.
(14) Broadley, M. R.; Bowen, H. C.; Cotterill, H. L.; Hammond, J. P.;
Meacham, M. C.; Mead, A.; White, P. J. Phylogenetic variation
in the shoot mineral concentration of angiosperms. J. Exp. Bot.
2004, 55, 321336.
(15) White, P. J.; Broadley, M. R. Biofortifying crops with essential
mineral elements. Trends Plant Sci. 2005, 10, 586593.
(16) Watanabe, T.; Broadley, M. R.; Jansen, S.; White, P. J.; Takada,
J.; Satake, K.; Takamatsu, T.; Tuah, S. J.; Osaki, M. Evolutionary
control of leaf element composition in plants. New Phytol. 2007,
174, 516523.
(17) Broadley, M. R.; Willey, N. J.; Wilkins, J.; Baker, A. J. M.; Mead,
A.; White, P. J. Phylogenetic variation in heavy metal accumula-
tion in angiosperms. New Phytol. 2001, 152, 927.
(18) Broadley, M. R.; Willey, N. J.; Mead, A. A method to assess
taxonomic variation in shoot caesium concentrations among
flowering plants. Environ. Pollut. 1999, 106, 341349.
(19) Willey, N. J.; Fawcett, K. Species selection for phytoremediation
of 36Cl/35Cl using angiosperm phylogeny and inter-taxa dif-
ferences in uptake. Int. J. Phytorem. 2005, 7, 295306.
(20) Willey, N. J.; Fawcett, K. A phylogenetic effect on strontium
concentrations in angiosperms. Environ. Exp. Bot. 2006, 58,
258269.
(21) Willey, N. J.; Fawcett, K. Inter-taxa differences in root uptake
of 103/106Ru by plants. J. Environ. Radioact. 2006, 86, 227240.
(22) Willey, N. J.; Tang, S.; Watt, N. Predicting inter-taxa differences
in plant uptake of 134/137Cs. J. Environ. Qual. 2005, 34, 1478
1489.
(23) Grime, J. P. Plant Strategies, Vegetation Processes and Ecosystem
Properties, 2nd ed.; John Wiley and Sons: Chichester, United
Kingdom; 2003.
(24) Pierce, S.; Vianelli, A.; Cerabolini, B. From ancient genes to
modern communities: the cellular stress response and evolution
of plant strategies. Func. Ecol. 2005, 19, 763776.
(25) Grime, J. P. In Chernobyl: The Governments Reaction Vol II.
Minutes of Evidence and Appendices: Evidence to Agriculture
Select Committee. Her Majestys Stationery Office: London, 1988;
399, 403..
(26) Thompson, R.; Welham, S. J. REML analysis of mixed models,
In The Guide to GenstatPart 2. Statistics. Payne, R. W., Ed.;
VSN International: Oxford, United Kingdom, 2001; pp 413-503.
(27) Grime, J. P.; Hodgson, J. G.; Hunt, R. Comparative Plant Ecology;
Unwin Hyman: Chichester, United Kingdom, 1988.
(28) Marschner, H. Mineral Nutrition of Higher Plants, 2nd Ed.;
Academic Press: London, 1995.
(29) Gahoonia, T. S.; Ali, O.; Sarker, A.; Nielsen, N. E.; Rahman, M. M.
Genetic variation in root traits and nutrient acquisition of lentil
genotypes. J. Plant Nutr. 2006, 29, 643655.
(30) Sheppard, S. C.; Evenden, W. G. The assumption of linearity in
soil and plant concentration ratios: An experimental evaluation.
J. Environ. Radioact. 1988, 7, 221247.
(31) Hodgeson, J. G.; Wilson, P. J.; Hunt, R.; Grime, J. P.; Thompson,
K. Allocating C-S-R plant functional types; A soft approach to
a hard problem. Oikos 1999, 85, 282294.
(32) Mattina, M. I.; Iannucci-Berger, W.; Eitzer, B. D.; White, J. C.
Rhizotron study of Cucurbitaceae: Transport of soil bound
chlordane and heavy metal contaminants differ with genera.
Environ. Chem. 2004, 1, 8689.
(33) Wang, X.; White, J. C.; Gent, M. P. N.; Iannucci-Berger, W.; Eitzer,
B. D.; Mattina, M. I. Phytoextraction of weathered p,p-DDE by
zucchini (Cucurbita pepo) and cucumber (Cucumis sativus)
under different cultivation conditions. Int. J. Phytorem. 2004,
6, 363385.
(34) Gent, M. P. N.; Parrish, Z. D.; White, J. C. Nutrient uptake among
subspecies of Cucurbita pepo L. is related to exudation of citric
acid. J. Am. Soc. Hort. Sci. 2005, 130, 782788.
(35) McCutcheon, S. C., Schnoor, J. L., Eds. Phytoremediation:
Transformation and Control of Contaminants; Wiley Inter-
science: New York, 2003.
(36) Willey, N., Ed. Phytoremediation: Methods and Reviews; Humana
Press: Totowa, New Jersey, 2007.
ES071531R
VOL. 42, NO. 6, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 2167