Introduction
Trees are ecosystem engineers, modifying soil physico-chemical properties,
nutrient cycling and microbial communities (Vesterdal and Raulund-Rasmussen,
1998; Aponte et al., 2011, 2013). The behaviour of different elements in the
afforested biogeochemical landscape will depend on their sources, their
chemical bonding properties and their degree of biological control (Ladanai
et al., 2010).
Chemical composition of forest superficial soil is influenced mainly by
leaf-fall quality, but also by atmospheric deposition, leaching and root
exudates (Aponte et al., 2011). In contaminated soils, the potentially toxic
elements can be partially immobilized by afforestation; this soil
remediation measure is known as “phytostabilization” (Bolan et al., 2011).
The Guadiamar Green Corridor (SW Spain) is a large-scale example of
phytostabilization. After a mine spill (in 1998) that contaminated about
4000 ha, soil was cleaned up and remediated, and later the site was
afforested with several native shrub and tree species (Domínguez et
al., 2008). In this study we explored the chemical composition of five
ecosystem compartments in the established mixed forest: canopy leaves,
forest floor, tree roots, topsoil and deep soil.
The objectives were to characterize the chemical heterogeneity at canopy
level, originated by planting different tree species, and to evaluate the
strength of their footprint on soil properties, 16 years after plantation.
Specific questions were, what was the rank of variability among chemical
elements by compartment? What was the mean chemical variability by ecosystem
compartment? Was there a significant difference among tree species for every
chemical element by ecosystem compartment? In particular, was there a
significant tree-species effect on soil?
Variability of the concentration of 23 chemical elements in five
ecosystem compartments: mean, maximum and minimum values of CV. Tree-species
effect indicated by the number of chemical elements showing significant
variability due to tree-species identity, according to the ANOVA or
Kruskal–Wallis results.
Ecosystem compartment
Coefficient of variation
Tree-species effect
Mean
Maximum
Minimum
Canopy leaves
78.6
226.4
6.6
23
Forest floor
65.3
130.2
11.6
20
Roots
61.8
134.9
7.4
22
Superficial soil
25.0
86.2
11.6
3
Deep soil
21.9
65.4
11.1
0
Material and methods
The study area was the Guadiamar Green Corridor (Sevilla, Spain). Its climate is
Mediterranean, with mild rainy winters and warm dry summers. Average
annual temperature is 19 ∘C and annual rainfall is 448 mm.
In a plot of about 14 ha (37∘23.165′ N 6∘13.668′ W) we
randomly selected five replicated trees of each of seven species afforested
(35 tree samples in total). Three species were deciduous (Populus alba,
Celtis australis and Fraxinus angustifolia) and four
evergreen (Quercus ilex, Olea europaea, Ceratonia siliqua
and Pinus pinea). We sampled in 2014 fully expanded leaves, forest
floor (litter), excavated roots, and soil at two depths: topsoil (0–10 cm)
and deep soil (10–30 cm).
Soil properties within the plot were similar before afforestation, and
therefore we could infer tree effects in this common garden experiment. Soil
was Fluvisol type, had a loamy texture (42 % sand and 19 % clay),
was acidic (pH about 4), and was nutrient-poor (0.07 % N, 8 mg kg-1 available
P) (site 4 in Domínguez et al., 2010).
Samples of the three plant materials and two soil layers were processed and
analysed for the composition of 22 chemical elements (Al, As, B, Ba, Ca, Cd,
Co, Cr, Cu, Fe, K, Li, Mg, Mn, Na, Ni, P, Pb, S, Sr, V, Zn) by inductively coupled plasma optical emission spectrometry (ICP-OES), and
of C and N by elemental analysis (see general methods in Madejón et al.,
2006; Domínguez et al, 2008). A total of 23 elements were considered
for plant materials (Li was under detectable levels in plants) and also 23
for soil samples (B was under detectable levels in soils).
The variability in the concentration of each chemical element by ecosystem
level was measured by the coefficient of variation (CV). Elements under
biological control were expected to have lower variability in plant
compartments than in soil (Ladanai et al., 2010). Global chemical
heterogeneity at each ecosystem level was evaluated as the mean CV of the
23 measured elements by compartment. We expected higher heterogeneity at
superficial soil (more affected by leaf-fall) than at deeper soil.
The source of variation due to the tree species was tested by analysis of
variance (one-way ANOVA) for each chemical element and ecosystem
compartment. Previously data were tested for normality (Shapiro–Wilk
test) and homoscedasticity (Levene test) and log-transformed if necessary.
Alternatively a non-parametric Kruskal–Wallis test was applied. In addition
to the global comparison, the chemical composition of topsoil under
contrasted tree species was compared by pair-wise Student's t tests. All the
analyses were carried out with the STATISTICA v8.0 program (Stat Soft, Tulsa,
Oklahoma, USA).
Results and discussion
Leaves
Tree canopies had the highest variability in the concentration of 23
elements; the mean CV was 79 % (Table 1). Elements with the lowest CV
values were C (7 %), N (20 %) and K (26 %), in general those under
higher biological control. The highest variability was
shown by Cd (226 %), Zn (158 %), V (126 %) and Mn (123 %),
non-limiting elements, which excess uptake was poorly regulated (Ladanai et
al., 2010).
The identity of tree species was a major source of variation for the
concentration of all the 23 elements in leaves (Table 1). The presence of
accumulator species explained the high variability of certain elements. For
example, the higher accumulation of Cd and Zn in leaves of Populus and of Mn in
Quercus has previously been reported in the Guadiamar area (Madejón et al., 2004,
2006). We also found here a remarkable accumulation of several elements –
Ba, Ca, Cr, Fe, Na, Pb, Sr and V – in the leaves of Celtis. Each tree species had
a characteristic chemical composition determined by selective uptake of soil
elements, differential exclusion or accumulation, and transport to leaves,
which could reflect a separation in their “biogeochemical niche” (Sardans
and Peñuelas, 2014).
Forest floor
The mass accumulated under the tree canopy was the net result of litter
production and decomposition, and its chemical composition would reflect
both litterfall and decomposers (Vesterdal and Raulund-Rasmussen, 1998;
Ladanai et al., 2010). The forest floor heterogeneity was high (mean CV of
65 %). Minimum CV value was found for C (12 %) and Mg (20 %), whereas
the maximum variability was presented by non-essential elements Cd
(130 %), As (100 %) and Pb (98 %).
Forest floor mirrored the chemical patterns of the canopy, with peaks of Cd
and Zn under Populus; Mn under Quercus; or Ca, Na and Sr under Celtis. However the tree-induced
heterogeneity was attenuated, showing significant differences among tree
species for 20 out of 23 elements (Table 1).
Roots
The average variability of root chemical composition was also high (mean
CV of 62 %), with the minimum CV value for C (7 %) and P (30 %), those
elements under higher biological control. In contrast, the non-essential
elements Cr (135 %), As (94 %) and Pb (91 %) presented the highest
variability in roots.
Bivariate diagram of the concentration of Cd and S (log scale) in
the superficial soil under the 35 trees of 7 species. For illustrative
purpose two contrasted species, Populus alba (triangle) and
Pinus pinea (square), are indicated
that showed pair-wise significant differences in S (t value of 2.6, p=0.03)
and Cd (t value of -3.0, p=0.02).
![]()
Roots had also a characteristic tree-species signal with significant
among-species variance for 22 out of 23 elements (Table 1). However, the
pattern was quite different from leaves; for example, there was no
significantly higher accumulation of Cd and Zn in Populus roots. The highest
accumulation of elements – Al, As, Cr, Pb and V – was produced in the roots
of Celtis. In general, accumulation of non-essential elements in roots may reflect
detoxification mechanisms by their immobilization in cell walls
(Domínguez et al., 2009) or in the mycelium of mycorrhizal fungi (Wu et
al., 2016).
Superficial soil
The heterogeneity of chemical concentration of 23 elements in topsoil was
lower than in the three plant compartments; the mean CV was 25 % (Table 1).
Elements presenting lowest variability were V (12 %), P (13 %) and
Fe (13 %), whereas the highest values were found for S (86 %), Pb
(41 %) and As (36 %). Forest soil heterogeneity is originated by changes
in soil-forming factors, management and stochastic variation (Quilchano et
al., 2008).
Tree-species identity was the source of significant difference for only
three elements (Cd, Co and Li) in topsoil. Despite the strong differences
in chemical composition of leaves and roots among tree species the chemical
footprint in the topsoil was weak. However, contrasted species had
significant pair-wise differences in the soil underneath (detected by
Student's t test); for example topsoil under Pinus pinea trees had lower
concentrations of Cd, Co, Li, Mn, Ni and Zn but higher concentrations of S than soil under Populus alba (Fig. 1).
Probably the elapsed time (16 years) was not enough to detect a significant
and strong tree signal in the soil. In a review of 70 studies Li et al.
(2012) found that age after plantation was a major factor explaining the
dynamics of soil C (significantly increased after 30 years) and N
(significant after 50 years).
Deeper soil
The chemical composition of deeper soil (10–30 cm) was even less
heterogeneous; mean CV was 22 %. The least variable elements were Cu (11 %),
P (12 %) and V (12 %), whereas the highest heterogeneity was presented
by S (65 %), Na (37 %) and Cd (34 %). There was no significant effect
of the species of tree planted aboveground for any chemical element in that
deeper horizon (Table 1). In general, the tree effects on soil nutrients
tend to disappear with depth, and subsoil features are frequently used as
proxy of the original soil conditions (Aponte et al., 2013).