Variation in wood nutrients along a tropical soil fertility gradient
Katherine D. Heineman1, Benjamin L. Turner2 and James W. Dalling1,2
1Program for Ecology, Evolution and Conservation Biology, University of Illinois-Urbana Champaign, 505 S Goodwin Ave, Urbana, IL 61801, USA; 2Smithsonian Tropical Research Institute,
Apartado 0843-03092, Balboa, Ancon, Panama
Author for correspondence:
Katherine D. Heineman
Tel: +1 402 750 6075
Email: kheineman@life.illinois.edu
Received: 21 October 2015
Accepted: 21 January 2016
New Phytologist (2016)
doi: 10.1111/nph.13904
Key words: environmental gradients,
Fortuna Forest Reserve, functional traits,
nutrient limitation, phosphorus cycling,
stoichiometry, wood density.
Summary

 Wood contains the majority of the nutrients in tropical trees, yet controls over wood nutri-
ent concentrations and their function are poorly understood.

 We measured wood nutrient concentrations in 106 tree species in 10 forest plots spanning
a regional fertility gradient in Panama. For a subset of species, we quantified foliar nutrients
and wood density to test whether wood nutrients scale with foliar nutrients at the species
level, or wood nutrient storage increases with wood density as predicted by the wood eco-
nomics spectrum.

 Wood nutrient concentrations varied enormously among species from fourfold in nitrogen
(N) to > 30-fold in calcium (Ca), potassium (K), magnesium (Mg) and phosphorus (P). Com-
munity-weighted mean wood nutrient concentrations correlated positively with soil Ca, K,
Mg and P concentrations. Wood nutrients scaled positively with leaf nutrients, supporting the
hypothesis that nutrient allocation is conserved across plant organs. Wood P was most sensi-
tive to variation in soil nutrient availability, and significant radial declines in wood P indicated
that tropical trees retranslocate P as sapwood transitions to heartwood. Wood P decreased
with increasing wood density, suggesting that low wood P and dense wood are traits associ-
ated with tree species persistence on low fertility soils.

 Substantial variation among species and communities in wood nutrient concentrations sug-
gests that allocation of nutrients to wood, especially P, influences species distributions and
nutrient dynamics in tropical forests.
Introduction
Tropical rainforests commonly occur on strongly weathered soils
depleted in rock-derived nutrients such as calcium (Ca), potas-
sium (K), magnesium (Mg) and phosphorus (P). As a conse-
quence, productivity in tropical forests is assumed to be limited
by the availability of one or more soil nutrients (Vitousek, 1984;
Tanner et al., 1998; Wright et al., 2011). In addition to effects
on productivity, spatial heterogeneity in soil nutrient availability
can drive compositional variation in tropical tree communities at
local scales and regional scales (John et al., 2007; Baldeck et al.,
2013; Condit et al., 2013). Differences among species in nutrient
acquisition and allocation potentially underlie soil-based habitat
partitioning in tropical forests (Chapin, 1980; Fine et al., 2004;
Palmiotto et al., 2004; Baltzer et al., 2005; Andersen et al., 2012),
and exacerbate plant–soil feedbacks that influence nutrient avail-
ability via nutrient deposition and mineralization from plant
material (Wardle et al., 2004). Consequently, quantifying varia-
tion in tree species allocation of limiting nutrients over soil fertil-
ity gradients is critical to understanding community assembly
and biogeochemistry in tropical forests.
Tree nutrient-use strategies are most commonly evaluated
through the nutrient stoichiometry of fresh leaves. Soil-based
habitat distributions have been linked to foliar nutrient
concentrations, because tropical plant taxa associated with high
fertility soils contain greater concentrations of N and P in fresh
leaves than species associated with low fertility soils (Tanner,
1977; Andersen et al., 2012; Katabuchi et al., 2012; Dalling
et al., 2016). Despite shifts in community mean foliar chemistry
along soil fertility gradients (Vitousek et al., 1988; Han et al.,
2005; Fyllas et al., 2009; Ordo~nez et al., 2009; Hayes et al.,
2014), foliar nutrient concentrations are poorly constrained by
soil nutrient availability in tropical forests, as variation in foliar
nutrients among co-occurring species growing on the same soil
habitat is nearly as great as regional variation in species foliar
nutrients across soil fertility gradients (Townsend et al., 2007;
Fyllas et al., 2009). Given the taxonomic control over foliar
nutrient concentrations and the importance of tissue N and P
concentrations in predicting rates of leaf (Santiago, 2007; Bakker
et al., 2011) and wood (Weedon et al., 2009; Zanne et al., 2015)
decomposition, a trait-based approach leveraging taxon-specific
chemical attributes has been advocated for modeling global car-
bon turnover from leaf (Cornwell et al., 2008) and woody (Corn-
well et al., 2009) biomass. Although access to species-level
information is rapidly expanding owing to the emergence of
global trait databases (Kattge et al., 2011), data on chemical
attributes of wood are poorly represented in the literature (Chave
et al., 2009), despite wood containing roughly half of Ca, K, Mg,

 2016 The Authors
New Phytologist
 2016 New Phytologist Trust
New Phytologist (2016) 1
www.newphytologist.com
Research
N and P in live vegetation in tropical forests (Tanner, 1985;
Wang et al., 1991; Stanley & Montagnini, 1999; Bond, 2010).
Trait inventories evaluating tropical tree species wood nutrient
concentrations, in concert with more widely collected foliar and
woody attributes, are necessary to more fully understand the vari-
ability among species in whole-tree nutrient use and the implica-
tions of this variability for ecosystem processes.
The extent to which foliar traits can be used to predict woody
traits may depend on the range of soil habitats and taxonomic
groups represented in the analysis. For instance, a global meta-
analysis of 145 woody species found that stem and leaf nutrient
concentrations were significantly, albeit weakly, correlated for
both N (r2 = 0.20) and P (r2 = 0.22; Kerkhoff et al., 2006). How-
ever, this pattern may be driven in part by latitudinal gradients in
soil N and P availability, which are reflected in foliar nutrient
concentrations (Reich & Oleksyn, 2004), rather than constraints
across organs in nutrient allocation at the species level. Although
decomposition rates of species wood, leaf and root tissues are cor-
related both within sites and across global scales (Freschet et al.,
2013), species wood and leaf decomposition rates are decoupled
when angiosperm and gymnosperm taxa are analyzed separately
(Pietsch et al., 2014), suggesting that the allocation of nutrients
and secondary metabolites might not be constrained across
organs in co-occurring angiosperms. Furthermore, a multivariate
analysis of wood and leaf characteristics in neotropical tree species
found that physical wood and leaf traits load along orthogonal
axes of variation (Baraloto et al., 2010), indicating that the selec-
tion pressures that shape species tradeoffs among foliar traits
might differ from those controlling the evolution of woody traits.
Consequently, although we expect that species wood and leaf
nutrient concentrations should be correlated at global scales, this
relationship is not well understood in highly diverse tropical tree
communities.
Wood and leaves differ in functional attributes that might dis-
rupt a tight correlation between nutrient concentrations of these
organs. Whereas woody biomass could provide a well-defended
storage organ for nutrients, water and carbohydrates (Chapin
et al., 1990), leaves are a poorer storage organ due to their short
lifespan and vulnerability to herbivores. Susceptibility to her-
bivory increases with foliar N concentration (Mooney & Gul-
mon, 1982; Andersen et al., 2010), thereby creating a possible
constraint on maximum investment of N in leaf metabolism,
with consequences for other nutrients that are linked through sto-
ichiometry. Compared to young, fully expanded leaves, stem
nutrient concentrations display a greater relative increase in N
and P concentrations in response to experimental N and P addi-
tion in seedlings in Panama (Schreeg et al., 2014) and trees in
China (Mo et al., 2015), which may reflect a greater capacity for
nutrient storage in wood compared with leaves. A global meta-
analysis of 71 angiosperm species found not only that sapwood
Ca, K, Mg, N and P concentrations all vary by an order of mag-
nitude, but also, species vary considerably in their abilities to
resorb nutrients as sapwood transitions to heartwood (Meerts,
2002). Given the potential variation in nutrient storage and
remobilization in woody tissues among species, wood nutrients
may constitute an important dimension of functional variation in
tropical tree communities.
The scarcity of interspecific wood nutrient data limits our
understanding of co-variation between wood nutrients and other
plant functional traits in tropical forests. This paucity of informa-
tion was acknowledged in the development of the wood eco-
nomics spectrum (WES; Chave et al., 2009), a comprehensive
meta-analysis emphasizing the inverse relationship between tree
species wood density and mortality rates. The WES predicts
tradeoffs among wood traits that facilitate fast growth (large con-
duit diameter) and adaptations that promote survival (high wood
density, high storage capacity). In line with these predictions,
stem nonstructural carbohydrate storage has been linked to
increased wood density and decreased mortality rates in tropical
forest saplings (Poorter & Kitajima, 2007). To date, studies eval-
uating the relationship between wood chemistry and the growth–
survival axis have been limited to wood N and provide mixed
evidence: wood N correlated positively with wood density and
negatively with relative growth rate for 54 tree species in Panama
(Martin et al., 2014), whereas no relationship was found between
the density and N content of wood in 23 tree species in Uganda
(Becker et al., 2012). Given that species leaf N concentration cor-
relates negatively with wood density (Kraft et al., 2008) and posi-
tively with species diameter growth rates in tropical forests
(Poorter & Bongers, 2006), a positive relationship between wood
nutrients and wood density would signal substantial functional
decoupling among plant organs. To better understand how com-
ponents of species nutrient use strategies relate to life history
tradeoffs, wood nutrients must be evaluated with respect to foliar
nutrients and wood density in co-occurring species in high-
diversity tropical forests.
Here, we measured the concentrations of Ca, K, Mg, N and P
in the wood of 106 Panamanian tree species growing in 10 mon-
tane and lowland forests spanning a range of soil ‘available’ nutri-
ent concentrations comparable to the range observed throughout
the tropics (Gartlan et al., 1986; Baillie et al., 1987; Phillips et al.,
2003; Quesada et al., 2009). By exploiting this extreme soil gradi-
ent, we present the most robust examination to date of the natu-
ral variation in wood nutrient concentrations among species and
sites in tropical forests. We hypothesized that if taxonomic varia-
tion in nutrient allocation influences the distribution of tree
species across soil fertility gradients, then (1) community mean
wood nutrient concentrations should correlate with soil nutrient
availability, and (2) there should be substantial interspecific varia-
tion in wood nutrient concentrations within a site. We also evalu-
ated whether wood nutrient concentrations in 58 montane forest
species correlate with other species-specific functional traits,
including foliar nutrients and wood density. If nutrient allocation
to biomass is constrained across plant organs at the species level,
then species wood nutrient concentrations should increase with
species leaf nutrient concentrations, which are often negatively
correlated with wood density. Alternatively, we hypothesized that
if wood nutrient concentrations are proportional to the invest-
ment of nutrients into storage reserves, then tree species wood
nutrient concentrations should correlate positively with wood
New Phytologist (2016) 
 2016 The Authors
New Phytologist
 2016 New Phytologist Trustwww.newphytologist.com
Research
New
Phytologist2
density, which is associated with persistence and survival strate-
gies in tropical tree species.
Materials and Methods
Study site
We sampled foliar and woody tissue from six montane forest sites
located within the Fortuna Forest Reserve (19 500 ha) and the
adjacent Palo Seco Forest Protectorate (125 000 ha), henceforth
Fortuna, in western Panama (Fig. 1a). This region encompasses
old growth, lower montane forest, ranging between 700 and
1500 m above sea level (asl), with mean annual temperatures
varying between 19 and 23°C (Cavelier et al., 1997). There is
strong interannual and spatial variability in precipitation among
study sites, with annual rainfall ranging from 4000 to
9000 mm yr1. A distinct dry season occurs from January to
April, but evapotranspiration does not exceed rainfall during this
period (Cavelier et al., 1997), with monthly rainfall accumulation
exceeding 100 mm per month on average during the dry season
in all but one site (Table 1).
Twelve permanent 1-ha forest plots were established at For-
tuna in 2003 in which all trees > 5 cm diameter at breast height
(DBH) are mapped, measured and identified to species. Plant tis-
sue was sampled in six plots, chosen to maximize variation in soil
nutrient availability across three geological substrates: rhyolitic
tuff, andesite and porphyritic dacite (Andersen et al., 2010). Soil
pH ranges between 3.6 and 5.6 among sites, which coincides
with substantial variability in Ca, K, Mg, N and P (Table 1).
There is considerable floristic turnover among soil habitats, with
only 22% of species shared between soils developed on dacite
and rhyolite located < 15 km apart.
We sampled wood from four additional lowland sites in the
Panama Canal watershed, part of a network of 1-ha plots estab-
lished in the region by the Center for Tropical Forest Science
(Fig. 1; Pyke et al., 2001; Turner & Engelbrecht, 2011; Condit
et al., 2013). Two of the focal plots were located on peninsulas
adjoining the Panama Canal (P13 and P25), and two plots in
Soberanıa National Park were located along Pipeline Road (P06)
and Camino de Cruces (P24; plot codes from Pyke et al., 2001).
This region consists of semi-deciduous, seasonally moist forest,
receiving c. 2500 mm of annual rainfall and with a mean annual
Fig. 1 Locations of 10 permanent 1-ha forest plots in Panama where we sampled wood. Leaves were also collected from the six Fortuna plots. Panama
Canal Watershed sites are seasonally moist lowland forest located < 110m above sea level (asl). Fortuna sites are wet lower montane forest located
800–1300m asl. Plots are grouped by low, medium and high fertility based on soil chemical variables and geological parent material (see Table 1).

 2016 The Authors
New Phytologist
 2016 New Phytologist Trust
New Phytologist (2016)
www.newphytologist.com
New
Phytologist Research 3
T
ab
le
1
Si
te
lo
ca
ti
o
n
,
en
vi
ro
n
m
en
ta
lc
h
ar
ac
te
ri
st
ic
s
an
d
sp
ec
ie
s
ri
ch
n
es
s
o
f
1
0
1
-h
a
P
an
am
an
ia
n
fo
re
st
p
lo
ts
n
ea
r
w
h
ic
h
w
o
o
d
co
re
sa
m
p
le
s
w
er
e
co
lle
ct
ed
En
vi
ro
n
m
en
ta
lv
ar
ia
b
le
s
M
o
n
ta
n
e
fo
re
st
:
Fo
rt
u
n
a
Lo
w
la
n
d
fo
re
st
:
P
an
am
a
C
an
al
W
at
er
sh
ed
C
h
o
rr
o
A
H
o
n
d
a
A
Sa
m
u
d
io
P
al
o
Se
co
H
o
rn
it
o
A
lt
o
Fr
io
P
2
5
P
0
6
P
1
3
P
2
4
1
P
ro
te
ct
ed
ar
ea
Fo
rt
u
n
a
Fo
rt
u
n
a
Fo
rt
u
n
a
P
al
o
Se
co
Fo
rt
u
n
a
Fo
rt
u
n
a
A
C
P
So
b
er
an
 ıa
B
C
N
M
So
b
er
an
 ıa
La
ti
tu
d
e
(N
)
8
°4
5
0 4
2
″
8
°4
3
0 3
″
8
°4
3
0 5
2
″
8
°4
6
0 4
3
″
8
°4
0
0 2
6
″
8
°3
9
0 1
5
″
9
°4
0 4
2
″
9
°9
0 2
3
″
9
°1
1
0 1
6
″
9
°7
0 2
5
″
Lo
n
g
it
u
d
e
(W
)
8
2
°1
4
0 3
2
″
8
2
°1
4
0 2
2
″
8
2
°1
4
0 5
3
″
8
2
°1
1
0 5
3
″
8
2
°1
2
0 1
5
0 ’
8
2
°1
2
0 5
4
″
7
9
°4
7
0 5
6
″
7
9
°4
4
0 3
9
″
7
9
°4
9
0 1
6
″
7
9
°4
0
0 3
6
″
El
ev
at
io
n
(m
as
l)
1
1
0
0
1
1
5
5
1
2
1
5
8
7
8
1
3
3
0
1
1
0
0
1
1
0
3
0
5
5
5
0
2
A
n
n
u
al
ra
in
fa
ll
(m
m
yr

1
)
5
5
0
0
6
2
0
0
4
8
0
0
6
2
0
0
5
1
0
0
4
6
0
0
+
2
1
0
0
2
3
0
0
2
6
0
0
2
2
0
0
D
ry
se
as
o
n
ra
in
fa
ll
(m
m
p
er
m
o
n
th
)
3
5
1
3
8
1
2
1
5
4
4
5
2
0
3
9
1
1
2
4
1
4
9
1
9
7
1
3
1
3
So
il
p
ro
p
er
ti
es
G
eo
lo
g
ic
al
su
b
st
ra
te
R
h
yo
lit
e
R
h
yo
lit
e
A
n
d
es
it
e
A
n
d
es
it
e
D
ac
it
e
D
ac
it
e
R
h
yo
lit
e
M
ar
in
e
Se
d
im
en
t
(G
at
u
n
ci
llo
)
M
ar
in
e
Se
d
im
en
t
(C
ai
m
it
o
)
A
g
g
lo
m
er
at
e
B
u
lk
d
en
si
ty
(g
cm

3
)
0
.1
3
0
.2
9
0
.3
9
0
.4
1
0
.2
5
0
.6
6
0
.8
7
1
.0
5
0
.6
7
0
.5
9
p
H
3
.6
7
3
.5
8
4
.1
8
4
.3
7
5
.0
3
5
.6
2
4
.6
1
4
.3
6
6
.3
9
6
.3
6
In
o
rg
an
ic
n
it
ro
g
en
(N
;
m
g
cm

3
)
1
.0
1
3
.4
4
1
.8
1
.6
3
6
.3
2
2
.5
4
2
.4
6
3
.0
7
3
.0
1
R
es
in
p
h
o
sp
h
o
ru
s
(P
;
m
g
cm

3
)
0
.0
8
0
.2
2
0
.4
2
0
.4
3
2
.2
2
1
.3
9
0
.1
8
1
.2
8
5
.6
5
7
.7
5
M
eh
lic
h
ca
lc
iu
m
(C
a;
m
g
cm

3
)
9
7
8
2
2
4
9
1
3
5
1
3
5
8
3
3
8
8
3
2
3
7
9
3
7
0
6
1
6
6
2
M
eh
lic
h
p
o
ta
ss
iu
m
(K
;
m
g
cm

3
)
1
9
1
1
6
4
3
1
9
6
4
8
3
2
3
0
2
1
1
2
0
7
M
eh
lic
h
m
ag
n
es
iu
m
(M
g
;
m
g
cm

3
)
1
9
2
0
5
3
4
0
2
5
4
5
5
1
6
7
1
1
7
7
3
7
5
3
9
4
Fo
re
st
st
ru
ct
u
re
Sp
ec
ie
s
ri
ch
n
es
s
7
2
1
3
5
1
2
5
1
6
7
1
3
9
8
3
8
4
7
8
6
0
6
0
B
as
al
ar
ea
(m
2
h
a
1
)
3
2
4
4
3
5
3
2
5
5
4
3
2
0
1
9
2
5
3
1
St
em
d
en
si
ty
(s
te
m
s
h
a
1
)
1
0
4
2
8
1
3
8
0
5
6
8
1
7
1
5
1
0
5
0
3
0
2
4
8
4
4
2
9
3
5
5
Sp
ec
ie
s
sa
m
p
lin
g
Sp
ec
ie
s
co
re
d
1
3
2
2
1
9
1
8
1
4
1
7
1
1
9
9
7
B
A
re
p
re
se
n
te
d
b
y
co
re
d
tr
ee
s
7
1
%
6
8
%
4
5
%
3
0
%
7
2
%
5
4
%
4
1
%
2
8
%
6
4
%
2
1
%
1
P
ro
te
ct
ed
ar
ea
s:
Fo
rt
u
n
a
Fo
re
st
R
es
er
ve
(F
o
rt
u
n
a)
,
P
al
o
Se
co
Fo
re
st
P
ro
te
ct
o
ra
te
(P
al
o
Se
co
),
P
an
am
a
C
an
al
A
u
th
o
ri
ty
(A
C
P
),
So
b
er
an
 ıa
N
at
io
n
al
P
ar
k
(S
o
b
er
an
 ıa
),
an
d
B
ar
ro
C
o
lo
ra
d
o
Is
la
n
d
N
at
u
re
M
o
n
u
m
en
t
(B
C
N
M
).
2
A
n
n
u
al
an
d
d
ry
se
as
o
n
ra
in
fa
ll
d
at
a
at
A
lt
o
Fr
io
w
er
e
co
lle
ct
ed
fr
o
m
2
0
1
2
to
2
0
1
4
,
al
lo
th
er
Fo
rt
u
n
a
p
lo
ts
w
h
er
e
m
o
n
it
o
re
d
2
0
0
7
–2
0
1
4
.
R
ai
n
fa
ll
va
lu
es
fr
o
m
th
e
lo
w
la
n
d
si
te
s
re
p
re
se
n
t
a
m
u
lt
i-
ye
ar
av
er
ag
es
.
3
B
u
lk
d
en
si
ty
,
p
H
,
in
o
rg
an
ic
N
an
d
M
eh
lic
h
ca
ti
o
n
s
fo
r
Fo
rt
u
n
a
si
te
s
w
er
e
p
u
b
lis
h
ed
p
re
vi
o
u
sl
y
in
A
n
d
er
se
n
e
t
a
l.
(2
0
1
0
).
4
Fo
re
st
st
ru
ct
u
re
fo
r
st
em
s
>
1
0
cm
d
ia
m
et
er
at
b
re
as
t
h
ei
g
h
t
(D
B
H
)
ar
e
fr
o
m
P
yk
e
e
t
a
l.
(2
0
0
1
)
fo
r
lo
w
la
n
d
si
te
s
an
d
C
.
M
.
P
ra
d
a
&
J.
W
.D
al
lin
g
(u
n
p
u
b
lis
h
ed
),
fo
r
m
o
n
ta
n
e
si
te
s.
New Phytologist (2016) 
 2016 The Authors
New Phytologist
 2016 New Phytologist Trustwww.newphytologist.com
Research
New
Phytologist4
temperature of 27°C (Pyke et al., 2001). In the dry season, peri-
ods during which evapotranspiration exceeds rainfall are frequent
and influence the recruitment of species in this area (Engelbrecht
et al., 2006). Focal plots are classified as mature secondary forest
aged between 60 and 100 yr (Pyke et al., 2001). Canal watershed
plots were selected to maximize contrast along the soil fertility
gradient (Table 1). The underlying geological substrates of the
plots in order of increasing P availability are: rhyolitic tuff (P25),
Gatuncillo Formation (marine sediment, P06), Caimito Forma-
tion (marine sediment, P13) and agglomerate (P24) (Turner &
Engelbrecht, 2011).
Plant tissue sampling and analysis
Wood core samples were extracted from 301 individual trees
from 76 species at Fortuna and 104 trees from 30 species the
Canal Watershed. In all plots, we sampled 7–22 woody species
with the greatest basal area in each plot. We cored three trees
> 10 cm DBH per species using a 4.3-mm Hagl€of increment
borer. Cores were taken at breast height (1.3 m) to a depth of half
the DBH of the tree. Because coring canopy palms at Fortuna
with hard exteriors and soft interiors damaged the borers, palms
were not sampled in the Panama Canal Watershed. Conse-
quently, compared to other sites, the sampled species represented
a smaller proportion of plot basal area in P06 and P24 (Table 1)
where the palm genera Astrocaryum, Oenocarpus and Attalea make
up > 20% of the total basal area. Trees at Fortuna were cored
outside permanent forest plots, but within 100 m of the plot
boundary in February 2011. Lowland wood samples were col-
lected from trees located within 1-ha plots in July 2013. At For-
tuna, foliar tissue was collected for 58 of the species sampled for
wood in July 2010. Three fully expanded shade leaves were col-
lected with a pruning pole from three individuals per species.
Although functional trait protocols recommend the collection of
sun leaves for foliar nutrient analysis (Cornelissen et al., 2003),
we sampled shade leaves to maximize the number of species
included in our study, because 43% of tree species with individu-
als > 10 cm DBH in the Fortuna plots do not reach the canopy
(DBH > 30 cm). Analysis of foliar plasticity to light in 38 tropi-
cal tree species found that rank order in species mean N and P
concentrations is preserved across sun and shade leaves (Rozen-
daal et al., 2006), indicating that the choice of sun vs shade leaves
should should not affect how wood and leaf nutrients co-vary
across species. Wood and leaf samples were stored on ice until
processing. Each wood core was divided into segments ≤ 5 cm
and volume determined by Archimedes’ principle. Segments were
then dried to constant mass at 60°C and wood density was calcu-
lated as segment dry mass/fresh volume.
All leaf material collected for each tree, including petioles and
rachii, were ground together in a KLECO Tissue Pulverizer
(Kinetic Laboratory Equipment, Visalia, CA, USA). A mini-
Wiley Mill (Thomas Scientific, Swedesboro, NJ, USA) was used
to grind wood core samples in 5 cm segments to account for pos-
sible radial differences in wood attributes (Lachenbruch et al.,
2011). We used 5 cm as a critical threshold of change in wood
chemistry based on the steep observed decline in nonstructural
carbohydrate concentrations between 4.5 and 6 cm in wood cores
taken from lowland Panamanian tree species (W€urth et al.,
2005). We were unable to categorize segments visually as heart-
wood or sapwood because the heartwood/sapwood transition is
often gradual in tropical trees (Jordan & Kline, 1977). Nitrogen
concentrations in leaf and wood tissues were tested on a Costech
Elemental Analyzer (Valencia, CA, USA) for samples analyzed in
Illinois and a Thermo Flash 1112 Elemental Analyzer (Waltham,
MA, USA) for samples analyzed in Panama. A subset of samples
tested on both Elemental Analyzers also closely and linearly cor-
related (r2 = 0.92, n = 10), although a correction factor of 0.18
was applied to all samples tested in Panama to ensure consistency
in the two datasets. To prepare wood and leaf material for Ca K,
Mg and P analysis, samples were dry ashed at 550°C for 1 h and
the ash dissolved in 1M HNO3 (Karla, 1998). Base cations for
all samples and P in tissues collected at Fortuna were measured
using inductively coupled plasma-optical emission spectrometry
(ICP-OES) on an Optima 2000 DV (Perkin Elmer, Waltham,
MA, USA). A subset of Fortuna wood samples with P concentra-
tions below the ICP-OES detection limit and all lowland wood
samples were analyzed for P via automated molybdate colorime-
try using the Lachat Quickchem 8500 (Hach Ltd, Loveland, CO,
USA). For samples above the detection limit of the ICP analyzed
on both instruments, P measurements by spectrometry and col-
orimetry were closely correlated (r2 = 0.95, n = 10) with an inter-
cept that did not differ significantly from zero. We included
certified reference samples (NIST 1515, apple leaves) and inter-
nal laboratory control standards in all analyses.
Soil sampling and analysis
Soil cores were taken to a depth of 10 cm from 13 locations in
each 1-ha plot during the wet season at both lowland and mon-
tane sites. Bulk density was determined by drying a known vol-
ume of soil at 105°C. Soil pH was determined in a 1 : 2 soil to
deionized water ratio using a glass electrode. Total soil inorganic
N was calculated as the sum of soil nitrate and ammonium mea-
sured in 0.5M K2SO4 extracts and determined by automated
colorimetry on a Lachat Quikchem 8500 (Hach Ltd). Readily
exchangeable P, which approximates plant available P, hereafter
‘resin P’, was determined by extraction with anion-exchange
membranes (Turner & Romero, 2009). Base cations were
extracted in Mehlich-3 solution (Mehlich, 1984) with detection
by ICP–OES on an Optima 7300 DV spectrometer (Perkin-
Elmer).
Statistical methods
Response of site-mean wood chemistry to soil nutrient
availability For consistency, we limited our analyses of plot and
interspecific variation in wood nutrient concentrations and densi-
ties to data from the outer 5 cm of wood because the majority of
trees cored were not big enough to yield multiple 5 cm segments.
The community weighted mean (CWM) nutrient concentration
for each 1-ha plot was calculated as the average of species mean
wood nutrient concentrations (in the outer 5-cm annulus)

 2016 The Authors
New Phytologist
 2016 New Phytologist Trust
New Phytologist (2016)
www.newphytologist.com
New
Phytologist Research 5
weighted by basal area of each species sampled in the 1-ha plot
(trees > 10 cm DBH). Ordinary least squares (OLS) regression
was used to fit the relationship between CWM wood and plot
mean soil nutrient concentrations, and between species mean
wood nutrient concentrations and respective plot level soil nutri-
ent concentrations to determine how much variation in species
wood nutrient concentrations is explained by soil nutrients alone.
Wood nutrient concentrations and soil variables were log-
transformed before regression analyses to meet the assumption of
normality of errors.
Wood vs leaf scaling relationships We modeled the relation-
ship between species mean wood and leaf nutrients among
organs for 58 woody species sampled for both leaf and wood
nutrients at Fortuna. If a species was sampled in more than one
plot, we used the mean species value across all plots so that there
were no duplicate species in the analysis. We modeled wood-leaf
scaling as a power function (Y ~ aXb), which was transformed to
be evaluated as a linear relationship (log(Y) ~ b9 log(X) + log
(a)). If the exponent or slope of this relationship (b) differs from
one, then scaling is nonlinear, indicating that the nutrient con-
centration of one tissue is more constrained in one tissue vs the
other. We fitted the log–log relationship between species mean
leaf and wood values for each element using type II Major Axis
(MA) regression in the lmodel2 package in R (Legendre, 2011).
MA regression is recommended over OLS regression in this case
because there is similar measurement error in both variables. We
used 95% confidence intervals to determine if b differed signifi-
cantly from 1. Because we modeled mean wood nutrient concen-
trations as a function of leaf nutrient concentrations, b > 1
indicates that leaf nutrients are more constrained among species
than wood nutrients. We determined r2 values from OLS regres-
sion of each relationship to find the proportion of variance in
species mean wood nutrient concentrations explained by foliar
nutrient concentrations.
Because some species are more closely related in evolutionary
history than others, species trait values may be considered nonin-
dependent and therefore violate the assumption of linear regres-
sion. To determine if species wood and leaf nutrient
concentrations co-vary after accounting for evolutionary history,
we tested relationships for the phylogenetically independent con-
trasts (PICs) of log-transformed species mean wood and leaf
nutrient concentrations (Felsenstein, 1985). We constructed a
phylogenetic tree for the species in our study from the
Angiosperm Phylogeny Group super tree (APG III; http://www.
mobot.org/MOBOT/research/APweb) using Phylomatic v3.0
(Webb & Donoghue, 2005). We used the fossil-derived ages of
tree taxa listed in Wikstrom (Wikstr€om et al., 2001) to determine
the branch lengths of this tree using BLADJ in Phylocom (Webb
et al., 2008). Because the APG III tree is resolved to the family
level for most lineages, genera were drawn as polytomies nested
within families and species were drawn as polytomies within
genera. Before phylogenetic analyses, polytomies in the tree were
broken randomly. PICs for the log of each species trait value
were calculated using the ape package (Paradis et al., 2004) in R.
We used MA regression to fit the scaling relationship between
the PICs of wood and leaf nutrient concentrations. PIC models
were fitted through the origin as suggested in Garland et al.
(1992).
Wood nutrient concentrations vs wood density For 76 woody
species cored at Fortuna, we tested if species mean wood nutrient
concentrations co-vary with each other and with woody density
using MA regression. We log-transformed wood nutrient concen-
trations before regression analysis to meet the assumption of nor-
mality of errors. We performed the same regression analysis for
the PICs of log-transformed species mean wood nutrients and
nonlog-transformed wood density. The significance threshold of
regression tests was adjusted using the Bonferroni correction to
account for multiple comparisons. If a species was cored in > 1
plot, we used the average species mean across all plots so that
there were no duplicate species in the analysis.
Radial variation in wood nutrient concentrations For the 110
trees, we fitted the log–log relationship between the nutrient con-
centrations of the inner (5–10 cm in depth) vs outer (0–5 cm in
depth) wood annuli for each element using MA regression. We
used 95% confidence intervals to determine if the intercept dif-
fered from zero and the slope differed from 1. We also calculated
the coefficient of determination r2 from OLS regression to evalu-
ate how well the concentration of inner annuli can be predicted
from outer annuli.
We evaluated if species vary in radial patterns of wood nutri-
ent allocation for 18 species for which we had available data on
both inner and outer wood segments for ≥ 3 individuals per
species. For each tree, we calculated the percentage radial dis-
crepancy in wood nutrients as: (outer – inner)/outer9 100. We
determined the mean radial discrepancy for each species  1 SE
to determine if the change in wood nutrient concentrations
from the outermost segment to adjacent inner segment is < 0,
= 0, > 0.
Results
Interspecific and inter-site variation in wood nutrients
Of the elements measured in the outer 5 cm of wood across the
106 tree species sampled in this study, mean concentrations of N
were greatest (2557 70 lg g1), followed by Ca
(2082 160 lg g1), K (1622 80 lg g1), Mg (492 40 lg
g1) and P (111 7 lg g1; Table 2; Supporting Information
Table S1). Although wood nutrient concentrations varied consid-
erably among species and sites, the magnitude of this variability
was not consistent among elements (Table 2). The range in mean
wood N among species (fivefold) was less than the smallest range
of species means measured for rock-derived elements including P
(35-fold), K (36-fold), Ca (47-fold) and Mg (51-fold). When
species averages were weighted by basal area to calculate CWM
nutrient concentrations, there was also a greater range of values
among sites in wood P and cations compared to wood N
(Table 2): N (1.8-fold), Mg (2.5-fold), K (4.5-fold), Ca (7-fold)
and P (8.5-fold).
New Phytologist (2016) 
 2016 The Authors
New Phytologist
 2016 New Phytologist Trustwww.newphytologist.com
Research
New
Phytologist6
Wood nutrient concentrations vs soil nutrient availability
In line with our predictions, CWM wood nutrient concentra-
tions (a plot-level estimate of wood nutrient status) for Ca, K,
Mg and P were significantly correlated with the plot mean nutri-
ent concentrations in the topsoil (Fig. 2a–c,e). CWM wood P
concentration correlated most strongly with its respective soil
metric (r2 = 0.71), followed by Mg (r2 = 0.64), K (r2 = 0.58) and
Ca (r2 = 0.49). Although CWM wood N was not significantly
correlated with soil inorganic N (r2 = 0.00, P = 0.519; Fig. 2e),
soil resin P was a strong predictor of CWM wood N (r2 = 0.59,
P = 0.009), indicating that CWM wood N concentrations varies
along the fertility gradient.
When species-level wood nutrient means were modeled as a
function of plot-level soil nutrient concentrations (Fig. 2f–j;
Table S2), the relationship between wood Mg and soil Mg was
no longer significant (r2 = 0.01, P = 0.141). For all nutrients, the
slope of the species means wood vs soil regression was smaller in
magnitude than the slope of the CWM (basal-area weighted)
wood vs soil regression fit (Fig. 2; Table S2), suggesting that
species with high basal area better reflect local soil nutrient avail-
ability than rare or small stature species.
Wood vs leaf nutrient concentrations
For 58 montane tree species for which we analyzed both wood
and leaf nutrient concentrations, species mean leaf and wood
nutrient concentrations were significantly positively correlated
for all elements evaluated, supporting the hypothesis that nutri-
ent allocation is constrained across organs at the species level.
Among significant relationships, wood and leaf tissue chemistry
was most strongly correlated for P (r2 = 0.36; Fig. 3e) and most
weakly correlated for Mg (r2 = 0.18; Fig. 3c). For Ca, K, Mg and
P, the slope of the wood vs leaf relationship (b) was significantly
> 1, indicating that wood nutrient concentrations scale nonlin-
early with leaf nutrient concentrations (Table 3; Fig. 3). By con-
trast, the slope of the leaf vs wood regression did not differ from
1 for N (Table S2; Fig. 3d), indicating that N scales isometrically
between wood and leaf tissues.
When wood and leaf nutrient concentrations were corrected
for evolutionary history using PICs, the scaling relationship
remained significant for all nutrients but Mg (Table 3). Slopes of
PIC models did not differ from observed models for any nutrient
(Table 3). Scaling exponents of relationship between wood–leaf
PICs remained significantly > 1 for K and P, and the b value of
the wood–leaf PIC for Ca marginally overlapped with 1 (95%
CI = 0.99–2.10).
Table 2 Summary statistics of community-weighted mean (CWM) and
species mean wood nutrient concentrations for 106 tree species sampled
across six lowland and four montane forest plots
Nutrient
Community weighted
means (lg g1) Species means (lg g1)
Mean Min Max Mean Min Max
Calcium (Ca) 2038 823 5692 2082 271 12 613
Potassium (K) 1547 653 2935 1622 157 5771
Magnesium (Mg) 526 320 807 492 61 3123
Nitrogen (N) 2603 1857 3454 2557 1300 5800
Phosphorus (P) 135 44 339 111 19 668
N : P 34 14 59 35 4 181
Carbon (C) : N 203 148 267 209 80 381
C : P 7276 2175 14881 7510 729 50 400
50 200 1000 5000
10
00
20
00
50
00
C
W
M
 w
oo
d 
C
a 
(μ
g 
g–
1 )
Montane
Lowland
50 200 1000 5000
50
0
20
00
10
 0
00
Soil Mehlich Ca (mg cm–3)  
S
pe
ci
es
 m
ea
n 
w
oo
d 
C
a 
(μ
g 
g–
1 )
10 20 50 100
50
0
10
00
20
00
50
00
C
W
M
 w
oo
d 
K
 (μ
g 
g–
1 )
10 20 50 100
20
0
50
0
20
00
Soil Mehlich K (mg cm–3) 
S
pe
ci
es
 m
ea
n 
w
oo
d 
K
 (μ
g 
g–
1 )
20 50 200 1000
30
0
40
0
60
0
80
0
C
W
M
 w
oo
d 
M
g 
(μ
g 
g–
1 )
20 50 200 1000
50
20
0
10
00
Soil Mehlich Mg (mg cm–3) 
S
pe
ci
es
 m
ea
n 
w
oo
d 
M
g 
(μ
g 
g–
1 )
1 2 5
20
00
25
00
35
00
C
W
M
 w
oo
d 
N
 (μ
g 
g–
1 )
1 2 5
10
00
20
00
40
00
Soil inorganic N (mg cm–3)
S
pe
ci
es
 m
ea
n 
w
oo
d 
N
 (μ
g 
g–
1 )
0.05 0.20 1.00 5.00
50
10
0
20
0
C
W
M
 w
oo
d 
P
 (μ
g 
g–
1 )
0.05 0.20 1.00 5.00
20
50
20
0
50
0
50
0
Soil resin P (mg cm–3)
S
pe
ci
es
 m
ea
n 
w
oo
d 
P
 (μ
g 
g–
1 )
(a) (b) (c) (d) (e)
(f) (g) (h) (i) (j)
Fig. 2 Wood nutrient concentrations (of the outer 5 cm of sapwood) vs soil nutrient availability in four lowland (open circles) and six montane (closed
circles) forest plots in Panama. Soil nutrients were measured in the top 10 cm of soil and are expressed in volumetric units (mg cm3). Lines of log–log
linear regression models are presented where the relationship between wood and soil nutrients is significant. (a–e) Community weighted mean (CWM)
wood nutrient concentrations of all species sampled in each plot. Y bars, one basal area-weighted standard error. (f–j) Species mean wood nutrient
concentrations plotted as a function of plot-level soil nutrient concentrations. Axes are plotted on log scales. See Methods section for soil nutrient
extraction protocols.

 2016 The Authors
New Phytologist
 2016 New Phytologist Trust
New Phytologist (2016)
www.newphytologist.com
New
Phytologist Research 7
Wood nutrient concentrations vs wood density
We did not find support for the hypothesis that wood nutrient
concentrations increase with wood density. Species mean wood
density significantly declined as species mean wood P increased
and was not significantly associated with any other wood nutrient
(Table 4). Wood Ca, K, Mg, N and P were all significantly posi-
tively correlated with one another at critical a = 0.05 (Table 4).
After correcting for multiple comparisons, Ca was significantly
correlated with only N and P, and Mg was correlated with only P
(Table 4).
Correlation of the PICs of species woody traits gave qualita-
tively similar results to analysis of observed values at a = 0.05,
except that PIC of wood Mg was negatively correlated with PIC
of wood density (Table S3). After correcting for multiple com-
parisons, PIC of wood P remained significantly correlated with
all wood traits; however, the relationships of wood Ca, Mg, and
N with wood K were no longer significant, and wood Mg was no
longer correlated with wood Ca or wood density (Table S3).
Radial variation in wood nutrients
In the 110 tree cores examined, the concentration of macronutri-
ents in the outermost 5 cm of wood was strongly positively corre-
lated with nutrient concentrations in the adjacent 5–10 cm
segment for all elements (Fig. 4; Table S4). Nutrient concentra-
tions in the outer segment explained a greater proportion of the
variance than the inner segment for Ca (r2 = 0.74), Mg (r2 = 0.77)
and N (r2 = 0.75) compared to K (r2 = 0.53) and P (r2 = 0.61).
For K, N and P in MA regression models, intercept of the inner
vs outer relationship was significantly < 0 and the slope was signif-
icantly > 1 (Table S4), indicating that inner segments have lower
concentrations of K, N and P on average than the outer segments
at low element concentrations. However, this discrepancy
between segments declines or reverses with increasing element
concentration (Fig. 4). For 88 of 110 individuals, the outer core
segment had a higher wood P concentration than the inner core
segment. The Ca and Mg inner vs outer regression parameters
did not significantly differ from a 1 : 1 relationship (Table S4).
Species varied in the magnitude and the direction of radial dif-
ferences in nutrient concentrations between outer and inner
annuli. Of the elements examined, radial patterns in wood P were
most qualitatively consistent across species, because 14 of 18
species had significantly higher P concentrations in the outer vs
inner core. Wood P concentrations declined by 35% from the
outer to inner segments across all species, although species mean
radial P discrepancies ranged widely from 2% to 88% (Fig. 5).
By contrast, Ca and Mg concentrations were significantly lower
in the outer compared to the inner core for the majority of
species (Fig. 5), and the magnitude of this difference exceeded
50% for four of 18 species. However, qualitatively distinct radial
cation differences were present in other species (Fig. 5). For wood
N and K, there were similar numbers of species with radial differ-
ences < 0, = 0, or > 0.
5000 10 000
20
0
10
00
50
00
Leaf Ca (μg g–1) 
W
oo
d 
C
a 
(μ
g 
g–
1 ) 
–0.2 –0.1 0.0
–0
.2
0.
0
0.
2
PIC of log leaf Ca
P
IC
 o
f l
og
 w
oo
d 
C
a
5000 10 000
10
0
50
0
20
00
Leaf K (μg g–1) 
W
oo
d 
K
 (μ
g 
g–
1 )
–0.15 0.05
–0
.2
0.
.2
PIC of log leaf K
P
IC
 o
f l
og
 w
oo
d 
K
 
1000 5000
50
20
0
10
00
Leaf Mg (μg g–1) 
W
oo
d 
M
g 
(μ
g 
g–
1 ) 
–0.2 0.0
0.1
0.2
–0
.3
–0
.1
0.
1
PIC of log leaf Mg
P
IC
 o
f l
og
 w
oo
d 
M
g
10 000 20 000 35 000
10
00
20
00
40
00
Leaf N (μg g  ) –1
W
oo
d 
N
 (μ
g 
g–
1 ) 
–0.15 0.05
–0
.1
5
–0
.0
5
0.
05
PIC of log leaf N
P
IC
 o
f l
og
 w
oo
d 
N
500 1000 2000
20
50
20
0
Leaf P (μg g–1)
W
oo
d 
P
 (μ
g 
g–
1 ) 
–0.15
–0.05
0
0
–0.05
–0.05 0.05
–0
.3
–0
.1
0.
1
PIC of log leaf P
P
IC
 o
f l
og
 w
oo
d 
P
(a)
(b)
(c)
(d)
(e)
(f)
(g)
(h)
(i)
(j)
Fig. 3 Scaling relationships of species mean leaf and wood nutrient
concentrations for 58 tree species sampled in western Panama. Lines
represent major axis regression model fits. Scaling relationships were fit for
log-transformed observed species mean wood and leaf nutrient
concentrations (a–e; closed circles) and for the phylogenetically
independent contrasts (PIC) of logged species mean values (f–j; open
circles). Wood nutrients were analyzed for the outermost 5 cm annulus of
sapwood.
New Phytologist (2016) 
 2016 The Authors
New Phytologist
 2016 New Phytologist Trustwww.newphytologist.com
Research
New
Phytologist8
Discussion
Variation in wood nutrients among species and sites
Wood nutrient concentrations varied substantially among 106
tree species sampled along a regional soil fertility gradient in
Panama. For some nutrients, observed values encompassed the
entire range of wood nutrient concentrations reported in tropical
tree species to date. Wood N concentrations (Table 2) fell within
the range of values reported previously for tropical forest species
(400–6900 lg g1; Becker et al., 2012; Mascaro et al., 2012;
Martin et al., 2014). The same was true for the observed range of
species mean wood Ca concentrations in this study, which fell
within the range of the values reported for sapwood in the Meerts
(2002) meta-analysis of 93 temperate and tropical angiosperm
species (60–15 000 lg g1). By contrast, the range in wood P
concentrations observed (19–668 lg g1) here exceeded the
range of previously reported for sapwood P concentrations in
wild angiosperms (20–615 lg g1; Meerts, 2002; Mascaro et al.,
2012). The range in species wood K and Mg observed here
exceeded the maximum previously reported values for sapwood
K (160–4500 lg g1) and Mg (80–1290 lg g1) reported in
Meerts (2002). Furthermore, the average of species mean wood
Ca, K and P concentrations in this study all exceeded the ‘typical
range’ of wood nutrient concentrations reported in Chave et al.
(2009), which is a frequently cited reference for wood traits, indi-
cating that calculations of forest wood nutrient stocks based on
this review may be underestimated.
Community weighted mean (CWM) wood nutrient concen-
trations also differed widely among plots, and paralleled soil
nutrient availability for all elements except N. Because the con-
centrations of macronutrients strongly covaried in both woody
Table 4 Major axis regression model fits of pairwise combinations of log-transformed species mean wood nutrient (calcium (Ca), potassium (K),
magnesium (Mg), nitrogen (N) and phosphorus (P); lg g1) and wood density (WD; g cm3) measured in 76 tree species from six forest sites at Fortuna
(df = 74)
x variable y variable
Intercept Slope
r2 P1Mean 2.5 97.5 Mean 2.5 97.5
WD Ca 11.2 32.9 0.4 31.8 11.7 44.5 0.02 0.249
K 21.2 12.9 22.5 24.1 51.5 9.7 0.02 0.179
Mg 18.7 11.8 96.1 22.2 176.2 10.4 0.04 0.081
N 16.8 10.8 2.1 15.7 17.1 5.3 0.01 0.302
P 10.9 8.2 25.4 11.2 36.2 6.6 0.10 0.005*
Ca K 2.21 2.75 5.22 0.70 0.29 1.39 0.10 0.006*
Mg 2.73 21.42 2.53 1.18 0.46 3.78 0.07 0.024*
N 5.90 4.85 6.88 0.26 0.12 0.41 0.16 < 0.001**
P 0.03 3.60 2.44 0.62 0.28 1.12 0.12 0.002**
K Mg 5.86 23.11 0.34 1.60 0.84 3.97 0.11 0.005*
N 5.42 4.15 6.58 0.32 0.16 0.50 0.18 < 0.001**
P 2.10 4.93 0.06 0.90 0.62 1.29 0.30 < 0.001**
Mg N 6.65 5.82 7.45 0.19 0.06 0.34 0.10 0.007*
P 0.61 1.14 1.99 0.66 0.42 0.96 0.26 < 0.001**
N P 11.72 17.61 8.02 2.08 1.60 2.84 0.41 < 0.001**
1P values: *, significant at P < 0.05; **, significant correcting for multiple comparisons (P < 0.0033).
Table 3 Major axis regression model fits of the scaling relationship between observed species mean wood and leaf nutrient concentrations (log(leaf) ~ log
(a) + log(wood)9 b) and the phylogenetically independent contrasts (PICs) of species mean wood and leaf concentrations (PIC of log leaf ~ PIC of log
wood9 b) for 58 tree species sampled at Fortuna Forest Reserve
Nutrient Model n
Intercept: log (a) Slope: b
r2 PMean 2.5 97.5 Mean 2.5 97.5
Calcium (Ca) Observed 58 10.61 5.35 1.36 1.93 1.36 3.00 0.33 < 0.001
PIC 56 1.40 0.99 2.10 0.36 < 0.001
Potassium (K) Observed 58 11.03 22.30 5.33 1.97 1.36 3.19 0.30 < 0.001
PIC 56 1.92 1.24 3.50 0.24 < 0.001
Magnesium (Mg) Observed 58 17.72 45.64 8.76 2.93 1.82 6.39 0.18 < 0.001
PIC 56 5.61 1.95 8.14 0.03 0.223
Nitrogen (N) Observed 58 4.47 11.45 0.35 1.25 0.83 1.95 0.31 < 0.001
PIC 56 1.01 0.84 1.23 0.49 < 0.001
Phosphorus (P) Observed 58 10.03 17.42 5.99 2.10 1.52 3.17 0.36 < 0.001
PIC 56 2.22 1.85 2.67 0.54 < 0.001

 2016 The Authors
New Phytologist
 2016 New Phytologist Trust
New Phytologist (2016)
www.newphytologist.com
New
Phytologist Research 9
biomass and the soil available pools, it is difficult to determine if
the ecosystem sequestration of nutrients in woody biomass is pro-
portional to the soil availability of each nutrient independently,
or if the increases in the availability of one limiting element
increases the uptake of nonlimiting elements to maintain stoi-
chiometric balance (Marschner, 1995). For example, CWM wood
N might correlate significantly with soil P, but not soil N,
because N investment in wood is constrained by P limitation. By
contrast, the Stability of Limiting Nutrients Hypothesis (Han
et al., 2011) would posit that the low variability in wood N
among species and along environmental gradients is evidence that
N is more limiting than other elements, because increased avail-
ability of limiting elements should result in increased growth, not
increased tissue nutrient concentrations. Given that a long-term
factorial nutrient addition experiment in the Panama canal water-
shed has found that N, P and K all limit aspects of plant growth
(Kaspari et al., 2008; Wright et al., 2011), and that responses to
N addition in western Panama vary among species (Adamek
et al., 2009), it seems likely that numerous interacting forces
influence the uptake and subsequent sequestration of individual
nutrients in wood. Nonetheless, the result that rock-derived
macronutrients in wood generally track the availability of these
nutrients in the soil has implications for ecosystem and commu-
nity processes.
Although soil nutrient concentrations were strong predictors
of community mean wood nutrients at the plot level, wide varia-
tion in species mean nutrient concentrations in a given habitat
suggest that species were unlikely to be excluded completely from
a habitat based on nutrient allocation patterns, either physiologi-
cally or competitively. However, comparison of the slopes of
wood vs soil regression models indicates that basal area CWM
wood nutrient values are more sensitive to soil nutrient variation
than the unweighted average across species, providing evidence
that tradeoffs related to nutrient allocation underlie shifts in
species abundance across soil gradients. Alternatively, this pattern
could occur if canopy trees differ systematically in wood nutrient
allocation from smaller stature trees. In a Jamaican montane
forest, Tanner (1985) attributed the significantly lower nutrient
concentrations of taller trees relative to short trees as a potential
reason that tall trees were able to attain high biomass on nutrient
impoverished soils. For 42 tree species in Bolivia, maximum tree
height has been linked to wood anatomical traits including wood
density, vessel diameter, and hydraulic conductance (Poorter
et al., 2010), which could have indirect effects on wood nutrient
uptake and storage. Given that dominant canopy species dispro-
portionately influence estimates of forest aboveground biomass
(Slik et al., 2013; Bastin et al., 2015), efforts to quantify forest
nutrient stocks in living biomass should prioritize the sampling
of species with high basal area in a given community.
The strength of the relationship between species mean wood
nutrient concentrations and soil nutrients were strongest for Ca
and P, which were also the soil nutrients most closely associated
with tree species distributions in lowland Panama (Condit et al.,
2013). This pattern provides evidence that differences among
species in the acquisition and use of Ca and P is partially
200 500 1000 5000
20
0
50
0
20
00
10
 0
00
Ca outer annulus (μg g–1)
C
a 
in
ne
r a
nn
ul
us
 (μ
g 
g–
1 )
50 100 200 500 1000 5000
20
50
20
0
10
00
50
00
K outer annulus (μg g–1)
K
 in
ne
r a
nn
ul
us
 (μ
g 
g–
1 )
50 100 200 500 2000 5000
50
20
0
50
0
20
00
Mg outer annulus (μg g–1)
M
g 
in
ne
r a
nn
ul
us
 (μ
g 
g–
1 )
1000 2000 3000 5000
10
00
20
00
50
00
N outer annulus (μg g–1)
N
 in
ne
r a
nn
ul
us
 (μ
g 
g–
1 )
2 5 10 20 50 200 500
2
5
10
50
20
0
10
00
P outer annulus (μg g–1)
P
 in
ne
r a
nn
ul
us
 (μ
g 
g–
1 )
(a)
(b)
(c)
(d)
(e)
Fig. 4 Scaling relationship of nutrient concentrations in ‘outer’ 5 cm of
annulus of wood and adjacent ‘inner’ 5–10 cm annulus for 110 trees cored
in Panama. Black lines represent major axis (MA) regression model fits of
the scaling relationship between nutrient concentrations in inner and outer
(log(inner) ~ log(a) + log(outer)9 b). Blue dashed lines represent a 1 : 1
relationship.
New Phytologist (2016) 
 2016 The Authors
New Phytologist
 2016 New Phytologist Trustwww.newphytologist.com
Research
New
Phytologist10
responsible for their distribution across edaphically heteroge-
neous landscapes. Small within-species sample size and high
species turnover among sites prevented us from assessing how
much of this response was explained by taxonomic or environ-
mental controls. In Amazonian tree species, foliar P and Ca con-
centrations are more sensitive to environmental variation in soil
availability than N and Mg concentrations (Fyllas et al., 2009),
and experimental nutrient addition in tropical forests show that
both foliar and wood P concentrations respond more strongly to
nutrient addition than do N concentrations (Harrington et al.,
2001; Ostertag, 2010; Schreeg et al., 2014; Mo et al., 2015).
Wood Ca and P may also be more sensitive than other elements
to soil conditions due to respective variation in the soil available
nutrient pools, because the ranges of soil available Ca (115-fold)
and P (90-fold) were vastly greater than the variation in Mg (38-
fold), K (20-fold) and inorganic N (sixfold) among the 10 plots
sampled (Table 1). However, the heterogeneity in Ca and P in
this region may provide a basis for the evolution of tradeoffs in
Ca and P allocation strategies to optimize fitness in high vs low
resource environments. Similarly, foliar P is the primary leaf trait
differentiating tree species adapted to habitats differing widely in
soil nutrient availability in Borneo (Baltzer & Thomas, 2010).
Therefore, P allocation may be a key component of edaphic spe-
cialization niches in tropical regions worldwide. In regions with
less spatial variation in soil nutrient availability than Borneo or
Panama, weaker selection for tradeoffs in nutrient allocation
might results in less variation among species and sites in wood
nutrient concentrations.
Wood–leaf nutrient scaling
The scaling relationship between species mean wood and leaf tis-
sues was significant for all nutrients for species examined along
the Fortuna nutrient gradient, supporting findings of previous
meta analyses which have shown that N and P allocation is con-
strained across plant organs (Kerkhoff et al., 2006; Agren, 2008).
However, our results differ from previous observations that wood
and leaf nutrients concentrations are not closely correlated within
angiosperm taxa (Pietsch et al., 2014; Zanne et al., 2015). Signifi-
cant scaling between the PICs wood and leaf Ca, K, N and P
indicate that the apparent functional coordination of nutrient
allocation across organs is not simply a consequence of evolution-
ary history. Although wood and leaf nutrients are correlated for
all elements, the shape of scaling relationships differed among
nutrients. Nonlinear scaling of wood and leaf nutrients for all ele-
ments except N indicated that leaf nutrients are more constrained
than wood nutrients at high concentrations, perhaps because allo-
cation of nutrients to woody repositories increases when nutrients
are no longer limiting to photosynthesis. The dynamics of N
storage may differ from other elements because plants generally
store N as organic amino acids or proteins (Chapin et al., 1990)
and the storage of organic N may incur a substantial carbon cost
relative storage of inorganic molecules (Millard, 1988). By con-
trast, for P, which can be stored as inorganic phosphate in vac-
uoles (Sinclair & Vadez, 2002), the wood-leaf scaling exponent
was c. 2, meaning that, for example, a 10% increase in foliar P
corresponds to a 20% increase in wood P. The apparent
−200
−100
0
La
ura
ce
ae
 sp
.
1
E. 
pa
na
me
ns
is
P. 
ole
ifol
ius
O. 
me
xic
an
a
P. 
jura
na
M.
 m
elin
on
ian
a
V. 
se
bife
ra
M.
 ap
eta
la
S. 
rad
lko
fer
i
H. 
ob
lon
ga
P. 
arb
ore
a
P. 
ten
uifo
lium
P. 
se
ssi
lis
T. 
ve
rs
ico
lor
C. 
bra
sili
en
se
T. 
gu
ian
en
sis
V. 
gu
ate
ma
len
sis
C. 
alli
od
ora
R
ad
ia
l d
isc
re
pa
nc
y 
in
w
o
o
d 
K 
(%
)
−100
−50
0
C. 
bra
sili
en
se
E. 
pa
na
me
ns
is
P. 
arb
ore
a
M.
 ap
eta
la
La
ura
ce
ae
 sp
.
1
S. 
rad
lko
fer
i
C. 
alli
od
ora
P. 
jura
na
T. 
gu
ian
en
sis
V. 
se
bife
ra
M.
 m
elin
on
ian
a
O. 
me
xic
an
a
P. 
ole
ifol
ius
H. 
ob
lon
ga
P. 
se
ssi
lis
V. 
gu
ate
ma
len
sis
T. 
ve
rs
ico
lor
P. 
ten
uifo
lium
R
ad
ia
l d
isc
re
pa
nc
y 
in
w
o
o
d 
Ca
 (%
)
(a) (b)
9 spp < 0
6 spp = 0
3 spp > 0
Mean = –23.8% ± 23.2%
6 spp < 0
7 spp = 0
5 spp > 0
Mean = –14.3% ± 17.1%
−150
−100
−50
0
50
La
ura
ce
ae
 sp
.
1
E. 
pa
na
me
ns
is
C. 
bra
sili
en
se
P. 
arb
ore
a
S. 
rad
lko
fer
i
M.
 ap
eta
la
P. 
jura
na
H. 
ob
lon
ga
T. 
ve
rs
ico
lor
P. 
ole
ifol
ius
M.
 m
elin
on
ian
a
V. 
se
bife
ra
P. 
ten
uifo
lium
T. 
gu
ian
en
sis
P. 
se
ssi
lis
C. 
alli
od
ora
O. 
me
xic
an
a
V. 
gu
ate
ma
len
sis
R
ad
ia
l d
isc
re
pa
nc
y 
in
w
o
o
d 
M
g 
(%
)
−100
−50
0
La
ura
ce
ae
 sp
.
1
M.
 m
elin
on
ian
a
V. 
se
bife
ra
T. 
ve
rs
ico
lor
P. 
jura
na
P. 
ole
ifol
ius
V. 
gu
ate
ma
len
sis
E. 
pa
na
me
ns
is
H. 
ob
lon
ga
P. 
se
ssi
lis
P. 
arb
ore
a
C. 
bra
sili
en
se
C. 
alli
od
ora
O. 
me
xic
an
a
P. 
ten
uifo
lium
M.
 ap
eta
la
S. 
rad
lko
fer
i
T. 
gu
ian
en
sis
R
ad
ia
l d
isc
re
pa
nc
y 
in
w
o
o
d 
N 
(%
)
0
30
60
90
P. 
arb
ore
a
M.
 m
elin
on
ian
a
E. 
pa
na
me
ns
is
P. 
jura
na
P. 
se
ssi
lis
T. 
ve
rs
ico
lor
V. 
se
bife
ra
H. 
ob
lon
ga
O. 
me
xic
an
a
S. 
rad
lko
fer
i
P. 
ole
ifol
ius
M.
 ap
eta
la
V. 
gu
ate
ma
len
sis
La
ura
ce
ae
 sp
.
1
T. 
gu
ian
en
sis
P. 
ten
uifo
lium
C. 
bra
sili
en
se
C. 
alli
od
ora
R
ad
ia
l d
isc
re
pa
nc
y 
in
w
o
o
d 
P 
(%
)
(c) (d) (e)
10 spp < 0
5 spp = 0
3 spp > 0
Mean = –23.2% ± 10.0%
5 spp < 0
6 spp = 0
7 spp > 0
Mean = –2.2% ± 6.6%
0 spp < 0
4 spp = 0
14 spp > 0
Mean: 35.6% ± 6.4%
Fig. 5 Species mean radial discrepancy in calcium (Ca), potassium (K), magnesium (Mg), phosphorus (P) and nitrogen (N) concentrations (a–e) between
the outermost annulus of the stem (0–5 cm in depth) and the adjacent inner annulus (5–10 cm in depth) for 18 tree species in Panama. Radial discrepancy
was calculated as (outer – inner)/outer9 100. Species mean values SE are reported in each panel.

 2016 The Authors
New Phytologist
 2016 New Phytologist Trust
New Phytologist (2016)
www.newphytologist.com
New
Phytologist Research 11
accumulation of P in wood at high foliar P concentrations may
reflect the evolutionary importance of P storage reserves. Com-
pared to other macronutrients, P is particularly immobile in soil
solution and is spatially and temporally variable in its availability
to plants (Lambers et al., 2008). This indicates that trees may be
under selection to allocate excess P to storage to mitigate P limi-
tation when the P demands of plant growth exceed P supply from
soil.
Correlation of wood nutrients with life history parameters
We did not find support for the hypothesis from the wood eco-
nomics spectrum (WES) that wood nutrient storage increases
with wood density (Chave et al., 2009). These results are in con-
trast to the findings of Martin et al. (2014) in which wood N was
positively correlated with wood density along this axis in tree
species in the Panama Canal watershed, but corroborate the find-
ings of Becker et al. (2012), who found no relationship between
wood N and wood density among tree species in Uganda. Given
that wood and leaf nutrients were positively correlated, wood
nutrients likely reflect whole plant allocation strategies or access
to soil resources rather than active allocation of nutrients to wood
storage reserves at the expense of other organs. In fact, wood P
declined with increasing wood density in our montane forest,
suggesting that low biomass P concentrations and dense wood
are traits related to the survival of tropical tree species on low fer-
tility habitats. Although previous studies have reported that tree
communities on low fertility soils have higher wood density
(Muller-Landau, 2004; Chave et al., 2006) and lower wood P
concentrations (Tanner, 1985) than communities on more fertile
soils, this study is the first to our knowledge to test a functional
relationship between these traits in tropical tree species. The sig-
nificant relationship between the PICs of wood density and wood
P indicate that this pattern is not solely a consequence of evolu-
tionary history and might represent coordinated evolution of
traits that facilitate survival on low resource habitats.
Radial variation in wood nutrient concentrations
Wood nutrient concentrations of the outermost 5 cm of the
trunk strongly predicted nutrient concentrations of the adjacent
inner segment for all nutrients, suggesting that radial variation in
wood nutrients does not qualitatively influence patterns of inter-
specific or intersite variation in wood nutrients. Retranslocation
of P from sapwood as it transitions to heartwood appears to be
widespread among Panamanian trees, because wood P was lower
in the inner annulus than the outer annulus for 76% of individu-
als cored. Although this study did not evaluate differences
between heartwood and sapwood explicitly, our results are consis-
tent with Meerts (2002), who found that heartwood concentra-
tions were significantly lower than sapwood concentrations for
59 of 64 tree species. Radial variation in wood Ca, K, Mg and N
concentrations were more idiosyncratic across species. Although
Meerts (2002), a dataset of predominately temperate species,
found that wood N and K concentrations were higher in sap-
wood than heartwood for the majority of species, there was no
consistent pattern across species or individuals for these elements
in our study. Given that the increased remobilization efficiency
of P compared to N from senesced leaves is used as a primary
indicator that tropical forests are more P-limited than temperate
forests (Vitousek, 1984), the increased retranslocation efficiency
of P from sapwood relative to other elements may be evidence
that the trees in our study were primarily P-limited. Because trees
retain metabolically inactive wood in their stems, retranslocation
of nutrients may be easier to quantify at the tree and species level
in wood cores than in leaf litter, and should perhaps be utilized
more often in studies of plant nutrient dynamics. Furthermore, if
radial P translocation is greater in tropical forests than other
ecosystems, it may exacerbate the allometric decline in whole-
plant N : P and C : P ratios (Elser et al., 2010), which are used to
calculate ecosystem nutrient stocks. As this study provides only a
cursory evaluation of radial variation in wood nutrients, studies
explicitly examining both the effect of continuous radial variation
and heartwood/sapwood status on wood nutrients across species
and sites are needed to improve our understanding of the func-
tion of wood nutrients.
Ecosystem implications of variation in wood nutrients
The marked variation in wood nutrient concentrations both
among and within soil habitats observed here reinforces the idea
that a trait-based approach using species-specific nutrient data
could improve estimates of ecosystem processes in diverse tropi-
cal forests (Cornwell et al., 2008). The scaling relationships of
wood and leaf nutrients suggest that leaf nutrient values in
global plant trait databases have the potential to be used to
assign wood nutrient content used in models predicting carbon
turnover from woody biomass (Cornwell et al., 2009). Given
that tissue nutrient concentrations may be more important than
wood anatomy and exogenous environmental factors in deter-
mining wood decomposition rates (Zanne et al., 2015), the
eightfold range in community mean wood P concentrations
observed here could translate to substantial differences in carbon
residence time along fertility gradients. Not only are wood nutri-
ent concentrations important for predicting the dynamics of
coarse woody debris, but also nonlinear scaling of species wood
and leaf P concentrations suggest that trees store excess P in
wood, which has implications for how tropical trees will respond
to future global change scenarios. Understanding the extent to
which P can be remobilized from woody tissues could improve
predictions of how the growth of tropical trees will respond to
the alleviation of other limiting factors via CO2 fertilization and
N deposition.
Conclusions
Wood stores more biomass than any other plant organ, and the
fate of carbon and nutrients in the wood of tropical trees has par-
ticularly important implications of global biogeochemical cycles.
Our study is among the first to demonstrate the enormous vari-
ability in wood nutrient concentrations among tropical tree
species, and the strong relationship between community mean
New Phytologist (2016) 
 2016 The Authors
New Phytologist
 2016 New Phytologist Trustwww.newphytologist.com
Research
New
Phytologist12
wood nutrients and soil resource availability. These results sug-
gest that allocation of limiting nutrients to woody biomass is an
important functional characteristic influencing species distribu-
tions and biogeochemistry along edaphic gradients in tropical
forests.
Acknowledgements
The authors are grateful for funding the National Science Foun-
dation Doctoral Dissertation Improvement Grant (DEB-
1311379), Clark Research Support Grant, Tinker Summer
Research Fellowship, and Program for Ecology, Evolution, and
Conservation Summer Research Grant. We thank the Center for
Tropical Forest Science for providing the census data for the 1-ha
plots in the Panama Canal Watershed, and the Smithsonian
Tropical Research Institute (STRI) for logistical support. We
thank Carmen Velasquez, Fredy Miranda and Robert Heineman
for field assistance, Arturo Morris Valdes for identification of
trees in the Fortuna Forest Reserve, and Dayana Agudo and Paola
Escobar for assisting with chemical analysis of wood at the STRI
Soil Lab. We thank also Claire Baldeck, Adriana Corrales, Jen-
nifer Jones and two anonymous reviewers for comments that
improved this manuscript.
Author contributions
K.D.H. and J.W.D. designed research and sampling protocol.
K.D.H. carried out field data collection. K.D.H. and B.L.T. par-
ticipated in chemical analyses of plant tissues. B.L.T. contributed
soil data for this study. K.D.H. wrote the manuscript with input
from J.W.D. and B.L.T.
References
Adamek M, Corre MD, H€olscher D. 2009. Early effect of elevated nitrogen
input on above-ground net primary production of a lower montane rain forest,
panama. Journal of Tropical Ecology 25: 637–647.
Agren GI. 2008. Stoichiometry and nutrition of plant growth in natural
communities. Annual Review of Ecology, Evolution, and Systematics 39: 153–170.
Andersen KM, Corre MD, Turner BL, Dalling JW. 2010. Plant–soil
associations in a lower montane tropical forest: physiological acclimation and
herbivore-mediated responses to nitrogen addition. Functional Ecology 24:
1171–1180.
Andersen KM, Endara MJ, Turner BL, Dalling JW. 2012. Trait-based
community assembly of understory palms along a soil nutrient gradient in a
lower montane tropical forest. Oecologia 168: 519–531.
Baillie IC, Ashton PS, Anderson JAR, Fitzpatrick EA, Tinsley J. 1987. Site
characteristics and the distribution of tree species in mixed dipterocarp forest
on tertiary sediments in central Sarawak, Malaysia. Journal of Tropical Ecology
3: 201–220.
Bakker MA, Carre~no-Rocabado G, Poorter L. 2011. Leaf economics traits
predict litter decomposition of tropical plants and differ among land use types.
Functional Ecology 25: 473–483.
Baldeck CA, Harms KE, Yavitt JB, John R, Turner BL, Valencia R, Navarrete
H, Davies SJ, Chuyong GB, Kenfack D et al. 2013. Soil resources and
topography shape local tree community structure in tropical forests. Proceedings
of the Royal Society of B 280: 20122532.
Baltzer JL, Thomas SC. 2010. A second dimension to the leaf economics
spectrum predicts edaphic habitat association in a tropical forest. PLoS ONE 5:
e13163.
Baltzer JL, Thomas SC, Nilus R, Burslem DFR. 2005. Edaphic specialization in
tropical trees: physiological correlates and responses to reciprocal
transplantation. Ecology 86: 3063–3077.
Baraloto C, Paine CET, Poorter L, Beauchene J, Bonal D, Domenach A-M,
Herault B, Pati~no S, Roggy J-C, Chave J. 2010. Decoupled leaf and stem
economics in rain forest trees. Ecology Letters 13: 1338–1347.
Bastin JF, Barbier N, Rejou-Mechain M, Fayolle A, Gourlet-Fleury S,
Maniatis D, de Haulleville T, Baya F, Beeckman H, Beina D et al. 2015.
Seeing Central African forests through their largest trees. Scientific Reports
5: 13 156.
Becker GS, Braun D, Gliniars R, Dalitz H. 2012. Relations between wood
variables and how they relate to tree size variables of tropical African tree
species. Trees 26: 1101–1112.
Bond WJ. 2010. Do nutrient-poor soils inhibit development of forests? A
nutrient stock analysis. Plant and Soil 334: 47–60.
Cavelier J, Jaramillo M, Solis D, de Leon D. 1997.Water balance and nutrient
inputs in bulk precipitation in tropical montane cloud forest in Panama.
Journal of Hydrology 193: 83–96.
Chapin FS. 1980. The mineral nutrition of wild plants. Annual Review of Ecology
and Systematics 11: 233–260.
Chapin FS, Schulze E-D, Mooney HA. 1990. The ecology and economics of
storage in plants. Annual Review of Ecology and Systematics 21: 423–447.
Chave J, Coomes D, Jansen S, Lewis SL, Swenson NG, Zanne AE. 2009.
Towards a worldwide wood economics spectrum. Ecology Letters 12: 351–
366.
Chave J, Muller-Landau HC, Baker TR, Easdale TA, Steege HT, Webb CO.
2006. Regional and phylogenetic variation of wood density across 2456
neotropical tree species. Ecological Applications 16: 2356–2367.
Condit R, Engelbrecht BMJ, Pino D, Perez R, Turner BL. 2013. Species
distributions in response to individual soil nutrients and seasonal drought
across a community of tropical trees. Proceedings of the National Academy of
Sciences, USA 110: 5064–5068.
Cornelissen J, Lavorel S, Garnier E, Diaz S, Buchmann N, Gurvich D, Reich P,
Ter Steege H, Morgan H, Van Der Heijden M. 2003. A handbook of
protocols for standardised and easy measurement of plant functional traits
worldwide. Australian Journal of Botany 51: 335–380.
Cornwell WK, Cornelissen JHC, Allison SD, Bauhus J, Eggleton P, Preston
CM, Scarff F, Weedon JT, Wirth C, Zanne AE. 2009. Plant traits and wood
fates across the globe: rotted, burned, or consumed? Global Change Biology 15:
2431–2449.
Cornwell WK, Cornelissen JHC, Amatangelo K, Dorrepaal E, Eviner VT,
Godoy O, Hobbie SE, Hoorens B, Kurokawa H, Perez-Harguindeguy N et al.
2008. Plant species traits are the predominant control on litter decomposition
rates within biomes worldwide. Ecology Letters 11: 1065–1071.
Dalling JW, Heineman KD, Lopez OR, Wright SJ, Turner BL 2016. Nutrient
availability in tropical rain forests: the paradigm of phosphorus limitation In:
Goldstein G, Santiago L, eds. Tropical tree physiology: adaptations and responses
in a changing environment. Basel, Switzerland: Springer.
Elser JJ, Fagan WF, Kerkhoff AJ, Swenson NG, Enquist BJ. 2010. Biological
stoichiometry of plant production: metabolism, scaling and ecological response
to global change. New Phytologist 186: 593–608.
Engelbrecht BMJ, Dalling JW, Pearson TRH, Wolf RL, Galvez DA, Koehler T,
Tyree MT, Kursar TA. 2006. Short dry spells in the wet season increase
mortality of tropical pioneer seedlings. Oecologia 148: 258–269.
Felsenstein J. 1985. Phylogenies and the comparative method. American
Naturalist 1: 1–15.
Fine PVA, Mesones I, Coley PD. 2004.Herbivores promote habitat
specialization by trees in amazonian forests. Science 305: 663–665.
Freschet GT, Cornwell WK, Wardle DA, Elumeeva TG, Liu W, Jackson BG,
Onipchenko VG, Soudzilovskaia NA, Tao J, Cornelissen JH. 2013. Linking
litter decomposition of above- and below-ground organs to plant–soil
feedbacks worldwide. Journal of Ecology 101: 943–952.
Fyllas NM, Patino S, Baker TR, Bielefeld Nardoto G, Martinelli LA, Quesada
CA, Paiva R, Schwarz M, Horna V, Mercado LM et al. 2009. Basin-wide
variations in foliar properties of amazonian forest: phylogeny, soils and climate.
Biogeosciences 6: 2677–2708.

 2016 The Authors
New Phytologist
 2016 New Phytologist Trust
New Phytologist (2016)
www.newphytologist.com
New
Phytologist Research 13
Garland T, Harvey PH, Ives AR. 1992. Procedures for the analysis of
comparative data using phylogenetically independent contrasts. Systematic
Biology 41: 18–32.
Gartlan JS, Newbery DM, Thomas DW, Waterman PG. 1986. The influence of
topography and soil phosphorus on the vegetation of korup forest reserve,
cameroun. Vegetatio 65: 131–148.
Han W, Fang J, Guo D, Zhang Y. 2005. Leaf nitrogen and phosphorus
stoichiometry across 753 terrestrial plant species in china. New Phytologist 168:
377–385.
Han W, Fang J, Reich PB, Ian Woodward F, Wang Z. 2011. Biogeography and
variability of eleven mineral elements in plant leaves across gradients of climate,
soil and plant functional type in china. Ecology Letters 14: 788–796.
Harrington RA, Fownes JH, Vitousek PM. 2001. Production and resource use
efficiencies in n- and p-limited tropical forests: a comparison of responses to
long-term fertilization. Ecosystems 4: 646–657.
Hayes P, Turner BL, Lambers H, Laliberte E. 2014. Foliar nutrient
concentrations and resorption efficiency in plants of contrasting nutrient-
acquisition strategies along a 2-million-year dune chronosequence. Journal of
Ecology 102: 396–410.
John R, Dalling JW, Harms KE, Yavitt JB, Stallard RF, Mirabello M, Hubbell
SP, Valencia R, Navarrete H, Vallejo M et al. 2007. Soil nutrients influence
spatial distributions of tropical tree species. Proceedings of the National Academy
of Sciences, USA 104: 864–869.
Jordan CF, Kline JR. 1977. Transpiration of trees in a tropical rainforest. Journal
of Applied Ecology 14: 853–860.
Karla YP. 1998. Handbook of reference methods for plant analysis. Boca Raton, FL,
USA: CRC Press.
Kaspari M, Garcia MN, Harms KE, Santana M, Wright SJ, Yavitt JB. 2008.
Multiple nutrients limit litterfall and decomposition in a tropical forest. Ecology
Letters 11: 35–43.
Katabuchi M, Kurokawa H, Davies SJ, Tan S, Nakashizuka T. 2012. Soil
resource availability shapes community trait structure in a species-rich
dipterocarp forest. Journal of Ecology 100: 643–651.
Kattge J, Diaz S, Lavorel S, Prentice I, Leadley P, B€onisch G, Garnier E,
Westoby M, Reich PB, Wright I. 2011. TRY – a global database of plant
traits. Global Change Biology 17: 2905–2935.
Kerkhoff AJ, Fagan WF, Elser JJ, Enquist BJ. 2006. Phylogenetic and growth
form variation in the scaling of nitrogen and phosphorus in the seed plants.
American Naturalist 168: E103–E122.
Kraft NJB, Valencia R, Ackerly DD. 2008. Functional traits and niche-based tree
community assembly in an amazonian forest. Science 322: 580–582.
Lachenbruch B, Moore JR, Evans R. 2011. Radial variation in wood structure
and function in woody plants, and hypotheses for its occurrence. In: Meinzer
FC, Lachenbruch B, Dawson TE, eds. Size- and age-related changes in tree
structure and function. Dordrecht, the Netherlands: Springer Science &
Business Media, 121–164.
Lambers H, Raven JA, Shaver GR, Smith SE. 2008. Plant nutrient-acquisition
strategies change with soil age. Trends in Ecology & Evolution 23: 95–103.
Legendre P. 2011. Lmodel2: Model II regression. R package version 1.7-0.
Marschner H. 1995. Functions of mineral nutrients: macronutrients.Mineral
Nutrition of Higher Plants 2: 379–396.
Martin AR, Erickson DL, Kress WJ, Thomas SC. 2014.Wood nitrogen
concentrations in tropical trees: phylogenetic patterns and ecological correlates.
New Phytologist 204: 484–495.
Mascaro J, Hughes RF, Schnitzer SA. 2012. Novel forests maintain ecosystem
processes after the decline of native tree species. Ecological Monographs 82: 221–
228.
Meerts P. 2002.Mineral nutrient concentrations in sapwood and heartwood: a
literature review. Annals of Forest Science 59: 10.
Mehlich A. 1984. Mehlich 3 soil test extractant: a modification of mehlich
2 extractant. Communications in Soil Science & Plant Analysis 15: 1409–
1416.
Millard P. 1988. The accumulation and storage of nitrogen by herbaceous plants.
Plant, Cell & Environment 11: 1–8.
Mo Q, Zou B, Li Y, Chen Y, Zhang W, Mao R, Ding Y, Wang J, Lu X, Li X.
2015. Response of plant nutrient stoichiometry to fertilization varied with
plant tissues in a tropical forest. Scientific Reports 5: 14605.
Mooney HA, Gulmon SL. 1982. Constraints on leaf structure and function in
reference to herbivory. BioScience 32: 198–206.
Muller-Landau HC. 2004. Interspecific and inter-site variation in wood specific
gravity of tropical trees. Biotropica 36: 20–32.
Ordo~nez JC, Van Bodegom PM, Witte JPM, Wright IJ, Reich PB, Aerts R.
2009. A global study of relationships between leaf traits, climate and soil
measures of nutrient fertility. Global Ecology and Biogeography 18: 137–149.
Ostertag R. 2010. Foliar nitrogen and phosphorus accumulation responses after
fertilization: an example from nutrient-limited hawaiian forests. Plant and Soil
334: 85–98.
Palmiotto PA, Davies SJ, Vogt KA, Ashton MS, Vogt DJ, Ashton PS. 2004.
Soil-related habitat specialization in dipterocarp rain forest tree species in
borneo. Journal of Ecology 92: 609–623.
Paradis E, Claude J, Strimmer K. 2004. APE: analyses of phylogenetics and
evolution in r language. Bioinformatics 20: 289–290.
Phillips OL, Vargas PN, Monteagudo AL, Cruz AP, Zans M-EC, Sanchez WG,
Yli-Halla M, Rose S. 2003.Habitat association among Amazonian tree species:
a landscape-scale approach. Journal of Ecology 91: 757–775.
Pietsch KA, Ogle K, Cornelissen JH, Cornwell WK, B€onisch G, Craine JM,
Jackson BG, Kattge J, Peltzer DA, Penuelas J. 2014. Global relationship of
wood and leaf litter decomposability: the role of functional traits within and
across plant organs. Global Ecology and Biogeography 23: 1046–1057.
Poorter L, Bongers F. 2006. Leaf traits are good predictors of plant performance
across 53 rain forest species. Ecology 87: 1733–1743.
Poorter L, Kitajima K. 2007. Carbohydrate storage and light requirements of
tropical moist and dry forest tree species. Ecology 88: 1000–1011.
Poorter L, McDonald I, Alarcon A, Fichtler E, Licona JC, Pe~na-Claros M,
Sterck F, Villegas Z, Sass-Klaassen U. 2010. The importance of wood traits
and hydraulic conductance for the performance and life history strategies of 42
rainforest tree species. New Phytologist 185: 481–492.
Pyke CR, Condit R, Aguilar S, Lao S. 2001. Floristic composition across a
climatic gradient in a neotropical lowland forest. Journal of Vegetation Science
12: 553–566.
Quesada CA, Lloyd J, Schwarz M, Baker TR, Phillips OL, Pati~no S,
Czimczik C, Hodnett MG, Herrera R, Arneth A et al. 2009. Regional and
large-scale patterns in amazon forest structure and function are mediated by
variations in soil physical and chemical properties. Biogeosciences Discussions
6: 3993–4057.
Reich PB, Oleksyn J. 2004. Global patterns of plant leaf n and p in relation to
temperature and latitude. Proceedings of the National Academy of Sciences, USA
101: 11001–11006.
Rozendaal D, Hurtado V, Poorter L. 2006. Plasticity in leaf traits of 38 tropical
tree species in response to light; relationships with light demand and adult
stature. Functional Ecology 20: 207–216.
Santiago LS. 2007. Extending the leaf economics spectrum to decomposition:
evidence from a tropical forest. Ecology 88: 1126–1131.
Schreeg LA, Santiago LS, Wright SJ, Turner BL. 2014. Stem, root, and older
leaf N: P ratios are more responsive indicators of soil nutrient availability than
new foliage. Ecology 95: 2062–2068.
Sinclair TR, Vadez V. 2002. Physiological traits for crop yield improvement in
low N and P environments. Plant and Soil 245: 1–15.
Slik JW, Paoli G, McGuire K, Amaral I, Barroso J, Bastian M, Blanc L, Bongers
F, Boundja P, Clark C et al. 2013. Large trees drive forest aboveground
biomass variation in moist lowland forests across the tropics. Global Ecology and
Biogeography 22: 1261–1271.
Stanley WG, Montagnini F. 1999. Biomass and nutrient accumulation in
pure and mixed plantations of indigenous tree species grown on poor soils
in the humid tropics of Costa Rica. Forest Ecology and Management 113:
91–103.
Tanner EVJ. 1977. Four montane rain forests of Jamaica: a quantitative
characterization of the floristics, the soils and the foliar mineral levels, and a
discussion of the interrelations. Journal of Ecology 65: 883–918.
Tanner EVJ. 1985. Jamaican montane forests: nutrient capital and cost of
growth. Journal of Ecology 73: 553–568.
Tanner EVJ, Vitousek PM, Cuevas E. 1998. Experimental investigation of
nutrient limitation of forest growth on wet tropical mountains. Ecology 79: 10–
22.
New Phytologist (2016) 
 2016 The Authors
New Phytologist
 2016 New Phytologist Trustwww.newphytologist.com
Research
New
Phytologist14
Townsend AR, Cleveland CC, Asner GP, Bustamante MMC. 2007. Controls
over foliar N: P ratios in tropical rain forests. Ecology 88: 107–118.
Turner BL, Engelbrecht BM. 2011. Soil organic phosphorus in lowland tropical
rain forests. Biogeochemistry 103: 297–315.
Turner BL, Romero TE. 2009. Short-term changes in extractable inorganic
nutrients during storage of tropical rain forest soils. Soil Science Society of
America Journal 73: 1972.
Vitousek PM. 1984. Litterfall, nutrient cycling, and nutrient limitation in
tropical forests. Ecology 65: 285–298.
Vitousek PM, Matson PA, Turner DR. 1988. Elevational and age gradients
in hawaiian montane rainforest: foliar and soil nutrients. Oecologia 77:
565–570.
Wang D, Bormann FH, Lugo AE, Bowden RD. 1991. Comparison of nutrient-
use efficiency and biomass production in five tropical tree taxa. Forest Ecology
and Management 46: 1–21.
Wardle DA, Bardgett RD, Klironomos JN, Set€al€a H, Van Der Putten WH,
Wall DH. 2004. Ecological linkages between aboveground and belowground
biota. Science 304: 1629–1633.
Webb CO, Ackerly DD, Kembel SW. 2008. Phylocom: software for the analysis
of phylogenetic community structure and trait evolution. Bioinformatics 24:
2098–2100.
Webb CO, Donoghue MJ. 2005. Phylomatic: tree assembly for applied
phylogenetics.Molecular Ecology Notes 5: 181–183.
Weedon JT, Cornwell WK, Cornelissen JHC, Zanne AE, Wirth C, Coomes
DA. 2009. Global meta-analysis of wood decomposition rates: a role for trait
variation among tree species? Ecology Letters 12: 45–56.
Wikstr€om N, Savolainen V, Chase MW. 2001. Evolution of the angiosperms:
calibrating the family tree. Proceedings of the Royal Society of London B 268:
2211–2220.
Wright SJ, Yavitt JB, Wurzburger N, Turner BL, Tanner EVJ, Sayer EJ,
Santiago LS, Kaspari M, Hedin LO, Harms KE et al. 2011. Potassium,
phosphorus, or nitrogen limit root allocation, tree growth, or litter production
in a lowland tropical forest. Ecology 92: 1616–1625.
W€urth MK, Pelaez-Riedl S, Wright SJ, K€orner C. 2005. Non-structural
carbohydrate pools in a tropical forest. Oecologia 143: 11–24.
Zanne AE, Oberle B, Dunham KM, Milo AM, Walton ML, Young DF. 2015. A
deteriorating state of affairs: how endogenous and exogenous factors determine
plant decay rates. Journal of Ecology 103: 1421–1431.
Supporting Information
Additional supporting information may be found in the online
version of this article.
Table S1 Species mean wood Ca, K, Mg, N and P concentrations
for 106 trees species sampled in this study.
Table S2 Summary of log–log linear regression models testing
the relationship between community weighted mean (CWM)
wood nutrient concentrations and soil nutrient availability and
species mean wood nutrient concentrations and soil nutrient
availability.
Table S3Major axis regression model fits of pairwise combina-
tions of the phylogenetically independent contrasts (PIC) of
species mean wood nutrient (Ca, K, Mg, N and P) and wood
density (WD) measured in 76 tree species from six forest sites at
Fortuna.
Table S4Major axis regression model fits of the scaling relation-
ship between nutrient concentrations in the outermost 5 cm
annulus of wood vs the adjacent 5–10 cm annulus of wood cores
(log(inner) ~ log(a) + log(outer)9 b) for 110 trees in Panama.
Please note: Wiley Blackwell are not responsible for the content
or functionality of any supporting information supplied by the
authors. Any queries (other than missing material) should be
directed to the New Phytologist Central Office.

 2016 The Authors
New Phytologist
 2016 New Phytologist Trust
New Phytologist (2016)
www.newphytologist.com
New
Phytologist Research 15