How cellulose-based leaf toughness and lamina density
contribute to long leaf lifespans of shade-tolerant species
Kaoru Kitajima1,2, Anna-Maria Llorens2, Carla Stefanescu1, Marta Vargas Timchenko2, Peter W. Lucas3 and
S. Joseph Wright2
1Department of Biology, University of Florida, Gainesville, FL 32611, USA; 2Smithsonian Tropical Research Institute, Balboa, Panama; 3Department of Bioclinical Sciences, Kuwait University,
Kuwait
Author for correspondence:
Kaoru Kitajima
Tel: +1 352 3924234
Email: kitajima@ufl.edu
Received: 31 January 2012
Accepted: 29 April 2012
New Phytologist (2012) 195: 640?652
doi: 10.1111/j.1469-8137.2012.04203.x
Key words: anti-herbivory defence,
cellulose, herbivory, lamina density, leaf
lifespan, leaf toughness, seedling survival,
shade tolerance.
Summary
? Cell wall fibre and lamina density may interactively affect leaf toughness and leaf lifespan.
Here, we tested this with seedlings of 24 neotropical tree species differing in shade tolerance
and leaf lifespan under standardized field conditions (140?867 d in gaps; longer in shade).
We quantified toughness with a cutting test, explicitly seeking a mechanistic linkage to fibre.
? Lamina density, but not fracture toughness, exhibited a plastic response to gaps vs shade,
while neither trait was affected by leaf age. Toughness corrected for lamina density, a recently
recognized indicator of material strength per unit mass, was linearly correlated with cellulose
content per unit dry mass.
? Leaf lifespan was positively correlated with cellulose and toughness in shade-tolerant
species but only weakly in gap-dependent species. Leaf lifespan was uncorrelated with lamina
thickness, phenolics and tannin concentrations. In path analysis including all species, leaf
lifespan was directly enhanced by density and toughness, and indirectly by cellulose via its
effect on toughness. Different suites of leaf traits were correlated with early seedling survival
in gaps vs shade.
? In conclusion, cellulose and lamina density jointly enhance leaf fracture toughness, and
these carbon-based physical traits, rather than phenolic-based defence, explain species differ-
ences in herbivory, leaf lifespan and shade survival.
Introduction
Forest trees differ widely in demographic, morphological and
physiological traits in relation to light environments favourable
for juvenile recruitment and persistence (Bazzaz, 1979; Swain &
Whitmore, 1988; Welden et al., 1991; Poorter, 2006). The func-
tional basis for such species differences in regeneration light
demands may be understood in terms of light utilization strate-
gies (Grime, 1979; Givnish, 1988; Kohyama, 1991) and resist-
ance to natural enemies (Coley, 1983; Augspurger, 1984; Dalling
et al., 1997). These differences are often summarized in terms of
the contrasts between shade-tolerant and gap-dependent guilds of
species. In shaded forest understoreys where light availability
limits potential carbon gain, long leaf lifespan is necessary to
recoup leaf construction costs, and thus it is critical that leaves
are well defended against biotic and abiotic hazards (Bloom
et al., 1985; Coley et al., 1985; Endara & Coley, 2011). By con-
trast, in tree fall gaps, natural selection favours the competitive
advantage achieved by rapid height growth and turnover of leaves
with high photosynthetic capacity (Bongers & Pompa, 1990).
Such contrasting selective regimes are partially responsible for the
ubiquitous trait correlations among leaf mass per area (LMA),
nitrogen concentration (N), photosynthetic rates and leaf lifespan
(Reich et al., 1992; Wright et al., 2004). These leaf economic
spectrum traits, as well as leaf traits that enhance defence against
biotic and abiotic hazards, are considered central to the func-
tional trait syndrome associated with the growth-mortality
trade-offs observed for juveniles of forest tree species (Kitajima,
1994; Pacala et al., 1996; Gilbert et al., 2006; Sterck et al., 2006;
Wright et al., 2010).
Leaf defence can be achieved by physical or chemical properties
of molecules that make up leaves, the effectiveness of which may
depend on hazard agents and environments. Physical defence
traits include those based on toughening or hardening of tissues
or a deterrent geometry such as spines and thorns (Grubb, 1986,
1992; Lucas et al., 2000). Toughness-based defence, in particu-
lar, reflects quantities, types and arrangements of cell wall
components. Chemical defence molecules that operate as toxins,
by contrast, may be effective in smaller quantities, but can be
circumvented by co-evolved natural enemies (Coley et al., 1985).
Comparing multiple putative physical and chemical defence
traits across 46 tropical tree species, Coley (1983) found that
toughness (measured as punch resistance) and cellulose content
were the two traits that best explain species difference in
Research
640 New Phytologist (2012) 195: 640?652
www.newphytologist.com
 2012 The Authors
New Phytologist  2012 New Phytologist Trust
herbivory rates and leaf lifespan. Although not examined by
Coley (1983), LMA and its two components, thickness and lam-
ina density, may also enhance toughness and leaf lifespan (Wright
et al., 2004; Kitajima & Poorter, 2010; Onoda et al., 2011).
However, mechanistic relationships between these traits remain
ambiguous without proper quantification in a manner consistent
with conceptual and methodological advances in the understand-
ing of leaf toughness (Lucas et al., 1991; Aranwela et al., 1999;
Sanson et al., 2001). Once trait definitions and measurement
units are properly formulated, it becomes possible to analyse
the relationship between matter distribution and mechanical
properties.
In mechanical terms, a leaf lamina is a complex heterogeneous
structure containing cells with distinct shapes and cell wall archi-
tectures. Wall component molecules (e.g. cellulose microfibrils,
hemicelluloses, pectins, lignin and wall proteins) are combined in
different ways in different cell types to result in contrasting
mechanical properties within (e.g. tough and rigid veins vs elastic
mesophylls) and among leaves (e.g. sun vs shade leaves, slow- vs
fast-growing species). Here, we recognize this structural hierarchy
via a priori hypotheses illustrated in Fig. 1. Matter allocation
patterns can be quantified per unit dry mass, per unit volume or
per unit surface area. These three methods of standardizing
matter allocation patterns correspond to three ways of normalizing
mechanical measurements of resistance against fracture (Onoda
et al., 2011), as summarized in Table 1 for fracture resistance
quantified with cutting tests. Work-to-shear (Ws) is a measure of
structural resistance expressed per unit cut length, and as such
it reflects both lamina thickness (T ) and fracture toughness
(s, defined as work to fracture per unit cross sectional area). s is
expected to increase with lamina density (q, mass per unit vol-
ume), because when q is greater, more dry matter is encountered
per cross-sectional area during the cutting operation. Fracture
toughness corrected for this density effect (hereafter, density-
corrected toughness, c) corresponds to the average resistance per
unit dry mass of materials comprising the leaf. We predict that
values of c correlate with cell wall fibre content per unit dry mass
(Fig. 1, top).
Operationally, the three measures of toughness, Ws, s and c
are determined from each other as below (Onoda et al., 2011).
Ws ? T  s Eqn 1
s ? q  c Eqn 2
In a given leaf, these two relationships are absolute. But, there
is no a priori reason to know what proportion of interspecific
variation in Ws is attributable to species differences in s vs T
(Eqn 1), and what proportion of interspecific variation in s is
attributable to species differences in c vs q (Eqn 2). In a global
overview, Onoda et al. (2011) found that c explains 60?70% of
variation in Ws, most variation being within sites. However, the
mechanistic basis for variation in c or ecological correlates of c
remains unexplored.
Because of the statistical nonindependence of the three mea-
sure of toughness as described above, it is not possible to test the
full conceptual model illustrated in Fig. 1, but several recent
studies shed light on the predicted causal relationships. In juve-
nile leaves of evergreen tree species, Ws, s and q, but not T, cor-
relate with palatability, leaf lifespan, growth, survival and
regeneration light requirements (Kitajima & Poorter, 2010; Lusk
et al., 2010). By contrast, Westbrook et al. (2011) asked what
makes a leaf tough from an evolutionary perspective by compar-
ing phylogenetic independent contrasts for leaf toughness and
fibre contents of subcanopy individuals of 197 species of trees
and shrubs. Their results show that s correlates with q and %cel-
lulose, and that s, q and %cellulose, but not T, covary with juve-
nile survival in shade. None of these three studies explicitly asked
how density-corrected toughness, c, might correlate with leaf
lifespan, growth, survival and habitat preference. Phenotypic
plasticity of toughness-related traits in response to light environ-
ment and leaf age also remains unexplored. It is well known that
young expanding leaves are less tough than mature leaves (Coley,
1983; Choong, 1996; Grubb et al., 2008), but leaf age effects
after full expansion are unknown even though leaf lifespan may
be exceedingly long for juveniles of shade-tolerant trees.
In order to evaluate interspecific variation and phenotypic
plasticity of leaf toughness-related traits, we planted 24 tree
species in replicate gap and understorey common garden experi-
ments in a Panamanian moist forest. We examined correlations
among functional and demographic traits in relation to their
regeneration light demands, following the conceptual framework
illustrated in Fig. 1 that relates carbon-based leaf defences includ-
ing mechanical (Ws, s, c, T and q) and chemical (phenols and
C-based physical defence C based chemical
Ma?er
alloca?on
Mechanical
proper?es
-
defence
Phenolics
HerbivoryFracture resistance of
Dry mass per unit
l ( )
Cell-wall fiber per
unit dry mass
TanninsFracture resistance of
each unit dry mass (? )
Leaf lifespan
each unit volume (? )
Fracture resistance per
unit cut length (Ws )
vo ume ?
Dry mass per unit
leaf area (LMA)
Lamina
thickness(T)
Fig. 1 A conceptual framework for the relationships among traits that
contribute to carbon-based physical and chemical defence of leaves. No a
priori association is expected between physical and chemical defence
traits, while strong a priori relationships are expected among physical
defence traits (see text). We developed alternative path models (e.g.
Fig. 4, Supporting Information Fig. S2) based on these a priori expecta-
tions on direct and indirect effects of mechanical defence traits on leaf
demography. For example, we predicted that lamina density and cell wall
fibre content may reduce herbivory and enhance leaf lifespan, but their
effects may be direct as well as indirect through their effects on toughness
(i.e. s and c).
New
Phytologist Research 641
 2012 The Authors
New Phytologist  2012 New Phytologist Trust
New Phytologist (2012) 195: 640?652
www.newphytologist.com
tannins) defences. We compared alternative path models to
evaluate predicted multivariate relationships among tough-
ness-related traits and leaf lifespan as illustrated in Fig. 1. By
doing so, we could test whether fibre and lamina density had
direct effects on leaf lifespan in addition to their indirect effects
mediated through toughness. Based on the result of Westbrook
et al. (2011), we expected that %cellulose, but not other fibre
components, would correlate with density-corrected toughness,
and that this cellulose-based toughness should correlate positively
with leaf lifespan and juvenile survival in shade, both within and
across gap- and shade-tolerant guilds. The results addressed the
following questions:.
? How do Ws and its underlying components, T, s, q, and c,
and cell wall fibre contents, covary across species, leaf ages, and
light environments (gap vs shaded understorey)?
? Which aspects of carbon-based defence correlate with herbiv-
ory level and leaf lifespan across species? More specifically, we
were interested in comparing carbon-based chemical defence
(phenolics and tannins) with fibre content and mechanical resist-
ance (Ws, s and c).
? How do these leaf-level traits differ between regeneration
guilds (gap-dependent vs shade-tolerant) and in relation to
seedling survival in gaps and shade?
Materials and Methods
Site and species
The study was conducted in the Barro Colorado Nature Monu-
ment (BCNM) in Panama. Leigh et al. (1982) describe the flora,
ecology and environment of the seasonal moist tropical forests of
the BCNM. Annual rainfall averages 2600 mm, 90% of which
falls between May and December. We established six common
gardens in 20-m tall secondary forest on the Buena Vista Penin-
sula in May 2002. Each garden was approximately rectangular
(7?8 m ? 8?10 m) and was enclosed by a metal-mesh fence to
exclude ground-dwelling vertebrates. Three gardens were estab-
lished in natural treefall gaps whose edges were cleared and
widened to maintain the same-sized opening throughout the
experiment. The remaining three gardens were located in shaded
understorey. We took canopy photos at five positions per garden,
and calibrated them against photosynthetic active radiation
(PAR) measured with quantum sensors (LI-185; LICOR,
Lincoln, NE, USA) for a minimum of 3 d in the centre of each
garden. The mean PAR during the wet season was 0.5?0.8% of
above-canopy PAR (0.14?0.18 mol d)1) in understorey shade
gardens and 23.4%, 37.7% and 50.7% of above-canopy PAR
(6.7, 11.3 and 11.5 mol d)1) in gap gardens.
The 24 study species (Table 2) were common tree species,
including gap-dependent species rarely observed in the shaded
understorey and shade-tolerant species whose seedlings and
saplings were frequent in the shaded understory of old growth
forest in BCNM (Condit et al., 2006; Comita et al., 2010;
K. Kitajima, A.-M. Llorens, C. Stefanescu et al. unpublished data
and field observations; see Supporting Information Notes S1 for
more information). The 24 species ranged broadly in leaf lifespan
and seedling survival in shade (Table 2). Seedlings were trans-
planted into the six common gardens between April 2002 and
May 2003. Seeds were collected from forests within the BCNM,
and germinated under 30% or 1% light availability in plastic
trays filled with washed sand and forest soil from the top 30-cm
of the soil profile. After radicle emergence and before full expan-
sion of the first leaf, 6?12 seedlings of each species germinated in
high and low light were transplanted to two randomly chosen
sections of each gap and understorey garden, respectively. Seed-
lings were planted at least 30 cm from one another. Mortality
within the first week post-transplant was negligible as seedlings
were transplanted during the rainy season.
Survival census for leaves and seedlings
Survival of all transplanted seedlings was recorded by weekly cen-
sus. Production and survival of leaves were monitored on a subset
of seedlings (? 3 individuals per garden per species). At each
monthly census, we numbered newly expanded leaves sequen-
tially with a nontoxic waterproof marker, and recorded
presence ? absence of previously marked leaves. Both censuses
continued until July 2006. For early seedling survival, we
Table 1 Summary of material distribution traits expected to influence leaf mechanical properties, in terms of leaf fracture resistance measured with a
cutting test
Matter distribution traits Corresponding fracture resistance traits
Matter distribution
standardization Traits examined in this study
Mechanical measurement
normalization Traits examined in this study
Quantity per unit dry mass %cellulose, %lignin, %ADF, %NDF Fracture toughness divided by density c: density-corrected toughness
(mJ g)1 m)
Quantity per unit volume q: Lamina density (g cm)3),
cellulose per volume (g cm)3)
Work to fracture per unit cross-sectional
area of cut
s: fracture toughness
(J m)2)
Quantity per unit surface
area
LMA (g m)2) Work to fracture per unit cut length Ws: work-to-shear (J m
)1)
Material distribution traits are grouped by their standardization schemes in the left side, whereas the fracture resistance trait that should be compared with
each type of matter distribution trait is listed in the right within the same row. See Fig. 1 for the expected interrelationships among these traits.
ADF, acid detergent fibre; NDF, neutral detergent fibre; LMA, leaf mass per area.
642 Research
New
Phytologist
 2012 The Authors
New Phytologist  2012 New Phytologist Trust
New Phytologist (2012) 195: 640?652
www.newphytologist.com
calculated the proportion of seedlings alive 90 d post-transplant,
pooling conspecific seedlings across gardens within each light
environment (Table 2). The proportion of seedlings surviving
90 d in shade gardens was positively correlated with the shade
tolerance index of saplings (1?4 cm diameter at breast height)
used in the community-wide analysis of Comita et al. (2010)
(n = 22, r = 0.67, P = 0.008). The age at leaf death was analysed
using the Kaplan?Meier method, which accounts for censored
leaves that were removed from the census due to harvest or death
of the plant. Table S1 presents sample sizes for leaf demography,
which ranged from 125 to 1273 leaves per species.
Measurements of herbivory and leaf disks across leaf age
gradients
Leaf traits were determined between May and July 2004 in the
BCNM at the middle of the field campaign (June 2004), when
seedling age ranged from 272 to 749 d among species (aver-
age = 572 d). Because leaf traits may change with leaf age (e.g.
Kitajima et al., 1997, 2002; Mediavilla et al., 2008), we pres-
elected five pairs of leaves of similar age per plant to cover the full
range of leaf ages within each individual. For each preselected
leaf, the areal proportion removed by herbivores was measured
nondestructively with a gridded transparent sheet. Levels of her-
bivory were represented by two variables ? the average percentage
of area missing (henceforth, %area) and the percentage of leaves
with holes and ?or missing edges (henceforth, % leaf damage).
One leaf disk (1.41 cm2 each) was collected from each leaf
using a cork borer, avoiding major veins as much as possible.
Disks were placed between moist filter paper in plastic bags in a
cooler chilled with ice and brought back to the laboratory.
Subsequent monitoring of leaf survival detected no significant
difference in leaf demography between intact leaves and sampled
leaves. After determining fresh mass, one disk from each pair of
same-aged leaves was used for measurements of toughness, lam-
ina thickness and dry mass contents, while a second disk was used
for methanol extraction to quantify phenolics and tannin. In
gaps, all 24 species were sampled. In shade, seedlings were too
small for many species and only four species with sufficient
seedling size and number could be included.
Fracture toughness was measured with the cutting method
developed by Lucas & Pereira (1990), using a pair of scissors
mounted on a portable universal tester (Darvell et al., 1996).
Lamina work-to-shear (Ws, J m
)1) was obtained by dividing total
work by the length of the cutting pass. Lamina thickness (T) was
measured using an analogue thickness gauge (Teclock SM112,
Table 2 Study species and their ecological characteristics, including regeneration guild (T, shade-tolerant; G, gap-dependent), median leaf lifespan and
seedling survival in gap and shade common gardens, and mean %leaf-damage (proportion of leaves showing herbivory damage) in gap gardens
Species Family Regeneration guild
Median leaf
lifespan (d)1 Seedling survival2
%leaf-damageGap Shade Gap Shade
Aspidosperma spruceanum Apocynaceae T 867 > 1074 0.96 0.96 23.0
Calophyllum longifolium Clusiaceae T 701 1253 0.71 0.79 45.8
Tetragastris panamensis Burseraceae T 672 1044 0.85 1.00 72.7
Gustavia superba Lecythidaceae T 643 > 1170 0.92 0.75 47.0
Trichilia tuberculata Meliaceae T 643 > 1260 1.00 0.94 19.5
Brosimum alicastrum Moraceae T 614 1028 0.67 0.90 45.8
Randia armata Rubiaceae T 531 957 0.97 0.97 29.0
Posoqueria latifolia Rubiaceae T 523 1104 0.91 1.00 63.3
Virola surinamensis Myristicaceae T 504 1091 0.93 1.00 74.8
Pouteria reticulata Sapotaceae T 447 > 991 0.96 0.95 21.6
Beilschmiedia pendula Lauraceae T 420 1119 0.83 0.85 54.4
Lacistema aggregatum Salicaceae G 391 584 1.00 0.88 27.7
Pachira sessilis Malvaceae T 336 553 0.88 0.97 78.2
Eugenia nesiotica Myrtaceae T 307 636 0.94 0.56 32.4
Dipteryx oleifera Fabaceae G 280 na 0.94 na 95.0
Vochysia ferruginea Vochysiaceae G 280 728 0.93 0.63 50.2
Tabernaemontana arborea Apocynaceae T 252 441 1.00 0.97 65.8
Pentagonia macrophylla Rubiaceae G 252 483 0.83 0.67 73.8
Tocoyena pittieri Rubiaceae T 252 593 1.00 0.96 23.6
Genipa americana Rubiaceae T 224 532 0.89 0.94 76.2
Hybanthus prunifolius Violaceae T 224 350 1.00 0.88 52.3
Jacaranda copaia Bignoniaceae G 224 na 0.71 0.18 81.5
Anacardium excelsum Anacardiaceae G 161 167 0.94 0.61 77.3
Ceiba pentandra Malvaceae G 140 na 0.87 0.00 66.2
Species are listed in decreasing order of median leaf lifespan in gaps. See the Materials and Methods section for details.
1For species with long leaf lifespan, > x d, where x indicates the maximum leaf age recorded at the end of the study (proportion of leaves alive at the
indicated age: A. spruceanum: 87%, G. superba: 55%, T. tuberculata: 61%, and P. reticulata: 81%, according to Kaplan?Meier estimates of survivorship
curves); na, not available due to high seedling mortality in shade.
2Proportion of seedlings that survived for 90 d after planting to gaps and shade common gardens.
New
Phytologist Research 643
 2012 The Authors
New Phytologist  2012 New Phytologist Trust
New Phytologist (2012) 195: 640?652
www.newphytologist.com
Nagano, Japan). Lamina fracture toughness (s, J m)2) was calcu-
lated by dividing Ws by T. All cut pieces were collected and dried
at 65C to determine dry mass. Leaf mass per area (LMA) was
determined by dividing disk dry mass by disk area. Leaf tissue
density (q) was determined by dividing LMA by T. Density-
corrected toughness was calculated by dividing s by q (Eqn 2).
Analysis of phenolics and tannin followed the micro-analytical
method described in detail by Lucas et al. (2001) and Dominy
et al. (2003), after combining leaf disks to obtain 0.1 g fresh
sample per plant (see Notes S1 for further details).
Nitrogen and fibre contents
Leaf nitrogen concentrations were determined from a separate set
of leaf disks sampled from 10 species with sufficiently large plants
in gaps as part of a study focusing on leaf age effects on photosyn-
thetic physiology (Stefanescu, 2006). After drying at 60C to
constant weight, nitrogen content per unit mass (%N) was deter-
mined using an elemental analyzer (Costech Analytical Model
4010, Valencia, CA, USA). Because of strong negative effects of
leaf age, %N was estimated for a standardized age of 30 d from
species-specific regressions of %N vs leaf age for comparison
across species.
Cell wall fibre contents were determined for 21 species with a
method modified from Van Soest (1963) to determine neutral
detergent fibre (%NDF), acid detergent fibre (%ADF) and
%lignin on an ash-free mass basis. This used dried leaf samples
from 3- to 4-yr-old plants that were destructively harvested in
2005?2006 from the same common garden experiment (see
Notes S1 for further details). The %hemicellulose (= %NDF )
%ADF) value represents the water-soluble fibre fraction consist-
ing mostly of hemicelluloses and pectin, which provide pliable
cross-links between cellulose microfibrils (Carpita & Gibeaut,
1993). The %cellulose (= %ADF ) %lignin) value is a measure
of the relative abundance of cellulose microfibrils which influence
the directional tensile strength of primary and secondary cell
walls (Carpita & Gibeaut, 1993). By contrast, large chemically
recalcitrant lignin molecules consist of irregular conjugations of
many phenolic molecules and provide permanent and rigid
cross-linkages within the secondary cell wall (Taiz & Zeiger,
2002). The structural combination of these heterogeneous mate-
rials with contrasting mechanical properties determines the
toughness of leaf tissues (Jeronimidis, 1980; Lucas et al., 2000).
Fibre contents per unit volume (e.g. g cellulose cm)3) were also
examined for potential correlation with nonfibre traits.
Statistical analysis
We evaluated path models based on the a priori relationships
shown in Fig. 1, but because Ws, s and c were not measured
independently (Eqns 1 and 2), we compared alternative models,
each including only one of these three traits. The overall model
fit and the significance of each causal path were examined using
structural equation modelling software (AMOS v5, AMOS
Development Corporation; Grace, 2006). Path coefficients (b)
were estimated with log-ratio maximum likelihood. The total
effect of one variable on another is the total of path coefficients
for direct paths plus the products of path coefficients for indirect
paths via other variables. Model fit was assessed with a v2-test
that determines whether hypothesized trait correlations deviate
significantly from observed correlations (Grace, 2006). Non-
significance (P > 0.05) meant failure to reject the path model.
Alternative causally plausible models were evaluated through
addition and subtraction of individual paths and variables, and
evaluated with the Akaike Information Criterion (AIC).
Phylogenetically independent contrasts (PICs) were obtained
with Phylomatic software available online (http://www.phylo
diversity.net/phylomatic/), based on the APG3-derived megatree.
One polytomy reduced the total number of PICs that could be
analysed to 22 for the 24 species. Pairwise PIC correlations were
calculated with the Phylocom v. 4.1, Analysis of Traits module
(Webb et al., 2008), assuming uniform branch lengths. Because
observed variation in leaf toughness and its relationship with leaf
lifespan may be largely due to variations between gap-dependent
and shade-tolerant guilds rather than within each guild (Dominy
et al., 2008), we used ANCOVA to evaluate how trait correla-
tions may be influenced differently for the two regeneration
guilds. All standard statistical analyses (Pearson correlations,
ANOVA and ANCOVA) were performed with JMP v8.0 (SAS
Institute, Cary, NC, USA).
Results
Species differences in leaf lifespan, herbivory and leaf traits
The 24 species varied widely in leaf lifespan, herbivory damage
and seedling survival (Table 2). The median leaf lifespan varied
from 160 to 867 d and from 167 to > 1260 d for seedlings
growing in gaps and in shade, respectively. In all species, leaf
lifespan was always longer for shade-grown seedlings than
gap-grown seedlings (Table 2, paired t = 8.0, P < 0.001). For
17 species for which median leaf lifespan could be determined
in both environments, there was a strong correlation between
the two environments (r = 0.90, P < 0.001, Fig. S1).
Shade-tolerant species tended to have longer leaf lifespan than
gap-dependent species within a given environment, but there
was substantial variation within each guild and overlap
between guilds (Table 2, Fig. S1). Herbivory also exhibited
wide ranges (minimum to maximum of species means), both
in terms of proportion of leaves showing damage
(%leaf-damage, 19.5?95%, Table 2) and proportion of area
missing (%area, 0.3?41.3%, Table 3).
Leaf functional traits measured for gap-grown seedlings varied
widely among species with negligible effects of leaf age except for
%N (see Tables 3 and S1 for species mean values and ANCOVA
results for the effects of species, leaf age and their interaction).
For toughness-related traits, species means ranged two- to
eight-fold as follows: lamina thickness (0.13?0.29 mm), density
(0.24?0.42 g cm)3), LMA (32.8?86.4 g m)2), Ws (0.019?0.155
J m)1), s (83.5?525.5 J m)2), %cellulose (9.7?33.6%), and
%N (1.38?3.07%, at 30 d after leaf full expansion). Many of the
predicted correlations amongst these traits (Fig. 1) were
644 Research
New
Phytologist
 2012 The Authors
New Phytologist  2012 New Phytologist Trust
New Phytologist (2012) 195: 640?652
www.newphytologist.com
significant and similar in strength and direction for species mean
and PIC values (Table 4), as explained further in the next sec-
tion. For traits that were not related to toughness, species means
also ranged widely (Table S1) as follows: NDF (37.1?68.5%),
lignin (8.1?25.9%), phenolics (1.2?23.1 mg gallic acid equiva-
lent g)1 dry mass), tannins (0?18.7 quebracho tannin equiva-
lent g)1 dry mass; 15 species had no detectable tannin with the
radial diffusion method).
Correlates of leaf toughness at material, tissue and
structural levels
Cell wall fibre contents per unit mass, including %hemicellulose,
%cellulose and %lignin, varied independently of each other
(Table S2). As expected (Fig. 1), density-corrected toughness (c)
correlated strongly and positively with %cellulose (r = 0.84,
P < 0.001, Fig. 2a). By contrast, c had a marginally significant,
negative correlation with %hemicellulose (r = ) 0.44, P = 0.04)
and no significant correlation with the other fibre contents
(Tables 4, S2). Although %NDF is often considered to be a good
approximation of cell wall dry mass fraction (approximately,
%NDF = %hemicellulose + %cellulose + %lignin), %NDF was
not correlated with c (r = ) 0.33, P = 0.60).
As expected (Fig. 1), fracture toughness (s) was correlated with
c, %cellulose, cellulose : N ratio, and cellulose per unit volume
(Table 4). Unexpectedly, s did not have a significant pairwise
correlation with lamina density, q (r = 0.33, P = 0.11). Yet, s
was more strongly correlated with cellulose per unit volume than
with %cellulose (Table 4, Fig. 2b), presumably because the
former incorporates the joint effects of q and %cellulose. q varied
independently of T and %cellulose (Table 4, P > 0.4), while T
and %cellulose were weakly correlated (Table 4, P = 0.02). As
expected (Fig. 1), work-to-shear (Ws) was correlated with both s
(r = 0.94) and T (r = 0.80).
Phenotypic plasticity of toughness-related traits with leaf
age and light environment
Leaf age had a nonsignificant or negligibly small effect on all leaf
traits of gap-grown seedlings except for %N (which declined with
leaf age, Table 3) and leaf size (which was smaller for older leaves
that had developed when seedlings were smaller, Table S1). Light
environment, on the other hand, affected several tough-
ness-related traits in four to seven shade-tolerant species for
which leaf traits were also determined in shade (see Table S3 for
species means for shade-grown plants and the results of two-way
Table 3 Species means for selected leaf traits relevant for toughness for gap-grown juveniles (listed in the order of median leaf lifespan as in Table 2)
Species %herbivory
Lamina thickness
(T, mm)
Lamina density
(q, g cm)3)
LMA
(g m)2)
Work-to-shear
(Ws, J m
)1)
Fracture
toughness
(s, J m)2)
Density-corrected
toughness
(c, mJ g)1 m) %Cellulose %Nitrogen
A. spruceanum 0.3 0.22 0.36 78.1 0.097 438.6 1.22 28.5 1.71
C. longifolium 5.8 0.29 0.29 86.4 0.155 525.5 1.80 33.6 1.38
T. panamensis 6.9 0.16 0.37 57.5 0.069 429.0 1.15 21.3 1.71
G. superba 10.4 0.15 0.37 52.9 0.050 349.7 0.95 23.2 2.89
T. tuberculata 6.8 0.16 0.42 68.7 0.048 289.4 0.69 13.2 2.17
B. alicastrum 4.4 0.21 0.32 66.6 0.052 253.4 0.79 16.3
R. armata 3.1 0.19 0.33 61.0 0.039 209.3 0.63
P. latifolia 18.3 0.21 0.34 71.6 0.060 284.3 0.84 21.2
V. surinamensis 11.1 0.15 0.31 47.9 0.035 223.8 0.71 17.6 1.88
P. reticulata 6.0 0.14 0.38 52.4 0.038 267.4 0.70 21.1
B. pendula 5.7 0.21 0.38 76.6 0.070 345.8 0.91 16.7
L. aggregatum 8.2 0.15 0.25 35.5 0.040 273.8 1.11 30.4
P. sessilis 9.2 0.18 0.29 53.2 0.050 273.7 0.95 22.5
E. nesiotica 15.4 0.13 0.35 42.6 0.035 268.6 0.77 19.8
D. oleifera 41.3 0.17 0.24 41.1 0.040 244.4 1.01 23.5
V. ferruginea 4.8 0.18 0.23 42.3 0.045 252.6 1.11 17.9 2.96
T. arborea 11.6 0.13 0.28 32.8 0.019 147.8 0.53 14.2 2.93
P. macrophylla 7.2 0.23 0.24 54.4 0.074 321.9 1.34 27.1
T. pittieri 1.6 0.15 0.33 47.7 0.026 189.7 0.57
G. americana 23.9 0.17 0.27 42.6 0.029 170.8 0.63 12.5 2.08
H. prunifolius 10.0 0.13 0.27 34.9 0.013 101.8 0.37 18.8
J. copaia 4.6 0.14 0.35 46.7 0.011 83.5 0.24 9.7
A. excelsum 9.4 0.18 0.28 47.2 0.054 301.8 1.07 1.83
C. pentandra 7.6 0.15 0.24 36.5 0.019 122.0 0.51 13.5 3.07
Sp <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001
Leaf age 0.77 0.10 0.15 0.52 0.92 0.74 0.26 <0.001
Interaction 0.04 0.45 0.63 0.06 0.47 0.44 0.03 <0.001
Shown at the bottom are P-values for ANCOVAs used to evaluate species differences, effects of leaf age, and species ? leaf age interactions for traits
determined with leaves of known age and for ANOVA for %cellulose, which was determined at plant level. See Supporting Information Table S1 for
additional trait values.
LMA, leaf mass per area.
New
Phytologist Research 645
 2012 The Authors
New Phytologist  2012 New Phytologist Trust
New Phytologist (2012) 195: 640?652
www.newphytologist.com
ANOVA on the effects of light environment and species). LMA,
q, %cellulose and %lignin were greater for gap-grown seedlings,
while c and %phenolics were greater for shade-grown seedlings
(P < 0.001 for all). Despite these plastic changes, neither T, s
nor Ws differed significantly between gap- and shade-grown
seedlings (P > 0.05). In terms of two measures of herbivory,
%leaf-damage was greater in gaps than shade (P = 0.03), while
%area did not differ significantly between gaps and shade. When
species?environment interactions were significant (c, %cellulose,
%phenolics, %lignin), the direction of phenotypic changes
between gap and shade was consistent across species but the
magnitudes of change differed among species.
Correlations of leaf lifespan with toughness-related traits
Median leaf lifespan of gap-grown seedlings was positively corre-
lated with all measures of toughness and cellulose, but not with
phenolics and tannin (Table 5). Across 24 species, leaf lifespan
was strongly correlated with s (r = 0.73) and Ws (r = 0.73), but
only marginally with c (r = 0.40) and T (r = 0.40). Leaf lifespan
was also positively correlated with q (r = 0.60) and cellulose per
volume (r = 0.77), which were also the only two traits correlated
with %leaf-damage (r = ) 0.46 and ) 0.48, respectively;
Table 5). Median leaf lifespan was negatively correlated with
%leaf-damage (r = ) 0.45, P = 0.027). Fibre fractions other
than cellulose (NPE, lignin, ash) were not significantly correlated
with leaf lifespan or %leaf-damage.
Cellulose provided the material-basis for leaf toughness, as
indicated by the strong linear relationships after appropriate nor-
malization (Fig. 2). Shade-tolerant and gap-dependent guilds
exhibited an identical relationship between %cellulose and
density-corrected toughness (Fig. 2a) and between cellulose per
unit volume and fracture toughness (Fig. 2b). However, the rela-
tionships of leaf lifespan with cellulose and fracture toughness dif-
fered between shade-tolerant and gap-dependent guilds (Fig. 3).
This was true whether cellulose content was expressed per unit dry
mass or volume (Fig. 3a,b) and whether fracture toughness was
corrected for density or not (Fig. 3c,d). The positive effect of
cellulose contents and fracture toughness on leaf lifespan was
steeper for shade-tolerant species than for gap-dependent species.
Gap-dependent species may have moderately tough leaves, but
Table 4 Pairwise correlations for leaf traits relevant for leaf toughness and nutritional quality, using species mean values for gap-grown plants
%Cellulose %lignin Cellulose : N
Density-
corrected
tough. (c)
Lamina
density (q)
Cellulose
per volume
Fracture
tough. (s)
Lamina
thick
(T)1
Leaf mass
per area1
Work-to-
shear1
n 21 21 10 24 24 21 24 24 24 24
%Cellulose 0.39 0.92 0.90 0.86 0.59 0.45 0.83
%lignin 0.28 0.34 )0.12 0.25 0.21 0.27 0.16 0.28
Cellulose : N 0.93 0.34
Density-corrected toughness (c) 0.84 0.33 0.86 )0.13 0.82 0.90 0.71 0.48 0.91
Lamina Density (q) )0.16 )0.01 0.34 )0.15 0.46 0.33 0.07 0.65 0.23
Cellulose per volume 0.83 )0.24 0.90 0.71 0.40 0.92 0.58 0.69 0.85
Fracture toughness (s) 0.69 0.29 0.87 0.87 0.33 0.86 0.73 0.76 0.96
Lamina thickness (T)1 0.49 0.31 0.80 0.64 )0.07 0.41 0.61 0.84 0.85
Leaf mass per area1 0.31 0.25 0.85 0.41 0.58 0.60 0.70 0.77 0.79
Work-to-shear1 0.68 0.36 0.91 0.88 0.19 0.75 0.94 0.80 0.76
Pearson correlation coefficients are shown below the diagonal for species-level correlations (number of species, n, indicated in the first row), and above the
diagonal for phylogenetically independent contrast correlations. Significance of correlation is indicated by bold italic (P < 0.001), bold non-italic (P < 0.01)
and italic (P < 0.05). For cellulose : N ratio, PIC correlation cells are blank because nitrogen-leaf age relationship was determined only for 10 species.
1log-transformed prior to analysis.
Fig. 2 The relationship of toughness with cellulose contents for leaves of
seedlings of 21 tropical tree species: (a) density-corrected toughness
against cellulose per unit mass, and (b) fracture toughness against cellulose
per unit volume. Open and closed symbols represent species mean values
for gap-dependent species and shade-tolerant species, respectively.
ANCOVA shows no significant effects of guilds on the slope and elevation
of either relationship. See Table 4 for correlations across gap and shade
guilds.
646 Research
New
Phytologist
 2012 The Authors
New Phytologist  2012 New Phytologist Trust
New Phytologist (2012) 195: 640?652
www.newphytologist.com
their leaf lifespan was shorter than shade-tolerant species with
similarly tough leaves (Fig. 3c,d).
Correlations of leaf traits with seedling survival
Different leaf traits were correlated with early seedling survival in
understorey vs gaps. Shade survival of seedlings was positively
correlated with leaf lifespan, cellulose per unit volume and frac-
ture toughness (Table 5). Gap survival of seedlings was negatively
correlated with lamina thickness, LMA and %leaf-damage. These
pairwise correlations were also supported by PICs (Table 5). The
positive relationship of shade survival with leaf lifespan was
steeper for gap-dependent species than for shade-tolerant species
(ANCOVA, guild main effect = ns, lifespan as covariate significant
at P = 0.02, guild ? lifespan interaction significant at P = 0.02).
Direct and indirect effects of toughness-related traits on
leaf lifespan
We compared alternative path models (Fig. 4) to evaluate direct
and indirect effects of cellulose content, lamina density and
toughness on median leaf lifespan. Each alternative model
included only one measure of toughness c, s or Ws, because they
were not determined independently of each other. Figure 4
shows the best-fit model that included c (Model A), s (Model B)
and Ws (Model C) in which the linkage between cellulose and
toughness were evaluated.
Model A predicts that cellulose per unit mass directly enhances
c, but only enhances leaf lifespan indirectly via its effect on c.
Adding a direct and nonsignificant path from cellulose to leaf
lifespan (b = 0.28, P = 0.14) to Model A would minimally affect
the support (DAIC = 0.1; Fig. S2a). Adding lamina thickness
(T) to Model A, with its covariance with cellulose per unit mass
(r = 0.49, P = 0.04) and an insignificant direct path to leaf life-
span would significantly reduce the model fit (v2 = 10.0,
P = 0.08, DAIC = 13.3).
In Model B, s is included instead of c. As hypothesized in
Fig. 1, s depended on lamina density and cellulose per unit mass.
Lamina density also had a significant direct effect on leaf lifespan
in addition to its indirect positive effect via s. The data did not
support the addition of T to Model B (P < 0.001). If cellulose
per unit mass was replaced with cellulose per unit volume ?
because the latter incorporates the effect of lamina density ? there
was no significant path from lamina density to s (Fig. S2b).
Models A and B were similarly supported by the data (DAIC =
1.4), although Model B explained a slightly greater proportion of
interspecific variation in leaf lifespan than Model A.
In Model C, Ws is chosen as the mechanical measure. As
hypothesized in Fig. 1, Ws was influenced by both cellulose and
T, covariance of which must be included for the model fit. Alter-
native models modified from Model C by dropping either T or
cellulose (Fig. S2c,d, respectively) also fit the data.
In summary, comparisons across the three alternative models
in Fig. 4 show that lamina density always had a significant direct
Table 5 Pairwise correlations of ecological traits (median leaf lifespan, percentage of leaves damaged by herbivory, seedling survival in shade and gaps)
with leaf traits
n r P nPIC rPIC
Correlation of median leaf lifespan with
Density-corrected toughness 24 0.40 0.053 22 0.49
Cellulose per mass 21 0.44 0.045 19 0.60
Cellulose : nitrogen ratio 10 0.80 0.005
Fracture toughness 24 0.72 <0.001 22 0.77
Lamina density 24 0.60 0.002 22 0.64
Cellulose per volume 21 0.77 <0.001 19 0.60
Log (work-to-shear) 24 0.66 <0.001 22 0.72
Log (LMA) 24 0.76 <0.001 22 0.80
Log (Lamina thickness) 24 0.43 0.054 22 0.57
%Phenolics 23 0.01 NS 22 )0.18
%Tannin 23 0.24 NS 22 0.17
%Leaf-damage 24 )0.45 0.027 22 )0.53
Correlation of %leaf-damage with
Cellulose per volume 21 )0.48 0.028 19 )0.38
Lamina density 24 )0.46 0.024 22 )0.62
Correlation of shade seedling survival with
Median leaf lifespan 23 0.49 0.017 22 0.44
Cellulose per volume 21 0.40 0.080 19 0.18
Fracture toughness 23 0.35 0.100 22 0.33
Correlation of gap seedling survival with
Log (lamina thickness) 24 )0.52 0.010 22 )0.55
Log (LMA) 24 )0.44 0.030 22 )0.46
%leaf-damage 24 )0.42 0.039 22 )0.20
Shown are Pearson correlation coefficients (r, followed by level of significance P) and for phylogenetic independent contrasts (rPIC). Only correlations with
r > 0.35 are shown, except for correlation of leaf lifespan with phenolics and tannin. n, number of species; nPIC, number of PICs. Three traits that reflect
lamina thickness (work-to-shear, LMA, lamina thickness) are log-transformed to improve normality. Significance of correlation is indicated by bold italic
(P <0.001), bold non-italic (P <0.01) and italic (P ? 0.05).
New
Phytologist Research 647
 2012 The Authors
New Phytologist  2012 New Phytologist Trust
New Phytologist (2012) 195: 640?652
www.newphytologist.com
positive effect on leaf lifespan, while cellulose had a significant
indirect effect via toughness regardless of how the latter was nor-
malized (c, s or Ws). Lamina thickness, by contrast, had no direct
effect on leaf lifespan, but did have a significant positive effect on
Ws (Fig. 4c).
Discussion
Leaf mechanical properties and structural carbon allocation
Both cellulose and lamina density contributed to leaf toughness
and enhanced leaf lifespan (Figs 2?4), especially when their
effects were considered together. Lamina density was a significant
determinant of fracture toughness, but only when cellulose effect
was simultaneously taken into account (Table 4, Fig. 4).
Furthermore, the direct positive effect of lamina density on leaf
lifespan was stronger when its indirect effect through toughness
was simultaneously accounted for (Fig. 4), compared to a simple
pairwise correlation (Table 5). The densest material component,
cellulose, seems of overwhelming importance in setting the level
of density-corrected toughness (Fig. 2a; Table 4), consistent with
the results of Alvarez-Clare & Kitajima (2007) and Westbrook
et al. (2011). Although all cell wall components are probably
brittle in isolation, our results show that in the composite con-
struction of cell walls, cellulose provides the indispensable skele-
ton. By contrast, density-corrected toughness was uncorrelated
with lignin and hemicellulose (Table S2) and only weakly with
%ADF (= %cellulose + %lignin) (r = 0.47, P = 0.03). This con-
trasts with a commonly held view in ecology that lignin contrib-
utes to toughness (Cornelissen et al., 1999). Lignin has a role in
stiffening the cell wall via water exclusion (increasing bending
resistance; Alvarez-Clare & Kitajima, 2007) and is vital for eco-
system processes via its chemical properties, variously slowing
digestion in animal guts, litter decomposition and nutrient
cycling (Aerts, 1997; Cornelissen et al., 2004; Kurokawa &
Nakashizuka, 2008). However, the lack of correlation between
%lignin and toughness, also reported from Malaysian tropical
forest (Kurokawa & Nakashizuka, 2008), means that toughness
may not be an easy-to-measure ?soft trait? to substitute for ?hard
traits? relevant to nutrient cycling (Diaz et al., 2004).
Phenotypic plasticity of leaf lifespan and carbon-based
defence
Many leaf traits exhibit substantial phenotypic plasticity in
response to leaf age, ontogenetic stages and the environmental
conditions in which they develop, although intraspecific variation
is generally believed to be smaller than interspecific variations
(Gotsch et al., 2010; Fajardo & Piper, 2011). We found that
lamina density and LMA increased significantly with leaf age in
some species, while toughness did not (Tables 3, S1). The leaf
age gradient was unfortunately confounded with the light gradi-
ent within the crown and by ontogeny in the current study (i.e.
older leaves were produced by much smaller seedlings). However,
a previous study of seedling leaves at the same position 6 months
apart (Alvarez-Clare & Kitajima, 2007) did not find any effects
of leaf age on toughness either.
This common garden study avoided the confounding of in situ
environment and habitat preference of the species. For example,
gap phenotypes of gap-dependent species are often compared
with shade phenotypes of shade-tolerant species in field-based
samplings (e.g. Gotsch et al., 2010; Onoda et al., 2011), but in
such comparisons it is ambiguous whether observed trait varia-
tions reflect evolutionary or plastic responses (Endara & Coley,
2011). Both the theory of optimal leaf lifespan and empirical
studies predict that LMA, lamina thickness, and phenolic and
tannin contents increase with higher light availability, while leaf
lifespan decreases with light availability (Kikuzawa, 1984;
Fig. 3 The effects of cellulose contents and toughness on
median leaf lifespan for the two regeneration guilds of
tropical trees, gap-dependent (open symbols with broken
trend line) and shade-tolerant (closed symbols with solid
line). The physical defence traits (ANCOVA results
associated with each) are: (a) cellulose per unit dry mass
(guild P = 0.006; covariate P = 0.02; interaction, not
significant (ns)), (b) cellulose per unit volume (guild
P = 0.05; covariate P = 0.006; interaction, ns),
(c) density-corrected toughness (guild P < 0.001;
covariate P = 0.003; interaction P = 0.03), and
(d) fracture toughness (guild P = 0.001; covariate
P = 0.003; interaction P = 0.03).
648 Research
New
Phytologist
 2012 The Authors
New Phytologist  2012 New Phytologist Trust
New Phytologist (2012) 195: 640?652
www.newphytologist.com
Ellsworth & Reich, 1992; Shure & Wilson, 1993; Ackerly &
Bazzaz, 1995; Oguchi et al., 2006; Barbehenn & Constabel,
2011). Our results matched most of these expectations, with
shorter leaf lifespan, but greater LMA and leaf density in
gap-grown than shade-grown plants.
Plastic responses to light environments may not coincide with
adaptive responses to light environments (Shure & Wilson,
1993; Lusk et al., 2008; Endara & Coley, 2011). Hence, adapta-
tion to shade with longer leaf lifespan was accompanied by an
increase in tissue density and toughness (Figs 3, 4), while accli-
mation to shade resulted in a decrease in tissue density and %cel-
lulose, and inconsistent and small changes in fracture toughness
(Table S3). Perhaps, the plastic response of fracture toughness
(s = q ? c) to light is difficult to predict because density (q) gen-
erally decreases with shading (Niinemets, 2001; Oguchi et al.,
2006; Kitajima & Poorter, 2010), while density-corrected tough-
ness (c) tends to increase with shading (thus ?shade leaves punch
above their weight?; Lusk et al., 2010; Onoda et al., 2011).
Importance of cellulose-based toughness for long leaf
lifespan
Despite recognition of the ecological importance of leaf tough-
ness for 40 years or more (Feeny, 1970), few studies have explic-
itly examined toughness at different levels of the leaf structural
organization. Being so heterogeneous, with different toughness
values for different tissues, toughness at different levels may not
show the same ecological trend. Density-corrected toughness (c)
correlated closely with %cellulose, and c and tissue density
together contribute to fracture toughness and leaf lifespan
(Fig. 4). It appears that fracture toughness and work-to-shear are
similarly related to herbivory and leaf lifespan (Table 4). A
similar conclusion was reached by Dominy et al. (2008) who
re-analysed the leaf lifespan data of Coley (1983) with new mea-
surements of fracture toughness, instead of the penetrometry in
Coley?s original study. Although lamina thickness contributes to
Ws, it shows little relationship with the demographic characteris-
tics of species (Table 5, also Wright et al., 2001; Kitajima &
Poorter, 2010; Westbrook et al., 2011; but see Funk & Throop,
2010).
Carbon investment in cellulose microfibrils and its resulting
toughness predicted species differences in herbivory damage and
leaf lifespan, while lignin, hemicellulose and nonstructural
carbon-based defence such as phenolics and tannins did not.
Only four species overlapped between the current study and 46
species studied by Coley (1983). In both studies, leaf toughness
and cellulose per unit mass strongly correlated with leaf lifespan,
while lignin and phenolics per unit mass did not. Greater impor-
tance of toughness or cellulose on herbivory and leaf lifespan,
compared to phenolics or lignin, is also reported for temperate
tree species (Matsuki & Koike, 2006; Mediavilla et al., 2008;
Pearse, 2011). A recent meta-analysis (Endara & Coley, 2011)
also found that phenolics and tannin show no generalizable rela-
tionships with toughness, herbivory level or leaf lifespan, perhaps
because individual tannins and phenolic compounds vary greatly
in their properties and functions (Close & McArthur, 2002;
Lokvam & Kursar, 2005; Barbehenn & Constabel, 2011).
Cell wall fibre may reduce herbivory not just because fibre
increases toughness, but also because it reduces nutritional values,
which are also affected by protein and water contents (Peeters,
2007; Clissold et al., 2009). But, %NDF (total nondigestive fibre
contents) were uncorrelated with herbivory and leaf lifespan.
Other studies also found that herbivore choice and level are often
influenced more by toughness than by fibre contents and digest-
ibility (Hill & Lucas (1996) for primates, Clissold et al. (2004)
for insects, Kurokawa & Nakashizuka (2008) for field herbivory
rates). One interesting idea is that nutritional values may corre-
late better with herbivory when expressed per unit work to frac-
ture (Clissold et al., 2009). Hence, Lusk et al. (2010) proposed
%cell content (= 100 ) %NDF) divided by density-corrected
toughness as an indicator of the relative nutritional value per bite
effort. In three of our four shade-tolerant species, this nutritional
value index was higher for gap-grown seedlings than shade-grown
seedlings similar to the finding by Lusk et al. (2010). However,
this index was not correlated with %leaf-damage within the gap,
Cellulose per 0.84 Density-corrected
(a) Model A
Median leaf
lifespan
Lamina
density (g cm?3)
unit mass (g g?1)
0.67
0.56
toughness (mJ g?1m)
? 2 = 2.69
P = 0.44
R2 = 0.70
Fracture
toughness (Jm?2)
0.80
(b) Model B
R2 = 0.77 AIC = 16.7
R2 0 76
Cellulose per
unit mass (g g?1)
Median leaf
lifespan
Lamina
density (g cm?3) 0.43
0.66
Model C
? 2 = 2.06
P = 0.36
AIC = 18.1R2 = 0.80
= .0.47
(c)
Lamina
Cellulose per
unit mass (g g?1)
0.42
0.49
0.63
Work-to-shear
(Jm?1)
R2 = 0 77
? 2 = 3.73
P = 0.44
AIC = 25.7
Thickness (mm)
Median leaf
lifespanLamina
density (g cm?3) 0.48 R2 = 0.79
0.32 0.61
.
Fig. 4 Three alternative path models supported by interspecific correla-
tions among 21 tropical tree species. Each model tests a subset of the
conceptual model (Fig. 1), including only one measure of mechanical
resistance against fracture, either (a) density-corrected toughness (Model
A), (b) fracture toughness (Model B), or (c) work-to-shear (Model C). The
overall model fit is indicated by likelihood v2 (i.e. the data do not reject the
illustrated models when P > 0.1), and selected from alternative models by
AIC comparisons (see Supporting Information Fig. S2 for alternative
models with similar levels of overall fits). R2 values represent the propor-
tion of interspecific variance in toughness and median leaf lifespan
explained by each model. Unidirectional arrows indicate positive direct
paths, with numbers indicating path coefficients (standardized partial
regression coefficients, b). Bidirectional arrows indicate significant correla-
tions between two exogenous variables (P < 0.05). All paths are supported
at P < 0.001 except for a thin arrow in Model C (lamina density to
work-to-shear, P = 0.02). See Tables 4 and 5 for pairwise correlations.
New
Phytologist Research 649
 2012 The Authors
New Phytologist  2012 New Phytologist Trust
New Phytologist (2012) 195: 640?652
www.newphytologist.com
and only weakly with median leaf lifespan across 21 species
(r = ) 0.46, P = 0.03).
Leaf trait syndromes adaptive to gaps vs shade
Leaf lifespan and seedling survival were associated with leaf func-
tional traits, but two unexpected results emerged. First, different
sets of leaf traits were correlated with seedling survival in gaps vs
shade (Table 5). Leaf lifespan, leaf toughness and lamina density
were positively correlated with seedling survival in shade, but
none of these traits were significant for seedling survival in gaps.
Not surprisingly, a higher level of herbivory meant lower survival
in gaps (Table 5). Lamina thickness and LMA, which might be
associated with sclerophylly (Grubb, 1986; Wright et al., 2001),
had no correlation with %leaf-damage and unexpectedly negative
effects on gap survival. We determined seedling survival for the
first 3 months after germination, during which we observed that
species with high mortality in gaps exhibited stunted growth sug-
gestive of stress from excessive light and heat. Perhaps, constitu-
tively thick leaves with high lamina density may limit efficient
CO2 exchange (Niinemets et al., 2009) and safe channelling of
excess radiation energy.
Second, the relationship of leaf lifespan with toughness dif-
fered between gap-dependent and shade-tolerant guilds (Fig. 3),
even though the mechanistic link between cellulose and toughness
was identical across the two groups (Fig. 2). Re-analysing the data
of Coley (1983), Dominy et al. (2008) observed that the positive
correlation of toughness with leaf lifespan and herbivory reflects
species differences between shade-tolerant and pioneer guilds, but
not within the shade-tolerant guild. By contrast, we found that
leaf toughness was associated with leaf lifespan within the latter
guild, similar to findings by Lusk et al. (2010) and Numata
et al. (2003). Density-corrected toughness, the demographic
implication of which was examined explicitly for the first time,
had a strong positive effect on leaf lifespan for shade-tolerant
species, but this was much weaker for gap-dependent species
(Fig. 3c). Thus, at intermediate-to-high levels of biomass
investment to cellulose (Fig. 3a), shade-tolerant species had
greater leaf lifespans. This is perhaps because shade-tolerant and
gap-dependent species may differ in anatomical arrangements of
thick-walled cells and the ratio of mesophyll, vein and epidermal
tissues. Such shifts in the relationship between leaf lifespan and
leaf toughness between ecological guilds can also be found in
relation to adaptation to other environmental factors. Wright &
Westoby (2002) found that the relationship of leaf lifespan with
both LMA and work-to-shear differed between species from low-
and high-rainfall regions. It is interesting that longer leaf lifespan
and toughness were only weakly associated within gap guild
(Fig. 3). Perhaps, gap-dependent species are selected to have
similarly rapid turnover of short-lived leaves to achieve fast growth
(Kikuzawa, 1991; Ackerly, 1999; Portsmuth & Niinemets, 2007),
while species difference in cellulose contents among gap-
dependent species may reflect differences in leaf display and
architecture. Future studies that integrate anatomy with
mechanical properties, such as Onoda et al.?s (2008) approach for
acclimation to light and nutrient, may reveal important
anatomical traits responsible for adaptive and acclimation
responses of leaf traits to environments.
Conclusion
Carbon allocation to cellulose, but not other cell wall compo-
nents or phenolics and tannin, contributed to leaf toughness.
Cellulose contents and density-corrected toughness were strongly
associated with leaf lifespan within the shade-tolerant guild, but
less so in gap-dependent species, which tended to possess
shorter-lived leaves at the same %cellulose content and den-
sity-corrected toughness. Lamina density had both direct and
indirect effects via toughness on herbivory level (negative) and
leaf lifespan (positive). Toughness and long leaf lifespan were
associated with better survival in shade, but not in gaps. Lamina
thickness was uncorrelated with shade tolerance or leaf lifespan,
but a combination of thick lamina and high LMA negatively
influenced gap survival. Overall, our results place greater impor-
tance on cellulose-based physical defence than phenolic-based
defence or nutritional values as correlates of species differences in
herbivory and leaf lifespan associated with life history strategies
and regeneration light requirements.
Acknowledgements
We thank: Roberto Cordero and Sebastian Bernal for the field
data collection; Ann Hagerman, Michelle Mack, Kris Callis and
Anne Baker for making tannin and fibre analyses possible; and
Amy Zanne for help with phylogenetic analysis. The study was
supported by NSF 0093033 to K.K. and by the Smithsonian
Tropical Research Institute.
References
Ackerly D. 1999. Self-shading, carbon gain and leaf dynamics: a test of alternative
optimality models. Oecologia 119: 300?310.
Ackerly DD, Bazzaz FA. 1995. Leaf dynamics, self-shading and carbon gain in
seedlings of a tropical pioneer tree. Oecologia 101: 289?298.
Aerts R. 1997. Climate, leaf litter chemistry and leaf litter decomposition in
terrestrial ecosystems: a triangular relationship. Oikos 79: 439?449.
Alvarez-Clare S, Kitajima K. 2007. Physical defence traits enhance seedling
survival of neotropical tree species. Functional Ecology 21: 1044?1054.
Aranwela N, Sanson G, Read J. 1999. Methods of assessing leaf-fracture
properties. New Phytologist 144: 369?393.
Augspurger CK. 1984. Seedling survival of tropical tree species; interactions of
dispersal distance, light gaps, and pathogens. Ecology 65: 1705?1712.
Barbehenn RV, Constabel CP. 2011. Tannins in plant-herbivore interactions.
Phytochemistry 72: 1551?1565.
Bazzaz FA. 1979. Physiological ecology of plant succession. Annual Review of
Ecology and Systematics 10: 351?371.
Bloom AJ, Chapin FS, Mooney HA. 1985. Resource limitation in plants-an
economic analogy. Annual Review of Ecology and Systematics 16: 363?392.
Bongers F, Pompa J. 1990. Leaf dynamics of seedlings of rain forest species in
relation to canopy gaps. Oecologia 82: 122?127.
Carpita NC, Gibeaut DM. 1993. Structural models of primary-cell walls in
flowering plants: consistency of molecular-structure with the
physical-properties of the walls during growth. Plant Journal 3: 1?30.
Choong MF. 1996. What makes a leaf tough and how this affects the pattern of
Castanopsis fissa leaf consumption by caterpillars. Functional Ecology 10:
668?674.
650 Research
New
Phytologist
 2012 The Authors
New Phytologist  2012 New Phytologist Trust
New Phytologist (2012) 195: 640?652
www.newphytologist.com
Clissold FJ, Sanson GD, Read J. 2004. Indigestibility of plant cell wall by the
Australian plague locust, Chortoicetes terminifera. Entomologia Experimentalis Et
Applicata 112: 159?168.
Clissold FJ, Sanson GD, Read J, Simpson SJ. 2009. Gross vs. net income: how
plant toughness affects performance of an insect herbivore. Ecology 90:
3393?3405.
Close DC, McArthur C. 2002. Rethinking the role of many plant phenolics ?
protection from photodamage not herbivores? Oikos 99: 166?172.
Coley PD. 1983. Herbivory and defensive characteristics of tree species in a
lowland tropical forest. Ecological Monographs 53: 209?233.
Coley PD, Bryant JP, Chapin FS III. 1985. Resource availability and plant
anti-herbivore defence. Science 230: 895?899.
Comita LS, Muller-Landau HC, Aguilar S, Hubbell SP. 2010. Asymmetric
density dependence shapes species abundances in a tropical tree community.
Science 329: 330?332.
Condit R, Ashton P, Bunyavejchewin S, Dattaraja HS, Davies S, Esufali S,
Ewango C, Foster R, Gunatilleke I, Gunatilleke CVS et al. 2006. The
importance of demographic niches to tree diversity. Science 313: 98?101.
Cornelissen JHC, Perez-Harguindeguy N, Diaz S, Grime JP, Marzano B,
Cabido M, Vendramini F, Cerabolini B. 1999. Leaf structure and defence
control litter decomposition rate across species and life forms in regional floras
on two continents. New Phytologist 143: 191?200.
Cornelissen JHC, Quested HM, Gwynn-Jones D, Van Logtestijn RSP, De Beus
MAH, Kondratchuk A, Callaghan TV, Aerts R. 2004. Leaf digestibility and
litter decomposability are related in a wide range of subarctic plant species and
types. Functional Ecology 18: 779?786.
Dalling JW, Swaine MD, Garwood NC. 1997. Soil seed bank community
dynamics in seasonally moist lowland tropical forest, Panama. Journal of
Tropical Ecology 13: 659?680.
Darvell BW, Lee PKD, Yuen TDB, Lucas PW. 1996. A portable fracture
toughness tester for biological materials. Measurement Science & Technology 7:
954?962.
Diaz S, Hodgson JG, Thompson K, Cabido M, Cornelissen JHC, Jalili A,
Montserrat-Marti G, Grime JP, Zarrinkamar F, Asri Y et al. 2004. The plant
traits that drive ecosystems: evidence from three continents. Journal of
Vegetation Science 15: 295?304.
Dominy NJ, Grubb PJ, Jackson RV, Lucas PW, Metcalfe DJ, Svenning JC,
Turner IM. 2008. In tropical lowland rain forests monocots have tougher
leaves than dicots, and include a new kind of tough leaf. Annals of Botany 101:
1363?1377.
Dominy NJ, Lucas PW, Wright SJ. 2003. Mechanics and chemistry of rain
forest leaves: canopy and understorey compared. Journal of Experimental Botany
54: 2007?2014.
Ellsworth DS, Reich PB. 1992. Leaf mass per area, nitrogen content and
photosynthetic carbon gain in Acer saccharum seedlings in contrasting forest
light environments. Functional Ecology 6: 423?435.
Endara MJ, Coley PD. 2011. The resource availability hypothesis revisited: a
meta-analysis. Functional Ecology 25: 389?398.
Fajardo A, Piper FI. 2011. Intraspecific trait variation and covariation in a
widespread tree species (Nothofagus pumilio) in southern Chile. New Phytologist
189: 259?271.
Feeny P. 1970. Seasonal changes in oak leaf tannins and nutrients as a cause of
spring feeding by winter moth caterpillars. Ecology 51: 565?581.
Funk JL, Throop HL. 2010. Enemy release and plant invasion: patterns of
defensive traits and leaf damage in Hawaii. Oecologia 162: 815?823.
Gilbert B, Wright SJ, Muller-Landau HC, Kitajima K, Hernandez A. 2006. Life
history trade-offs in tropical trees and lianas. Ecology 87: 1281?1288.
Givnish TJ. 1988. Adaptation to sun and shade: a whole-plant perspective.
Australian Journal of Plant Physiology 15: 63?92.
Gotsch SG, Powers JS, Lerdau MT. 2010. Leaf traits and water relations of
12 evergreen species in Costa Rican wet and dry forests: patterns of
intra-specific variation across forests and seasons. Plant Ecology 211:
133?146.
Grace JB. 2006. Structural equation modeling and natural systems. Cambridge,
UK: Cambridge University Press.
Grime JP. 1979. Plant strategies and vegetation processes. New York, NY, USA:
John Wiley.
Grubb PJ. 1986. Sclerophylls, pachyphylls and pycnophylls: the nature and
significance of hard leaf surfaces. In: Juniper B, Southwood R, eds. Insects and
the plant surface. London, UK: Edward Arnold, 137?150.
Grubb PJ. 1992. A positive distrust in simplicity lessons from plant defences and
from competition among plants and among animals. Journal of Ecology 80:
585?610.
Grubb PJ, Jackson RV, Barberis IM, Bee JN, Coomes DA, Dominy NJ, De la
Fuente MAS, Lucas PW, Metcalfe DJ, Svenning JC et al. 2008. Monocot
leaves are eaten less than dicot leaves in tropical lowland rain forests:
correlations with toughness and leaf presentation. Annals of Botany 101:
1379?1389.
Hill DA, Lucas PW. 1996. Toughness and fiber content of major leaf foods of
Japanese macaques (Macaca fuscata yakui) in Yakushima. American Journal of
Primatology 38: 221?231.
Jeronimidis G. 1980. The fracture-behavior of wood and the relations between
toughness and morphology. Proceedings of the Royal Society of London Series B
208: 447?460.
Kikuzawa K. 1984. Leaf survival of woody plants in deciduous broad-leaved
forests. 2. Small trees and shrubs. Canadian Journal of Botany 62: 2551?
2556.
Kikuzawa K. 1991. A cost-benefit analysis of leaf habit and leaf longevity of trees
and their geographical pattern. American Naturalist 138: 1250?1263.
Kitajima K. 1994. Relative importance of photosynthetic traits and allocation
patterns as correlates of seedling shade tolerance of 13 tropical trees. Oecologia
98: 419?428.
Kitajima K, Mulkey SS, Samaniego M, Wright SJ. 2002. Decline of
photosynthetic capacity with leaf age and position in two tropical pioneer tree
species. American Journal of Botany 89: 1925?1932.
Kitajima K, Mulkey SS, Wright SJ. 1997. Decline of photosynthetic capacity
with leaf age in relation to leaf longevities for five tropical canopy tree species.
American Journal of Botany 84: 702?708.
Kitajima K, Poorter L. 2010. Tissue-level leaf toughness, but not lamina
thickness, predicts sapling leaf lifespan and shade tolerance of tropical tree
species. New Phytologist 186: 708?721.
Kohyama T. 1991. A functional model describing sapling growth under a
tropcial forest canopy. Functional Ecology 5: 83?90.
Kurokawa H, Nakashizuka T. 2008. Leaf herbivory and decomposability in a
Malaysian tropical rain forest. Ecology 89: 2645?2656.
Leigh EG, Rand AS, Windsor DM. 1982. The ecology of a tropical forest: seasonal
rhythms and long-term changes. Washington, DC, USA: Smithsonian
Institution Press.
Lokvam J, Kursar TA. 2005. Divergence in structure and activity of phenolic
defences in young leaves of two co-occurring Inga species. Journal of Chemical
Ecology 31: 2563?2580.
Lucas PW, Beta T, Darvell BW, Dominy NJ, Essackjee HC, Lee PKD, Osorio
D, Ramsden L, Yamashita N, Yuen TDB. 2001. Field kit to characterize
physical, chemical and spatial aspects of potential primate foods. Folia
Primatologica 72: 11?25.
Lucas PW, Choong MF, Tan HTW, Turner IM, Berrick AJ. 1991. The
fracture-toughness of the leaf of the dicotyledon Calophyllum inophyllum L.
(Guttiferae). Philosophical Transactions of the Royal Society of London Series B
334: 95?106.
Lucas PW, Pereira B. 1990. Estimation of the fracture-toughness of leaves.
Functional Ecology 4: 819?822.
Lucas PW, Turner IM, Dominy NJ, Yamashita N. 2000. Mechanical defences to
herbivory. Annals of Botany 86: 913?920.
Lusk CH, Onoda Y, Kooyman R, Gutierrez-Giron A. 2010. Reconciling
species-level vs plastic responses of evergreen leaf structure to light gradients:
shade leaves punch above their weight. New Phytologist 186: 429?438.
Lusk CH, Reich PB, Montgomery RA, Ackerly DD, Cavender-Bares J. 2008.
Why are evergreen leaves so contrary about shade? Trends in Ecology &
Evolution 23: 299?303.
Matsuki S, Koike T. 2006. Comparison of leaf life span, photosynthesis and
defensive traits across seven species of deciduous broad-leaf tree seedlings.
Annals of Botany 97: 813?817.
Mediavilla S, Garcia-Ciudad A, Garcia-Criado B, Escudero A. 2008. Testing the
correlations between leaf life span and leaf structural reinforcement in 13
New
Phytologist Research 651
 2012 The Authors
New Phytologist  2012 New Phytologist Trust
New Phytologist (2012) 195: 640?652
www.newphytologist.com
species of European Mediterranean woody plants. Functional Ecology 22:
787?793.
Niinemets U. 2001. Global-scale climatic controls of leaf dry mass per area,
density, and thickness in trees and shrubs. Ecology 82: 453?469.
Niinemets U, Diaz-Espejo A, Flexas J, Galmes J, Warren CR. 2009. Role
of mesophyll diffusion conductance in constraining potential
photosynthetic productivity in the field. Journal of Experimental Botany 60:
2249?2270.
Numata S, Kachi N, Okuda T, Manokaran N. 2003. Leaf herbivory and defences
of dipterocarp seedlings in the Pasoh forest reserve. In: Okuda T, Manokaran
N, Matsumoto Y, Niiyama K, Thomas SC, Ashton PS, eds. Pasoh: ecology of a
lowland rain forest in southeast Asia. Tokyo, Japan: Springer, 495?505.
Oguchi R, Hikosaka K, Hiura T, Hirose T. 2006. Leaf anatomy and light
acclimation in woody seedlings after gap formation in a cool-temperate
deciduous forest. Oecologia 149: 571?582.
Onoda Y, Schieving F, Anten NPR. 2008. Effects of light and nutrient
availability on leaf mechanical properties of Plantago major: a conceptual
approach. Annals of Botany 101: 727?736.
Onoda Y, Westoby M, Adler PB, Choong AMF, Clissold FJ, Cornelissen JHC,
Diaz S, Dominy NJ, Elgart A, Enrico L et al. 2011. Global patterns of leaf
mechanical properties. Ecology Letters 14: 301?312.
Pacala SW, Canham CD, Saponara J, Silander JA, Kobe RK, Ribbens E. 1996.
Forest models defined by field measurements: estimation, error analysis and
dynamics. Ecological Monographs 66: 1?43.
Pearse IS. 2011. The role of leaf defensive traits in oaks on the preference and
performance of a polyphagous herbivore, Orgyia vetusta. Ecological Entomology
36: 635?642.
Peeters PJ. 2007. Leaf biomechanical properties and the densities of herbivorous
insect guilds. Functional Ecology 21: 246?255.
Poorter L. 2006. Leaf traits are good predictors of plant performance across 53
rain forest species. Ecology 87: 1733?1743.
Portsmuth A, Niinemets U. 2007. Structural and physiological plasticity in
response to light and nutrients in five temperate deciduous woody species of
contrasting shade tolerance. Functional Ecology 21: 61?77.
Reich PB, Walters MB, Elsworth DS. 1992. Leaf life-span in relation to leaf,
plant and stand characteristics among diverse ecosystems. Ecological
Monographs 62: 365?392.
Sanson G, Read J, Aranwela N, Clissold F, Peeters P. 2001. Measurement of leaf
biomechanical properties in studies of herbivory: opportunities, problems and
procedures. Austral Ecology 26: 535?546.
Shure DR, Wilson LA. 1993. Patch-size effects on plant phenolics in successional
openings of the Southern Appalachians. Ecology 74: 55?67.
Stefanescu CC. 2006. An evaluation of the cost-benefit analysees of leaf lifespan using
seedlings of tropical tree species. Gainesville, FL, USA: University of Florida.
Sterck FJ, Poorter L, Schieving F. 2006. Leaf traits determine the growth-survival
trade-off across rain forest tree species. American Naturalist 167: 758?765.
Swain MD, Whitmore TC. 1988. On the definition of ecological species groups
in tropical rain forest. Vegetatio 75: 81?86.
Taiz L, Zeiger E. 2002. Plant physiology, 2nd edn. Sunderland, MA, USA: Sinauer
Associates.
Van Soest PJ. 1963. Use of detergents in the analysis of fibrous feeds. II. A rapid
method for the determination of fiber and lignin. Journal of the Association of
Official Agricultural Chemists 46: 829?835.
Webb C, Ackerly DD, Kembel SW. 2008. Phylocom: software for the analysis of
phylogenetic community structure and character evolution. Version 4.0.
[WWW document] URL http://phylodiversity.net/phylocom/. [accessed 6
September 2009].
Welden CW, Hewett SW, Hubbell SP, Foster RB. 1991. Sapling survival,
growth, and recruitment: relationship to canopy height in a neotropical forest.
Ecology 72: 35?50.
Westbrook JW, Kitajima K, Burleigh JG, Kress WJ, Erickson DL, Wright SJ.
2011. What makes a leaf tough? Patterns of correlated evolution between leaf
toughness traits and demographic rates among 197 shade-tolerant woody
species in a Neotropical forest. American Naturalist 177: 800?811.
Wright SJ, Kitajima K, Kraft NJB, Reich PB, Wright IJ, Bunker DE, Condit R,
Dalling JW, Davies SJ, Diaz S et al. 2010. Functional traits and the
growth-mortality trade-off in tropical trees. Ecology 91: 3664?
3674.
Wright IJ, Reich PB, Westoby M. 2001. Strategy shifts in leaf physiology,
structure and nutrient content between species of high- and low-rainfall and
high- and low-nutrient habitats. Functional Ecology 15: 423?434.
Wright IJ, Reich PB, Westoby M, Ackerly DD, Baruch Z, Bongers F,
Cavender-Bares J, Chapin T, Cornelissen JHC, Diemer M et al. 2004. The
worldwide leaf economics spectrum. Nature 428: 821?827.
Wright IJ, Westoby M. 2002. Leaves at low versus high rainfall: coordination of
structure, lifespan and physiology. New Phytologist 155: 403?416.
Supporting Information
Additional supporting information may be found in the online
version of this article.
Fig. S1 Relationship of median leaf lifespan of shade-grown
seedlings with leaf lifespan and toughness of gap grown seedlings.
Fig. S2 Alternative models to the path models shown in Fig. 4
with adequate fit.
Table S1 Species means for additional leaf traits measured with
gap-grown seedlings
Table S2 Correlations between cell wall fibre components and
density-corrected toughness
Table S3 Leaf traits measured with seedlings grown in understo-
rey shade gardens
Notes S1 Additional methodological details on regeneration
guild classification and trait measurements.
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.
652 Research
New
Phytologist
 2012 The Authors
New Phytologist  2012 New Phytologist Trust
New Phytologist (2012) 195: 640?652
www.newphytologist.com