Are functional traits good predictors of demographic rates? Evidence from five neotropical forests.

Abstract

A central goal of comparative plant ecology is to understand how functional traits vary among species and to what extent this variation has adaptive value. Here we evaluate relationships between four functional traits (seed volume, specific leaf area, wood density, and adult stature) and two demographic attributes (diameter growth and tree mortality) for large trees of 240 tree species from five Neotropical forests. We evaluate how these key functional traits are related to survival and growth and whether similar relationships between traits and demography hold across different tropical forests. There was a tendency for a trade-off between growth and survival across rain forest tree species. Wood density, seed volume, and adult stature were significant predictors of growth and/or mortality. Both growth and mortality rates declined with an increase in wood density. This is consistent with greater construction costs and greater resistance to stem damage for denser wood. Growth and mortality rates also declined as seed volume increased. This is consistent with an adaptive syndrome in which species tolerant of low resource availability (in this case shade-tolerant species) have large seeds to establish successfully and low inherent growth and mortality rates. Growth increased and mortality decreased with an increase in adult stature, because taller species have a greater access to light and longer life spans. Specific leaf area was, surprisingly, only modestly informative for the performance of large trees and had ambiguous relationships with growth and survival. Single traits accounted for 9-55% of the interspecific variation in growth and mortality rates at individual sites. Significant correlations with demographic rates tended to be similar across forests and for phylogenetically independent contrasts as well as for cross-species analyses that treated each species as an independent observation. In combination, the morphological traits explained 41% of the variation in growth rate and 54% of the variation in mortality rate, with wood density being the best predictor of growth and mortality. Relationships between functional traits and demographic rates were statistically similar across a wide range of Neotropical forests. The consistency of these results strongly suggests that tropical rain forest species face similar trade-offs in different sites and converge on similar sets of solutions.

Full text

Page 1

Ecology, 89(7), 2008, pp. 1908–1920
? 2008 by the Ecological Society of America
ARE FUNCTIONAL TRAITS GOOD PREDICTORS OF DEMOGRAPHIC
RATES? EVIDENCE FROM FIVE NEOTROPICAL FORESTS
L. POORTER,1,2,11S. J. WRIGHT,3H. PAZ,4D. D. ACKERLY,5R. CONDIT,3G. IBARRA-MANRI´QUEZ,4K. E. HARMS,3,6
J. C. LICONA,2M. MARTI´NEZ-RAMOS,4S. J. MAZER,7H. C. MULLER-LANDAU,3,8M. PEN˜A-CLAROS,2C. O. WEBB,9
AND I. J. WRIGHT10
1Forest Ecology and Forest Management Group and Resource Ecology Group, Centre for Ecosystem Studies, Wageningen University,
P.O. Box 47, 6700 AA Wageningen, The Netherlands
2Instituto Boliviano de Investigacio´n Forestal, Casilla 6204, Santa Cruz, Bolivia
3Smithsonian Tropical Research Institute, Apartado 0843–03092, Balboa, Ancon, Republic of Panama
4Centro de Investigaciones en Ecosistemas, Universidad Nacional Auto ´noma de Mexico, Campus Morelia, Antigua Carretera a
Patzcuaro 8701, 58190, Morelia, Michoacan, Mexico
5Department of Integrative Biology, 3060 Valley Life Sciences Building, University of California, Berkeley, California 94720-3140 USA
6Department of Biological Sciences, Louisiana State University, Baton Rouge, Louisiana 70803-1715 USA
7Department of Ecology, Evolution and Marine Biology, University of California, Santa Barbara, California 931067 USA
8Department of Ecology, Evolution, and Behavior, 100 Ecology Building, 1987 Upper Buford Circle, St. Paul, Minnesota 55108 USA
9Arnold Arboretum of Harvard University, Harvard University Herbaria, 22 Divinity Avenue, Cambridge, Massachusetts 02138 USA
10Department of Biological Sciences, Macquarie University, New South Wales 2109 Australia
Abstract.
traits vary among species and to what extent this variation has adaptive value. Here we
evaluate relationships between four functional traits (seed volume, specific leaf area, wood
density, and adult stature) and two demographic attributes (diameter growth and tree
mortality) for large trees of 240 tree species from five Neotropical forests. We evaluate how
these key functional traits are related to survival and growth and whether similar relationships
between traits and demography hold across different tropical forests.
There was a tendency for a trade-off between growth and survival across rain forest tree
species. Wood density, seed volume, and adult stature were significant predictors of growth
and/or mortality. Both growth and mortality rates declined with an increase in wood density.
This is consistent with greater construction costs and greater resistance to stem damage for
denser wood. Growth and mortality rates also declined as seed volume increased. This is
consistent with an adaptive syndrome in which species tolerant of low resource availability (in
this case shade-tolerant species) have large seeds to establish successfully and low inherent
growth and mortality rates. Growth increased and mortality decreased with an increase in
adult stature, because taller species have a greater access to light and longer life spans. Specific
leaf area was, surprisingly, only modestly informative for the performance of large trees and
had ambiguous relationships with growth and survival.
Single traits accounted for 9–55% of the interspecific variation in growth and mortality rates
at individual sites. Significant correlations with demographic rates tended to be similar across
forests and for phylogenetically independent contrasts as well as for cross-species analyses that
treated each species as an independent observation. In combination, the morphological traits
explained 41% of the variation in growth rate and 54% of the variation in mortality rate, with
wood density being the best predictor of growth and mortality. Relationships between
functional traits and demographic rates were statistically similar across a wide range of
Neotropical forests. The consistency of these results strongly suggests that tropical rain forest
species face similar trade-offs in different sites and converge on similar sets of solutions.
A central goal of comparative plant ecology is to understand how functional
Key words:
species coexistence; specific leaf area; trees; tropical rain forest; wood density.
demography; functional ecology; height; life-history theory; seed mass; shade tolerance;
INTRODUCTION
Functional traits are attributes of species that
influence their vital rates (survival, growth, and repro-
duction) and, ultimately, fitness (Ackerly 2003). A
central goal of plant ecology is to understand how
functional traits vary among species and to what extent
this variation has adaptive value. The study of
interspecific variation in plant traits has generated
important insights into the occurrence of trait trade-
offs and trait syndromes (Lambers and Poorter 1992,
Grime et al. 1997, Reich et al. 2003), the classification of
plants into functional groups (Dı´az and Cabido 1997,
Grime 2001), and the consequences of these trade-offs
Manuscript received 7 February 2007; revised 29 August
2007; accepted 21 September 2007; final version received 7
November 2007. Corresponding Editor: K. T. Killingbeck.
11E-mail: lourens.poorter@wur.nl
1908

Page 2

and functional groups for ecosystem functioning (ter
Steege and Hammond 2001, Dı´az et al. 2004). Several
key traits have been proposed to be universally
important for plant performance and to represent
relatively independent aspects of plant ecological ‘‘strat-
egy.’’ Among these, seed size, specific leaf area (leaf area
per unit leaf mass; SLA), wood density (dry mass per
unit fresh volume; WD), and plant height at maturity
(Hmax) have been shown to play a central role in plant
regeneration and functional ecology, and they represent
relatively easily measured characteristics that can be
obtained for large numbers of species (Westoby 1998,
Weiher et al. 1999).
Seed size strongly influences reproduction and estab-
lishment (Mazer 1990, Paz et al. 2005). A large seed
provides the new germinant with sufficient reserves to
establish successfully under low-resource conditions
(Kitajima 2002), to make a large seedling with a greater
chance of escaping size-dependent mortality, and to
recover from damage due to herbivory or falling debris
(Harms and Dalling 1997). By contrast, small-seeded
species produce a greater number of seeds per unit of
reproductive effort (Moles et al. 2004) and are better
colonizers of sites that are ephemeral in space and time
(Dalling et al. 1998).
Specific leaf area is the light-capturing foliar area per
unit of leaf biomass invested (see Plate 1). Species with
high SLA tend to have high nutrient concentrations and
mass-based photosynthesis and respiration rates (Reich
et al. 1992, Wright et al. 2004, Poorter and Bongers
2006). Interspecific variation in seedling growth rate is
largely driven by variation in SLA (Poorter and Van der
Werf 1998, Wright and Westoby 1999). Low SLA leaves
tend to be thick and dense, and thus physically robust
and less attractive to herbivores than leaves with high
SLA (Coley 1983, Wright and Westoby 2002). Conse-
quently, low SLA leaves tend to be longer-lived, which
by itself may lead to longer plant life spans (Sterck et al.
2006).
Wood density represents the biomass invested per unit
wood volume. Low WD can contribute to higher stem
growth rate because more volume is produced per unit
biomass (King et al. 2005). Conversely, high-density
wood tends to be constructed of small cells with thick
walls and limited intercellular space (Castro-Dı´ez et al.
1998). This makes stems more resistant to breakage (van
Gelder et al. 2006) and to fungal and pathogen attack
(Augspurger 1984), thus contributing to enhanced plant
survival (Muller-Landau 2004). Finally, Hmaxplays a
fundamental role in access to light (Westoby 1998,
Poorter et al. 2005). Light is a unidirectional resource
and competition for light is highly asymmetric. Taller
plant species intercept, on average, more light and thus
potentially realize faster growth rates.
Although the functional value of these traits forms
one of the central paradigms of comparative plant
ecology (Westoby 1998, Reich et al. 2003), the relation-
ships between these traits and vital rates such as growth
and mortality have rarely been evaluated under field
conditions. This is especially the case for long-lived and
slow-growing organisms such as trees for which the
determination of vital rates requires monitoring large
populations for long time periods.
Relationships among functional traits and vital rates
may reflect direct effects, but it is more likely that both
form part of a suite of coevolved traits (Reich et al.
2003). Life-history theory predicts that species special-
ized for ephemeral high-resource environments should
have a strong ability to colonize and to preempt
resources by having a copious production of small
seeds, fast growth, small size at first reproduction, and
rapid completion of their life cycle (Pianka 1970). In
contrast, species specialized for stable low-resource
environments should have large offspring, high survival,
large size at reproduction, and long life span. Under
low-resource conditions, selection favors allocation to
storage and defense to maintain accumulated resources
(Kitajima 1996). In tropical rain forests these two life-
history extremes are represented by pioneer species that
have high growth and mortality rates and regenerate in
gaps and shade-tolerant species that have slow growth
and low mortality rates that enable survival in the
shaded understory (Condit et al. 1996, Wright et al.
2003). These life-history extremes also give rise to the
growth/survival trade-off that is frequently observed
among tropical tree species (Kitajima 1994, Wright et al.
2003).
This study is the first to quantify relationships
between vital rates (growth and mortality) and key
functional traits (seed volume, SLA, WD, and Hmax)
across a large group of species in the post-regeneration
stage and to assess whether similar trait–rate relation-
ships exist in different plant communities. Data were
collected for ‘‘large’’ trees (mostly between 10 and 50 cm
diameter) of 240 species from five Neotropical forests.
We pose two central questions. First, are functional
traits known to be good predictors of demographic rates
in the regeneration stage also good predictors of
performance in the post-regeneration stage? Second,
are there general trait–rate relationships across all five
forests? Based on life-history theory and empirical
studies (as cited here above) we predicted that (1)
growth and mortality rates should decrease with an
increase in seed size; (2) growth and mortality rates
should increase with an increase in SLA; (3) growth and
mortality rates should decrease with higher wood
density; (4) growth rates should decrease with an
increase in Hmaxbecause species from disturbed micro-
sites should be small and grow fast, and mortality rates
should decrease with Hmaxbecause species adapted to
low-resource environments should be tall and long-lived;
and (5) relationships between functional traits and
demographic rates should be consistent across sites
because these relationships express general convergent
evolutionary patterns of adaptive value. All interspecific
July 20081909FUNCTIONAL TRAITS AND DEMOGRAPHIC RATES

Page 3

relationships were examined using interspecific correla-
tions and phylogenetically independent contrasts.
METHODS
Research sites and species
Functional traits and demographic rates were com-
piled from permanent sample plots located in five
Neotropical rainforests (Table 1). The forests span the
latitudinal range of Neotropical rainforests, from 188 N
(Los Tuxtlas, Mexico) to 158 S (La Chonta, Bolivia).
Annual rainfall varies from 1580 to 4725 mm across
sites, the length of the dry period (number of months
with mean rainfall ,100 mm) varies from 0 to 7 months,
and the mean annual temperature varies from 24.88 to
26.28C. At least one demographic rate and one
functional trait were measured for 240 woody tree
species, representing 146 genera, 46 families, and 21
orders. The delineation of genera followed Kew Botanic
Gardens, and the delineation of higher taxonomic levels
followed the Angiosperm Phylogeny Group (informa-
tion available online).12Palms were excluded from the
study because most have fixed trunk diameters and little
secondary growth.
Functional traits
We evaluated four functional traits; seed volume (SV),
specific leaf area (SLA), wood density (WD), and
maximum adult height (Hmax). Site-specific data were
obtained from the literature or collected in the field for
each species. Measurement protocols occasionally varied
among sites; hence, a concerted effort was made to
standardize data to facilitate comparisons among sites.
Here we describe methods of trait measurement only
briefly; further details are reported by Wright et al.
(2007).
Seed volume.—At most sites SV was calculated from
seed lengths and widths, assuming an elliptical shape.
For the Mexican sites, seeds including the seed coat were
then dried and weighed. At Barro Colorado Island
(BCI) seed dry mass was taken from the literature
(diaspore masses, Augspurger 1986; simple seed masses
and diaspore masses, Kitajima 1992; diaspore masses
without wings, Dalling et al. 1998) or obtained from
seeds collected in the field (endosperm and embryo);
then, seed mass was converted to volume using an
allometric equation based on the Los Tuxtlas data
(log10[mean seed volume] ¼ 0.043 ? 1.027log10[mean
seed mass], r2¼ 0.93, n ¼ 272) (Wright et al. 2007).
Specific leaf area.—Leaves were collected from 5–10
m tall trees at Los Tuxtlas (Bongers and Popma 1990),
from 6–16 m tall trees at La Chonta (Rozendaal et al.
2006), and from canopy trees at Fort Sherman and BCI.
Sun leaves were collected at all sites except BCI, where a
mixture of sun and shade leaves was collected. Specific
leaf area was calculated as the leaf blade area divided by
the leaf dry mass, excluding the petiole. For compound-
leaved species the rachis was included for La Chonta or
only winged rachae were included for BCI and Fort
Sherman. For some BCI species SLA was measured on
leaf discs rather than whole leaves. Whole-leaf SLA was
estimated for these species using the allometric equation
for species with SLA for both discs and whole leaves
(log10[whole-leaf SLA] ¼ 0.0129 ? 0.972log10[leaf disc
SLA], r2¼ 0.91, n ¼ 101) (Wright et al. 2007).
Wood density.—Wood density came from various
published sources (Los Tuxtlas, Barajas-Morales 1985;
BCI, Muller-Landau 2004, Chave et al. 2006). At La
Chonta wood samples of tall species were collected from
trees between 20 and 50 cm dbh (measured at 1.3 m
above the ground surface); samples for shorter species
were taken from plants close to the maximal dbh for
each species. Fresh wood volume was determined with
the water displacement method, after which samples
were oven-dried at 708C and weighed.
Maximum adult height.—At Los Tuxtlas, Hmaxwas
visually estimated for the five largest trees or obtained
from Ibarra-Manrı´quez and Sinaca (1995). At BCI and
Chajul, Hmax was either measured directly, estimated
from maximum dbh using a regression equation for all
species pooled, or taken (for BCI only) from Croat
(1978). At La Chonta, plant height was measured with a
TABLE 1. Environmental characteristics of the five rain forest sites.
Site
Forest
type LatitiudeLongitude
MAT
(8C)
Rainfall
(mm/yr)
Length of dry
period (mo) PSP (ha)Period (yr)
La Chonta, Bolivia1
MSE158450S 628600W24.815807 81 ha, .40 cm;
40.5 ha, .20 cm;
12 ha, .10 cm
50 ha, .1 cm
14 ha, .10 cm
6 ha, .1 cm
5 ha, .10 cm
2–4 (1)
BCI, Panama2
Chajul, Mexico3
Fort Sherman, Panama4
Los Tuxtlas, Mexico5
MSE
ME
ME
wet
98100N
168040N
98170N
188350N
798850W
908450W
798580W
938060W
26.2
24.0
26.2
24.6
2632
3000
3057
4725
4
4
3
0
5 (5)
10 (1)
3 (3)
8 (8)
Notes: Forest type (MSE, moist semi-evergreen; ME, moist evergreen), latitude, longitude, mean annual temperature (MAT),
annual rainfall, length of dry period (number of months with ,100 mm rain/mo), size of permanent sample plots (PSP) with the
minimum diameter at breast height of trees included, and the monitoring period used to calculate demographic rates (measurement
interval in parentheses) are given. The sites are ordered according to decreasing seasonality. Environmental data are (with some
modifications) from Wright et al. (2007). Superscripts refer to the source for the site description. 1, Pen ˜ a-Claros et al. (2008); 2,
Leigh et al. (1982); 3, Ibarra-Manrı´quez and Martı´nez-Ramos (2002); 4, Condit et al. (2004); 5, Bongers et al. (1988).
12hhttp://www.mobot.org/MOBOT/research/APweb/i
L. POORTER ET AL.1910 Ecology, Vol. 89, No. 7

Page 4

clinometer for 25 mature trees per species and regressed
against diameter using the asymptotic formula of
Thomas (1996). Hmaxwas calculated for the tree with
the largest diameter, using species-specific regression
equations (Poorter et al. 2006).
Demographic rates
Demographic rates were determined for trees between
10 and 50 cm dbh for mapped plots at each site (Table
1). Plot size varied from 5 to 81 ha, measurement
intervals from one to eight years, and the total
evaluation period from two to 10 years. All plots
supported mature forest without signs of recent human
disturbance, excepting La Chonta, where low-impact
logging of 1 tree/ha occurred in the 1970s. Annual
mortality rates (MR) were calculated as (log[N0] ?
log[Nt])/time for species with 20 or more individuals,
where N0and Ntare the numbers of individuals at the
first and last measurement, respectively. Relative stem
diameter growth rates (RGR, in millimeters per
millimeter per year) were calculated as (log[dbht] ?
log[dbh0])/time, where dbh0 and dbht are the tree
diameters at the first and last measurement, respectively.
Mean RGR was calculated for all species that had 10 or
more individuals. For Chajul and Los Tuxtlas, trees
with resprouts were excluded.
Interspecific comparisons might potentially be ham-
pered if growth is size dependent and if size distributions
vary among species. We evaluated the potential magni-
tude of this problem for BCI by comparing absolute
growth rates standardized to 25 cm dbh with the average
RGR for the 10–50 cm dbh interval. We fit the
logarithm of absolute growth rates to the logarithm of
initial diameter for species with 20 or more individuals in
the 10–50 cm dbh interval. We then used the log–log
regression equations to estimate the absolute growth
rate of 25 cm dbh trees. This estimated absolute growth
rate and the average RGR observed for trees in the 10–
50 cm DBH interval were strongly correlated (Pearson’s
r ¼ 0.95, P , 0.001, n ¼ 73), indicating that size-
dependent growth does not bias the analysis.
Analyses
We used three types of analyses (correlation, AN-
COVA, and multiple regression) to provide different
insights. First, Pearson correlations were used to
evaluate interspecific relationships between demographic
rates and functional traits at each site (a so-called
‘‘cross-species analysis’’). This analysis identifies statis-
tically significant demographic-trait relationships and
permits a first comparison of the direction and strength
(r2) of these relationships within and across sites. These
within-site correlations are also free of any between-site
biases that might be introduced by the small differences
in census intervals used to estimate vital rates and the
differences in protocols used to measure traits. Loga-
rithmic transformations of RGR and SV and an angular
transformation of MR satisfied the normality assump-
tion. Twenty species had zero mortality; for these species
the population sizes were apparently too small (median
n ¼ 26) to get a reliable estimate of mortality rate, and
they were therefore excluded from MR analyses.
Studentized residuals were .3 in three instances, and
these cases were excluded.
Second, we used ANCOVA models to test the null
hypothesis that all sites showed similar trait–rate
relationships. In these models, RGR or MR were
dependent variables, functional traits were covariates,
and site was a grouping factor with five levels. A
significant site 3 covariate interaction would indicate
that the slope of the trait–rate relationship differs among
sites. When the interaction was insignificant the homo-
geneity of slopes assumption of ANCOVA was met and
the analysis was repeated without the insignificant
interaction term to determine whether the intercept of
the trait–rate relationship differed among sites. Gener-
alized linear interactive modeling (GLIM) enabled a
binomial error and logit link function for analyses
involving MR and a normal error and an identity link
function for analyses involving RGR. We weighted
species by the number of individuals used to determine
RGR or MR for the ANCOVA analyses because the
precision of vital rate estimates increases with the sample
size.
Third, forward multiple regressions were performed
to evaluate relationships between demographic rates and
combinations of functional traits. The multiple regres-
sion analyses were performed for all five sites combined
(with site as a dummy variable) and for BCI and La
Chonta only (because too few species had all four
functional traits measured at the three remaining sites).
Trait–rate correlations were also assessed using
phylogenetically independent contrasts (PIC). In cross-
species analyses, each species contributes a single data
point. In PIC analyses, each independent contrast
(evolutionary divergence) contributes a data point,
allowing one to ask (1) whether evolutionary shifts in
one trait have tended to be associated throughout
history with evolutionary shifts in another trait and (2)
whether results from cross-species analyses largely
reflect trait differences among clades (Westoby 1999).
A phylogenetic tree describing the hypothesized evolu-
tionary relationships among species was constructed
using Phylomatic (Webb and Donoghue 2005; based on
the conservative angiosperm phylogeny, C20040402).
Where relationships among genera within a given family
were not fully resolved they were treated as polytomies.
Relationships among species within genera were also
treated as polytomies. Analyses were conducted with the
‘‘Analysis of Traits’’ (version 3.0) module in Phylocom
version 3.22 (Webb et al. 2006). Phylogenetic branch
lengths were set to 1 and polytomies were resolved to
provide one contrast, following Pagel (1992). A corre-
lation coefficient was then calculated between the set of
PICs for each pair of traits. Significance tests in these
analyses use N ? 1 degrees of freedom, where N is the
July 20081911 FUNCTIONAL TRAITS AND DEMOGRAPHIC RATES

Page 5

number of internal nodes (i.e., PICs) providing con-
trasts.
RESULTS
Relative growth rate and MR were measured for 278
(236 different species) and 198 (173 species) species–site
combinations, respectively. Maximum height, SV, WD,
and SLA were measured for 268, 266, 191, and 132 of
these species–site combinations, respectively. There was
considerable variation in sample size for the four
functional traits and the two demographic rates both
within and among sites (Table 2).
MR vs. RGR
Mortality rate and RGR were significantly positively
correlated for all sites combined (Fig. 1) and for La
Chonta alone (r¼0.44, P¼0.008, n¼36). For the other
sites the relationship was positive but not significant
(Fort Sherman, r¼0.29, P¼0.17, n¼24; BCI, r¼0.12,
P¼0.24, n¼94; Chajul, r¼0.33, P¼0.086, n¼28; Los
Tuxtlas, r¼0.07, P¼0.83, n¼12). To check whether all
sites show a similar relationship between MR and RGR,
we did an ANCOVA analysis with site as factor and
RGR as covariate and weighted for the number of
individuals per species. Mortality rate did not vary with
site (F4,184¼2.06, P¼0.088) and RGR (F1,184¼2.77, P
¼ 0.098), but there was a significant site 3 RGR
interaction (F4,184¼ 12.5, P ¼ 0.044).
Relationships between functional traits and RGR
Relative growth rate was significantly negatively
correlated with SV when all sites were combined and
also for La Chonta and Chajul alone (Table 2, Fig. 2,
Appendix A). In the ANCOVAs, the interactions
between site and the covariates or functional traits were
always insignificant, the homogeneity of slopes assump-
tion of ANCOVA was always satisfied, and the analyses
were performed again without the insignificant site–
covariate interactions. This final step of the ANCOVA
indicated that intercepts differed significantly among
sites for the RGR–SV relationship (Table 3). Chajul and
La Chonta had significantly lower intercepts and lower
RGR than the other three forests (post hoc test, P ,
0.05). Relative growth rate was also significantly lower
at Chajul and La Chonta than at the other sites when the
other functional traits were used as covariates (Table 3).
Relative growth rate was significantly negatively
correlated with SLA when all sites were combined
(Table 2). This significant relationship was caused by the
combination of relatively high SLA and low RGR for
species from La Chonta. The ANCOVA takes such site
differences into account. In the ANCOVA, site had a
highly significant effect on RGR, SLA was also
significant, and the slope of the RGR–SLA relationship
was reversed and became positive (Table 3). Deciduous
species had a slightly higher SLA (12.6 mm2/mg) than
evergreen species (11.5 mm2/mg; ANOVA, F1,101¼4.11,
TABLE 2.
(MR) for five rain forest sites, for all sites combined (All), for all sites combined but taking average values for species that occur
at several sites (Species means), and for phylogenetically independent contrasts (Ind. contrasts).
Pearson correlation coefficients between functional traits and relative diameter growth rate (RGR) and mortality rate
Rain forest site
SV–RGRSV–MR SLA–RGRSLA–MRWD–RGRWD–MRHmax–RGRHmax–MR
rNrNrNrNrNrNrNrN
La Chonta
BCI
Chajul
Fort Sherman
Los Tuxtlas
All
Species means
Ind. contrasts
?0.43
?0.02 125 ?0.36
?0.33
0.10 38 ?0.10
?0.28
?0.15 260 ?0.43 189
?0.15 219 ?0.42 164
?0.18
45 ?0.46
40 ?0.70
12 ?0.74
36
92
23
23
15
0.16
0.24
?0.07
?0.03
0.36
?0.20
?0.17
?0.33
34
44
14
25
11
128
107
55
0.40
?0.14
?0.03
0.27
0.27
0.03
0.05
?0.13
28
34
8
17
14
101
87
48
?0.45
?0.35
?0.19
?0.58
?0.10
?0.37
?0.38
?0.15
46
81
21
27
11
186
149
70
?0.30
?0.15
?0.40
0.25
?0.58
?0.22 139
?0.19 115
?0.27
36
63
12
14
14
0.30
0.34
0.28
0.55
?0.62
0.38
0.38
0.34
46
133
41
30
12
262
220
97
?0.26
?0.16
?0.14
?0.23
0.12
?0.08
?0.11
0.01
36
95
26
19
15
191
166
7293 ?0.25 7258
Notes: N indicates the number of species. Functional traits included are seed volume (SV), specific leaf area (SLA), wood density
(WD), and maximum adult height (Hmax). Logarithms of RGR and seed volume and the angular transformation of MR achieved
normality. Significant correlations are in boldface type (P , 0.05) or in boldface italic type (P , 0.01).
FIG. 1.
relative diameter growth rate (RGR) of tree species at five
Neotropical forest sites (gray triangles, La Chonta; gray circles,
Barro Colorado Island; black circles, Fort Sherman; gray
diamonds, Chajul; black diamonds, Los Tuxtlas). Note the
logarithmic scale for RGR. Pearson correlation coefficient and
significance level are shown.
*** P , 0.001.
Relationship between mortality rate (MR) and
L. POORTER ET AL.1912Ecology, Vol. 89, No. 7

Page 6

P ¼ 0.045). To evaluate the possibility that the RGR–
SLA relationship differed between evergreen and decid-
uous species, we redid the ANCOVA with site and
deciduousness as factors, SLA as a covariate, and the
interaction between deciduousness and SLA. All main
factors were significant, but the interaction effect was
insignificant (F1,101¼ 0.04, P ¼ 0.84), indicating that
evergreen and deciduous species have similar relation-
FIG. 2.
functional traits of tree species at five Neotropical forest sites (gray triangles, La Chonta; gray circles, Barro Colorado Island; black
circles, Fort Sherman; gray diamonds, Chajul; black diamonds, Los Tuxtlas). Functional traits included are seed volume (SV;
measured in mm3prior to log-transformation), specific leaf area (SLA), wood density (WD), and maximum adult height (Hmax).
Note the logarithmic scale for RGR. The Pearson correlation coefficient is shown if significant.
* P , 0.05; ** P , 0.01; *** P , 0.001.
Trait correlations between relative diameter growth rate (RGR), angular-transformed mortality rate (MR), and
July 2008 1913FUNCTIONAL TRAITS AND DEMOGRAPHIC RATES

Page 7

ships between SLA and RGR. The interaction between
deciduousness and SLA was also insignificant when MR
was the dependent variable (F1,76¼ 1.97, P ¼ 0.16).
Relative growth rate was significantly negatively
correlated with WD when all sites were combined and
also for three of the five sites alone (Table 2). Relative
growth rate was significantly positively correlated with
Hmaxwhen all sites were combined and also for three of
the five sites alone (Table 2). These results for WD and
Hmaxwere confirmed by the ANCOVAs in which both
site and the covariate effects were highly significant
(Table 3).
To determine which trait was the best predictor of
interspecific variation in RGR, we used stepwise
multiple regression analyses, which included site as a
dummy variable (Table 4). The first variable included
was site, which explained 48% of the variation in RGR.
The second variable entered was WD, which had a
negative effect on RGR and explained an additional
11% of the variation. Seed volume, SLA, and Hmaxdid
not contribute significantly to the model. The overall
model predicted RGR reasonably well, explaining 59%
of the variation. The model was confined to 110 species
because SLA was available for relatively few species. We
therefore repeated the multiple regression analysis
without SLA, thus expanding the sample size to include
178 species. Wood density was again an important
factor, but this time Hmaxand SV were also significant.
Maximum height had a positive effect and SV a negative
effect on RGR (Table 4). The multiple regression
TABLE 3.
tree species from five rain forest sites.
Relationships between relative stem diameter growth rate (RGR), mortality rate (MR), and four functional traits for
Dependent variable
and covariate
Site main effectCovariate
Sloper2
Site
F df
v2
F df
v2
LC BCICHFS LT
RGR
SV
SLA
WD
Hmax
MR
SV
SLA
WD
Hmax
25.0***
30.0***
21.8***
26.5***
4, 261
4, 134
4, 182
4, 258
6.4*
9.5**
18.1***
18.4***
1, 261
1, 134
1, 182
1, 258
?0.047
0.014
?0.53
0.008
0.32
0.49
0.37
0.33
a
a
a
a
b
b
b
b
a
a
a
a
b
b
b
b
b
b
b
b
4, 187
4, 96
4, 135
4, 189
27.2***
19.6***
22.6***
23.2***
1, 187
1, 96
1, 135
1, 189
36.4***
1.2NS
21.8***
14.2***
?0.309
0.03
?1.91
?0.026
0.22
0.14
0.24
0.16
a
a
a
a
b
b
b
b
ab
ab
ab
ab
ab
ab
ab
b
ab
ab
ab
ab
Notes: Functional traits included are seed volume (SV), specific leaf area (SLA), wood density (WD), and maximum adult height
(Hmax). Dependent variables are RGR and MR, site is a grouping factor, and the functional traits are covariates in ANCOVA
models implemented with generalized linear interactive modeling (GLIM). All interactions between site and functional traits were
insignificant, the homogeneity of slopes assumption of ANCOVA was satisfied, and the analyses were redone without the
interaction term. The F values (for RGR), v2(for MR), P values, degrees of freedom, the slope of the covariate, and coefficients of
determination of the full model (r2) are shown. Sites (LC, La Chonta; BCI, Barro Colorado Island; CH, Chajul; FS, Fort Sherman;
LT, Los Tuxtlas) followed by a different letter are significantly different (Bonferroni, P , 0.05). Relative growth rate and SV were
log-transformed, and MR was angular transformed prior to analyses.
* P , 0.05; ** P , 0.01; *** P , 0.001; NS, P . 0.05.
TABLE 4.
traits.
Stepwise multiple regression relating relative diameter growth rate (RGR) and mortality rate (MR) to site and functional
Variable
Including SLAExcluding SLA
RGR MR RGR MR
br2
Pbr2
Pbr2
Pbr2
P
Site
SV
SLA
WD
Hmax
Model
df model
df error
—
—
—
0.476
—
—
0.109
—
***
—
—
***
—
———
***
***
***
**
—0.302
0.021
***
*
—0.118
0.146
***
***
?0.019
?0.007
?0.166
?0.002
0.067
0.144
0.276
0.056
?0.032
?0.501
0.006
?0.018
?0.106
?0.002
?0.724
—
0.055
0.031
***
**
0.088
0.059
***
***
0.585
5
104
***0.543
4
82
***0.409
7
170
***0.411
7
128
***
Notes: Functional traits included were seed volume (SV), specific leaf area (SLA), wood density (WD), and maximum adult
height (Hmax). Dependent variables were weighted by sample size. The analysis was repeated twice, with SLA and without SLA.
Regression slopes (b), partial r2, and significance levels (P) are shown. The number of species for RGR and MR are 110 and 87 for
the analysis with SLA, and 178 and 136 for the analysis without SLA. Relative growth rate and SV were log-transformed, and MR
was angular-transformed prior to analyses. When the site main effect was significant, the corresponding r2is shown. A dash
indicates that independent variables were not significant and were excluded from the final model.
* P , 0.05; ** P , 0.01; *** P , 0.001.
L. POORTER ET AL.1914Ecology, Vol. 89, No. 7

Page 8

analysis was also carried out for the two sites (La
Chonta and BCI) that had complete data for a sufficient
number of species. Relative growth rate was negatively
related to WD for La Chonta (r2¼ 0.29) and negatively
related to WD, SLA, and SV for BCI (r2¼ 0.42;
Appendix C).
Relationships between functional traits and MR
Mortality rate was significantly negatively related to
SV when all sites were combined and for four of the five
sites alone (Fig. 2, Table 2, Appendix B). Seed volume
also had a highly significant negative effect on MR in the
ANCOVA (Table 3). La Chonta had a significantly
lower intercept and therefore lower MR than BCI (post
hoc tests, P , 0.05), and the same results were obtained
when other functional traits were used as covariates.
Mortality rate was unrelated to SLA for the combined
sites and positively related to SLA for La Chonta alone
(Tables 2 and 3). Mortality rate was significantly
negatively related to wood density when all sites were
combined and for Los Tuxtlas alone (Tables 2 and 3).
Finally, MR was significantly negatively related to Hmax
for all sites combined, but only when intersite differences
in MR were incorporated in the ANCOVA model
(Table 3). There were no significant interactions between
site and any of the covariates or functional traits in the
four ANCOVA analyses.
Wood density entered the stepwise multiple regression
model first and explained 28% of the variation in MR
(Table 4). Specific leaf area, SV, and Hmaxexplained an
additional 14%, 7%, and 6% of the variation in MR,
respectively. Only site did not contribute significantly to
the multiple regression model. The relationships between
MR and the four functional traits were all negative.
Overall, the model explained 54% of the variation in
MR. The multiple regression analysis was repeated
without SLA, thus including a larger number (178) of
species. Again the three remaining functional traits had
significant negative effects on MR. Multiple regression
analysis at individual sites showed that MR was
negatively related to WD and Hmax(r2¼ 0.50) for La
Chonta and that MR was negatively related to SV and
SLA (r2¼ 0.24) for BCI (Appendix C).
Independent contrasts
Correlations based on independent contrasts were
generally similar to the cross-species correlations pre-
sented in Table 2. Two significant cross-species correla-
PLATE 1.
to enhance the leaf display per unit of leaf biomass invested. Photo credit: M. Pen ˜ a-Claros.
Pourouma cecropiifolia leaf capturing sunlight in the La Chonta forest, Bolivia. A high specific leaf area is important
July 20081915FUNCTIONAL TRAITS AND DEMOGRAPHIC RATES

Page 9

tions (RGR vs. SV and RGR vs. WD) were no longer
significant when phylogeny was taken into account;
however, for the RGR–SV relationship the magnitude
of the phylogenetic correlation was larger than for the
cross-species correlation, suggesting that the change in
significance was due to lower power. Correlation
coefficients for RGR vs. SLA and MR vs. WD also
increased in magnitude with independent contrasts.
DISCUSSION
The four morphological traits were each significantly
correlated with relative growth rates and/or mortality
rates, suggesting either direct functional effects or
indirect correlations mediated by species life histories.
Analyses based on independent contrasts did not lead to
consistent reductions in the magnitude of the correla-
tions, indicating that the relationships arose through
repeated evolutionary divergences and functional con-
vergences.
Comparison among sites: similar trade-offs,
different vital rates
The relationships between demographic rates and
functional traits were statistically similar across a wide
range of Neotropical forests (Table 3). This result
strongly suggests that species from different sites face
similar trade-offs and converge on similar sets of
solutions (cf. ter Steege and Hammond 2001, Reich et
al. 2003, Wright et al. 2004). La Chonta and Chajul had
slower growth rates than the other sites (Table 3). La
Chonta is more seasonal than the other sites (Table 1),
and growth might slow during the seven dry months,
leading to lower annual growth rates. Slow growth at La
Chonta might also be attributed to the relatively poor
soils in the Amazon compared to Central America
(Powers 2005). Slow growth rates at Chajul might also be
explained by soils. Most Chajul trees grew on low hills
with sandy, acidic, phosphorous-poor soil or on karst
with shallow soils and poor water-holding capacity.
Seed volume
Growth and mortality rates decreased with larger SV
both within and across forest sites, in agreement with
our first prediction (see Introduction). Seed size is an
important component of plant regeneration strategies.
Seed size directly affects seedling survival and indirectly
affects seedling growth, as small-seeded species have
limited seed reserves and must deploy roots and leaves
rapidly to become autotrophic. Many small-seeded
species have photosynthetic cotyledons (Ibarra-Manrı´-
quez et al. 2001) and high SLA and leaf area ratio
(LAR), which all fuel growth. Correlations between
RGR and seed size are therefore especially high just
after germination (Paz et al. 2005), but disappear over
time as the initial SLA and LAR differences vanish
(Baraloto et al. 2005, Poorter and Rose 2005). Seed size
is also correlated with the growth and mortality rates of
large trees because seed size is closely linked with a suite
of traits that enhance shade tolerance (Troup 1921,
Hammond and Brown 1995, Osunkoya 1996). Shade-
tolerant species have large seeds to regenerate success-
fully in the shade and inherently slower growth and
mortality rates compared to light-demanding species.
Not all species fit this paradigm. Shade-tolerant species
can have minute seeds, and pioneer species can have
large seeds (Metcalfe and Grubb 1995). The consistent
negative relationships observed between large tree
mortality and seed volumes suggest, however, that such
outliers are rare (Table 2).
Specific leaf area
The relationship between SLA and the growth and
mortality of large trees was ambiguous. Relative growth
rate was negatively related to SLA when sites were
pooled because La Chonta tended to have the highest
SLA and lowest RGR values. The ANCOVA incorpo-
rates such site differences, and RGR–SLA was reversed
and was positive in the ANCOVA, which is consistent
with our second hypothesis. However, in the multiple
regression analysis for the combined sites, SLA had no
effect on RGR and a negative effect on MR, which is
inconsistent with our second hypothesis. The ambiguous
and modest effect of SLA on demographic rates is
surprising because SLA is considered to be a key trait
influencing plant performance (e.g., Lambers and
Poorter 1992, Grime et al. 1997, Wright et al. 2004)
and is routinely included in screening studies of plant
traits (e.g., Weiher et al. 1999). Growth was expected to
scale positively with SLA, because SLA indicates the
efficiency of biomass investment for light interception.
Mortality rate was expected to scale positively with
SLA, because high SLA leaves are less tough and suffer
from higher herbivory rates (Coley 1983). Studies with
seedlings under controlled conditions invariably show
that SLA is a good predictor of interspecific variation in
growth and survival (e.g., Kitajima 1994). Poorter and
Bongers (2006) evaluated the performance of saplings of
53 rain forest species in the field and found that SLA
was a reasonable predictor of interspecific variation in
height growth (r2¼ 0.18) and mortality (r2¼ 0.26).
Specific leaf area may be important for the performance
of small plants, as leaf removal has a large impact on
their survival (e.g., Pearson et al. 2003), but less
important for the performance of large trees that have
substantial reserves to recuperate from photosynthetic
biomass loss (Wu ¨ rth et al. 2005). Similarly, SLA may be
more important for light interception of saplings for
which it determines to a large extent the total leaf area
and leaf area ratio, but less so for large trees, for which
leaf area and light interception are largely determined by
branching patterns, the number of meristems, and tree
architecture (Sterck and Bongers 2001).
Wood density
Both growth and mortality rates declined with higher
WD, in line with the third prediction. The negative
L. POORTER ET AL.1916 Ecology, Vol. 89, No. 7

Page 10

relationship between RGR and WD was significant for
three of our five sites (0.12 ? r2? 0.34), and the two
sites for which this relationship was insignificant had the
smallest sample sizes. In the all-species regression and in
the ANCOVA, WD was negatively correlated with both
RGR and MR. Similar negative relationships between
absolute growth rates of trees and WD have been
reported from tropical rain forests from Guyana (r2¼
0.25), Surinam (r2¼ 0.37) and Borneo (r2¼ 0.19,
summarized in ter Steege 2003), Amazonia (r2¼ 0.13,
Nascimiento et al. 2004), and Malaysia (r2¼ 0.51, King
et al. 2006). Wood density was the best predictor of
RGR (Table 4, Appendix A), presumably because it is
directly related to the construction costs of the wood.
All else being equal, species with dense wood grow more
slowly than do species with soft wood, and wood density
therefore sets an upper limit to potential diameter
growth. Interspecific variation in wood density is mainly
caused by spaces within and between cells. Soft-wooded
species also grow more rapidly because their large vessel
diameters enhance hydraulic conductance and water
transport capacity and, hence, photosynthetic rates
(Santiago et al. 2004).
Mortality rate has been found to be negatively related
to WD for seedlings growing under controlled condi-
tions (r2¼ 0.50, Kitajima 1994), for saplings growing in
the field (r2¼ 0.19, Muller-Landau 2004), and in this
study for larger-sized trees (r2¼0.05, all sites combined,
Table 2). High-density wood is less vulnerable to stem
breakage caused by falling debris or strong winds (Putz
et al. 1983) and less vulnerable to stem rot and
pathogens (Augspurger 1984). Small seedlings and
saplings are more vulnerable to damage from falling
debris and pathogens than are large trees, which might
explain why WD is a better predictor for MR in the
juvenile phase. Finally, WD is closely related to shade
tolerance (van Gelder et al. 2006). Shade-tolerant species
may combine well-defended, long-lived tissues with a
long life span and low inherent mortality rates (Loehle
2000). Osunkoya (1996) found wood density to be the
best predictor of light requirements for regeneration
among 89 Australian rain forest tree species. Foresters
have long been aware of this association and use wood
density to classify species into different shade tolerance
groups.
Maximum height
Growth increased with an increase in Hmax both
within and among forests, which contradicts the
prediction made on the basis of life-history theory.
Growth consistently increased with Hmaxat three sites
and decreased at one site (Los Tuxtlas). Maximum
height was available for just 12 Los Tuxtlas species, and
one (Nectandra ambigens) was an outlier; when this
outlier was removed the Hmax–RGR relationship was
insignificant for Los Tuxtlas. Life-history theory pre-
dicts that organisms with larger Hmax should have
slower growth rates. This relationship is apparent when
small, annual herbaceous plants are compared with
large, long-lived trees, but is not found when only trees
are compared. Studies from Malaysian rain forest also
indicate that the growth rate of trees increases with Hmax
(Thomas 1996, King et al. 2006). Light increases
exponentially with height in forest canopies, and taller
species may, on average, intercept a disproportionate
amount of light compared to smaller species (Poorter et
al. 2005). This, in combination with their larger crown
area (Poorter et al. 2006), may allow taller species to
realize faster growth rates than shorter species, when
compared at similar diameters. Tall species may also
have higher inherent growth rates because they are more
demanding of light (Poorter et al. 2005, Sheil et al.
2006).
Mortality rate decreased with an increase in Hmaxfor
all sites combined in the ANCOVA and multiple
regression analysis (Tables 3 and 4), which is consistent
with life-history theory (tall, long-lived species should
have low inherent mortality rates). Similar results have
been observed for an Ecuadorian rain forest (Korning
and Balslev 1994). There are two other possible
explanations for the negative Hmax–MR relationship.
First, small species might have higher mortality rates
because they are close to their maximal size for the
diameter range considered (10–50 cm) and many
individuals might be senescent (cf. King et al. 2006).
Environmental conditions might also contribute to the
higher mortality rates observed among smaller species if
small plants have less access to resources and hence
higher mortality. As a consequence, smaller species have
faster population turnover, with higher recruitment and
mortality rates (Kohyama et al. 2003).
Growth vs. mortality and the relative importance
of functional traits
There were fewer significant correlations between
functional traits and MR (six of 20 comparisons) than
between functional traits and RGR (nine of 20), and the
explained variation was also much lower (r2¼0.14–0.24
for MR compared to r2¼ 0.32–0.49 for RGR; Table 3).
The relationships with MR are probably weaker because
of the difficulties to get an accurate estimate of MR.
Mortality is a highly stochastic, slow process, and
therefore either a large sample size or a long evaluation
period is needed to get accurate mortality estimates.
Given this limitation, it is surprising how well four
functional traits predicted mortality rates when all sites
were combined (overall r2¼0.54). Wood density was the
best predictor of MR, followed by SLA, SV, and Hmax.
All functional traits had negative effects on MR, which
is consistent with our predictions for all traits except
SLA. When the multiple regression analysis was carried
out for individual sites, only WD and Hmax were
significant for La Chonta and only SLA and SV were
significant for BCI. Discrepancies among these three
regression analyses are probably due to the smaller
number of species that could be included for La Chonta
July 20081917 FUNCTIONAL TRAITS AND DEMOGRAPHIC RATES

Page 11

and BCI (ca. 30 species per site had complete trait data).
Smaller sample sizes reduce statistical power and
increase the possibility that unusual species might have
influenced the results.
The trade-off between growth and survival might
contribute to the coexistence of forest trees (Kitajima
and Poorter 2008), although this trade-off may be
stronger for small seedlings and saplings (Kitajima 1994,
Wright et al. 2003, Poorter and Bongers 2006) than for
large trees (Fig. 1). Wood density was also the best
predictor of RGR (Table 4) and was the only functional
trait considered that directly contributed to the growth/
survival trade-off for large trees, as it affected survival
positively and growth negatively (Nascimiento et al.
2005). In contrast, leaf traits are also important
determinants of the growth/survival trade-off in the
seedling stage (Kitajima 1994, Poorter 1999, Sterck et al.
2006). Wood density may fulfill such a central role,
because it is closely related to shade tolerance and to a
range of plant functions, including defense, stability,
water relations, carbon gain, and growth (Santiago et al.
2004).
Small seedlings vs. large trees
Results from the literature suggest that trait–rate
relationships are much stronger for small plants than for
large trees (as found in this study). One reason is that
interspecific variation in demographic rates is much
larger for small plants than for large trees, which makes
it statistically easier to detect trait–rate relationships. A
second reason is that tree size tends to swamp more
subtle trait differences. As mentioned above, SLA may
be the strongest determinant of total leaf area, light
interception, and growth in the seedling stage, but crown
size and branching pattern may be the strongest
determinant in the adult stage. A third reason that
trait–rate relationships might be stronger for seedlings
than for adults concerns the contrasting light environ-
ments in which seedlings of different species regenerate
in the field (Wright et al. 2003); for example, seedlings of
pioneers have high population-level mortality rates
because all individuals that establish in the shaded
understory die and high population-level diameter
growth rates because those individuals that survive do
so in gaps, where they can realize high growth rates. In
contrast, adult trees tend to experience more similar
irradiance levels and hence, more similar growth and
mortality rates because they are all in the (sub)canopy.
Data from this study come from five well-studied
tropical forest sites. In putting together the data matrix,
the number of species with complete data decreased
dramatically with the number of variables included. We
know surprisingly little about functional traits and
demographic rates of large suites of tree species.
Comparative, standardized research on large sets of
species is therefore needed to advance our understanding
of the salient features that drive species coexistence,
community composition, and dynamics.
ACKNOWLEDGMENTS
We thank all the people who one way or another have
contributed data to this study and who have made the
tremendous effort to maintain the long-term demographic
plots at the different forest sites. We also thank Robin Chazdon
and Kaoru Kitajima for helpful comments on the manuscript.
This paper is the result of the NCEAS working group ‘‘Life-
history variation and community structure in Neotropical
rainforest communities: ecological and phylogenetic influenc-
es.’’ NCEAS is supported by NSF grant DEB-94-21535, UCSB,
and the State of California. L. Poorter was supported by Veni
grant 863.02.007 from NWO and a fellowship from the
Wageningen Graduate School Production Ecology and Re-
source Conservation. Data from Chajul and Los Tuxtlas were
obtained from PAPIIT-DGAP-UNAM and MacArthur Foun-
dation grants to M. Martı´nez-Ramos and with the technical
support of J. Rodrı´guez-Velazquez.
LITERATURE CITED
Ackerly, D. D. 2003. Community assembly, niche conservatism,
and adaptive evolution in changing environments. Interna-
tional Journal of Plant Sciences 164:165–184.
Augspurger, C. K. 1984. Seedling survival of tropical tree
species: interactions of dispersal distance, light-gaps, and
pathogens. Ecology 65:1705–1712.
Augspurger, C. K. 1986. Morphology and dispersal potential of
wind-dispersed diaspores of neotropical trees. American
Journal of Botany 73:353–363.
Barajas-Morales, J. 1985. Wood structural differences between
trees of two tropical forests in Mexico. IAWA Bulletin 30:
559–586.
Baraloto, C., P. M. Forget, and D. E. Goldberg. 2005. Seed
mass, seedling size, and neotropical seedling establishment.
Journal of Ecology 93:1156–1166.
Bongers, F., and J. Popma. 1990. Leaf characteristics of the
tropical rain forest flora of Los Tuxtlas, Mexico. Botanical
Gazette 151:354–365.
Bongers, F., J. Popma, J. Meave del Castillo, and J. Carabias.
1988. Structure and floristic composition of the lowland rain
forest of Los Tuxtlas, Mexico. Vegetatio 74:55–80.
Castro-Dı´ez, P., J. P. Puyravaud, J. H. C. Cornelissen, and P.
Villar-Salvador. 1998. Stem anatomy and relative growth
rate in seedlings of a wide range of woody plant species and
types. Oecologia 116:57–66.
Chave, J., H. Muller-Landau, T. R. Baker, T. A. Easdale, H. ter
Steege, and C. O. Webb. 2006. Regional and phylogenetic
variation of wood density across 2456 Neotropical tree
species. Ecological Applications 16:2356–2367.
Coley, P. D. 1983. Herbivory and defensive characteristics of
tree species in a lowland tropical forest. Ecological Mono-
graphs 53:209–233.
Condit, R., S. Aguilar, A. Hernandez, R. Perez, S. Lao, G.
Angehr, S. P. Hubbell, and R. B. Foster. 2004. Tropical
forest dynamics across a rainfall gradient and the impact of
an El Nino dry season. Journal of Tropical Ecology 20:51–
72.
Condit, R., S. P. Hubbell, and R. B. Foster. 1996. Assessing the
response of plant functional types to climatic change in
tropical forests. Journal of Vegetation Science 7:405–416.
Croat, T. B. 1978. Flora of Barro Colorado Island. Stanford
University Press, Stanford, California, USA.
Dalling, J. W., S. P. Hubbell, and K. Silvera. 1998. Seed
dispersal, seedling establishment and gap partitioning among
pioneer trees. Journal of Ecology 86:674–689.
Dı´az, S., and M. Cabido. 1997. Plant functional types and
ecosystem function in relation to global change: a multiscale
approach. Journal of Vegetation Science 8:463–474.
Dı´az, S., et al. 2004. The plant traits that drive ecosystems:
evidence from three continents. Journal of Vegetation
Science 15:295–304.
L. POORTER ET AL.1918 Ecology, Vol. 89, No. 7

Page 12

Grime, J. P. 2001. Plant strategies, vegetation processes, and
ecosystem properties. John Wiley and Sons, Chichester, UK.
Grime, J. P., et al. 1997. Integrated screening validates primary
axes of specialisation in plants. Oikos 79:259–281.
Hammond, D. S., and V. K. Brown. 1995. Seed size of woody
plants in relation to disturbance, dispersal, soil type in wet
Neotropical forests. Ecology 76:2544–2561.
Harms, K. E., and J. W. Dalling. 1997. Damage and herbivory
tolerance through resprouting as an advantage of large seed
size in tropical trees and lianas. Journal of Tropical Ecology
13:617–621.
Ibarra-Manrı´quez, G., M. Martı´nez-Ramos, and K. Oyama.
2001. Seedling functional types in a lowland rain forest in
Mexico. American Journal of Botany 88:1801–1812.
Ibarra-Manrı´quez, G., and S. Sinaca Colin. 1995. Lista
florı´stica comentada de la estacio ´ n de biologı´a tropical
‘‘Los Tuxtlas’’, Veracruz, Me ´ xico. Revista de Biologı´a
Tropical 43:75–115.
King, D. A., S. J. Davies, and M. N. Nur Sapardi. 2006.
Growth and mortality are related to adult tree size in a
Malaysian mixed dipterocarp forest. Forest Ecology and
Management 223:152–158.
King, D. A., S. J. Davies, M. N. Nur Supardi, and S. Tan. 2005.
Tree growth is related to light interception and wood density
in two mixed dipterocarp forests of Malaysia. Functional
Ecology 19:445–453.
Kitajima, K. 1992. The importance of cotyledon functional
morphology and patterns of seed reserve utilization for the
physiological ecology of neotropical tree seedlings. Disserta-
tion. University of Illinois, Urbana-Champaign, Illinois,
USA.
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. 1996. Ecophysiology of tropical tree seedlings.
Pages 559–597 in S. S. Mulkey, R. L. Chazdon, and A. P.
Smith, editors. Tropical forest plant ecophysiology. Chap-
man and Hall, New York, New York, USA.
Kitajima, K. 2002. Do shade-tolerant tropical tree seedlings
depend longer on seed reserves? Functional growth analysis
of three Bignoniaceae species. Functional Ecology 16:433–
444.
Kitajima, K., and L. Poorter. In press. Functional basis for
resource niche differentiation by tropical trees. In W. P.
Carson and S. A. Schnitzer, editors. Tropical forest
community ecology. Blackwell, Oxford, UK.
Kohyama, T., E. Suzuki, T. Partomihardjo, T. Yamada, and T.
Kubo. 2003. Tree species differentiation in growth, recruit-
ment and allometry in relation to maximum height in a
Bornean mixed dipterocarp forest. Journal of Ecology 91:
797–806.
Korning, J., and H. Balslev. 1994. Growth rates and mortality
patterns of tropical lowland tree species and the relation to
forest structure in Amazonian Ecuador. Journal of Tropical
Ecology 10:151–166.
Lambers, H., and H. Poorter. 1992. Inherent variation in
growth rate between higher plants: a search for physiological
causes and ecological consequences. Advances in Ecological
Research 23:187–261.
Leigh, E. G., A. S. Rand, and D. M. Windsor. 1982. The
ecology of a tropical forest: seasonal rhythms and long-term
changes. Smithsonian University Press, Washington, D.C.,
USA.
Loehle, C. 2000. Strategy space and the disturbance spectrum: a
life history model for tree species coexistence. American
Naturalist 156:14–33.
Mazer, S. J. 1990. Seed mass of Indiana dune genera and
families—taxonomic and ecological correlates. Evolutionary
Ecology 4:326–357.
Metcalfe, D. J., and P. J. Grubb. 1995. Seed mass and light
requirements for regeneration in South-east Asian rain forest.
Canadian Journal of Botany 73:817–826.
Moles, A. T., D. S. Falster, M. R. Leishman, and M. Westoby.
2004. Small-seeded species produce more seeds per square
metre of canopy per year, but not per individual per lifetime.
Journal of Ecology 92:384–396.
Muller-Landau, H. C. 2004. Interspecific and inter-site varia-
tion in wood specific gravity of tropical trees. Biotropica 36:
20–32.
Nascimiento, H. E. M., W. F. Laurence, R. Condit, S. G.
Laurence, S. D’Angelo, and A. C. Andrade. 2005. Demo-
graphic and life history correlated for Amazonian trees.
Journal of Vegetation Science 16:625–634.
Osunkoya, O. O. 1996. Light requirements for regeneration in
tropical forest plants: taxon-level and ecological attribute
effects. Australian Journal of Ecology 21:421–229.
Pagel, M. D. 1992. Method for the analysis of comparative
data. Journal of Theoretical Biology 156:431–442.
Paz, H., S. J. Mazer, and M. Martı´nez-Ramos. 2005.
Comparative ecology of seed mass in Psychotria (Rubiaceae):
within- and between-species effects of seed mass on early
performance. Functional Ecology 19:707–718.
Pearson, T. H., D. F. R. P. Burslem, R. E. Goeriz, and J. W.
Dalling. 2003. Interactions of gap size and herbivory on
establishment, growth and survival of three species of
neotropical pioneer trees. Journal of Ecology 91:785–796.
Pen ˜ a-Claros, M., E. M. Peters, M. J. Justiniano, F. Bongers, G.
Blate, T. S. Fredericksen, and F. E. Putz. 2008. Regeneration
of commercial tree species following silvicultural treatments
in a moist tropical forest. Forest Ecology and Management
255:1283–1293.
Pianka, E. R. 1970. On r- and K-selection. American Naturalist
104:592–597.
Poorter, H., and A. van der Werf. 1998. Is inherent variation in
RGR determined by LAR at low light and by NAR at high
light? Pages 309–336 in H. Lambers, H. Poorter, and M. M.
I. van Vuuren, editors. Inherent variation in plant growth:
physiological mechanisms and ecological consequences.
Backhuys, Leiden, The Netherlands.
Poorter, L. 1999. Growth responses of fifteen rain forest tree
species to a light gradient: the relative importance of
morphological and physiological traits. Functional Ecology
13:396–410.
Poorter, L., and F. Bongers. 2006. Leaf traits are good
predictors of plant performance across 53 rain forest species.
Ecology 87:1733–1743.
Poorter, L., F. Bongers, F. J. Sterck, and H. Wo ¨ ll. 2005.
Beyond the regeneration phase: differentiation of height-light
trajectories among tropical tree species. Journal of Ecology
93:256–267.
Poorter, L., L. Bongers, and F. Bongers. 2006. Architecture of
54 moist forest tree species: traits, trade-offs, and functional
groups. Ecology 87:1289–1301.
Poorter, L., and S. A. Rose. 2005. Light-dependent changes in
the relationship between seed mass and seedling traits: a
meta-analysis for rain forest tree species. Oecologia 142:378–
387.
Powers, J. S., K. K. Treseder, and M. T. Lerdau. 2005. Fine
roots, arbuscular mycorrhizal hyphae and soil nutrients in
four Neotropical rain forests: patterns across large geograph-
ic distances. New Phytologist 165:913–921.
Putz, F. E., P. D. Coley, K. Lu, A. Montalvo, and A. Aiello.
1983. Uprooting and snapping of trees: structural determi-
nants and ecological consequences. Canadian Journal of
Forest Research 13:1011–1020.
Reich, P. B., M. B. Walters, and D. S. Ellsworth. 1992. Leaf
life-span in relation to leaf, plant, and stand characteristics
among diverse ecosystems. Ecological Monographs 62:365–
392.
July 20081919 FUNCTIONAL TRAITS AND DEMOGRAPHIC RATES

Page 13

Reich, P. B., I. J. Wright, J. Cavender-Bares, J. M. Craine, J.
Oleksyn, M. Westoby, and M. B. Walters. 2003. The
evolution of plant functional variation: traits, spectra, and
strategies. International Journal of Plant Sciences 164:143–
164.
Rozendaal, D. M. A., V. H. Hurtado, and L. Poorter. 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, L. S., G. Goldstein, F. C. Meinzer, J. B. Fisher, K.
Machado, D. Woodruff, and T. Jones. 2004. Leaf photosyn-
thetic traits scale with hydraulic conductivity and wood
density in Panamanian forest canopy trees. Oecologia 140:
543–550.
Sheil, D., A. Salim, J. Chave, J. Vanclay, and W. D.
Hawthorne. 2006. Illumination–size relationships of 109
coexisting tropical forest tree species. Journal of Ecology
94:494–507.
Sterck, F. J., and F. Bongers. 2001. Crown development in
tropical rain forest trees: patterns with tree height and light
availability. Journal of Ecology 89:1–13.
Sterck, F. J., L. Poorter, and F. Schieving. 2006. Leaf traits
determine the growth–survival trade-off across rain forest
tree species. American Naturalist 167:758–765.
Ter Steege, H. 2003. Long-term changes in tropical tree
diversity. Studies from the Guiana Shield, Africa, Borneo,
and Melanesia. Tropenbos Series 22. Tropenbos Internation-
al, Wageningen, The Netherlands.
Ter Steege, H., and D. S. Hammond. 2001. Character
convergence, diversity, and disturbance in tropical rain forest
in Guyana. Ecology 82:3197–3212.
Thomas, S. C. 1996. Asymptotic height as a predictor of growth
and allometric characteristics in Malaysian rain forest trees.
American Journal of Botany 83:556–566.
Troup, R. S. 1921. The silviculture of Indian trees. Volume 1.
Oxford University Press, Oxford, UK.
Van Gelder, H. A., L. Poorter, and F. J. Sterck. 2006. Wood
mechanics, allometry, and life-history variation in a tropical
rain forest tree community. New Phytologist 171:367–378.
Webb, C. O., D. D. Ackerly, and S. Kembel. 2006. Phylocom.
Software for the analysis of community phyloge-
netic structure and character evolution. hhttp://www.
phylodiversity.net/phylocom/i
Webb, C. O., and M. J. Donoghue. 2005. Phylomatic: tree
assembly for applied phylogenetics. Molecular Ecology
Notes 5:181–183.
Weiher, E., A. van der Werf, K. Thompson, M. Roderick, E.
Garnier, and O. Eriksson. 1999. Challenging Theophrastus: a
common core list of plant traits for functional ecology.
Journal of Vegetation Science 10:609–620.
Westoby, M. 1998. A leaf-height-seed (LHS) plant ecology
strategy scheme. Plant and Soil 199:213–227.
Westoby, M. 1999. Generalization in functional plant ecology:
the species sampling problem, plant ecology strategies
schemes, and phylogeny. Pages 847–872 in F. I. Pugnaire
and F. Valladares, editors. Handbook of functional plant
ecology. Marcel Dekker, New York, New York, USA.
Wright, I. J., and M. Westoby. 1999. Differences in seedling
growth behaviour among species: trait correlations across
species, and trait shifts along nutrient compared to rain
gradients. Journal of Ecology 87:85–97.
Wright, I. J., and M. Westoby. 2002. Leaves at low versus high
rainfall: coordination of structure, lifespan and physiology.
New Phytologist 155:403–416.
Wright, I. J., et al. 2004. The worldwide leaf economics
spectrum. Nature 428:621–827.
Wright, I. J., et al. 2007. Relationships among ecologically-
important dimensions of plant trait variation in 7 Neotrop-
ical forests. Annals of Botany 99:1003–1015.
Wright, S. J., H. C. Muller-Landau, R. Condit, and S. P.
Hubbell. 2003. Gap-dependent recruitment, realized vital
rates, and size distributions of tropical trees. Ecology 84:
3174–3185.
Wu ¨ rth, M. K. R., S. Pela ´ ez-Riedl, S. J. Wright, and C. Ko ¨ rner.
2005. Non-structural carbohydrate pools in a tropical forest.
Oecologia 143:11–24.
APPENDIX A
Figures showing trait correlations between relative diameter growth rate and functional traits of tree species at five Neotropical
forest sites (Ecological Archives E089-111-A1).
APPENDIX B
Figures showing trait correlations between mortality rate and functional traits of tree species at five Neotropical forest sites
(Ecological Archives E089-111-A2).
APPENDIX C
Multiple regression relating relative diameter growth rate and mortality rate to functional traits for La Chonta and Barro
Colorado Island (Ecological Archives E089-111-A3).
L. POORTER ET AL.1920 Ecology, Vol. 89, No. 7

Page 14

Ecological Archives E089-111-A1
L. Poorter, S. J. Wright, H. Paz, D. D. Ackerly, R. Condit, G. Ibarra-Manríquez, K. E. Harms, J. C. Lincona, M. Martínez-
Ramos, S. J. Mazer, H. C. Muller-Landau, M. Peña-Claros, C. O. Webb, and I. J. Wright. 2008. Are functional traits good
predictors of demographic rates? Evidence from five Neotropical forests. Ecology 89:1908–1920.
Appendix A. Figures showing trait correlations between relative diameter growth rate (RGR) and functional traits of tree species at five Neotropical forest
sites.

FIG. A1. Trait correlations between relative diameter growth rate (RGR) and functional traits of tree species at five Neotropical forest
sites. Note that the axes of RGR and seed volume were log-transformed prior to analysis. Pearson correlation is shown if significant.
*P < 0.05, **P < 0.01, ***P < 0.001. Outlying species are indicated with three-letter abbreviations from genus name, and the first letter
from species name.
[Back to E089-111]
http://www.esapubs.org/Archive/ecol/E089/111/appendix-A.htm

Page 15

Ecological Archives E089-111-A1
L. Poorter, S. J. Wright, H. Paz, D. D. Ackerly, R. Condit, G. Ibarra-Manríquez, K. E. Harms, J. C. Lincona, M. Martínez-
Ramos, S. J. Mazer, H. C. Muller-Landau, M. Peña-Claros, C. O. Webb, and I. J. Wright. 2008. Are functional traits good
predictors of demographic rates? Evidence from five Neotropical forests. Ecology 89:1908–1920.
Appendix A. Figures showing trait correlations between relative diameter growth rate (RGR) and functional traits of tree species at five Neotropical forest
sites.

FIG. A1. Trait correlations between relative diameter growth rate (RGR) and functional traits of tree species at five Neotropical forest
sites. Note that the axes of RGR and seed volume were log-transformed prior to analysis. Pearson correlation is shown if significant.
*P < 0.05, **P < 0.01, ***P < 0.001. Outlying species are indicated with three-letter abbreviations from genus name, and the first letter
from species name.
[Back to E089-111]
http://www.esapubs.org/Archive/ecol/E089/111/appendix-A.htm